Bushfire Risk Assessment at the Urban-Bush Interface (Ubi) in Sydney, Australia: an Integrated Modelling Approach

BUSHFIRE RISK ASSESSMENT AT THE URBAN-BUSH INTERFACE (UBI) IN SYDNEY, AUSTRALIA: AN INTEGRATED MODELLING APPROACH

Daminda Thushara Solangaarachchi BSc (Hons), MSc (GIS & RS)

A thesis submitted in fulfilment of the requirements for the degree of Doctor of Philosophy

School of Physical, Environmental & Mathematical Sciences The University of New South Wales Australian Defence Force Academy Canberra, ACT, 2600, Australia 31 August 2012 ii

ORIGINALITY STATEMENT

‘I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis.

I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.’

Signed ……………………………………………......

Daminda Thushara Solangaarachchi

31 August 2012

iii

ABSTRACT

Bushfire Risk Assessment at the Urban-Bush Interface (UBI) in Sydney, Australia: An Integrated Modelling Approach

Bushfires are one of the major threats to the environment and human systems in Australia. The recent 2009 Black Saturday bushfires in Victoria claimed more than 2,029 homes and 173 lives, and demonstrated that fire management authorities need to rethink their current risk and emergency management approaches. Rapid population growth and land use change at the urban-bush interface combined with favourable weather conditions for bushfires are causing a rapid increase in vulnerability of communities exposed to bushfires. Identifying vulnerability and risk before an event occurs are essential steps towards efficient and effective risk management. Global initiatives such as the United Nations International Decade for Natural Disaster Reduction (IDNDR) and Hyogo Framework for Action (HFA) have highlighted the importance of research for formulating the overall value of disaster risk reduction through national and local risk assessments. In order to achieve that, it is necessary to ‘measure’ the existing level of risk and the potential future risk that may be encountered in future bushfire events. In Australia, various institutions and agencies have developed a variety of bushfire risk assessment models. However, many of these models focus primarily on the hazard component of risk, which is mainly based on physical factors such as weather, fuel, and topography. A risk assessment model that integrates both the hazard component and the elements of vulnerability such as social vulnerability, physical vulnerability, and emergency response and coping capacity is yet to be developed. Risk assessments often use objective, quantifiable approaches. However, assessing the objective level of risk itself is not enough for efficient risk management decision-making. Understanding subjective judgements of residents living at urban-bush interface, the factors affecting their decisions and their perceptions of bushfire risk and attitudes towards current bushfire management strategies is also an important step towards effective bushfire risk management. Despite the considerable effort that has been directed towards encouraging bushfire preparedness in Australia, research on public perceptions of bushfire management strategies to reduce bushfire risk is relatively rare.

This thesis develops a multifaceted understanding of vulnerability and risk based on a holistic approach to risk. In this research, the hazard component is recognised as the product of the probability of occurrence and the severity of an event. Vulnerability is shown to arise from the inherent socioeconomic conditions of households, the exposure and physical succesptibility of locations and a community’s capactity to respond and cope with hazard events. Risk is identified as a function of hazard and vulnerability. To understand these different dimensions, a mixed methods approach was utilised in this thesis. A quantitative method was developed for a multidisciplinary evaluation of risk that assesses its different components individually and then combines them algorithmically. A GIS-based, Fuzzy Multi Criteria Evaluation (FMCE) method was utilized to integrate the components of risk. Such techniques also enable appropriate means of quantification and visualization of complex data in map form. Qualitative methods were primarily used to investigate subjective questions such as perceptions, household and community level preparedness activities. Household surveys and semi- structured interviews with local residents, community fire volunteers, local council members and others who participated in responses to the fires were conducted to capture such information. Exploratory data analysis was performed to understand these subjective judgements and the results were presented in graphical format. This thesis demonstrates the fundamental importance of understanding the multi- dimensional characteristics of risk in managing bushfire risk at the urban bush interface. The results revealed the spatial variation of composite risk as well as the elements of risk; hazard and vulnerability. It identified important physical and socioeconomic dimensions of vulnerability and the response and coping capacities of the communities. These variations help to prioritise different disk reduction initiatives in different areas. It also found different perceptions and attitudes of residents towards bushfire management activities. This information could help to further modify the risk reduction measures in order to address specific household and community level issues. The overall results of this thesis will provide a framework to strengthen the risk reduction measures that engage in anticipating future disaster risk, reducing existing exposure, hazard, or vulnerability, and improving community capacities to cope with hazard events.

PUBLICATIONS RELATED TO THIS THESIS

Journal Article Solangaarachchi D., Griffin A.L., Doherty M.D., Social Vulnerability in the Context of Bushfire Risk at the Urban Bush Interface in Sydney: A Case Study of the Blue Mountains and Ku-ring-gai Local Council Areas, Natural Hazards: Journal of the International Society for the Prevention and Mitigation of Natural Hazards, DOI: 10.1007/s11069-012-0334-y

vii

viii

ACKNOWLEDGEMENTS

First and foremost I want to thank my supervisor, Dr. Amy L. Griffin for her valuable guidance and encouragement throughout the research process. She provided constructive comments and helped me to navigate obstacles and recognise the opportunities during the process. She has always being patient and willing to listen. I am fortunate to have had the opportunity to study under her guidance. I also wish to express profound respect to my co-supervisor Michael D. Doherty for his constructive suggestions and guidance in improving this research.

This work would not have been possible without the interest of the members of the public in Ku-ring-gai and the Blue Mountains who received and responded to my survey. Most people were happy to participate in the survey, and even some people who didn’t participate found it useful. I would like to thank them for assisting me in gaining an in-depth knowledge of my study area.

The support I got from the local councils and NSW Fire and Rescue (NSWFR) was very helpful. I would like to express my thanks to Greg Buckley (NSWFR), Dr. Jenny Scott and Jennie Cramp (Ku-ring-gai Council), and Peter Belshaw (Blue Mountains City Council) for helping me to complete this research. The key data and information used for this study was provided by Local councils, Land and Property Management Authority, NSW Rural Fire Services and NSW Fire and Rescue. Without these data, this study would not have been possible. Their provision of data is greatly appreciated.

This research was financially supported by School of Physical, Environmental and Mathematical Sciences at the UNSW Canberra and, I extend my thanks and appreciation to the school for its support. My appreciation also extends to the PEMS administration team for the many ways they helped throughout my research. I am indebted to Ms Julie Kesby for her excellent help with regards to referencing, thesis formatting, and gathering resources for my study.

Special thanks to my dear colleagues in the Geographers’ room, especially Solomon, Dustin and Vijai. So many good ideas were born with what started as random ix conversations in the room. I will always remember the time we spent discussing, arguing, and laughing.

I am deeply grateful to my parents for teaching me the value of education, and who sacrificed the best times of their lives for my education. I would not have come this far without their support. My brothers, thank you for being my best friends during my journey. Your love and support has given me the strength I needed to walk along this path. My heartfelt thanks also go to my father-in-law who have always supported me and encouraged me to pursue my goals wherever they took me.

A final word of thank to my wife, Beyandi, for love, encouragement, support and good food. Thank you for being understanding during those late nights and early mornings. Without your continuous support, love and patience I would not have made it.

LIST OF ABBREVIATIONS

ABS Australian Bureau of Statistics AHP Analytical Hierarchy Process CCD Census Collection Districts CFU Community Fire Unit DEM Digital Elevation Model GIS Geographic Information Systems IDNDR International Decade for Natural Disaster Risk Reduction IDW Inverse Distance Weight IPCC Intergovernmental Panel on Climate Change KDE Kernel Density Estimation LGA Local Government Area LPMA The Land and Property Management Authority MCE Multi Criteria Evaluation NSW RFS New South Wales Rural Fire Services NSWFR New South Wales Fire and Rescue OWA Ordered Weighted Average PCA Principle Component Analysis SOVI Social Vulnerability Index SWS Static Water Supply UBI Urban Bush Interface UNISDR International Strategy for Disaster Risk Reduction WLC Weighted Linear Combination

xii

TABLE OF CONTENTS

ORIGINALITY STATEMENT ...... iii

ABSTRACT ...... v

PUBLICATIONS RELATED TO THIS THESIS ...... vii

ACKNOWLEDGEMENTS ...... ix

LIST OF ABBREVIATIONS ...... xi

TABLE OF CONTENTS ...... xiii

LIST OF FIGURES ...... xix

LIST OF TABLES ...... xxiii

CHAPTER ONE: INTRODUCTION ...... 1

1.1 Background ...... 1

1.2 Disasters, Risks, Vulnerability ...... 2

1.3 Bushfires in Australia ...... 3

1.4 Risk Assessment ...... 6

1.5 Motivation for this Research ...... 7

1.6 Objectives ...... 10

1.7 Research Benefits ...... 11

1.8 Structure of the Thesis ...... 11

CHAPTER TWO: LITERATURE REVIEW ...... 15

2.1 Introduction ...... 15

2.2 Natural Hazards and Disasters ...... 16

2.3 Vulnerability ...... 18 2.3.1 Vulnerability as exposure to the hazard ...... 21 2.3.2 Vulnerability as (in)capability to cope ...... 21 2.3.3 Vulnerability as exposure and response ...... 22

xiii

2.4 Conceptual Frameworks for Understanding Vulnerability ...... 23

2.5 Risk ...... 26

2.6 Conceptual Frameworks for Risk ...... 28

2.7 Disaster Risk Management ...... 37

2.8 Disaster Risk Management Frameworks ...... 40 2.8.1 Australian/New Zealand Risk Management Framework ...... 43

2.9 Bushfire Risk Management Framework ...... 45

2.10 Summary ...... 49

CHAPTER THREE: HAZARD RISK AND VULNERABILITY ASSESSMENT 51

3.1 Introduction ...... 51

3.2 Risk Assessment ...... 51

3.3 Elements of Risk ...... 55

3.4 Hazard Assessment ...... 56

3.5 Vulnerability Assessment ...... 56

3.6 Elements of Vulnerability ...... 58

3.7 Integrated Risk Assessment ...... 61

3.8 GIS and Risk Assessment ...... 63

3.9 Scale of Risk Assessment ...... 65

3.10 Uncertainty in Risk Assessment ...... 66

3.11 Risk Assessment and Risk Perception ...... 66

3.12 Study Area ...... 67 3.12.1 Blue Mountains LGA ...... 69 3.12.2 Ku-ring-gai LGA...... 71

3.13 Summary ...... 72

CHAPTER FOUR: BUSHFIRE HAZARD ASSESSMENT ...... 75

4.1 Overview ...... 75

4.2 Introduction ...... 75

xiv

4.3 Bushfire Hazard ...... 76

4.4 Bushfire Hazard vs. Bushfire Risk ...... 78

4.5 Bushfire Hazard Assessment...... 79

4.6 The Conceptual Framework for Bushfire Hazard Assessment ...... 81 4.6.1 Ignition Probability ...... 82 4.6.2 Fire Severity ...... 82

4.7 GIS and Spatial Modelling in Risk Assessment ...... 83 4.7.1 Spatial Multi Criteria Evaluation Model ...... 86

4.8 Materials and Methods ...... 92 4.8.1 Ignition Probability ...... 93 4.8.2 Fire Severity ...... 95 4.8.3 Exploratory Data Analysis ...... 96 4.8.4 Recurrence Interval ...... 96

4.9 Results ...... 97 4.9.1 Fire History Database ...... 97 4.9.2 Exploratory Data Analysis ...... 100 4.9.2.1 Fire History ...... 100 4.9.2.2 Recurrence Interval ...... 102 4.9.3 Bushfire Hazard in Ku-ring-gai ...... 104 4.9.4 Bushfire Hazard in the Blue Mountains ...... 107

4.10 Discussion ...... 111

4.11 Conclusions ...... 111

4.12 Summary ...... 112

CHAPTER FIVE: SOCIAL VULNERABILITY IN THE CONTEXT OF BUSHFIRE RISK AT THE URBAN BUSH INTERFACE IN SYDNEY ...... 113

5.1 Overview ...... 113

5.2 Introduction ...... 113

5.3 Social Vulnerability ...... 115 5.3.1 Social Vulnerability Assessment ...... 116 5.3.2 Social Vulnerability Indicators ...... 119

5.4 Methodology ...... 120

5.5 Results ...... 124 5.5.1 Dimensions of Social Vulnerability in Ku-ring-gai ...... 125 5.5.2 Dimensions of Social Vulnerability in the Blue Mountains ...... 127 5.5.3 Social Vulnerability Index and Mapping ...... 130

5.6 Discussion ...... 137

5.7 Conclusions ...... 140

5.8 Summary ...... 141

CHAPTER SIX: BUSHFIRE RISK ASSESSMENT: AN INTEGRATED MODELLING APPROACH ...... 143

6.1 Overview ...... 143

6.2 Introduction ...... 143

6.3 Bushfire Risk and Conceptual Frameworks...... 144

6.4 Integrated Approach ...... 147

6.5 Bushfire Risk Assessment - MCE Model ...... 148 6.5.1 Standardisation Weighting and Identifying Evaluation Rules ...... 150

6.6 Vulnerability Assessment ...... 152 6.6.1 Exposure and Physical Susceptibility ...... 152 6.6.2 Social and Economic Fragilities ...... 166 6.6.3 Response and Coping Capacities ...... 166 6.6.4 Integrated Vulnerability ...... 177

6.7 Integrated Risk ...... 182

6.8 Conclusions ...... 186

6.9 Summary ...... 188

CHAPTER SEVEN: UNDERSTANDING COMMUNITIES AT THE URBAN BUSH INTERFACE FOR BUSHFIRE PREPARATION, RESPONSE AND RECOVERY ...... 189

7.1 Introduction ...... 189

7.2 Treating Bushfire Risk at the UBI ...... 191

7.3 Social Construction of Bushfire Risk...... 192

7.4 What Makes Communities More Vulnerable and Less Resilient? ...... 194

7.5 Social Factors Shape the Level of Vulnerability and Resilience to Bushfires ...... 197 7.5.1 Perception of Bushfire Risk ...... 199 7.5.2 Risk Knowledge, Awareness ...... 200 7.5.3 Risk Communication...... 202 7.5.4 Experience and Local Environment ...... 204 7.5.5 Community Strength ...... 206

xvi

7.5.6 Available Resources ...... 208 7.5.7 Institutional Arrangements ...... 210 7.5.8 Shared Responsibility...... 211

7.6 Community Based Bushfire Management in the Study Area ...... 216

7.7 Household Level Activities in the Study Area ...... 218

7.8 Role of the Fire Management Authorities in the Study Area ...... 219

7.9 Methodology ...... 220 7.9.1 Household Survey ...... 221 7.9.2 Interviews with Community Members and Key Personnel ...... 222 7.9.3 Data Analysis ...... 224

7.10 Results ...... 226 7.10.1 Household Survey ...... 226 7.10.1.1 Demographics of Respondents ...... 226 7.10.1.2 Knowledge and Bushfire Awareness ...... 236 7.10.1.3 Household Level of Preparedness ...... 245 7.10.1.4 Household Level Preparedness and Related Issues ...... 253 7.10.1.5 Community Level Preparedness ...... 255 7.10.1.6 Community Preparedness and Related Issues ...... 264 7.10.1.7 Shared Responsibilities ...... 265 7.10.1.8 Response and Recovery Actions ...... 269 7.10.2 Household Interviews and Focus Group Discussions ...... 274

7.11 Discussion and Limitations ...... 282

7.12 Conclusions ...... 283

CHAPTER EIGHT: CONCLUSIONS OF THE RESEARCH ...... 287

8.1 Introduction ...... 287

8.2 Summary of Results ...... 287

8.3 Future Research ...... 292

8.4 Use of this Study ...... 294

REFERENCES ...... 297

APPENDIX I: PARTICIPANT INFORMATION STATEMENT ...... 329

APPENDIX II: HOUSEHOLD SURVEY QUESTIONNAIRE ...... 331

APPENDIX III: PAIRWISE COMPARISON TOOL ...... 342

xvii

xviii

LIST OF FIGURES

Figure 1: Concept of risk ...... 30

Figure 2: The risk triangle ...... 30

Figure 3: Conceptual framework of earthquake disaster risk ...... 32

Figure 4: Conceptual framework of risk ...... 33

Figure 5: Conceptual framework for risk that takes a holistic approach to disaster risk assessment and management ...... 34

Figure 6: Disaster impact framework ...... 36

Figure 7: UNISDR framework for disaster risk reduction ...... 41

Figure 8: The BBC conceptual framework...... 42

Figure 9: AS/NZS Risk management process ...... 44

Figure 10: ‘Cities Project’ interpretation of the risk management process ...... 47

Figure 11: Interpretation of the bushfire risk assessment within the risk management process ...... 48

Figure 12: Risk assessment process...... 53

Figure 13: Elements of vulnerability and risk in an integrated risk assessment...... 62

Figure 14: Urban Bush Interface ...... 69

Figure 15: Map showing the study areas and the Greater Sydney region...... 71

Figure 16: Conceptual framework for bushfire hazard assessment...... 82

Figure 17: Spatial distribution of ignition points in Ku-ring-gai...... 98

Figure 18: Spatial distribution of ignition points in the Blue Mountains...... 99

Figure 19: Number of ignitions in Ku-ring-gai (1980-2007)...... 100

Figure 20: Number of ignitions in the Blue Mountains (1980-2006)...... 101

Figure 21: Annual area burnt in Ku-ring-gai (1986-2007)...... 101

Figure 22: Annual area burnt in the Blue Mountains (1986-2006)...... 102

Figure 23: Recurrence interval and the annual burnt area in Ku-ring-gai...... 103 xix

Figure 24: Recurrence interval and the annual burnt area in the Blue Mountains...... 103

Figure 25: Input maps for the bushfire hazard map (Ku-ring-gai)...... 105

Figure 26: Bushfire hazard map for the Ku-ring-gai area...... 106

Figure 27: Hazard zonation map for the Ku-ring-gai area...... 107

Figure 28: Input maps for the bushfire hazard map (Blue Mountains)...... 108

Figure 29: Bushfire hazard map of the Blue Mountains area...... 109

Figure 30: Bushfire hazard zonation map of the Blue Mountains area...... 110

Figure 31: Social vulnerability by CCD at the UBI based on SoVI...... 131

Figure 32: Maps showing distribution of individual component scores by CCD in the Ku-ring-gai LGA (components 1-3)...... 133

Figure 33: Maps showing distribution of individual component Scores by CCD in the Ku-ring-gai LGA (components 4-6)...... 134

Figure 34: Maps showing distribution of individual component scores by CCD in the Blue Mountains LGA (components 1-3)...... 135

Figure 35: Maps showing distribution of individual component scores by CCD in the Blue Mountains LGA (components 4-6)...... 136

Figure 36: Integrated bushfire risk assessment model...... 148

Figure 37: MCE model for bushfire risk assessment. Adapted from...... 149

Figure 38: Different sigmoid fuzzy membership functions ...... 151

Figure 39: Fuzzy criterion maps of the indicators of exposure and physical susceptibility in Ku-ring-gai...... 159

Figure 40: Fuzzy criterion maps of the indicators of exposure and physical susceptibility in the Blue Mountains...... 161

Figure 41: Exposure and physical susceptibility maps - Ku-ring-gai...... 164

Figure 42: Exposure and physical susceptibility map - Blue Mountains...... 165

Figure 43: Exposure and physical susceptibility zonation map - Blue Mountains...... 165

Figure 44: Fuzzy criterion maps of the indicators of emergency response and coping capacity in Ku-ring-gai...... 171

Figure 45: Fuzzy criterion maps of the indicators of emergency response and coping capacity in the Blue Mountains...... 173

Figure 46: Emergency response and coping capacity maps - Ku-ring-gai...... 176

Figure 47: Emergency response and coping capacity maps - Blue Mountains...... 177

Figure 48: Integrated vulnerability map - Ku-ring-gai...... 179

Figure 49: Integrated vulnerability zonation map - Ku-ring-gai...... 180

Figure 50: Integrated vulnerability map - Blue Mountains...... 181

Figure 51: Integrated vulnerability zonation map - Blue Mountains...... 182

Figure 52: Integrated risk map - Ku-ring-gai...... 183

Figure 53: Integrated risk zonation map - Ku-ring-gai...... 184

Figure 54: Integrated risk map - Blue Mountains...... 185

Figure 55: Integrated risk zonation map - Blue Mountains...... 185

Figure 56: Hazard and risk relationship (Paton & Johnston 2006)...... 195

Figure 57: Framework of community wildfire preparedness ...... 212

Figure 58: Social cognitive preparation model ...... 213

Figure 59: Bushfire preparedness model...... 215

Figure 60: Interviewing community members and key personnel...... 223

Figure 61: Gender demographics comparison with ABS data...... 229

Figure 62: Age demographics comparison with ABS data...... 230

Figure 63: Level of household income/week among respondents, in comparison with ABS data...... 231

Figure 64: Household type comparison with ABS data...... 232

Figure 65: Home ownership comparison with ABS data...... 233

Figure 66: Employment status comparison...... 234

Figure 67: Level of education of adults...... 235

Figure 68: Number of people at home during day and nighttime...... 236

Figure 69: Years of residence in their current suburb...... 237

Figure 70: Reason for living in Ku-ring-gai or the Blue Mountains...... 238

Figure 71: Previous bushfire experience...... 240 xxi

Figure 72: Sources of information...... 242

Figure 73: Awareness of fire ban rules and regulations...... 244

Figure 74: Perceived level of bushfire risk...... 245

Figure 75: Knowledge of the month when the bushfire season starts...... 246

Figure 76: Reasons for not conducting bushfire preparedness activities...... 248

Figure 77: Information that encourages bushfire preparedness activities...... 249

Figure 78: Bushfire preparedness - household level activities...... 250

Figure 79: Discussion of preparedness issues with fire management authorities...... 251

Figure 80: Perceived levels of preparedness...... 253

Figure 81: “I feel like I belong to the local community” ...... 255

Figure 82: Source of social connection...... 256

Figure 83: Community help each other in an emergency ...... 257

Figure 84: Community discussions about bushfire preparedness activities...... 258

Figure 85: Participation in CFU/fireguard programmes...... 260

Figure 86: Participating in community level activities...... 261

Figure 87: Best practice for bushfire preparedness and mitigation involvement...... 262

Figure 88: Access to resources...... 264

Figure 89: Perceptions of responsibility for community protection and bushfire management activities...... 267

Figure 90: Community protection and bushfire management activities...... 268

Figure 91: Immediate contact point...... 270

Figure 92: My family would evacuate in an event of life threatening bushfire...... 271

Figure 93: When is the right time to evacuate? ...... 272

Figure 94: Point of evacuation...... 273

xxii

LIST OF TABLES

Table 1: Differences between risk assessment and risk perception ...... 67

Table 2: Social vulnerability indicators and variables used in the PCA ...... 122

Table 3: Social Vulnerability Index comparison ...... 124

Table 4: Social vulnerability and its dimensions – Ku-ring-gai LGA ...... 125

Table 5: Social vulnerability and its dimensions – Blue Mountains LGA ...... 128

Table 6: Indicators of exposure and physical susceptibility ...... 154

Table 7: Pairwise comparison matrix of indicators of exposure and physical susceptibility ...... 156

Table 8: Indicators of exposure and physical Susceptibility - Fuzzy membership functions, evaluation criteria and weights ...... 157

Table 9: Vegetation communities in Ku-ring-gai and the Blue Mountains ...... 158

Table 10: Indicators of response and coping capacity ...... 167

Table 11: Pairwise comparison matrix of indicators of emergency response and coping capacity ...... 169

Table 12: Indicators of exposure and physical susceptibility - Fuzzy membership functions, evaluation criteria and weights ...... 169

Table 13: Pairwise comparison matrix of vulnerability factors ...... 178

Table 14: Factors of vulnerability - Fuzzy membership functions, evaluation criteria and weights ...... 178

Table 15: Variables and derived composite variables ...... 225

Table 16: Demographic profile of the respondents ...... 228

Table 17: Communication of total fire ban information ...... 243

Table 18: When do you normally start to prepare for bushfires? ...... 247

Table 19: Use of bushfire assessment tools ...... 252

Table 20: Community protection and bushfire management activities ...... 259

Table 21: Areas of community preparedness that respondents believe need further development ...... 263 xxiii

Table 22: Reason to not to evacuate ...... 274

xxiv

Chapter One: Introduction

1.1 Background

Every year natural disasters threaten the sustainability of human systems, disrupting their use of resources and threatening human lives. The Great East Japan earthquake occurred on 11 March 2011, 130 km off Japan’s eastern coast, causing a tsunami that, together with the earthquake, may have killed more than 20,000 people (UNISDR 2011). In 2010, western Russia experienced its hottest summer in 100 years. Around 800,000 hectares in western Russia were affected by severe wildfires between July and September. This resulted in the deaths of more than 50 civilians and fire fighters. Some 2,000 houses burnt down, and more than 30 villages were completely destroyed (UNISDR 2011). The earthquake that struck Haiti on 12 January 2010 translated into a massive disaster which caused 222,517 fatalities (UNISDR 2011).

In 2008, numerous major disasters threatened human development gains across the world. In May, tropical cyclone Nargis caused an estimated 140,000 deaths in Myanmar and China’s most powerful earthquake since 1976 affected Sichuan and parts of Chongqing, Gansu, Hubei, Shaanxi and Yunnan provinces, killing at least 87,556 people, injuring more than 365,000 and affecting more than 60 million people (UNISDR 2009a). In August 2008, the Kosi River in Bihar, India, broke through an embankment and changed its course 120 km eastwards, rendering useless more than 300 km of flood defences that had been built to protect towns and villages. Flowing into supposedly floodsafe areas, the flood affected 3.3 million people in 1,598 villages located in 15 districts. It was characterized as the worst flood in the area for 50 years (UNISDR 2009a).

In light of escalating disaster losses, the international community has prioritized promoting safer communities while campaigning for effective disaster risk reduction strategies. Global initiatives such as the United Nations’ International Decade for Natural Disaster Reduction (IDNDR) in 1999 and the Hyogo Framework for Action (HFA) 2005-2015, which is an outcome strategy from the 2005 World Conference, have

1 highlighted the importance of the research mandate for formulating the overall value of disaster risk reduction (UNISDR 2005; United Nations 1999). The latest initiative, the Hyogo Framework for Action (HFA), identified five key areas that promote disaster risk reduction. Those are governance, risk assessment, knowledge and education, risk management and vulnerability reduction, and disaster preparedness and response. Among other priorities, the HFA defines the identification, assessment and monitoring of disaster risks as one of the most important steps in disaster risk reduction.

“The starting point for reducing disaster risk and for promoting a culture of disaster resilience lies in the knowledge of the hazards and the physical, social, economic and environmental vulnerabilities to disasters that most societies face, and of the ways in which hazards and vulnerabilities are changing in the short and long term, followed by action taken on the basis of that knowledge” (UNISDR 2005, p. 7).

1.2 Disasters, Risks, Vulnerability

In addition to these international initiatives, the scientific community has also put much effort into increasing its understanding of the concepts of vulnerability and risk in order to develop planning strategies and assessment tools to reduce potential losses. Disasters are often viewed as complex interactions between a hazard and a vulnerable human population, which often result from political, economic, and development failures (Oliver-Smith 1996; Smith & Petley 2009; Villagrán de León 2008). A common understanding of disasters is hampered by the various definitions of hazard, vulnerability and risk, derived from concepts and theories from various schools of thought. Risk assessment depends on understanding these key, underlying terms. Moreover, understanding risk factors and communicating risk to decision makers and the general public still remain significant challenges (Cardona et al. 2012). Disaster risk is derived from the combination of physical hazards and the vulnerabilities of exposed elements (Cardona et al. 2012). Therefore, the disaster risk management community often emphasises that the association between hazard and vulnerability determines both probabilities and consequences. This association often depends on the physical, social and geographic conditions that prevail in the area of interest (Lavell et al. 2012).

Hazard is defined as the possible future occurrence of a natural or human induced physical event that may cause impacts on exposed populations (Cardona et al. 2012). Often a hazard is considered to be an external element of risk. The probability of occurrence is an estimate of how often a hazard event occurs. A review of historic events and favourable environmental conditions for the event assists with this determination.

Vulnerability is broadly defined as the potential for loss, and is an essential concept in natural hazards research. Vulnerability is generally perceived as a predisposition of societies to be affected by hazards, including the incapacity to cope with an event that results in adverse effects on those exposed (Cardona et al. 2012; Villagrán de León 2006). It arises from the present conditions of communities that may be exposed in the future. In the field of natural hazards, linkages between unsafe conditions and different dimensions of vulnerability have been investigated using integrated approaches. Integrated vulnerability assessments go beyond traditional vulnerability modelling to provide a wider and more comprehensive explanation of vulnerability, one which differentiates exposure and physical susceptibility, social and economic fragilities, and lack of ability to cope and recover as different dimensions of vulnerability (Cardona & Barbat 2000; Cardona & Hurtado 2000; Cardona et al. 2012).

1.3 Bushfires in Australia

Bushfires are one of the major threats to the environment and human systems in Australia. The recent 2009 Black Saturday bushfires in Victoria claimed more than 2029 homes and 173 lives, and demonstrated that fire management authorities need to rethink their current bushfire risk and emergency management approaches. In Australia the majority of bushfire impacts occur at the urban-bush interface (UBI) (Buxton et al. 2011; Dixon 2005). &#0 , 31&,2#0$ !# ! , #"#$',#" 1b ,7 0# 5�#1203!230#10#1'"#,2' *Q',"3120' *Q0#!0# 2'-, *-0 %0'!3*230 * 0#*-! 2#" "( !#,2-0 +-,%!-+ 312' *# 31&* ,"$3#*1c -220#**TRRWQ.TSSRTClimate change scenarios predict that south-eastern Australia will experience increased temperatures, decreased precipitation, and high winds, which will result in increased bushfire intensity, longer fire weather seasons, and increased frequency of extreme or

3 catastrophic fire risk days (Blue Mountains Bush Fire Coordinating Committee 2008; Hennessy et al. 2005). With population growth and urbanization in areas of high fire risk along with insufficient bushfire management activities, it becomes clear that high consequence events are likely to become more frequent and more intense (Pitman et al. 2007).

According to Collins (2005), fires in the UBI have risen to prominence for three reasons; the occurrence of destructive fires in the area; rapid growth of the population in areas that are biophysically hazardous; and inadequate mitigation efforts of federal, state and local governments and private land management. The urban interface is a complicated environment for fire agencies to successfully operate within. Challenges include the sheer number of properties at risk, the dynamic nature of fires, a lack of community awareness and education, large-scale self-evacuation, narrow streets, failure of essential services such as electricity and gas, and lack of interagency communication.

In traditional bushfire models, fires are started by lightning strikes in rural locations. The development of large scale fires often results from adverse weather conditions and large volumes of combustible vegetation, causing the fires to make runs that threaten urban communities in their path (Bradstock et al. 1998). These urban interface fires develop quickly, and have the potential to overrun local fire fighting resources (Lowe 2008). In these scenarios, planning and developing mitigation and response strategies prior to any bushfire event play an important role in managing hazard impacts.

Fires at the UBI pose the greatest challenges facing Australian fire management agencies today (Preston et al. 2009). To utilize scarce response resources most efficiently, it is important to have prior knowledge about the level of vulnerability and risk that are posed to communities at the UBI. Such information can also help to identify the most vulnerable populations, and to minimize potential losses from future bushfire events by undertaking planning and preparation activities. Losses can be minimized if bushfire management activities are designed to target the most vulnerable people.

The UBI in the Sydney metropolitan area is one of the more bushfire-prone densely populated areas in Australia. The level of bushfire risk is being increased by an ever growing population and the concentration of high value properties and industrial and commercial assets (Chen 2005). The potential for future increases in bushfire risk is of particular concern in the area with projections of climate change that suggest the region is likely to experience more fire-weather days (Preston et al. 2009). The 1994 bushfires covered 75 percent of all bushland in the Sydney metropolitan area, caused significant economic losses, disrupted essential services, and caused both injury and death (Dixon 2005). The Black Christmas fires in 2001 destroyed 121 homes, and the 2002 fires destroyed 10 homes (EMA Disaster Database 2012). All of these events caused significant economic loss. Bushfires pose significant but largely unmeasured risks to people and their property (Bradstock et al. 2008). Bushfire risk is not simply determined by the severity of fire event. It is rather a product of the nature of fires and the vulnerability of the communities that are exposed to fires (Whittaker 2008). Therefore deeper analysis of this interaction is required for effective fire management.

In Australia, disaster and emergency management agencies have proposed a stronger focus on anticipation, mitigation, and recovery and resilience in order to achieve safer, more sustainable communities (Gabriel 2009). Such disaster management activities should be driven by better knowledge of vulnerability and risk assessments. Identifying levels of hazard, vulnerability and risk as well as the factors that influence them and their interactions facilitates the development of effective disaster risk reduction strategies that target the needs of specific groups. Thus, these activities are more effective at the local level because the impact of disasters is a product of interactions between local scale conditions such as weather, vegetation and fuel loads, topography, social characteristics, level of preparedness and emergency response and coping capacities. Furthermore, risk mitigation and emergency response measures such as planning and preparedness at the local level are often driven by collective actions of residents and local councils (Preston et al. 2009).

Despite the many unforgettable bushfire events that have occurred in Australia, and their significant impacts, the lack of a risk assessment framework has been a primary

5 limitation to quantitative risk assessment (Middelmann 2007). This hinders the use of comprehensive bushfire risk assessments in the planning process. Thus there is limited awareness of the geography of bushfire risk, implications of climate change for future bushfire hazards, changing patterns of social vulnerability, or how to develop appropriate adaptive responses (Preston et al. 2009). The Victorian Bushfires Royal Commission (2009) also highlighted the importance for effective emergency response of having information on vulnerable populations. Such information can guide emergency managers to recognise the specific needs of vulnerable people such as those who might need evacuation assistance or separate consideration, particularly on high fire risk days (Victorian Bushfires Royal Commission 2009).

1.4 Risk Assessment

Within the hazard research community, risk assessment can be defined differently in various contexts. However, it primarily concerns the degree to which a population, the built environment, and socioeconomic activities are susceptible to damage from a hazard event and includes physical aspects of hazard events such as location, magnitude, frequency and process (Chen et al. 2003). Although different disciplines conceptualise risk in different ways, the literature suggests that there is a common emphasis on the interaction between the hazard agent and a vulnerable community. In most of the research, the equation risk = hazard x vulnerability has been used to elaborate the relationship between these three concepts. As hazards and vulnerability are spatially distributed, risk is inherently a spatial phenomenon, and risk assessment should address both the degree of risk and its spatial variations.

Risk assessments often use an objective quantifiable approach. However, assessing objective levels of risk alone is not sufficient for efficient risk management decision- making. Quantitative risk assessments need to be complemented with qualitative approaches (Cardona et al. 2012). The public often views and evaluates risk differently from researchers and experts. Understanding how the public constructs their perceptions of risk can greatly improve risk communication and direct risk reduction strategies most appropriately (Cottrell et al. 2009). This emphasizes that taking only a theoretical approach to analysing bushfire risk is not sufficient to make decisions; the possible mix

6 of underestimating risk and overconfidence in facing a risk may increase the vulnerability of the community. Therefore it is important to identify factors that influence the decisions of residents living at the UBI about bushfire management. Understanding subjective judgements of residents living at UBI, the factors affecting their decisions and their perception of bushfire risk, and attitudes towards current bushfire management strategies are also important steps towards effective bushfire risk management.

Despite the fact that considerable effort has been directed towards encouraging bushfire preparedness in Australia, research on public perceptions of bushfire management strategies to reduce bushfire risk is uncommon. Available studies focus on a particular region or bushfire management strategy (Bushnell et al. 2007; Cottrell et al. 2008; Lowe et al. 2008). Social, cultural and institutional processes influence household and community level risk management and preparedness activities (Bushnell et al. 2007; Miceli et al. 2008; Paton et al. 2000). Therefore people from one social group may not perceive risk in the same way as others, and even within the same community, people perceive risk differently (Paton et al. 2000). An understanding of community perceptions and attitudes will support the development of improved risk communication and risk reduction strategies (Cottrell et al. 2008; Paveglio et al. 2009).

1.5 Motivation for this Research

Local governments are one of the key governance scales for bushfire risk management. In the context of bushfires, most bushfire preparedness activities are conducted at the local level. While a range of institutional arrangements and policy measures for mitigation of bushfire risk have been implemented within Australian local governments, the exploration of bushfire risk at the local level is not well addressed (Preston et al. 2009). Nevertheless, local governments need information on bushfire risk in order to understand the implications of bushfire risk management in the context of local conditions as well as guidance on how such information can be incorporated into local level planning, bushfire risk and emergency management, and resource management, in order to facilitate the implementation of relevant policies and measures.

It is difficult to implement preparedness and response activities across the UBI in an equitably distributed manner because there are never enough resources to cover all areas. To manage bushfires, fire management agencies have to work in the areas that are considered to be at highest risk of damage from bushfires. Easy identification of areas where high-risk populations live and spatial comparisons to prioritize high-risk areas across the UBI allows agencies to funnel resources to those areas. If hazard, and the dimensions of vulnerability can be incorporated into a risk assessment model that is capable of determining risk, the effectiveness of emergency management may be increased. Such an integrated model will answer questions like: Where is the hazard high? Where are the most vulnerable people living? Why are they vulnerable? Where is risk high? Where would potential losses be high? Furthermore, preparedness activities can be designed to address specific issues (i.e., reduce social vulnerability, reduce physical vulnerability or increase community capacity). This contributes to risk reduction and ultimately reduces the possibility of future disasters.

At present, at the local government level, bushfire management and planning rely heavily on ‘bushfire prone area’ maps that are developed using vegetation buffering (NSW Rural Fire Service 2006). These ‘bushfire prone’ maps only provide information about proximity to flammable vegetation. Therefore, to understand the level of risk at the UBI, a more holistic approach is needed. The assessment of bushfire risk at the local level is complex and challenging as the different elements of risk need to be assessed individually and then integrated through some kind of algorithm to produce a holistic picture (Villagrán de León 2006). Thus far, at the local level there is no consensus on how to address risk holistically while measuring the elements of risk.

To address this issue, a variety of bushfire risk assessment models have been developed by various institutions and agencies. Many of them have only focused on the hazard component of risk, and are therefore based mainly on physical factors such as weather, fuel, and topography. Existing wildfire risk models only show the area of assets or value potentially impacted by fires. Such approaches do not provide information on the pre- existing condition of vulnerability. Therefore, tools for assessing bushfire risk that integrate hazard and pre-existing conditions of vulnerability for a particular geographic area need to be developed. To prepare an effective disaster mitigation plan, planners 8 need tools that simultaneously address all of the dimensions of vulnerability. To address this need, planners and government agencies need to develop simplified models to understand communities and to assist communities with the process of risk and vulnerability assessment.

This thesis develops a multifaceted understanding of bushfire vulnerability and risk at the UBI based on a holistic approach to risk. Based on the developed risk framework, a spatial bushfire risk assessment is performed. In this research, hazard is recognised as the probability of occurrence and the severity of the event. Vulnerability is shown to arise from the inherent socioeconomic conditions of households, exposure and physical susceptibility of locations, and the level of response and coping capacities within communities. Risk is identified as a function of hazard and vulnerability.

A methodology for bushfire risk assessment is presented for the Local Government Area (LGA) level, using two LGAs. A GIS-based, Fuzzy Multi Criteria Evaluation (FMCE) method was utilized to assess and integrate the individual elements of risk. Such techniques also provide a means of quantification and visualization of complex data in map form. Specific conditions in the LGA, data availability and the bushfire management setting are taken into account.

This research also highlights the importance of understanding different perceptions and attitudes of residents towards bushfire management activities. Such information helps to further modify risk reduction measures in order to address specific household and community level issues. Qualitative methods were primarily used to investigate questions such as subjective perceptions and household and community level preparedness activities. Household surveys and semi-structured interviews with local residents, community fire volunteers, local council members and others who participated in responses to fires were conducted to capture such information. Exploratory data analysis was performed to understand these subjective judgements. The overall results of this thesis provide an advanced framework to strengthen risk reduction measures that engage in anticipating future disaster risk, reducing existing exposure, hazard, or vulnerability, and improving community capacities to live with hazards. 9

1.6 Objectives

This research is devoted to developing a conceptual model of risk that provides a holistic perspective on bushfire risk assessment. It incorporates hazard; social vulnerability; exposure and physical susceptibility; and response and coping capacity at the UBI as well as information on a variety of social and cognitive processes that affect judgements, perceptions, and behaviour with respect to bushfire management and preparedness activities at the local level. The key objectives of the research are to:

1. Develop and implement an integrated bushfire risk assessment framework that would assist existing bushfire risk management processes.

2. Develop a GIS-based integrated bushfire risk assessment model to diagnose the bushfire risk at the Urban Bush Interface (UBI).

3. Understand household and community characteristics that influence subjective decisions about bushfire management activities at the local level.

To accomplish these objectives, the following tasks were carried out:

Review and analyse existing definitions, frameworks and models of hazard, vulnerability, risk and risk management to understand the most suitable conceptual and analytical framework for bushfire risk assessment in the context of bushfire risk management. Bushfire hazard assessment and mapping based on bushfire hazard occurrence probabilities and severity in the study area using the available bushfire history data. Develop a social vulnerability index for bushfires at the census collection district level using available census data. Assessment of vulnerability to bushfires based on physical exposure and susceptibility, social vulnerability, and response, recovery, and coping capacities at the UBI using a GIS- based Fuzzy Multi Criteria Evaluation model. Assessment of bushfire risk at the UBI based on the Hazard-Risk model.

A household level assessment was performed using a mixed methods approach in order to identify factors affecting community and household level decision making processes that influence local level bushfire management activities at the UBI.

1.7 Research Benefits

The findings of this research contribute to bushfire risk mitigation and reduction at the UBI in the Sydney metropolitan area. Integrated bushfire risk assessment will help to differentiate hazard, vulnerability and risk. Major contributions of this research are: 1. The development of a conceptual framework for bushfire risk assessment in the context of bushfire risk management. 2. The development of a bushfire hazard assessment model using fire history data. 3. The development of a social vulnerability index to understand the socioeconomic conditions at the UBI using census data. 4. The development of a GIS-based fuzzy multicriteria evaluation model to analyse bushfire risk at the local government level using available data sources. 5. Contribution to local level bushfire risk management processes by understanding spatial variations in bushfire hazard, risk and vulnerability across the study area. 6. The development of a household and community bushfire preparedness model. 7. Contribute to household and community level bushfire management processes by describing the different perceptions, attitudes and behaviours of residents living at the UBI.

1.8 Structure of the Thesis

This thesis is comprised of four parts: Part I (Chapters 1 to 3) lays the foundation that motivates this research into integrated bushfire risk assessment. The theoretical aspects, and conceptual and assessment frameworks that relate to the study are discussed. In Part II (Chapters 4 to 6), bushfire risk at the UBI is assessed using an integrated approach. The elements of risk are investigated and an integrated model is presented. Part III (Chapter 7) discusses household and community level bushfire management processes; the factors affecting the level of household and community level bushfire preparedness

11 are also assessed. Part IV (Chapter 8) concludes the overall research findings and highlights the importance of those findings. Furthermore, it describes a scope for future work related to the integrated bushfire risk assessment process for effective bushfire management.

Chapter 2 presents a literature review of definitions, conceptual frameworks, and their applications in disaster risk management. This is important because the lack of a common understanding of these basic concepts is a barrier to developing an integrated risk assessment framework. Different conceptual frameworks for vulnerability, risk and risk management are discussed, and a risk assessment framework for bushfire risk in the context of bushfire risk management in Australia is presented.

Chapter 3 provides an overview of hazard, vulnerability and risk analysis. It explains different assessment approaches and presents the analytical framework for the study based on the holistic risk assessment framework discussed in Chapter 2. It also provides an overview of the study areas.

In this research, bushfire hazard is considered to be an element of bushfire risk. Bushfire hazard assessment is presented in Chapter 4. It describes the methods used to analyse available data and generate fire hazard maps using information on ignition probability and fire severity. The result of the bushfire hazard assessment is used as an input for the overall risk assessment in Chapter 6.

Chapter 5 discusses the socioeconomic dimension of vulnerability. Social vulnerability at the UBI is modelled using a Social Vulnerability Index (SoVI). The results, including the development of the SoVI, data and indicator selection and factor identification, are presented.

Chapter 6 explores the integrated risk assessment process. It describes the analysis of overall vulnerability, including exposure and physical susceptibility, social vulnerability, and response and coping capacities. It then combines the hazard and vulnerability assessment results to develop an integrated bushfire risk map.

Chapter 7 investigates household and community level bushfire management processes. It presents a bushfire preparedness model and based on the household survey conducted, the different perceptions, attitudes and behaviours of residents living at the UBI are discussed.

Chapter 8 discusses the overall findings, limitations and challenges of this research. It further describes how useful these findings are for supporting effective bushfire management. It concludes by suggesting policy implications to minimise potential bushfire impacts and future work that is needed to improve the integrated bushfire risk assessment model developed in this thesis.

Chapter Two: Literature Review

2.1 Introduction

The number of disasters, number of people affected and the estimated economic losses generated by disasters have increased over the past decades (Hagenlocher 2012). Every recent disaster has illustrated the process of translating the impact through the level of exposure and vulnerability (UNISDR 2011). Therefore, the importance of policies, programmes and mechanisms to reduce exposure and vulnerability and to promote more resilient societies is widely recognised at all levels (Hagenlocher 2012; UNISDR 2005). Thus identification, assessment and monitoring of disaster risk have become key priorities among researchers and decision-makers. Contemporary research on natural hazards and disasters is multidisciplinary in nature. Natural scientists study the nature of the extreme events involved in hazards whereas social scientists study the human dimensions of both impacts and responses.

Geographers often try to provide a perspective that integrates physical and human dimensions of hazards while placing emphasis on the spatiotemporal distributions of hazards, their impacts and vulnerability (Alexander 1993). According to Cutter et al. (2000), hazards research in geography began when Harlan Barrows employed a human ecology approach in 1923 to find out how people and society adjust to extreme environmental events, such as floods. Later, Gilbert White (1936, 1945) argued against the conception of hazard as an isolated geophysical event and incorporated a social perspective on natural hazards to develop a human ecological framework that operates at the interface of both natural and human systems (Smith 2001). This broadened the field of natural hazards from a narrow focus on technological and engineering solutions to human adjustments (Burton et al. 1978). Since then, social scientists effectively began their own research on the interaction between natural hazards, people and society.

During the 1970s, hazards research split into three distinct perspectives (Smith & Petley 2009). Scientists who believed hazards are geophysical events extended their work on environmental control to improve predictions about extreme natural events and the

15 construction of physical works designed to resist them. Geographers continued to work on a hazard-based approach predicated on the unifying concept of human ecology and the notion of mitigating losses by adding various human adjustments to the existing use of physical control structures. Sociologists, on the other hand, adopted a disaster-based view with an emphasis on understanding the role of human behaviour at times of community crisis and the need to improve preparedness for such mass emergencies.

There has been considerable evolution of theories and concepts in hazards research in the decades since the 1970s. At present, hazards research considers more than just the hazards themselves. It considers hazards within their particular context such as geography of the event, physical properties of the hazard and the social, economic, political, spatial, temporal and organizational structure of the system where the hazard takes place (Cutter et al. 2000).

This chapter explores some of the key concepts in hazards research; hazard, vulnerability and risk. It reviews definitions and conceptual frameworks for vulnerability and risk in order to understand their multidisciplinary nature. It further discusses the risk management frameworks that help to minimise levels of exposure and risk. Finally, it highlights the importance of an integrated risk framework and develops a bushfire risk assessment framework for bushfire risk management based on a holistic approach to risk.

2.2 Natural Hazards and Disasters

The basic term ‘natural hazard’ implies a potentially damaging physical event, phenomenon or human activity that may cause a loss of life or injury, property damage, social and economic disruption, and/or environmental degradation (UNISDR 2004). “A hazard may be regarded as the pre-disaster situation, in which some risk of disaster exists, principally because the human population has placed itself in a situation of vulnerability” (Alexander 1993, p. 7). Different researchers have employed different approaches to define hazards and disasters and their relationship to many disciplines. The term hazard sometimes creates confusion and it has been used imprecisely and with different implicit meanings, but in addition, the term has evolved with understanding of

16 the components that interact to comprise hazardousness (Tobin & Montz 1997). Alexander (1993, p. 4) defines hazard in four ways based on previous literature: “A naturally occurring or man-made geologic condition or phenomenon that presents a risk or is a potential danger to life or property (American Geological Institute 1984); An interaction of people and nature governed by the co-existent state of adjustment of the human use system and the state of nature in the natural event system (White 1973); Those elements in the physical environment which are harmful to man and caused by forces extraneous to him (Burton and Kates 1964); The probability of occurrence within a specific period of time and within a given area of a potentially damaging phenomenon (UNDRO 1982)”.

From these four perspectives on hazards, it is clear that a hazard represents the potential for impact on human beings and their environment (Alexander 1993; Tobin & Montz 1997).

Burton et al. (1978) further elaborated upon and modified the concept of interactions between humans and extreme natural events. They argued that natural systems are neither benevolent nor maliciously motivated towards humans: they are neutral, in the sense that they neither prescribe nor set powerful constraints on what can be done with them. It is people who transform the environment into resources and hazards, by using natural features for economic, social, and aesthetic purposes. Tobin and Montz (1997) also employed a similar concept. They stated that a natural hazard represents the potential interaction between humans and extreme natural events and represents the potential or likelihood of an event. In these views, hazards are seen as a result arising from the interaction of social and natural systems. If the result of these interactions causes disruptions in the human environment, it will create natural hazards (Burton et al. 1993).

A natural disaster, in contrast to a hazard, can be defined as a rapid, instantaneous or profound impact of the natural environment upon the socio-economic system

(Alexander 1993). UNISDR (2009b) defined a disaster as “a serious disruption of the functioning of a society, causing widespread human, material, or environmental losses which exceed the ability of affected society to cope using only its own resources.” These disruptions in human society are mainly due to the interaction between a hazardous physical event and vulnerable social conditions (Lavell et al. 2012). Disasters are often classified sudden or slow onset (according to their speed), or natural or man-made (according to their cause). A disaster might also be defined as an extremely hazardous event that differs significantly from the norm and that disrupts the workings of society (Alexander 1993; Tobin & Montz 1997). However, there is no definite threshold that determines exactly what size of natural event can be used to categorically define an event as a disaster (Alexander 1993; Tobin & Montz 1997).

The natural and the human are inextricably bound together in almost all disaster situations (Wisner et al. 2004). Therefore, if there is no significant interaction between an extreme natural event and a human system, there will be no disaster (Alcántara- Ayala 2002; Alexander 1993; Burton et al. 1993). Villagrán (2006) elaborated upon this idea by noting that a disaster is preceded by at least two predispositions: the possibility that the triggering event takes place, usually called a hazard at this potential state; and a pre-existing vulnerability, the pre-disposition of people, processes, infrastructure, services, organisations, or systems to be affected, damaged, or destroyed by the event.

2.3 Vulnerability

Various definitions and terms exist for the term “vulnerability” in the hazard and disaster management literature. In general, vulnerability is identified as the propensity for or predisposition to be adversely affected by an event. The concept of vulnerability is used frequently in many diverse research and policy communities such as sociology, geography, engineering, natural sciences and anthropology. Recently its use has become more prominent in areas such as natural hazards, including climate change (Cutter et al. 2000; Simpson & Human 2008), poverty and human security (Chaudhuri et al. 2002; O’Brien & Leichenko 2007), environmental change and sustainability (Barnett et al. 2008; Metzger & Schröter 2006; Sands & Podmore 2000), and food security (Burg 2008). As a result, the concept of vulnerability has been employed within multiple

18 contexts, with multiple dimensions and at multiple geographical and temporal scales (Hufschmidt 2011). These different views on vulnerability have emerged as a consequence of the conceptual needs required to address particular aspects of the potential impacts of disasters (Villagrán de León 2006). Yet the various definitions of and meanings ascribed to the concept of vulnerability have hampered a common understanding of how to measure vulnerability.

The concept of vulnerability achieved prominence in the 1960s, when it was viewed as a measure of the propensity for the damage a structure could face from a hazard of a given intensity. This conceptualization of vulnerability developed out of the engineering perspective and it was predominantly based on structural and physical aspects of the environment (UNDRO 1980). In the late 1970s the concept of vulnerability acquired a hazard-centric perspective through the work of White (1973) and Burton et al. (1978). They stated that vulnerability is not just an attribute of physical structures but of social groups and therefore vulnerability is socially produced or at least influenced. This concept was used to understand the extent to which people experience a particular hazard differently: as a result of their likelihood of exposure (to the hazard) and their capacity to withstand the hazard, which is often in turn influenced by their socioeconomic circumstances (Cannon 1994; Schneiderbauer & Ehrlich 2004). Since then human aspects of vulnerability and its role in natural disasters have gained wide attention from scholars around the world and as a result, various definitions, concept and frameworks have emerged over the decades.

Timmerman (1981) conceptualised vulnerability as the degree to which a system reacts adversely to the occurrence of a hazardous event. This adverse reaction includes response, absorption and recovery from the event. According to this conceptualization, exposure to a hazardous event and the reaction of society to it can be seen as the key elements of vulnerability. Chambers (1989) described these elements as the internal and external sides of vulnerability. The external side of vulnerability is the set of shocks and stresses to which an individual or household is subjected, and the internal side is defencelessness, meaning a lack of ability to cope without experiencing a damaging loss.

While the theoretical basis of vulnerability progressed over the last two decades, discrepancies in the meaning of vulnerability have also arisen in different research and practical settings. Cutter (1996, p. 532) noted: “Despite more than a decade’s worth of collective research experience about the concept, vulnerability still means different things to different people.” Although there is a variety of different academic meanings for the term vulnerability, it also has a commonplace meaning: being prone to or susceptible to damage or injury (Wisner et al. 2004). Vulnerability has also been broadly defined as the potential for loss of property or life from environmental hazards (Cutter et al. 2000). However, this general definition of vulnerability does not specify the type of loss or the characteristics of the individuals, groups, or societies experiencing the loss. Cardona (2004) identified vulnerability as an internal risk factor that is mathematically expressed as the potential that the exposed system may be affected by the hazard phenomenon.

The UNISDR (2004) definition for vulnerability describes it as the conditions determined by physical, social, economic and environmental factors or processes that increase the susceptibility of a community to the impact of hazards. The physical aspect of vulnerability refers to susceptibilities of location and the built environment, and can be represented through factors such as population density, remoteness of a settlement, and the construction materials and design used for critical infrastructure and housing. Social factors include levels of health and well-being of individuals, gender ratios of the population, levels of literacy and education, the existence of peace and security, access to human rights, social equity, traditional values, beliefs, and organisational systems. Economic factors include poverty and the levels of individual, community, and national economic reserves, levels of debt, degrees of access to credits, loans, and insurance, and economic diversity. Environmental factors include natural resource depletion and degradation.

From the brief survey above, the literature clearly shows that there are many possible ways to categorize the various connotations and concepts related to vulnerability, based on how vulnerability has been conceived. The most common themes are vulnerability as exposure, vulnerability as (inadequate) coping capacity and vulnerability as the

20 intersection of exposure and (inadequate) coping capacity (Cutter 1996; Gall 2007; Villagrán de León 2006; Whittaker 2008).

2.3.1 Vulnerability as exposure to the hazard

In this perspective, vulnerability is referred to as direct exposure to a potential hazard; how the system deals with the hazard, its characteristics and its impacts. It mainly describes vulnerability as a physical property and is also viewed as the degree of loss resulting from the occurrence of a natural event (Buckle et al. 2000). Alexander (1993) stated that vulnerability is a function of the costs and benefits of inhabiting areas at risk of natural disasters. He emphasized that human occupancy of and humans’ activities in hazardous zones are the central domain of vulnerability. Exposure depends on proximity to the natural hazard, the rapidity of its onset, its duration, spatial impacts or areal extent, and the probability with which a hazard of a specific magnitude and frequency occurs (Cutter 1996). Cutter used the term biophysical vulnerability to express the idea of vulnerability as exposure. Although this view emphasises that vulnerability is only determined by the exposure to an external event, others argue that the characteristics of the human system and the relationship between human society and the external event also play a critical role (Dolan & Walker 2006).

2.3.2 Vulnerability as (in)capability to cope

This perspective views vulnerability as socially constructed and it is conceptualized as the pre-disaster state of the social system, which is determined by social conditions and historical circumstances. It is further described as (inadequate) coping responses, including a community’s lack of social resilience and resistance to hazards (Cutter 1996; Villagrán de León 2006). Wisner et al. (2004) explained this idea as the characteristics of a person or group and their situation that influence their capacity to anticipate, cope with, resist and recover from the impact of a natural hazard. They further described vulnerability as a result of social processes and structures that constrain access to resources, resulting in an inability to cope with hazard impacts. Differences in societal conditions also determine the level of exposure and thus the potential damage among different social groups (Cutter et al. 2003; Wu et al. 2002). As an example, low-income people might move into a flood prone area because of the

21 affordability of properties. Living in a flood zone might increase both their level of exposure and potential losses. In this perspective, researchers have emphasised the role of the socioeconomic position of individuals and communities within the system, which includes characteristics such as income, gender, race and age, irrespective of the physical properties of hazard exposure.

2.3.3 Vulnerability as exposure and response

In this view, both exposure and the underlying social conditions of a population that make it incapable of coping with hazard events are considered to be determinants of vulnerability. Therefore this perspective defines vulnerability as a relationship between physical events and the social characteristics of populations affected by these events, and integrates both the physical events and underlying causal characteristics of vulnerability (Dolan & Walker 2006). Along these lines, Watts & Bohle (1993) defined vulnerability in terms of exposure, capacity and potentiality. They emphasised that “the most vulnerable individuals, groups, classes and regions are those most exposed to perturbations, who possess the most limited coping capability, who suffer the most from crisis impact and who are endowed with the most circumscribed capacity for recovery” (Watts & Bohle 1993, p. 45). Bohle et al. (1994) elaborated upon this definition by suggesting that vulnerability is an aggregate measure of human welfare that integrates environmental, social, economic, and political exposure to a range of potential harmful perturbations. Villagrán (2006) further modified this definition by defining vulnerability as a multi-layered and multi-dimensional social space defined by the political, economic, and institutional capabilities of people in specific places and times.

An earlier expression of this idea can be found in Cutter et al.’s (2000) contention that the degree to which populations are vulnerable to hazards is not solely dependent upon the physical nature of the hazard and that social factors play a significant role. In their ‘hazards of place’ model, the hazard potential interacts with the underlying social fabric of the place to create social vulnerability. Social fabric includes socio-demographic characteristics as well as perceptions of and experience with risk and hazards. The geographic filter includes the site and situation of the place and its proximity to hazard sources and events, and interacts with hazard potential to produce the place’s

22 biophysical vulnerability. The social and biophysical vulnerability elements are mutually related and produce the overall vulnerability of the place.

Vulnerability in the context of climate change has been defined as “the degree to which a system is susceptible to, or unable to cope with, adverse effects of climate change, including climate variability and extremes. Vulnerability is a function of the character, magnitude, and rate of climate variation to which a system is exposed, its sensitivity, and its adaptive capacity” (McCarthy 2001, p. 995). In this definition, vulnerability is seen as an integrated measure of the expected magnitude of adverse effects to a system caused by a given level of certain external dimensions (exposure) and internal dimensions (sensitivity and adaptive capacity).

2.4 Conceptual Frameworks for Understanding Vulnerability

Over the years, different conceptual frameworks for understanding vulnerability have been developed to explain vulnerability in the context of natural hazards. Widely accepted vulnerability frameworks include the Pressure and Release (PAR) and Access models (Blaikie et al. 1994; Wisner et al. 2004), the Vulnerability in Sustainability Science model (Turner et al. 2003), the Vulnerability of Place model (Cutter 1996; Cutter et al. 2000), the IPCC model (Lavell et al. 2012; McCarthy 2001), the holistic perspective on vulnerability (Cardona & Barbat 2000; Cardona & Hurtado 2000), and other integrated models (Dolan & Walker 2006; Füssel 2007; UNISDR 2004).

The PAR model discusses the root causes of vulnerability (unsafe conditions) and various components and processes that create dynamic pressures that result in unsafe conditions. The PAR model of vulnerability recognises three elements that cause the progression of vulnerability; root causes, dynamic pressures and unsafe conditions. It starts with root causes, which include a lack of resources and power along with ideologies of political and economic systems. These root causes generate dynamic pressures such as a lack of skills or local institutions, and macro forces such as population change and urbanization. Dynamic pressures, in turn, drive a system towards unsafe conditions. Unsafe conditions are seen as a combination of factors involving the physical environment, the local economy, social relations, and public actions and

23 institutions. However, within the PAR model, this progression of vulnerability is taken as separate to and independent of hazard characteristics.

In their framework for vulnerability in sustainability science, Turner et al. (2003) viewed vulnerability as a component of the human environmental system. This system operates at multiple spatial, functional and temporal scales. They identified exposure, sensitivity and resilience as key elements in their vulnerability framework. Exposure is characterised by the frequency, magnitude and the duration of the hazard. Sensitivity is determined by both human conditions (population and economic structures) and environmental conditions (biophysical endowments). Resilience is characterised by coping responses, impact responses and adjustments, and adaptation responses. Sensitivity and resilience jointly determine the consequences of a hazard. In this framework, characteristics of the hazard are considered to be an element (exposure) that determines the level of vulnerability.

In the ‘vulnerability of place’ model, the interaction between biophysical vulnerability and social vulnerability creates place vulnerability (Cutter 1996). It also assumes vulnerability is a pre-existing condition and distinctive to particular places and times. This model recognises that hazard potential is lessened or amplified by the social fabric of society (social vulnerability) as well as place-based characteristics such as proximity to the hazard source (biophysical vulnerability). According to this framework hazard potential is filtered through both of these elements and the interaction and intersection of social and physical elements of vulnerability results in place vulnerability.

Vulnerability frameworks in the context of global climate change address three key factors; exposure, sensitivity and adaptive capacity (McCarthy 2001; Preston et al. 2009). In these frameworks exposure is seen as an external stressor, sensitivity as the responsiveness of a system to the external stressor, and adaptive capacity is viewed as the inherent property that defines capacity to deal with exposure (Dolan & Walker 2006; Preston et al. 2009). This type of vulnerability framework is only suitable for slow onset disasters such as drought and climate change because adaptation is considered to be a longer term and more-sustained adjustment (Lavell et al. 2012). In the context of a sudden onset disaster like bushfires, instead of adaptive capacities, 24 emergency response capabilities play an important role in minimising impacts. Although climate change frameworks include exposure within the notion of vulnerability, it might be problematic when interpreting the concept of risk in relation to vulnerability and the hazard. This is because the term ‘hazard’ is widely used in risk research to explain the specific hazard conditions while ‘exposure’ is used to represent hazard conditions in the context of climate change (Hufschmidt 2011). This needs to be considered in analyses of risk because the hazard component may therefore be double- weighted.

The holistic perspective on vulnerability provides a more comprehensive explanation of vulnerability. It presents vulnerability as a dynamic system that is characterised by physical exposure and susceptibility, socioeconomic fragility and a lack of resilience to cope and recover (Birkmann 2006b; Cardona & Barbat 2000; Cardona & Hurtado 2000). These factors determine both direct and indirect impacts of hazard events. Integrated frameworks often consider the inherent properties of physical and social environments as an interrelated and interdependent human-environment system (Dolan & Walker 2006). Within risk-hazard models, vulnerability is often seen as a separate element. Vulnerability is characterised by factors such as physical infrastructure, population, and social and political systems (Davidson 1997). The objective of integrated vulnerability frameworks is to include all relevant dimensions of vulnerability, i.e. economic, social, physical as well as some aspects of ecological vulnerability (Fuchs et al. 2011). In most of the integrated frameworks, vulnerability is characterised by physical exposure, fragility of the socio-economic system and lack of resilience to cope and recover (Dolan & Walker 2006). Other authors view the level of vulnerability as determined by the level of exposure and susceptibility (Birkmann 2006a).

Many of the above mentioned frameworks have highlighted vulnerability as the main driver that determines the existing and future level of risk (PAR, holistic perspective, IPCC-2012, sustainability science model and some integrated models). Therefore, risk is defined as a function of hazard and vulnerability (Birkmann 2006b).

Although a number of different social and environmental characteristics have been addressed in vulnerability frameworks, they do not represent the actual interaction between hazard and vulnerability. Often vulnerability frameworks have not been able to suggest specific vulnerability reduction measures such as hazard mitigation and emergency response. The lack of clarity about the relationship between hazard and vulnerability within these frameworks makes it difficult to understand the concept of disaster risk. With recent developments in natural hazard research, disasters are often viewed as a result of complex interactions between hazardous physical events and a vulnerable society (Lavell et al. 2012). The outcome of a disaster is characterised by the hazard phenomenon, the degree of vulnerability of exposed elements and their capacities to withstand disaster events (Granger 2003; Lavell et al. 2012). This interaction between hazardous conditions and the vulnerable elements needs to be understood in order to identify specific risk management strategies, either through hazard reduction or vulnerability reduction measures. Therefore, this study focuses on the concept of risk. Hazard is considered to be an external component of risk while vulnerability is considered an internal component of risk (Cardona 2004).

2.5 Risk

The concept of risk has gained favour in disaster management over the past two decades as it signifies the possibility of adverse effects in the future (Lavell et al. 2012). Risk is driven by the interaction of social environmental processes, from the combination of physical hazards and the vulnerabilities of exposed elements (Lavell et al. 2012). The term “risk” is multidisciplinary in nature and is used in natural hazards research with different definitions in various situations (Thywissen 2006). Nevertheless, risk seems to have two broad foci: event occurrence on one hand and the consequences of the event on the other hand (Bachmann 2001). Risk is often defined as probabilistic in nature (Bachmann & Allgower 2001; Blanchi et al. 2002; Brooks 2003), relating either to; the probability of occurrence of a hazard that acts to trigger a disaster or series of events with an undesirable outcome, or the probability of the outcome (loss) of an event or combining the probability of the hazard event with a consideration of the outcome of the hazard.

This concept of probability is mainly driven by the likelihood of hazard occurrence and the degree of loss. Concepts of risk that emphasise the hazard component are not rare in the hazard literature. Therefore the risk of these events is characterized by the magnitude of the hazard (including size and spread), the frequency and duration, and the history of hazard occurrence (Alwang et al. 2001; Schneiderbauer & Ehrlich 2004). The consequences are mainly measured in terms of causalities and economic loss and are mainly used in the insurance sector to estimate potential losses from future disasters.

In the context of natural disasters, the combination of probability of occurrence and consequences provides a quantitative representation of a qualitatively defined hazard (Lavell et al. 2012). Consequences are largely influenced by the nature of hazard and the vulnerability of the exposed elements. Hazard and vulnerability can each contribute to the probability of occurrence. Wisner et al. (2004) defined risk as a complex combination of vulnerability and hazard. The relationship between hazard and vulnerability that explains root causes, dynamic pressure and unsafe conditions shows the consequences and likelihood. UNISDR (2004, p. 16) provided a more comprehensive definition of risk as the “probability of harmful consequences, or expected loss of lives, people injured, property, livelihoods, economic activity disrupted (or environment damaged) resulting from interactions between natural or human induced hazards and vulnerable conditions”. These proposed concepts of risk lead to a conventional equation of risk given below:

Risk = Hazard x Vulnerability Equation (1)

However, the conventional equation has been further modified to elaborate upon interactions between hazard and vulnerability. Most efforts have added new elements while the main concept remains same. Crichton (1999), for example, proposed risk as a function of three elements, hazard, vulnerability, and exposure.

Risk = Hazard*Exposure*Vulnerability Equation (2)

Granger (1999) substituted the term ‘elements at risk’ for exposure. Risk = Hazard*Elements at Risk*Vulnerability Equation (3) 27

Both attempts have tried to identify the exposure in terms of lives or properties at risk in a given area as a separate component of vulnerability.

Recent publications define risk by incorporating terms such as coping capacity, emergency response and recovery capacity, and deficiencies in preparedness (Lavell et al. 2012). Coping capacities refer to the means by which people or organisations use available resources and capacities to face adverse consequences related to a disaster. In general, such capacities involve management of resources before, during, and after the disaster. Therefore it is suggested that there is an inverse relationship between risk and coping capacities (Villagrán de León 2006).

5- ǳ0') -$'$/4 Risk = *+$)"+$/$ . Equation (4)

However, the relationship between risk and capacities has also been described as a linear relationship. In such conceptualizations, coping capacities are assigned a negative value because of their negative relationship with the level of risk (Hahn 2003, p. 115).

Risk = Hazard + Exposure + Vulnerability - Coping Capacities Equation (5)

All of these efforts represent risk as a probability of harmful consequences, or expected losses resulting from interactions between natural or human-induced hazards and vulnerable conditions. However, there is no consensus on how to measure any of the components thus far. The assessment of risk itself is highly context-specific and rather complex, as the components need to be assessed individually and then combined based on the specific risk model used for the study.

2.6 Conceptual Frameworks for Risk

In addition to the numerous mathematical expressions of risk, various conceptual frameworks for risk are also available in the literature. Many of the conceptual approaches for risk had their origin in studies of technological hazards and some of them were adapted for natural disaster risk (Carreño et al. 2007). After Gilbert White (1964) initiated the concept of risk in natural hazards, sociologists, geographers and

28 engineers devoted their efforts to explaining how the term risk related to their disciplines. Since then sociologists began to focus on social responses to disasters, civil engineers’ focus was on physical and technical aspects of disasters and geographers’ focus was on natural hazard risk with an emphasis on a socio-ecological perspective (Carreño et al. 2007). Kates (1971) provided an example model of risk while considering the interactions between nature, humans and technology. When Whitman (1973) started damage assessment, methodologies were developed for physical risk assessment (Carreño et al. 2007).

Later, various analytical concepts and models of risk were developed to systematise its study. In recent decades, a more integrated vision of risk that combines both hazard and vulnerability has evolved. During the 1990’s, stimulated by the International Decade for Natural Disaster Reduction (IDNDR), many studies that dealt with disaster risk emerged around the world. The topic gained wide attention and it is being increasingly recognized as an important step in disaster risk reduction. It also highlighted that the terms hazard, vulnerability and risk have had different meanings and implications from both methodological and practical angles (Cardona, 2004).

The risk analysis guidelines developed by GTZ (2004) (Figure 1), identify hazard and vulnerability as the essential elements of risk. According to this framework, risk only arises when hazard and vulnerability coexist. This framework implements the mathematical model presented in equation 1. However, it does not elaborate on components of vulnerability, and exposure is considered to be a component of hazard, while self-protection capabilities are a component of vulnerability. The relationship of these factors and their contributions toward disaster risk reduction are addressed. Therefore applying this model in to this framework may be highly context specific. For an example, this model may be more suitable for individual level assessment rather than community level assessment.

Figure 1: Concept of risk (GTZ 2004, p. 17).

Several authors have portrayed risk as a triangle. This concept was initially developed by Crichton (1999). According to him, the area inside the triangle represents the risk and the sides of the triangle represent the independent factors that contribute to risk (Figure 2). Those independent factors include exposure, vulnerability and hazard.

Figure 2: The risk triangle (Crichton 1999, p. 130), Source: Middelmann (2007).

Birkmann (2006b) has discussed the triangular risk framework that was developed by Villagrán De León in 2004. Villagrán’s triangle of risk also consists of three components: vulnerability, hazard and deficiencies in preparedness. In this framework, exposure is not directly mentioned. However, he views exposure as a component of hazard. In risk triangles, each component of risk contributes in equal proportions, and changing any one of these three factors may change the level of overall risk.

These triangular frameworks are implementations of equation 2. Although some elements such as exposure and deficiencies in preparedness are seen as separate elements of risk, the relationship between these elements and the level of vulnerability and risk is not clear, nor does the model clarify the relationship between these elements and levels of vulnerability and risk. Therefore, how these elements contribute to practical implications is difficult to ascertain from these triangular models.

Davidson (1997) (Figure 3) introduced two new components into the traditional risk framework: external context and emergency response and recovery capabilities. His framework is directed to sudden onset disasters like earthquakes. In this framework hazard represents the geophysical phenomenon that serves as an initiating event (e.g., an earthquake). Vulnerability is conceptualised as the potential impact on physical structures and populations. Exposure describes the size of the area affected; it includes a list of everything that is subjected to effects of the hazard, including physical infrastructure, population, economy and social and political systems. External context is included to describe how damage to a certain place (e.g., a city) affects those outside that place (e.g., those in other cities). Emergency recovery and response capability describes how effectively and efficiently a city can respond to and recover from short and long-term impacts through formally organized activities that are performed after, before or during the disaster.

Earthquake Disaster Risk

External Emerg.Resp.& Hazard Exposure Vulnerability Context Recovery Cap.

Ground Physical Physical Economy Planning Shaking Infrastructure Infrastructure Transportation Resources Population Population Collateral Hazards Politics Mobility and Economy Economy Access Culture Social- Social- Political Political System System Figure 3: Conceptual framework of earthquake disaster risk (Davidson 1997, p. 30).

The above mentioned risk framework highlights five contributing elements. It considers risk to be the sum of hazard, exposure, vulnerability, external context and emergency response and recovery capacities. However, the same importance has been given to each category and that is questionable. The relationships between categories are not well explained and the model’s contribution towards disaster risk reduction is not illustrated. This model also relies on various dimensions related specifically to earthquakes (e.g., ground shaking). It may be difficult to apply this framework to other hazards mainly because of its complexity and a lack of data availability.

Bollin et al. (2004) adapted the conceptual framework discussed above to describe disaster risk at the community level in Latin America (Figure 4). Their framework has only four categories. It views risk as the sum of hazard, exposure, vulnerability and capacity and measures. Compared to Davidson’s framework, external context is not visible here. Hazard is described through an event’s probability and severity, while exposure is determined by structures, population and economy, vulnerability is characterized by physical, social, economic and environmental aspects and capacity and measures are defined through physical planning, and social, economic, and management

32 and institutional capacities. Although emergency response and recovery measures are not highlighted in this framework, relevant indicators that determine emergency response and recovery capacity are considered within the capacity and measures component. Bollin et al. (2004) have tried to apply Davidson’s framework in a different context. However, their framework has also failed to address the limitations identified in Davidson’s framework. Both Davidson’s and Anderson’s frameworks are based on the linear risk model discussed in the equation 4. Therefore, their integration of the risk factors remain questionable.

Disaster Risk

Capacity and Hazard Exposure Vulnerability Measures

Probability Structures Physical Physical planning Severity Population Social Social capacity Economy Economic Economic Environment capacity

Management

Figure 4: Conceptual framework of risk (Bollin et al. 2004).

Birkmann (2006b) discussed a conceptual framework for a holistic approach to risk that was developed by Cardona and Barbat (2000) (Figure 5). This framework is characterised by three categories of vulnerability factors. Exposure and physical susceptibility, which is designated as hard risk, and is viewed as being hazard-dependent. Fragility of the socioeconomic system, which is viewed as soft risk and is non hazard-dependent.

Lack of resilience to cope and recover, which is also defined as soft risk and is non hazard-dependent.

The overall level of risk is determined by the interaction between these vulnerability factors and the hazard. This framework further discusses the concepts of ‘‘hard risk’’ and ‘‘soft risk’’. Hard risk is the risk of potential damage to physical infrastructure and the built environment and is influenced by the exposure and physical susceptibility of the built environment. It can also be explained as direct, tangible or measurable risk. Soft risk is the potential socioeconomic impact on communities and organizations. It can be considered to include indirect, intangible and unmeasurable impacts, which are influenced by the socio-economic characteristics of the different localities (units) in the city and their disaster coping capacity or degree of resilience.

Figure 5: Conceptual framework for risk that takes a holistic approach to disaster risk assessment and management (Cardona & Barbat 2000). Source: (Birkmann 2006b, p. 33).

According to this framework, risk is determined by vulnerability conditions in a given hazard-prone area. This framework gives a broader understanding of risk while

34 intending to capture conditions influencing direct physical impacts (exposure and susceptibility) as well as indirect and at times intangible impacts (socio-economic fragility and lack of resilience) of potential hazard events. The relationship between hazard and vulnerability is well-defined and the consequences of the interaction are defined as risks. This framework also suggests a feedback loop that starts when there is a risk. This loop stimulates the risk management processes and corrective and prospective interventions (Birkmann 2006b).

Carreño et al. (2007) later developed a theoretical framework and holistic approach to disaster risk adapted from Cardona & Barbat (2000). In their model the meaning of hard and soft risk were revised into physical damage and impact factors. The level of risk is viewed as a function of the potential physical damage and the impact factor. The impact factor is determined by social and economic fragilities and lack of resilience, and physical damage is determined by the susceptibility of the exposed elements to a hazard and its potential intensity and frequency of occurrence. This framework also includes a feedback loop for corrective and prospective interventions to give an idea about the relationship between risk reduction measures and components of risk. Socio-economic fragility and a lack of resilience are a set of factors (related to indirect or intangible effects) that aggravate physical risk (potential direct effects). Thus, total risk depends on direct effects, or physical risk, and indirect effects.

Lindell et al. (2006), in their disaster impact framework (Figure 6), stated that disaster impact is determined by pre-impact conditions in the community (i.e., hazard exposure, social vulnerability and physical vulnerability), event-specific conditions, and emergency management interventions (i.e., improvised disaster responses, improvised disaster recovery, hazard mitigation and emergency preparedness practices). Although impacts have been divided into physical and social impacts, common elements of the holistic risk framework (Cardona & Barbat 2000) can also be seen in this framework. Hazard event characteristics represent the hazard conditions whereas other event- specific conditions and emergency management interventions identify the community’s capacity to prepare, respond and recover. Therefore, total impacts depicted in this framework describe the level of risk.

Figure 6: Disaster impact framework (Lindell et al. 2006), Source: Lindell (2011, p. 3)

Both the holistic frameworks and the disaster impact framework addressed issues of vulnerability and risk very well. They provide insight into the underlying factors of vulnerability, the relationship between hazard, vulnerability and risk, and the contribution of these relationships towards risk management. They also highlight the need for institutional structures, public policies, and actions to implement risk reduction strategies. The holistic approach to risk framework includes feedback loops, which connect the level of risk with the control system of risk management. The control system connects with interventions such as risk identification, risk reduction, disaster management and risk transfer that minimise the overall impact and risk. In the disaster impact framework, event-specific conditions are connected with emergency management interventions that minimise social and physical impacts. Both frameworks rely on the conventional risk model discussed in equation 1. Although different factors determine the level of vulnerability, the final risk level is derived from the interaction between hazard and vulnerability. They do not consider hazard and vulnerability to be equally important.

When considering practical applications, the holistic approach to risk framework is easier to implement than the disaster impact framework because the holistic approach

36 includes additional information such as corrective and prospective interventions through a risk management system. In a comprehensive risk assessment model, the key elements that have been considered are; hazard, exposure, vulnerability and capacities. In the holistic frameworks of risk, exposure and physical susceptibility, socioeconomic fragility and coping and recovery capacities are considered to be factors of vulnerability, while the interaction between hazard and vulnerable conditions determines the level of risk. This framework gives an insight into the overall risk in a given community while evaluating the key elements of risk and their interactions. Therefore, this research utilises a holistic framework for risk, which provides vital information that can be used when implementing risk reduction measures.

However, it is also important to identify the appropriate mode of using a risk framework within the risk management framework. It helps emergency planners to identify certain elements of risk in order to initiate specific risk reduction measures during the risk management process. The holistic framework is related with the AS/NZS risk management framework (Figure 9) because similar feedback loops are present in both frameworks. Therefore, the holistic framework of risk is considered to be the most suitable risk framework to provide a theoretical grounding for the context of bushfire risk assessment in Australia.

2.7 Disaster Risk Management

Just as there are many definitions of risk and risk assessment, there are many different definitions of risk management. Effective risk management succeeds in recognizing uncertainty associated with natural events and providing information on potential hazards, vulnerabilities and risks in order to make informed decisions for risk reduction measures (Pearce 2000). This has led to increased emphasis on better integration of disaster reduction programs with other development issues. But disaster risk reduction traditionally has yet to be given priority in sustainable development programmes (Yodmani 2001).

Until the late 1990’s, many emergency managers viewed disasters as extreme events created entirely by natural forces and the social and economic implications and causes

37 of these events were not considered (Yodmani 2001). Therefore, disaster management practices were largely considered primarily to consist of relief and response activities. Governments and relief agencies were more focused on relief and recovery activities rather than understanding the nature of the impacts of disasters. With recent developments in the understanding of risk as a combination of a physical event and the vulnerability of a society and the development of the risk reduction strategy paradigm, there has been a shift in disaster management programs from relief and recovery to mitigation and preparedness.

“Disasters are no longer seen as extreme events created entirely by natural forces but as manifestations of unresolved problems of development” (Yodmani 2001, p. 1).

There are deeper causes that drive people’s exposure to potential hazards: socioeconomic limitations, incapacities to mobilize resources, population pressure, migration to hazard-prone areas in search of better opportunities, and lack of policies and legislation (Villagrán de León 2008). The increasing frequency of hazard events, prevailing socioeconomic conditions, unplanned urbanization, growing demand for food, industrial goods and services, and increasing pressures on (and over-exploitation of) natural resources will make people more vulnerable to natural events. Therefore, it is important to understand the factors that are often interrelated with disaster management activities. Social, political, institutional, economical, physical and environmental factors are identified as the most important drivers of risk (Baas et al. 2008; Villagrán de León 2008). Disaster reduction strategies need to be implemented using a variety of measures, spanning all sectors of society at all levels in order to address these issues. “While better emergency response systems will save lives and properties, many of these losses can be avoided – or reduced – if appropriate policies and programmes are instituted to address the root causes and set in place mitigation, preparedness and response mechanisms that are effectively integrated into overall development planning” (Baas et al. 2008, p. 1)

In the current paradigm of disaster risk management, priority is given to the issue of risk reduction for vulnerable populations. To address this issue, a holistic approach focusing on risk and vulnerability has brought about the concept of risk reduction or disaster risk 38 management (UNISDR 2004). This approach has several interrelated components: hazard assessment, vulnerability analysis and enhancement of management capacity. The current paradigm of risk management approaches also addresses issues of social inequality in risk reduction. However, it is also recognized that the role of response and relief assistance during a disaster remains important and needs to be enhanced at all levels (UNISDR 2004).

UNISDR (2009b, p. 4) defines disaster risk management as the “systematic process of using administrative directives, organizations, and operational skills and capacities to implement strategies, policies and improved coping capacities in order to lessen the adverse impacts of hazards and the possibility of disaster.” Risk management is the process that minimizes, distributes or shares potentially adverse consequences of hazards (Tobin & Montz 1997). It aims to increase the coping capacity of societies and minimize their level of vulnerability. It includes measures for before, during and after disasters. Disaster risk management is an integral part of disaster management, focusing on the situation before the extreme natural event, and relating to its effects during and after the disaster through risk analysis. Disaster risk management is used as an instrument for reducing the risk of disasters primarily by reducing vulnerability and strengthening self-protection capabilities. It takes into account and links technical, social, political, socioeconomic, ecological and cultural components of risk to form an integrated system to enable risk management to reduce risk to an acceptable level which a society can cope with (Kohler et al. 2004).

Disaster Risk Management often combines prevention, mitigation and preparedness with response. It is used to refer to management of both pre-disaster risk and post- disaster risk; therefore the emergency management elements are also included in risk management frameworks. Elements of the risk management framework consider the possibilities available to minimize vulnerabilities and disaster risks throughout a society, to avoid the adverse impacts of hazards, within the broad context of sustainable development (Baas et al. 2008).

2.8 Disaster Risk Management Frameworks

Over the years, different risk management frameworks have been introduced to manage natural hazards. The common objective of the risk management process is an effective integration of the concepts of risk into sustainable development, planning and programming, with a special emphasis on prevention, mitigation and preparedness at all levels help to deal with natural hazards in an effective and efficient manner (Bründl et al. 2009). Disaster risk management is a threefold task: the analysis of risk, evaluation of risk and management of risk.

Within the disaster risk management framework the identification and analysis of risk concentrates on the interaction between ‘sources of risk’ and ‘elements at risk’. Disasters are considered to be the interface between extreme events and the population. Therefore disaster impacts are seen as a social product rather than a physical event (Salter 1997). Villagrán de León (2008) has explained disasters are a result of processes that generate or control the level of risk. To reduce risk, it is important to implement a variety of risk reduction measures spanning all sectors of society at all levels, from national to the local (Villagrán de León 2008). Therefore it is necessary to identify the sources of risk, the elements of risk as well as vulnerable entities, which contributes to controlling the level of risk.

UNISDR (2004) developed a framework for disaster risk reduction in the context of sustainable development (Figure 7). It describes the general context and primary activities of disaster risk management, including the elements necessary for any comprehensive disaster risk reduction strategy.

Figure 7: UNISDR framework for disaster risk reduction. Source: (UNISDR 2004, p. 15)

This framework identifies the relationship between vulnerability and hazard factors and the risk reduction measures within four spheres; environmental, economic, political and socio-cultural. It also provides an important overview of key thematic areas such as risk identification and assessment, governance, knowledge management, risk management and preparedness and emergency management, that need to be taken into account in disaster risk reduction. In this framework vulnerability/capability assessment and hazard assessment are considered to be the basis for risk assessment, and other risk reduction measures such as early warning and preparedness are key components of this framework. However, Birkmann (2006b) noted that this framework does not indicate

41 how reducing vulnerability can also reduce risk. He argued that vulnerability is placed outside the risk response and preparedness framework, which makes it difficult to understand the necessity of also reducing risk through vulnerability reduction and hazard mitigation. It also shows that early warning, preparedness and response could reduce the disaster impact, even though a link between the risk factors and the application of risk reduction measures is not included. Moreover, the conceptual framework does not give an answer as to whether exposure should be seen as a feature of the hazard or of vulnerabilities.

Birkmann (2006b) introduced a new conceptual framework for risk reduction called the BBC framework based on the work done by Bogardi and Birkmann (2004) and Cardona (1999 and 2001) (Figure 8). The framework was developed based on three broad discussions (Birkmann 2006b): How to link vulnerability, human security and sustainable development The need for an integrated approach to disaster-risk assessment Developing causal frameworks for measuring environmental degradation in the context of sustainable development

Figure 8: The BBC conceptual framework, Source: Birkmann (2006b, p. 34). 42

The BBC framework understands vulnerability as a dynamic process that goes beyond the estimation of damage and the probability of loss. It promotes a problem-solving perspective by simultaneously analysing probable losses and weaknesses of the various exposed, affected elements (e.g., social groups) and their coping capacities as well as potential intervention measures (feedback-loop system) within all three key spheres of sustainable development (social, economic and environmental).

Furthermore, in the BBC conceptual framework, vulnerability has not been viewed as an isolated feature. Rather, vulnerability assessment also has to take into account the specific hazard types that the vulnerable society, its economy and environment are exposed to, and the interaction of both hazard and vulnerability that leads to risk. This means that the BBC framework underlines the necessity to focus on social, environmental and economic dimensions of vulnerability, clearly linking and integrating the concept of sustainable development. It also stresses the fact that vulnerability assessment should take into account exposed, susceptible elements and coping capacities, which might have an important impact on the likelihood to suffer harm and injury due to a hazardous event. Although one should distinguish between vulnerable elements and coping capacity, there is a certain overlap (Birkmann 2006b).

The BBC framework considers three main thematic spheres (economic, social and environment) within which vulnerability and risk should be measured, thus reflecting the three pillars of sustainable development. It also consists of potential intervention tools and feedback loops that could help to reduce vulnerability in the social, economic and environmental spheres (Birkmann 2006b). However, when applying this within a risk assessment process, it is hard to differentiate between the three pillars and identify the overlaps. It also does not explain the relationship between natural phenomena and the hazard. Therefore this framework is more useful for disaster risk management than risk assessment.

2.8.1 Australian/New Zealand Risk Management Framework

Risk management is a structured systematic process or set of procedures for examining risk and making risk reduction decisions based on the examined risk. The

Australian/New Zealand ISO 3100 risk management process (Figure 9) provides a guide to establish and implement the risk management process in a given context. It is being widely adopted in fields such as financial investment, occupational health and safety, asset management and natural hazard risk management. A disaster risk management framework identifies the steps needed in order to implement meaningful disaster management strategies. These steps are critical in allocating resources for mitigation and developing emergency response plans (Pearce 2000). It is comprised of five key components. The risk management process often begins with establishing the context and identifying the nature of the risk. Identifying the nature of the risk is a critical step as there can be no risk management without having identified the risk and need for its management (Tobin & Montz 1997). The other steps are to analyse risk, evaluate risk and treat risk.

Figure 9: AS/NZS Risk management process, Source: Middelmann (2007).

Establishing context is about defining parameters to be considered to set the scope of the risk management process and manage risk. It should reflect the need for the protection of life and property from natural disasters. The objectives, policies and strategies to be achieved through the risk management process need to be clarified. It 44 requires consideration of external and internal factors at all scales, whether national, regional or local. External factors include social, political, economic, natural environmental factors; whereas internal factors include strategies, capabilities, resources and knowledge. The key drivers and trends that have impacts on the objectives and standards, guidelines and models to be adopted need to be clarified. It is also important to define roles and accountabilities of stakeholders (AS/NZS 2009).

Risk assessment is the overall process of risk identification, analysis and evaluation. In the risk identification process, target risks are found, recognized and described based on the past experiences, and the losses and severity observed in those events. It also involves identification of risk sources, events, causes and their partial consequences. Information such as historical data, theoretical analysis and expert knowledge can be utilized in this process. However, the uncertainty involved with the risk also needs to be addressed (AS/NZS 2009; Asian Disaster Reduction Center 2005). Risk treatment is necessary when the level of risk remains intolerable (InConsult 2009). The countermeasures that are taken to treat risk are; avoid risk, reduce risk, transfer and retain risk. Treating risk sources, modifying the likelihood of risk events, changing the consequences and sharing the elements of the risk are other important measures. All those measures should be properly planned. The plan should consider targeted risk criteria, priorities, procedures, and review mechanisms. The public needs to be given access to available plans and policies in order to increase their understanding of risks to which they are exposed (Asian Disaster Reduction Center 2005; InConsult 2009). Each step requires expert knowledge to make decisions. Therefore it is important to communicate and consult with stakeholders to monitor and review the risk management process. It will also help to establish expectations and ensure their needs are considered (Middelmann 2007).

2.9 Bushfire Risk Management Framework

Although there are many available risk management frameworks, practical assessment of risk in the context of bushfires remains challenging. Risk concepts need to be understood and used as a theoretical backbone that facilitates knowledge for bushfire risk management (Bründl et al. 2009). Bushfire risk management is difficult as bushfire

45 risk management and risk reduction strategies each need different sets of expert knowledge that are based on different sets of information. For example, bushfire managers may need to understand fire dynamics as well as the social cognitive factors that influence household level bushfire preparedness. As a result of the bushfire management process, areas at risk based on these diverse factors should be prioritized, and risk reduction strategies need to be identified and recommendations need to be made. It is also important to adopt AS/NZS risk management standards to gain recognition and acceptance for the standards in disaster risk management in Australia.

Granger (1999) presented a risk management framework that includes both the AS/NZS risk management process and the elements of risk (Figure 10). This framework was widely accepted and was the base model used in the ‘Cities Project’ that contributed to the International Decade for Natural Disaster Reduction (IDNDR) (Middelmann & Granger 2000). The project was conducted by Geoscience Australia. In their framework they identified hazard, elements at risk and vulnerability as the components of risk. This framework provides a good understanding of risk and risk reduction processes. It also highlighted scenario analysis, which was novel at that time. However, within this framework, the vulnerability of communities was not viewed holistically. They further emphasised the term ‘community acceptance’ to understand risk acceptance, explaining the perceived levels of risk among communities.

Levels of Elements at risk community and their acceptance vulnerability

History of Risk mitigation events and Scenario Safe, sustainable Probability and strategies and their analysis and prosperous process models response consequences communities options

Hazard phenomena and environment

Monitoring and Warning and surveillance forecasts

Figure 10: ‘Cities Project’ interpretation of the risk management process (Granger 1999).

Smith (2001) recognised that risk management is not effective if it ignores people’s concerns and fears. People can influence the community-level decision making process and sometimes decisions based only on an objective assessment may not be accepted by the community. It is also important to understand how the public constructs their perceptions of bushfire to improve risk communication, and direct risk reduction strategies most appropriately (Bushnell & Cottrell 2007a; Cottrell et al. 2008). Perceived risk is often not considered during the risk treatment process. Risk assessment and risk perception need to be combined for more effective risk reduction decision making processes (Smith 2001).

Renn (1992) also highlighted the importance of these two concepts in risk management. “If risk is seen as an objective property of an event or activity and measured as the probability of well defined adverse effects, the policy implications are obvious. Order risks according to “objective” measures of probability and magnitude of harm, and allocate resources to reduce the greatest risk first. If, on the other hand, risk is seen as a cultural or social construction, risk management activities would be set according to different criteria, and priorities should reflect social values and lifestyle preferences” (Renn 1992, p. 55).

This study utilises a modified version of a risk management framework proposed by Granger (1999) to interpret the risk assessment within the AS/NZS risk management process in the context of bushfires (Figure 11). To better understand exposed communities and vulnerabilities, Granger’s (1999) concept of risk is replaced by the holistic approach to risk proposed by Cardona & Barbat (2000). The concept of community threshold is viewed as a much broader concept, as the social cognitive factors that influence decisions about the level of vulnerability and risk (see Chapter 7). That is, cognitive processes, expectations and actions influence risk perceptions. In the employed framework, ‘shared responsibilities’ are recognised as an important component when implementing bushfire risk management strategies and they are added as a new element.

Hazard (Bushfire)

ExposedCommunities VulnerabilityFactors Risk Risk Exposureandphysical Damageto Historyof mitigation susceptibility built bushfiresand strategiesand environment consequences Socialandeconomic response fragilities Socioeconomic options impact Communitycapacities

Cognitiveprocesses, Risk expectationsandactions perception

Sharedresponsibilities

Monitorandreview Communicateandconsult

Risk Risk Risk RiskAnalysis Identification Evaluation Treatment

Figure 11: Interpretation of the bushfire risk assessment within the risk management process, after Cardona & Barbat48 (2000) and (Granger 1999).

2.10 Summary

This chapter discussed a number of useful definitions of hazard, vulnerability and risk. It reviewed theoretical frameworks for vulnerability and risk and discussed the use of these frameworks within the risk management process. Different risk management frameworks, including the AS/NZS risk management framework, were also reviewed and the importance of risk management was emphasised. The importance of adopting a holistic approach to risk in the bushfire management process was identified. As a result, a risk management framework that includes important elements of risk within the AS/NZ risk management framework was developed (Figure 11). In this framework, risk analysis was identified as a mandatory step in the risk management process. The next chapter will discuss the risk assessment process using the knowledge gained through reviewing the definitions and frameworks of risk and vulnerability from this chapter.

Chapter Three: Hazard Risk and Vulnerability Assessment

3.1 Introduction

Disasters are often viewed as a severe disruption to the normal functioning of a community or a society, and are a result of the interaction between hazardous physical events and vulnerable social conditions (Dilley et al. 2005; Lavell et al. 2012). A disaster occurs when an extreme event exceeds a community’s capacity to withstand that event (Cutter et al. 2008; Lindell & Prater 2003; Lindell et al. 2006). Community capacity refers to the positive features of a community that explain its ability to react to hazard events, its ability to reduce adverse effects and to return to the normal situation. These capacity characteristics may help to reduce the level of risk imposed by a hazard (Lavell et al. 2012). Disaster impacts can be either physical or social in nature. Initially, the physical impacts of a disaster are mainly determined by the level of preparedness and the emergency response capacities of the community, while social impacts are largely dependent on relief and recovery capacities (Lindell & Prater 2003). Understanding these pre-disaster conditions and processes is essential for effective disaster management (Lindell et al. 2006).

This chapter investigates the risk assessment process within the framework of a holistic approach to risk. In other words, it undertakes the risk identification and risk analysis steps depicted in Figure 11 (Chapter 2.9). It identifies hazard and vulnerability as the main components of risk assessment. Different definitions and approaches to measuring hazard, vulnerability and risk are discussed and the elements of vulnerability and risk are identified in the context of bushfires. Finally, an integrated approach to risk assessment is outlined to analyse bushfire risk at the UBI based on a holistic approach to risk. This chapter also discusses the importance of both objective risk assessment and subjective risk perception in the risk management process.

3.2 Risk Assessment

Risk assessment is considered to be the foundation for all the other elements of disaster risk management; it helps to identify what the problems are (Mazareanu 2007). It is a 51 complex process that includes various elements and processes. In the context of disaster management, the primary objective of disaster risk assessment is to use available information systematically to examine factors such as the probability of a hazard event occurring and the level of vulnerability of exposed elements at risk that have the potential to cause losses (Mazareanu 2007; Petak & Atkisson 1982). It is most useful at the preparedness and mitigation phases of the hazard cycle (Dilley et al. 2005).

When both the probability of occurrence and the expected losses are high, risk exists. On the other hand, risk is not associated with a very low probability of occurrence, or with minimal expected losses (Bankoff et al. 2004). In such assessments, therefore risk is considered to be a product of probability and loss. This is frequently applied in the insurance industry to measure and evaluate the cost of disasters (Smolka 2006). However the ‘cost’ of disasters is typically only measured in terms of tangible losses and does not count indirect losses and impacts. Thus it is more difficult to capture the cost of intangible losses such as fatalities, ill health and the loss of personal memorabilia (Granger 2003). Therefore cost estimation models only provide the monetary value of potential losses and do not count the interaction between the hazard and the vulnerable communities. In reality, the actual risk of a population is based on the interaction between the hazard and vulnerable communities. That is, the level of exposure, hazard, vulnerability and the capacity of the communities that are exposed to a particular hazard. Therefore alternative methods for risk models based on hazard, vulnerability, exposure and community capacities have been introduced (Davidson 1997; Granger 2003; Pearce 2000).

Risk assessment processes often involve both the science of measurement and the art of judgement (Alexander 1993). UNISDR (2004) defines risk assessment as “a methodology to determine the nature and extent of risk by analysing potential hazards and evaluating existing conditions of vulnerability that could pose a potential threat or harm to people, property, livelihoods and the environment on which they depend. The process of conducting a risk assessment is based on a review of both the technical features of hazards such as their location, intensity, frequency and probability; and also the analysis of the physical, social, economic and environmental dimensions of vulnerability and exposure, while taking particular account of the coping capabilities 52 pertinent to the risk scenarios” (p. 16). Figure 12 shows the elements and basic stages undertaken in a risk assessment process (UNISDR 2004).

In general, the risk assessment process is divided into three phases: identification of risk, analysis of risk and evaluation of risk (AS/NZS 2009; Petak & Atkisson 1982; UNISDR 2004). Identification of risk factors is usually the starting point of the risk assessment. Risk analysis is often comprised of assessment of risk factors (i.e., hazard, vulnerability and capacities). Hazard assessment helps to identify the probability of occurrence of a specific hazard as well as its frequency, intensity and the area of impact. Vulnerability analysis measures the pre-disaster condition of a human society, which identifies both susceptibilities and capacities. Risk analysis is done by contrasting the level of hazard on the one hand and vulnerability and capacities on the other. Risk evaluation is the process that investigates the social consequences associated with different levels of estimated risk. The results of risk evaluation are then directly linked with the risk management decision-making process.

Figure 12: Risk assessment process (UNISDR 2004, p. 63).

According to Petak & Atkisson (1982), risk assessment aims to identify societal risk, which may include the description of the problem associated with the natural hazard, its

53 impact and the side effects of the event. Risk estimation is targeted at producing a quantitative description of the population at risk. It uses scenario-based analysis, such as the consequences associated with various magnitude/event probability calculations and the integration of these magnitude/event probabilities into a quantitative measure of risk. Risk evaluation relies on providing policymakers with referents against which the possible significance of risk may be weighed.

AS/NZS (2009) has defined risk assessment as the systematic process of understanding the nature of and deducing the level of risk. It provides the basis for risk evaluation and decisions about risk treatment. Risk analysis may be undertaken to varying degrees of detail depending upon the risk, the purpose of the analysis, and the information, data and resources available. Analysis may be qualitative, semi-qualitative or quantitative, or a combination of these, depending on the circumstances. A risk analysis is usually conducted to identify adverse consequences, although it may also be used proactively to identify and prioritise potential opportunities (AS/NZS 2004).

There are two common types of approaches to risk assessment: the qualitative approach and the quantitative approach (Smith 2001). Qualitative risk analysis is a process of assessing the impacts of the identified risk factors. The definitive characteristic of qualitative models is the use of subjective indices, such as ordinal hierarchies: low- medium-high, vital-critical-important, benchmark, etc. Through this process, priorities are identified to mitigate potential risk factors, depending on the impact they could have (Mazareanu 2007). Qualitative research emphasises the qualities of entities, and processes and meanings that are not experimentally examined or measured in terms of quantity, amount, intensity or frequency (Denzin & Lincoln 2000). Therefore in qualitative risk assessment, the components of risk, which are basically hazard, elements at risk and vulnerability, are expressed verbally and the final result is in terms of ranked or verbal risk levels (ICG 2006). Qualitative risk can also be presented as a risk matrix with the qualitative hazard in the first dimension and qualitative consequences in the second dimension. Qualitative risk assessment is subjective in nature; the intimate relationship between the researcher and the study and the situational constraints shape the study (Denzin & Lincoln 2000).

Denzin & Lincoln (2000) have emphasised that the quantitative studies focus on the measurement and analysis of casual relationships between variables, not processes. Quantitative risk assessment involves quantification of risk components and com- putation of risk from these components. The purpose of quantitative risk assessment is to calculate a mathematical value for the risk, which enables improved risk communication and systematic decision making (Lee & Jones 2004).

However, in reality, the quantitative risk assessment process is complex because there is an additional need to understand the magnitude of the event and how it may affect risk outcomes (Smith 2001). In order to achieve that, the elements and sub-elements of risk need to be identified, assessed individually and combined based on a model, such as those discussed in Chapter 2.5.

3.3 Elements of Risk

The different risk models and conceptual frameworks discussed in Chapter 2 identified different elements as important for understanding risk. Those models are highly context specific and the elements that have been used can vary with the perception of their importance by researchers. The most prominent risk elements are ‘probability of occurrence’ and ‘consequences’ (AS/NZS 2004; Smith 2001). These elements are widely applied by the insurance industry in order to understand potential monetary losses. The key elements of the conventional equation for natural hazard risk are ‘hazard’ and ‘vulnerability’ (Blaikie et al. 1994; UNISDR 2004). After Crichton (1999), ‘exposure’ was also included in the conventional risk equation. Granger (1999), in his model, substituted ‘elements at risk’ for exposure. With the development of research on climate change, coping capacity, adaptive capacity, community resilience and deficiencies in preparedness are also being emphasised as factors that adjust the level of vulnerability and have been added into risk paradigms (Villagrán de León 2006). Sometimes scholars have considered exposure and capacities to be independent elements (Davidson & Lambert 2001; Zhijun et al. 2009) and sometimes they have seen them as subcomponents of vulnerability (Dolan & Walker 2006; Gallopín 2006; Preston et al. 2009). However, it is important to emphasise that hazard, vulnerability, exposure,

55 elements at risk, capacities and resilience are different concepts and those definitions relate to specific methodological approaches that facilitate the understanding of risk.

As discussed in Chapter 2, this study adopts a holistic approach to risk (Cardona & Barbat 2000), and within that framework, hazard and vulnerability are considered to be key elements of risk. Therefore, within that framework, hazard and vulnerability assessment are the key components of risk assessment.

3.4 Hazard Assessment

Hazard assessment is the first step of any risk assessment. It helps to identify external events that create risk. “The objective of hazard assessment is to identify the probability of occurrence of a specific hazard, in a specific future time period, as well as its intensity and area impact” (UNISDR 2004, p. 64). In the hazard assessment, hazard information associated with the human system is collected to evaluate and identify the significance of the hazard for the human system. The usual information collected includes: Local hazards - location & probability The extent to which they threaten local populations - severity Ease with which their effects can be averted - manageability (ADPC 2006)

Hazard assessments typically involve people who have scientific knowledge about a specific hazard type, as they deal with various technical aspects of hazards such as geological, geophysical and atmospheric conditions. The results of hazard assessment are usually displayed in intensity maps for different hazard scenarios that indicate the locations that are likely to be impacted. This component is described in Chapter 4.

3.5 Vulnerability Assessment

The objective of a vulnerability assessment is identifying, quantifying, and prioritizing the vulnerabilities in a given place at a given time. It is an important prerequisite for risk assessment, and plays an important role in identifying the inherent weaknesses of community that make it more susceptible to harm. Vulnerability analyses can be used extensively by disaster managers to delineate, characterise and assess potential impacts,

56 which leads improved mitigation strategies (Cutter 1996). The scientific community has put much effort into increasing our understanding of the concept of vulnerability and developing tools to assess the level of vulnerability within a population. As discussed in Chapter 2.3, different conceptual frameworks have been developed to understand vulnerability in the context of natural hazards. As a consequence, divergent methods of vulnerability assessment have emerged. Despite these different concepts and frameworks, some similarities can also be found. All these frameworks share a common view that disasters are a product not only of hazardous events but also of social, economic and political environments (Fuchs et al. 2011; Rashed et al. 2007). The central focus of vulnerability assessment is to identify the pre-disaster conditions that make people more susceptible to natural disasters.

The different concepts and frameworks for vulnerability have suggested various dimensions of vulnerability. UNISDR (2004) identified four dimensions of vulnerability that are relevant in the context of disaster reduction; physical, social, economic and environmental. Physical factors describe the exposure of vulnerable elements within a region; economic factors, the economic resources of individuals, population groups, and communities; social factors, the non-economic factors that determine the well-being of individuals, population groups, and communities, and environmental factors, the state of the environment within a region. Later on other dimensions such as political and psychological were added (Hufschmidt 2011; Paton & Johnston 2001). In natural hazards research, multiple dimensions are often utilised, with social, environmental and economic dimensions the most widely used (Bogardi & Birkmann 2004; Cardona et al. 2012).

The many dimensions of vulnerability have made vulnerability assessment very complicated. As a result, division of work between disciplines is visible and the results have proven that the vulnerability assessments are largely dependent on the purpose and the interest of the analysis (Fuchs et al. 2011). “There is a tendency for researchers to focus on only one aspect of vulnerability, depending on their expertise. Engineers tend to focus on building vulnerability models incorporating considerations such as structural type, building use, building codes and engineering assessment. Economists develop models focusing on economic losses and the impact of government expenditure on 57 recovery and mitigation options to reduce risk. Social scientists focus on people, communities, access to services and organisational and institutional measures” (Middelmann 2007, p. 37). In reality, a comprehensive assessment of risk should consider all aspects of vulnerability. It requires expertise in different disciplines with wide access to resources. However, such assessments are rare due to lack of resources and coordination between relevant institutions (Middelmann 2007).

The objective of this study is to understand the impact of natural hazards on human systems. Therefore the main focus is to identify their consequences for human societies.

3.6 Elements of Vulnerability

Since there is no standard set of elements to represent vulnerability, methodological approaches to vulnerability assessment are highly varied. Various elements have been used based on the context, multi-dimensional nature, spatial and temporal scale of analysis, and the dynamic nature of vulnerability (Hufschmidt 2011). Common elements include exposure, physical vulnerability, social vulnerability, biophysical vulnerability, sensitivity, resilience, and coping capacity/adaptive capacity.

Exposure refers to the elements that are subjected to hazard impacts (Middelmann 2007). Exposure is caused by the presence of human systems in hazard-prone geographical areas, which are thereby subjected to potential harm, loss or damage (Lavell et al. 2012; Lindell et al. 2006). In some literature, the magnitude of the physical process is also included in the measure of exposure. This may be problematic when working within the concept of risk as the magnitude of the physical process may already be accounted for in the assessment of the hazard itself (Hufschmidt 2011).

Physical vulnerability is defined as the vulnerability of the physical or built environment (2003). It is also sometimes described as structural fragility (Dilley et al. 2005). Social vulnerability refers to the pre-existing condition of the individual or a human system that can be viewed independently of a hazard (Cutter et al. 2009; Middelmann 2007). It further describes characteristics of the population that influence their preparedness, response and recovery capacities (Cutter et al. 2009). Biophysical

58 vulnerability is associated with the human occupancy of the hazard prone area and the characteristics (magnitude, frequency, duration) of a particular event (Cutter 1996), so could also be described as exposure. Sensitivity is defined as the degree to which a system is affected either adversely of beneficially by a hazard (Brooks 2003). The effect can be either direct or indirect. Sensitivity explains the impact of the hazard (severity) and it is not widely used in natural hazard research and mainly used in climate change research (Brooks 2003; Preston et al. 2009; Turner et al. 2003).

Within the hazard research community, the concepts of adaptive capacity and coping capacity are sometimes used interchangeably with resilience (Hufschmidt 2011; Thywissen 2006), though climate change researchers distinguish between these concepts. Each of these elements has an ability to modify the level of vulnerability. “Coping capacity encompasses those strategies and measures act directly upon damage during the event by alleviating or containing the impact by bringing about efficient relief, as well as those adaptive strategies that modify behaviour or activities in order to circumvent or avoid damaging effects” (Thywissen 2006, p. 38). Coping capacity is also recognised as a response capacity, while adaptive capacity is explained in a much broader sense as the process that can be channelled into mitigation, preparedness and recovery (Hufschmidt 2011).

Resilience is often seen as the capacity of a system to absorb disturbance and re- organize while undergoing changes to retain the same function, structure, identity and feedbacks (Berkes 2007; UNISDR 2004; Walker et al. 2004). In the context of hazards and disasters, resilience is defined in several ways. Engineering resilience is only discussed in term of the strength and flexibility of materials and structures while community resilience is defined as the ability of social units to mitigate hazards, contain the effects of disasters when they occur, and carry out recovery activities in ways that minimize social disruptions and mitigate the effects of future events (Bruneau et al. 2003; Cutter et al. 2010).

The above-mentioned elements are being widely used in many vulnerability assessments and frameworks in different disciplines (see Chapter 2.3). However, there is little agreement on the meaning of these elements. Use of these elements is mainly 59 based on the context, expertise and the perception of the user. Some elements are compatible with each other and some are not. For example, resilience may include both coping and adaptive capacities. The relationships between these elements are not well known and are complex, interactive and linked. Therefore, careful scrutinising is required to avoid misunderstanding.

In the holistic risk framework used in this study, Cardona & Barbat (2000) have identified three elements that make communities vulnerable: exposure and physical susceptibility, social and economic fragility, and lack of resilience and ability to cope or recover. Exposure and physical susceptibility are hazard-dependent elements such as the volume and concentration of elements in a given area. In this study, exposure and physical susceptibility represents both physical and environmental characteristics that influence the potential impact of a bushfire on a community. Such characteristics include building density, proximity to the hazardous area, topography, and land cover. Social and economic fragilities are non hazard-dependent elements that relate to existing societal conditions. Such conditions include poverty and marginality, health levels, dependency and social disparity. This element is identified as social vulnerability in this study (Chapter 5). Lack of resilience or ability to cope and recover represents weaknesses in the ability of communities to absorb the impact of hazard events, and their lack of capacity to prepare, respond and recover. Parameters such as local human resources, infrastructure and emergency facilities are considered in this element. In the context of bushfires, preparedness and response capacity are important in order to minimise the impact of the hazard on the community. At the UBI almost all residents have insured their properties, and have thus transferred the risk. Some of the conditions that influence recovery capacity, such as level of income, are addressed within the element of social vulnerability. Therefore, preparedness and response capacity is considered the central component of this study that addresses the characteristics of lack of resilience or ability to cope and recover (Chapter 6). The interaction between the level of hazard and the vulnerability conditions determined by its elements will provide an integrated view of risk.

3.7 Integrated Risk Assessment

Integrative holistic approaches to risk assessment that capture a greater range of dimensions of vulnerability and risk are recognised as useful tools to inform risk reduction strategies (Cardona et al. 2012). Integrated approaches to risk assessment combine the elements of risk, thus producing an integrated risk map (Granger 2003; Greiving et al. 2006). Various other integrated methods such as the hazard of place model (Cutter et al. 2000), non spatial models such as the urban earthquake disaster index (Davidson 1997), and the world risk index (UNU-EHS 2011), and spatial models (Rashed et al. 2007) are also available within the risk hazard framework. The most recent IPCC SREX report (Lavell et al. 2012) highlighted the importance of this integrated approach to disaster risk reduction.

This study uses an integrated view of risk that consists of two distinct elements; hazard and vulnerability. Hazard is implemented as the probability of occurrence of a future bushfire event as well as its likely intensity and area of impact. Vulnerability is conceived as a characteristic of the community that can be assessed through a combination of elements associated with the physical conditions of the geographic location and the built environment (Chapter 6), the social conditions of the population in that place (i.e., social vulnerability) (Chapter 5) and the community-level capacity to prepare for and respond to the hazard (Chapter 6). Elements of vulnerability and hazard are integrated to understand risk (Chapter 7) (Figure 13). Although vulnerability is considered an internal component of risk within the holistic approach to risk, shadows of the place based model of vulnerability (Cutter 1996) are visible within the concept of vulnerability in the framework employed in this research. However, in the initial place- based model, place vulnerability consists of bio-physical and social elements, while the framework used here consists of three elements: exposure and physical susceptibilities, social vulnerability and community preparedness and response capacity. It is important to identify these elements separately for implementing risk reduction measures because those elements explain the underlying causes of risk. Nevertheless, these elements overlap and are inextricably bound together in disaster situations.

Risk

Hazard Vulnerability

Exposure and Social Response and Physical Vulnerability Coping Susceptibility Capacities

Pre-impact conditions

Figure 13: Elements of vulnerability and risk in an integrated risk assessment.

The integrated risk assessment used in this research consists of three key components: hazard assessment, vulnerability assessment and integrated risk assessment. The objective of bushfire hazard assessment is to understand the spatial distribution of bushfire hazard based on potential ignition and severity. The resulting hazard map is used as an input for the integrated risk map. Vulnerability assessment is performed to understand the extent to which various elements contribute to vulnerability; exposure and physical susceptibility, social vulnerability and preparedness and response capacities. Separate assessment is performed to understand the contribution of each element to overall vulnerability and vulnerability maps are developed. The individual vulnerability maps are combined to develop an overall vulnerability map of the study area. Finally the hazard map and the overall vulnerability map is combined based on the conventional risk model discussed in Chapter 2 (equation 1) to develop the integrated risk map. The final integrated risk map along with the resulting hazard map, the overall vulnerability map and individual vulnerability element maps allows decision makers to identify areas that are highly hazardous and have a high degree of vulnerability as well as the underlying causes that make the area more vulnerable, such as physical exposure, social inequalities and lack of preparedness and response measures.

3.8 GIS and Risk Assessment

Since the physical properties of the hazard vary spatially and the populations that might be exposed to a particular hazard are also spatially distributed, risk is inherently spatial (Emmi & Horton 1995). Therefore, a risk assessment should address both the degree of risk as well as spatial variations in risk. Geographic Information Systems (GIS) is a tool that captures, manipulates, processes, analyses and visualises spatial data (O'Sullivan & Unwin 2010). It fulfils the required risk assessment tasks such as synthesising spatial data and mapping the spatial relationships between hazard phenomena, exposed elements and vulnerable communities (Zerger 2002). “GIS provides the analytical ‘engine’ which drives the natural hazard risk assessment process. It also provides a more potent form of risk communication through its capacity to provide a visual representation of risk situations” (Granger 1998, p. 15). Knowing the locations of hazard zones, the distribution of vulnerable groups, and understanding their spatial relationships is the key to developing preparedness planning and mitigation strategies. It is rather advantageous to fuse a philosophy of risk management and GIS as a decision-support tool in the risk management process (Zerger 2002). GIS has generated great benefits throughout the entire disaster management cycle, from preparedness, prevention, and mitigation through detection to response and recovery (Greene 2002).

GIS have been utilized in wide variety of risk assessment applications in different disciplines such as public health (Fleming et al. 2007; Rushton 2003), ground water and environmental pollution (Al-Adamat et al. 2003; Jerrett et al. 2001; Secunda et al. 1998) and natural hazards (Chen et al. 2003; Granger 2003; Zerger 2002). The common objective of these applications is to identify populations at risk with respect to the threat to which they are exposed. In natural hazard risk research, it is widely used for hazard mapping (Dhakal et al. 2000; Van Westen et al. 1997), vulnerability mapping (Cutter et al. 2000; Preston et al. 2009; Wu et al. 2002), and integrated risk mapping (Chen et al. 2003; Granger 2003). These maps can be valuable for risk communication, and evaluation of risk management strategies. Other GIS applications such as evacuation modelling, spatial decision support systems, monitoring systems and post-disaster

63 recovery systems have also received particular attention in the natural hazards literature (Zerger 1998).

Spatial relationships between hazard intensity and vulnerable populations are an important investigation in integrated risk assessment. Despite the development of the theories for an integrated approach to risk, examples from this field have mainly concentrated on physical hazard modelling within the GIS because critical variables contributing to susceptibility can be derived from spatial data sources. A GIS-based approach for integrating physical hazard models and human vulnerability factors into natural hazard risk assessment has not been well addressed. While Cutter et al. (2000) provided a detailed example of integrating biophysical and social vulnerability components in their “vulnerability of place” model, their work conceptualises potential hazards, their frequency, and their locational impacts as biophysical vulnerability, which is independent of hazard intensity.

The risk assessment conducted by Zhijun et al. (2009) used the hazard risk framework and included hazard, exposure, vulnerability and emergency response capacities as elements of risk. However, the hazard component of their assessment was only based on physical parameters and physical hazard models that predict the intensity of the hazard were not utilised. Chen et al. (2003) and Greiving et al.(2006) have proposed an integrated risk assessment method in GIS within the hazard risk framework which contains hazard and vulnerability as key elements. Their proposed models include the relationship between hazard intensities and vulnerable populations. However, they only proposed the model and it was not actually implemented.

Granger (2003) provided a more complete example of integrated risk assessment using GIS. In his framework, elements of risk that were considered are elements at risk, hazard and vulnerability. Hazard intensity was derived from storm tide hazard data and scenario levels and the spatial relationships between hazard, vulnerable communities and elements at risk were identified. The results provide material that is useful for future development planning, education of the community as to the risks they face, and the steps that they and public officials need to take to reduce those risks. In this study, GIS was used to understand the relationship between hazard and vulnerability. Hazard 64 assessment was performed using physical hazard models and integrated with the vulnerability.

Chen et al. (2003) identified three important steps to include in an integrated risk assessment process supported by GIS; data integration, risk assessment and risk decision-making. Data integration includes building a spatial database suitable for spatial analysis and modelling. In order to achieve that, different data types (spatial and non-spatial) gathered from various sources are combined. Physical environmental data are often acquired in spatial form either in raster or vector formats. Socioeconomic data includes population and household information that is readily available with census data. Other disaster management related data such as community-level activities and resources are available from relevant agencies and need to be collected and converted into usable formats. Risk assessment examines the individual aspects of hazard and vulnerability. Various GIS and spatial analysis techniques are employed in this step (see Chapter 6). Risk assessment directly contributes to risk reduction decision-making. During this process, useful risk management decisions are discussed and risk assessment tasks are identified.

3.9 Scale of Risk Assessment

Risk is a product of processes occurring at different spatiotemporal scales. Measuring vulnerability and risk at different scales helps to understand cross scalar dynamics in vulnerability and risk, from global to local or vice versa (Cutter et al. 2008). The level of detail of information and results vary with the scale. From global to local scales, the value of scenarios and data decreases while data accuracy and research effort required increase (Kaiser 2006). Cutter et al. (2008) have stated that the unit of analysis becomes a unique expression with differences in scale. At the regional level, units such as Gross Domestic Product (GDP) are used, while at the individual or household level, issues of livelihood and entitlements are more relevant.

Selecting an appropriate scale is crucial and often challenging as the relevance of a specific variable may change with scale and degree of data aggregation (Eakin & Luers 2006; Fekete et al. 2010; Turner et al. 2003). Data integration can be successful only if

65 spatial analysis units are compatible with each other (Chen et al. 2003). Therefore, scale of analysis plays an important role in identifying vulnerability and risk of people and places as the spatial distribution of vulnerability varies with scale. While sub-national scales provide insight into regional processes and patterns, finer scales are beneficial for better understanding and capturing the root causes of vulnerabilities (Fekete et al. 2010). In the context of disaster management, vulnerability assessments are often conducted at scales that coincide with the geographic levels at which disaster management decisions are taken (Cutter et al. 2008; Eakin & Luers 2006). The selection of appropriate scale is often influenced by the purpose, data availability and quality, construction and presentation of outputs, and the extent of the study area.

3.10 Uncertainty in Risk Assessment

Uncertainty is associated with any risk assessment. Therefore risk management plans have to deal with some extent of unpredictability regarding location, magnitude, intensity and loss (Gómez-Fernández 2000). Renn (1992) describes this as the distinction between reality and possibility. Any disaster management plan is developed based on a risk assessment. It is highly appropriate to use scenarios as a basis for developing disaster management plans due to the uncertainty associated with risk. By knowing what can happen and the level of risk involved, effective disaster response and recovery plans can be designed for each scenario. The relevant plan can be deployed in the event of a disaster according to the scenario. Furthermore, as each scenario indicates the magnitude and resources needed to manage the disaster, the response mechanism can be deployed in an efficient and effective manner. Scenario-based risk analysis also allows identification and prioritization of potential disaster areas. It can be used to design mitigation activities in order to reduce exposure and vulnerability through loss control activities.

3.11 Risk Assessment and Risk Perception

Although disaster risk is defined as the product of the likelihood of hazard event occurrence and vulnerable conditions, a variety of cognitive, cultural and social processes influence the perceptions of and judgments about risk and the allocation of efforts to address these risks (Lavell et al. 2012). Estimating risk based on probabilities

66 and potential consequences may not address the people’s concerns associated with the risk. “In some cases, as in vulnerability/capacity assessment exercises, risk perception may be formally included in the assessment process, by incorporating people’s own ideas and perceptions on the risks they are exposed to” (UNISDR 2004, p. 64).

Quantitative risk assessment is widely employed in hazard research and management. It is a quantitative method often carried out by experts using scientific models. It involves definitions of risk elements, value judgments and specific data sets (Smith 2001). It has the scientific attribute of repeatability, provided agreement can be reached on the risk estimates and is (relatively) independent of social and cultural processes (Cottrell et al. 2008; Kirkwood 1994). Individual perceptions of risk are determined by a number of factors including past experience, present attitudes, personality and values, and future expectations (Smith 2001). Understanding both risk perception and the scientific need for quantitative risk measurement and their relationships facilitates risk management actions and decisions (Cardona 2004). The differences between these two understandings of risk are summarised in Table 1.

Table 1: Differences between risk assessment and risk perception, Source: Smith (2001)

Phase of Analysis Risk Assessment Process Risk Perception Process Event monitoring Individual intuition Risk identification Statistical inference Personal awareness Magnitude/Frequency Personal experience Risk estimation Potential loss Intangible losses Risk scenario analysis Personality factors Risk evaluation Accepted level of risk Individual action Community policy analysis

3.12 Study Area

The UBI is a zone where urban development and activities overlap with natural or semi- natural bushland vegetation (Collins 2005; Lein & Stump 2009). Outside of Australia, this area is often described as the wildland-urban interface (WUI). In Australia the UBI population has increased over the time as there is a growing movement of population into UBI areas (Bushnell & Cottrell 2007b). Due to increasing human population levels and demands for both affordable housing and lifestyle blocks, urban development is 67 expanding deeper into bushland areas that are fire prone. This expansion has put many communities at increased bushfire risk. In particular, in recent years, bushfire risk in the UBI has increased because of the growing popularity of rural residential living and the preservation of inherently flammable natural areas within UBI development (Preston et al. 2009). People move to this area mainly due to quality of life issues such as the aesthetic value of the landscape, reduced crime, pollution and crowding, and convenience and affordability of the properties (Bushnell & Cottrell 2007b; Cottrell & King 2007).

At the UBI, a number of social, political and economic characteristics are interacting in close proximity, which makes the UBI a complex mosaic of local cultures and social adaptation (Paveglio et al. 2009). This mix of people, property, fuel load and increasingly frequent fire weather days has not been matched by appropriate bushfire preparedness efforts. This deficiency highlights the importance of including household and community preparedness in any risk management framework (Paton 2006c).

The UBI has become more vulnerable to potential bushfires due to rapid population growth, unplanned land use management, lack of awareness of risk, inadequate community involvement in bushfire management efforts, and lack of resources at the community level (Beringer 2000; Lowe 2008). The dynamic nature of the UBI makes it difficult for fire management authorities to operate at the UBI. Many people living in communities that are vulnerable to natural hazards also continue to demonstrate poor knowledge of preparedness, response and recovery measures (McIvor & Paton 2007). Even if the probability and intensity of bushfire hazard activity remains constant, rapid growth of population and economic and infrastructure development within the UBI environment will result in significant loss and disruption from a potential bushfire event (Paton 2006b). In order to minimise potential losses, bushfire risk management activities also need to be strengthened at the UBI. For this study UBIs within the Ku- ring-gai and Blue Mountains Local Government council areas were selected because both locations are at high biophysical risk of bushfire, yet they have differing socio- demographic characteristics.

Figure 14: Urban Bush Interface, Source: Aerial image - SKM, Photographs - Author.

3.12.1 Blue Mountains LGA

The Blue Mountains Local Government Area (LGA) is a major tourist destination in Australia because of its natural beauty and location at the western fringe of the Sydney Metropolitan Area (Figure 15). It is nestled within a larger World Heritage Area proclaimed in 2000 due to its outstanding level of unique natural values. The Blue Mountains LGA covers 1,433 km2, of which 70% comprises the Blue Mountains National Park. The LGA has a population of approximately 76,719 within 26 suburbs. The Blue Mountains suburbs can be classified as belonging to the Lower Blue Mountains (Lapstone, Glenbrook, Blaxland, Mount Riverview, Warrimoo, Sun Valley, Valley Heights, Springwood, Winmalee, Yellow Rock and Faulconbridge), Mid- mountains (Linden, Woodford, Hazelbrook, Lawson and Bullaburra) and Upper Blue Mountains (Wentworth Falls, Leura, Katoomba, Medlow Bath, Blackheath, Mount Victoria and Bell). However, the area’s population is largely concentrated in Springwood and Katoomba, the major town centres. The major industry in the Blue Mountains is tourism, which significantly contributes to the economy, industrial and cultural settings of the area.

A high proportion of the LGA’s permanent residents were born in Australia. Residents’ average taxable incomes are low compared to the broader Sydney metropolitan area. Moreover, the income differential between area residents and those of the Sydney metropolitan area is increasing (City of Blue Mountains 2009). There is a gradient of disadvantage across the Blue Mountains LGA (Index of Relative Socioeconomic Disadvantage score ranging from 810 to 11521), with people living in the Lower Blue Mountains tending to be more advantaged in terms of household income, educational achievement, and employment opportunities than those in the Upper Blue Mountains (Blue Mountains City Council 2012). This relationship is likely due to the Lower Blue Mountains’ greater access to the resources of the broader Sydney Metropolitan Region.

The Blue Mountains has numerous settlements situated in highly bushfire prone areas, with most communities located on exposed UBI in close proximity to bushland. Almost three-quarters of residential buildings (73%) in the LGA are in high bushfire risk areas (Chen 2005). Therefore, based on the percent of addresses located within bushfire-prone land, the Blue Mountains LGA ranks as the most bushfire-exposed LGA out of the 61 LGAs in the Greater Sydney Metropolitan Region (Chen 2005).

The bushfire season in the Blue Mountains runs from September to February. Favourable weather conditions associated with westerly or northwesterly winds, high temperatures and low humidity can lead to extreme fire behaviour with potentially catastrophic consequences. This poses threats to life and property, which creates challenges in emergency management and preparedness for bushfires. In addition, the area’s topography and limited road access makes the evacuation process extremely difficult in the event of bushfires (Burton & Laurie 2009). Because of the presence of tourists, on any given day there may also be large numbers of people who have little knowledge of bushfire risk present in the area.

The Blue Mountains area has on average 28 bushfires per year (Blue Mountains Bush Fire Coordinating Committee 2008). Of these, seven can be categorized as severe fires

1 Index of Relative Socio-economic Disadvantage focuses primarily on disadvantage, and is derived from census variables like low income, low educational attainment, unemployment, and dwellings without motor vehicles. Socio-Economic Indexes for Areas (SEIFA) 2006, Australian Bureau of Statistics (http://www.abs.gov.au/ausstats/[email protected]/mf/2033.0.55.001/).

70 that run longer than 24 hours. The main sources of ignition in the Blue Mountains are human activities and lightning. Human caused ignitions include deliberately lit fires and improperly supervised campfires. Ignition from lightning occurs due to thunderstorm activity.

Figure 15: Map showing the study areas and the Greater Sydney region.

3.12.2 Ku-ring-gai LGA

The Ku-ring-gai LGA is located in the northwest suburbs of Sydney (Figure 15) and lies 15 kilometres from Sydney’s CBD. It much more urbanized than the Blue Mountains LGA. The total council area spans 86 km2, of which 32 km2 are covered by bushland. The LGA is surrounded by national parks on three sides; Ku-ring-gai National Park to the east, Lane Cove National Park to the west, and Garigal National Park to the southwest.

The LGA has an estimated population of 108,135 within 17 suburbs (Australian Bureau of Statistics 2009). A majority of the residents of Ku-ring-gai are either Australian-born or from an English speaking background. They are well-established in the area, with most having lived there for more than 10 years. Most residents own their houses. In Ku-

71 ring-gai, a large proportion of residents are married and live in family households. Residents’ levels of education are high, but the level of full time employment is relatively low because a high proportion of the area’s residents are retired.

The Ku-ring-gai LGA is well-known for the quality of its primary and secondary school facilities. Major hospitals and nursing homes are also found in the area. In general, Ku- ring-gai’s socioeconomic indicators such as level of income, education and occupation are higher than the Sydney average (Deodhar 2004; Ku-ring-gai Council 2005; Ku-ring- gai Council 2006).

Ku-ring-gai is ranked as the third most vulnerable LGA to bushfire out of 61 LGAs in the Greater Sydney Metropolitan Region, based on the percent of addresses located within bushfire-prone land (Chen 2005). Bushfire season in Ku-ring-gai runs from October to March. During these months, high daytime temperatures and strong northwest winds typically prevail. Low humidity and dry vegetation along with high fuel loads and prevailing weather conditions influence bushfire behaviour in the area. There is a high probability of bushfire occurrence in the months of December and January (Hornsby/Ku-ring-gai Bushfire Management Committee 2008).

Ku-ring-gai has experienced severe bushfires that have caused significant damage to property and threatened lives. Ku-ring-gai and Hornsby Councils work together on bushfire management as the national parks border both areas. On average 40 bushfire incidents are reported per year across both areas. Every 7-10 years, a major bushfire event is expected. The main sources of ignition in Ku-ring-gai are human activities or lightning. Human caused ignitions include deliberate ignitions, campfires and debris burning (Hornsby/Ku-ring-gai Bushfire Management Committee 2008).

3.13 Summary

This chapter discussed different concepts of hazard, vulnerability and risk assessment. Based on the analytical frameworks of vulnerability and risk reviewed in Chapter 2, the elements of risk and vulnerability and the relationships between these elements were identified. An understanding of these elements is necessary to address the issue of

72 bushfire risk management and to develop an integrated method for assessing bushfire risk. After scrutinising the elements, an integrated risk assessment framework was outlined. The importance of quantitative risk assessment and risk perception was also highlighted (decision-making based on risk perception is more broadly discussed in Chapter 6). The use of GIS in the field of natural hazards was also discussed and its importance for hazard risk assessment was underscored. Finally, the study area of this research and the issues related to bushfire risk in the area were introduced.

The goal of this thesis is to develop an integrated model of bushfire risk for bushfire management. It also explores household level perceptions in the context of bushfire management to understand the relationship between risk perception and household preparation. The proposed risk assessment framework consists of two key elements of risk; hazard and vulnerability. Therefore understanding the level of bushfire hazard and vulnerabilities is the first step in understanding risk. The next chapter explores the hazard component of risk. It applies GIS-based multicriteria evaluation modelling in the context of bushfire risk assessment. A bushfire history database is used to obtain detailed description of bushfire events occurred in the study area. A spatial multicriteria evaluation technique is applied for modelling of the bushfire hazard. Results describe the spatial variation of bushfire hazard in the area. It is used as an input for bushfire risk modelling detailed in Chapter 6.

Chapter Four: Bushfire Hazard Assessment

4.1 Overview

This chapter address the hazard element of the overall risk assessment framework discussed in Chapter 3. As discussed in Chapter 2, bushfire hazard is considered to be the probability of occurrence of a future bushfire event as well as its likely intensity and scale of impact. In the bushfire hazard assessment process, this is defined as the ignition probability and fire severity. The bushfire hazard assessment employed in this study utilised a bushfire history database provided by the New South Wales Rural Fire Services (NSWRFS). Only events that occurred in the past 30 years were considered as they were the only years for which complete data were available. GIS and spatial modelling techniques were used to develop the bushfire hazard map. In Chapter 6 the resulting bushfire hazard map is used as an input to assess overall risk in the area.

4.2 Introduction

Natural disasters represent a potential threat to society and are a result of the interaction between humans and extreme natural events (Tobin & Montz 1997). As discussed in Chapter 2, hazard is considered to be the pre-disaster triggering factor, and is thus an important component of risk. Bushfire is a hazard that can be generated either naturally by lightning-caused ignition or by human-caused ignition. Bushfires become a risk when they interact with people and properties at the UBI (Blanchi et al. 2002). They are a part of Australian ecosystems. The occurrence of bushfires in Australia has steadily increased over time, with longer bushfire seasons and an increasing number of extreme fire weather days (Hennessy et al. 2005). Bushfire threat at the UBI has increased over the years and has had significant consequences such as those felt in the Canberra fires in 2003 and Black Saturday fires in Victoria in 2009. As a result, concerns about fire hazard mitigation and management have become a key policy issue for the public and policy makers (Ellis et al. 2004).

Fire management authorities have the capacity to suppress most bushfire events in normal circumstances. However sometimes fire behaviour exceeds the level of fire 75 fighting capacity (Lowe 2008). This happens for many reasons such as extreme weather conditions, fuel amount and types, and multiple ignitions that require more fire fighting resources. Therefore, it becomes necessary to mitigate these unexpected events efficiently in order to protect the lives and properties of people who are living in bushfire prone areas. To achieve that, bushfire hazard assessment models that produce robust, accurate and usable predictions of fire hazard are needed.

Researchers have put their effort into investigating how to best develop tools to assess bushfire hazard and risk. Widely used methods are; static bushfire susceptibility models (Atkinson et al. 2010; Lein & Stump 2009), bushfire simulation models (Finney & Station 1998; Gillian & Derek 2011), spatial interpolation techniques to estimate the ignition probability using fire history databases (Amatulli et al. 2007; Koutsias et al. 2004); and fire danger indexing, which is based upon the correlation between bushfires and weather conditions (Chuvieco & Salas 1996; Sharples et al. 2009). Most of these tools rely heavily on GIS techniques in order to visualise the spatial variation of the fire hazard. They also share a common objective of creating a better bushfire management system to minimize potential losses through prevention measures like fuel reduction burns that reduce the potential for fire ignition and propagation, reducing the level of exposure to fire, raising awareness of potential fire risk through community education, and increasing emergency response capacities (Vasilakos et al. 2007; Verde & Zêzere 2010).

4.3 Bushfire Hazard

Bushfires are a natural environmental phenomenon. They initiate as an ignition and propagate at various intensities, causing significant impacts to which the UBI can be subjected. Three things are needed to start a bushfire; availability of fuel, sufficient heat to cause and maintain ignition, and sufficient oxygen to sustain combustion (Granger & Hayne 2000; Hardy 2005). The intensity and destructive potential of a fire are determined by factors such as characteristics of the fuel, terrain and weather (Gillian & Derek 2011; Granger & Hayne 2000; Hardy 2005). These factors may change significantly over time. Therefore, in order to determine the level of bushfire risk, it is necessary to understand the external factors that determine the level of hazard.

The severity of and the loss from bushfires are mainly determined by three factors; changes in climate, fire management and land development (Murnane 2006). Changes in climate cause variations in fuel characteristics, soil moisture and fire weather conditions. Those factors control fire ignition and fire propagation. Fire management practices suppress bushfires by controlling fuel loads around UBIs, and increasing fire fighting resources and preparedness activities. Unplanned land development activities can cause expansion of the UBI, increase housing density and locate critical infrastructure facilities such as nursing homes at the UBI. Such initiatives lead to an increase in the level of bushfire risk at the UBI.

Bushfires cause damage through direct exposure, burning debris such as windblown sparks and embers, and radiant heat. Bushfire behaviour is determined by the point of ignition, weather conditions and intrinsic characteristics of the terrain. There can be one ignition or multiple ignitions. If there are multiple ignitions, fire severity may be determined by several fire fronts. Therefore, to understand this complex nature, it is important to have a bushfire hazard model that can be used to identify areas of bushfire hazard based on information that can be retrieved from past bushfire events.

Although extreme natural events have long been the primary research focus of the hazard research community, the recognition that hazards are not just physical events, but also include socially constructed situations, has broadened both the definition of hazard and geographers’ approaches towards understanding and ameliorating them (Cutter et al. 2000; Lavell et al. 2012). In this thesis, bushfire risk is considered to be derived from a combination of bushfire hazard and the vulnerabilities of exposed elements. The integrated risk framework used in this research suggests that the level of hazard and vulnerability drive the consequences of extreme bushfire events. Therefore, hazard is a key element of risk, and the hazard assessment is purely focused on the characteristics of the hazard, which help to identify where bushfire hazards occur, their severity and locational impacts, thus delineating hazard zones.

4.4 Bushfire Hazard vs. Bushfire Risk

For any given bushfire hazard assessment, the lack of clear definitions is an obstacle. In the area of bushfire research the terms “hazard” and “risk” are often ill defined, inconsistent and confused. A common understanding regarding concepts of bushfire hazard and risk does not exist (Bachmann & Allgower 2001; Verde & Zêzere 2010). If there is no common understanding of the terminology, the results of such assessments might be expressed in an erroneous way, which leads to inaccurate decision-making. In order to have successful bushfire management practices, a common terminology that is accepted by the research community is important.

Bachmann & Allgower (2001) have redefined the terminology that can be used in bushfire hazard and risk analysis. According to them bushfire hazard is “a process with undesirable outcomes” (p.28). It represents the probability of ignition and fire propagation. It does not include the expected impacts or outcome on the objects. Bushfire risk is mainly defined in terms of the outcome or the losses from the impact, which are mainly determined by the likelihood and consequences of the event (Atkinson et al. 2010; Atkinson et al. 2007; Bachmann & Allgower 2001; Shields & Tolhurst 2003). Fire risk has also been described as the chance that a fire might start at a specific location under specific circumstances, as affected by natural processes and the incidence of causative agents (Bachmann & Allgower 2001; Hardy 2005, p. 76). According to that definition, bushfire risk is the product of the probability of a bushfire and the expected bushfire damages. Applying this to bushfires, the likelihood component is the probability of a fire start (ignition) and spread (growth), and the consequence component is the impact of this fire starting, spreading and the interacting with the human system at the UBI. These interactions are influenced by more than simply the fire resistance of physical structures. Community level vulnerabilities and coping capacities also need to be considered as key components that influence the level of bushfire impact at the UBI.

This bushfire hazard assessment concentrates on fires that reach the UBI. Only the likelihood component of the risk equation is considered in this chapter. The probability of bushfire ignition and the severity of the bushfire event are used to understand

78 potential hazard levels at the UBI. The identified level of hazard is later integrated with the vulnerability components to identify the level of risk at the UBI (Chapter 6).

4.5 Bushfire Hazard Assessment

Natural hazards occur at various intensity levels. Hazard assessment requires understanding the probability of occurrence of a specific natural hazard at its various intensity levels and for its areas of impact (UNISDR 2004). The level of hazard intensity and the degree of vulnerability characterize the type and the magnitude of damage at any given location (Petak & Atkisson 1982). Hazard assessment requires information such as place of occurrence, intensity or magnitude of the hazard and the frequency of occurrence (Petak & Atkisson 1982; Tobin & Montz 1997). Various hazards are measured using different scales that are specific to the type of hazard. For example, flood hazard is usually measured using water depth, velocity and frequency. Bushfire hazard is measured by ignition, rate of fire spread, flame height and fire intensity (Pastor et al. 2003; Tolhurst et al. 2008).

Undertaking bushfire danger/threat assessments for efficient bushfire risk management was first introduced in the late 1960s (Shields & Tolhurst 2003). Initial methods were based on basic indices for fire danger, incorporating such features as ignition probability, fire size, resource damage rates and fuel loads. However, these pioneering methods were not spatial in nature (Hawkes et al. 1997). The McArthur Forest Fire Danger Meter and the McArthur Grassland Fire Danger Meter are some early examples of rating the fire danger level using weather and fuel variables (Granger & Hayne 2000). In the early 1990’s Geographical Information System (GIS) and spatial modelling techniques became more common and the importance of knowing spatial patterns of fire ignition and spread was widely recognized among fire managers. Models that determine the spatial extent of bushfire threat are used to make informed decisions about fuel management and fire prevention (Atkinson et al. 2010). Such bushfire hazard assessment models are known as “bushfire threat models” (Shields & Tolhurst 2003). Bushfire threat models are static in nature and do not show where bushfire can occur and in which direction fires will propagate, but rather they only show the areas that have higher potential for burning (Granger & Hayne 2000). These models are commonly

79 used among decision-makers because they are relatively simple, easy to use and do not rely heavily on many input variables. The results of such static models can be used to plan and implement specific risk reduction strategies.

However, these threat models do not include a bushfire history component and thus do not reflect the probability of occurrence. The output of these bushfire threat models is generated through aggregating the spatial layers and using ranking methods. The map overlay method does not capture the complexity of interactions between input variables. It also fails to answer some management-oriented questions such as those related to fire dynamics, potential loss, or resources required to manage a particular ‘threat’ output (Shields & Tolhurst 2003).

To overcome this issue, bushfire simulation models were introduced. These models are based on a function that combines several equations to calculate variables such as rate of fire spread, flame height, ignition probability and fuel consumption (Pastor et al. 2003). These quantitative models explain variability of bushfire hazard across spatial and/or temporal units of analysis (Mercer & Prestemon 2005). Such models increase knowledge of fire dynamics while producing a bushfire output from a set of inputs such as climate, weather, fuel, topography and fire management and suppression. Yet, these simulation models produce information about more than fire propagation. They also deliver fire intensity at each point, the presence of spot fires, etc. Fire propagation modelling depends on spatially and temporally varying factors such as bushfire fuels and behaviour characteristics to produce reliable fire spread predictions. Likelihood of fire occurrence at a specific location depends on ignition occurrence off site, fuels, topography, weather, and relative fire direction, allowing each fire to reach a location (Finney 2005). Although fire simulation models give the most nuanced assessments of fire hazard, the modelling process is often complex and relies heavily on the availability of numerous data inputs. Expert knowledge about physical processes of fires is required to run such models. The uncertainty associated with fire behaviour models also hinders their use for the decision-making process. Although fire behaviour models like FARSITE and BEHAVE are commonly used in decision-making processes in countries like the United States, in the Australian context, fire behaviour models (e.g., Phoenix)

80 are still at the research stage. Phoenix has been only applied to small areas and has not been widely applied in the decision-making process.

Bushfire ignition modelling is another common way of assessing bushfire hazard. It analyses bushfire ignitions using fire occurrence data. Fire occurrence can be defined as “the frequency of fires that have been reported and recorded within a finite area and historical period of time (e.g. number of fires/ha/year)” (Finney 2005, p. 98). Fire history databases with the geographic coordinates of the fires provide information on the spatial distribution of fire. They provide data to calculate the probability of fire occurrence in a given area. However, fire occurrence data only provides the ignition probability for a given location and hence does not reflect the probability of a fire at a given level of severity at a given geographic location. In order to measure the potential hazard it is therefore also important to estimate the severity of the given fire within the area of interest. Therefore, estimating ignition probability followed by fire severity is used in this study to assess bushfire hazard.

4.6 The Conceptual Framework for Bushfire Hazard Assessment

Figure 16 shows the conceptual framework that was developed to analyse bushfire hazard. It has two components; ignition probability and fire severity. Ignition probability is associated with the bushfire history and fire severity is associated with the proximity to previous fire events and the size of the previous fire events. This framework is based upon the likelihood component in the bushfire risk model developed by Shields & Tolhurst (2003). The proposed bushfire hazard assessment framework utilises spatial modelling techniques (also described in Chapter 6) that include an ignition model that shows ignition probabilities across the landscape and a distance model that describes the proximity of dwellings to past bushfire events across the landscape.

FireHazard

Probabilityof FireSeverity Occurence

Areaburnt Proximity

Figure 16: Conceptual framework for bushfire hazard assessment.

4.6.1 Ignition Probability

In this study, ignition probability is defined as the probability of fires occurring in an area. The history of fire events can be used to estimate the ignition probability for a given area. Ignitions may be related to the surrounding conditions (Mercer & Prestemon 2005). Therefore, events are likely to recur in the same location. Ignition points and their spatial relationships to variables can also be used to identify areas with high bushfire ignition risk (Chou et al. 1993). Geographic Information Systems are widely utilised as a tool to visualize the calculated ignition probability spatially across a landscape and over time. The output from the ignition probability analysis is presented in a map that represents the likelihood of being ignited by fire based on historic fire occurrence.

4.6.2 Fire Severity

When considering bushfire severity and the impact on an area, the distance between properties and the bushfire event is a critical factor (Chen 2005). It is important to predict the bushfire severity in a given area in order to initiate management plans to reduce the impact of potential events by identifying, quantifying and prioritizing properties at risk (Mercer & Prestemon 2005; Tolhurst et al. 2008). Fire intensity can also be determined based on the size of the fire, i.e. area burnt (Blanchi et al. 2002). This study utilises two key variables to determine the level of fire severity based on available data sources; the proximity of properties to past fire events and the area burnt.

4.7 GIS and Spatial Modelling in Risk Assessment

Integrated risk assessments often bring together several aspects of risk obtained from different approaches using various data sources (Fedra 1998). “Modelling natural disaster risk assessment is a complex task, involving a wide variety of processes which require large amounts of spatial and temporal thematic data and information coming from disparate sources. In this context, geography and Geographic Information Systems (GIS) can provide an ideal platform for the integration of the different data, their analysis and, ultimately, the development of disaster risk models for a region and its resident populations” (El Morjani et al. 2007, p. 9). Spatial models are used to develop our understanding of the effects and outcomes of unobservable processes that arise from the interaction between people and environment (Wainwright & Mulligan 2004).

A model is an abstraction of a real system that represents a complex reality in a simple way. It also provides a base for investigation of current processes and estimation of parameters as well as to evaluate whether these effects and outcomes are reproducible (Wainwright & Mulligan 2004). Due to the capability of examining and integrating spatial and thematic data retrieved from various sources quickly and effectively, spatial modelling has become one of the most powerful tools in environmental science and has been applied in areas such as land use change, natural resource management, environmental pollution and conservation, climate change and natural hazards. In the context of multidisciplinary research like natural hazards, spatial models provide the ability to integrate different results from different disciplines into a single product that summarises relationships and processes (Wainwright & Mulligan 2004). This approach is often not straightforward and accurate evaluation of processes and parameters is needed to obtain better results (Wainwright & Mulligan 2004). Spatial models amalgamate understanding-driven approaches with the application-driven approaches. Integrated natural hazard risk models have been developed and illustrated at the global (Dilley et al. 2005), national (Gunasekera 2009), and local levels (Abella & Van Westen 2007; Granger 2003) as well as at the community level with the help of community knowledge derived using a participatory GIS approach (Tran et al. 2009). In research on natural disaster risk, spatial models are widely applied in to hazards such as landslides, floods, earthquakes and forest fires.

Longley et al. (2010) have identified four different types of spatial models; static models and indicators; cartographic models and map algebra; individual and aggregate models; and cellular models. Static models do not vary over time and combine multiple inputs into a single output (Longley et al. 2010; Wainwright & Mulligan 2004). They are widely used in soil erosion modelling, ground water vulnerability modelling as well as natural hazard risk and vulnerability modelling (Abella & Van Westen 2007). Cartographic modelling and map algebra are based on mathematical combinations. Cartographic models have an ability to transform geographical characteristics into mathematical functions. A cartographic model is a set of interacting, ordered map operations that derive intermediate map data and manipulate them into a single output (Malczewski 2004). Map algebra provides a language to express a model that combines different output maps into one single map. In order use map algebra, model inputs and outputs need to be in raster form. Cartographic modelling and map algebra techniques are applied in studies such as urban growth simulations and land suitability modelling (Wainwright & Mulligan 2004). They are generic methods for organizing basic GIS operations into a complex spatial model (Malczewski 2004).

Individual and aggregate models and cellular models can be considered to be sophisticated geocomputational models (De Smith et al. 2009; O'Sullivan & Unwin 2010). These models are complex and expert knowledge and extensive research is required for their development. Individual models work at the individual level and are also known as agent-based models. Aggregated models are commonly called cellular automata models. Unlike static models, the results of these models are dynamic and vary over space and across time. In agent-based modelling an agent represents a human or other individual actors in a simulated real world (O'Sullivan & Unwin 2010). These models have been used to simulate many types of human and animal behaviour (Longley et al. 2010) as well as other physical processes such as bushfire propagation (Finney & Station 1998; Tolhurst et al. 2008) and land use change (Batty et al. 1999; White & Engelen 1993). In a cellular automata model, the earth is represented as a raster that has a number of possible states. The states change through time based on the execution of a set of rules that define the current state of a cell and its neighbours (Longley et al. 2010; O'Sullivan & Unwin 2010).

GIS is most successful when it is combined with mathematical modelling (Bonham- Carter 1994). Mathematical models are common, and states or the rate of change are often determined by the formally expressed mathematical rules (Wainwright & Mulligan 2004). Mathematical models can vary from simple equations to complex scenarios. Wainwright & Mulligan (2004) have classified mathematical models into three different categories; empirical models, conceptual models and physically based models. Empirical models are often based on the simplest mathematical functions that adequately fit the observed relationships between variables. These models do not require knowledge about physical laws or assumptions about the relationship between variables, but their results are often of great value as predictors or indicators. A conceptual model explains the observed behaviour based on preconceived notions of how the system incorporates parameter values that describe the observed relationship between the variables (Wainwright & Mulligan 2004). Conceptual models facilitate communication between experts in the decision process, which allows the provision of appropriate information to run the model. Conceptual models have a high level of explanatory depth compared to simple empirical models. Physically based models are derived from established physical principles and often produce results that are consistent with observations. These models explain relationships among variables as well as certain processes that transform inputs into outputs. Physically based models have good explanatory depth and greater degree of generality than conceptual and empirical models (Wainwright & Mulligan 2004).

GIS-based mathematical models are important tools for disaster management decision makers. As many mathematical models can forecast and provide real world details, such information can be used to develop risk reduction strategies. GIS-based analysis associated with mathematical modelling allows development of a common methodological approach for integrated risk assessment by merging elements of risk together (e.g., hazard and vulnerability). Integrated approaches are particularly useful for evaluating the spatial dependence of risks because they enable the combination of highly heterogeneous, spatially distributed variables such as population, critical infrastructure, etc. (Rigina & Baklanov 2002). In the context of natural hazards, risk- based decision-making is multidimensional and multidisciplinary in nature. It involves

85 socio-economic, environmental, physical and disaster management related factors, and decision tasks such as evaluation, prioritisation, and selection at different spatial and temporal scales (Chen et al. 2001).

The GIS-based Multicriteria Evaluation (MCE) approach is the most widely utilized technique to combine factors based on multiple decisions (Carver 1991; Chen et al. 2001; Malczewski 1999). It provides a powerful tool for spatial modelling, which allows the assessment of a particular region on the basis of multiple objectives and criteria, with supports for decision-making (Barredo & Bosque-Sendra 1998). The basic objective of MCE is to identify the best or most preferred alternative based on the conclusions derived from multiple criteria and conflicting objectives (Malczewski 1999; Yalcin & Akyurek 2004).

4.7.1 Spatial Multicriteria Evaluation Model

Spatial multicriteria evaluation is explained as “a process that transforms and combines geographic data and value judgements to obtain information for decision making” (Malczewski 2006, p. 703). It integrates the GIS’s capabilities of data acquisition, storage, retrieval, manipulation and analysis with multicriteria evaluation’s capabilities for aggregating spatial data and decision-makers’ preferences into unidimensional values for alternative decisions (Malczewski 1999). The terms multicriteria evaluation, multicriteria decision analysis, and multicriteria decision making are used interchangeably (Chen et al. 2001; Malczewski 1999). These different terminologies have been linked with GIS in various spatial decision support systems. Often a spatial multicriteria decision model contains; a set of geographically defined alternatives, a set of decision options which need to be considered by the decision maker, a set of evaluation criteria typically measured in different units; and a set of performance measures that are considered to be raw scores for each decision option against each criterion (Hajkowicz & Collins 2007; Malczewski 1999). Geographically defined alternatives are a collection of spatial objects and the results of the analysis mainly rely on the spatial distribution of those spatial objects (Malczewski 1999).

Although numerous approaches have been reported in the literature for the multicriteria evaluation process, six main steps are generally involved. Those are; Identify the problem and decision options Determine the evaluation criteria relevant to the decision-making problem (factors and constraints) Standardise the factors/criterion scores Determine the weight of criteria Aggregate the criteria (decision rules) Validate/verify the results

The overall objective of the MCE process involves integrating different components and requirements to develop a spatial decision support tool to address a specific problem with adequate consideration of decision-makers’ preferences. The specific objective serves as a guiding tool for structuring decision rules (Eastman et al. 1995). The decision is explained as the choice between alternatives that help to understand the problem, and alternatives represent different hypotheses, different causes of actions and different sets of features (Eastman et al. 1995). A criterion is considered to be the basis for a decision that measures the performance of decision options (Eastman et al. 1995; Hajkowicz & Collins 2007). Usually the evaluation criteria are determined based on the analysis of existing studies and judgements of experts who are knowledgeable about the specific problem (Hajkowicz & Collins 2007; Yalcin & Akyurek 2004).

In the standardisation process, criteria are transformed into a comparable, consistent numeric range (e.g. 0-1, 0-255, etc). This is important because criteria are measured using different scales and it is important to have an equivalent measure (Chen et al. 2001; Eastman et al. 1995). Widely applied standardisation techniques are; the maximum score procedure and the score range procedure, which are types of linear normalising techniques; value utility functions; and fuzzy set membership functions (Malczewski 1999). Fuzzy set theory allows more flexible MCE operations while continuity and uncertainty in the relationship between the criteria and the decision set is taken into consideration (Chen et al. 2001; Jiang & Eastman 2000).

Fuzzy set theory was designed to overcome the uncertainties of understanding linguistic or measured real-world, uncertain phenomena. Wood & Dragicevic (2007) have explained that these uncertainties could originate from non-statistical characteristics in nature - the absence of sharp boundaries in information. Fuzzy membership functions describe the main source of uncertainties involved in large-scale, complex decision- making processes. In the context of spatial multicriteria evaluation, given the continuous variation in many geographical phenomena, a fuzzy membership approach is appropriate for defining the boundaries between criteria because it retains complete information on partial memberships, giving due consideration to the uncertainty involved. This provides a more realistic standardisation approach compared to the other methods, especially when there is uncertainty inherent in the input data (Wood & Dragicevic 2007). Therefore numerous applications have utilized fuzzy set theory within the multicriteria evaluation process because of its capability for handling unquantifiable/qualitative criteria and obtaining quite reliable results (Wood & Dragicevic 2007).

In MCE, weighting is done based on the relative importance or preference for each criterion. Weights of individual criteria are often determined by a group of experts and decision-makers based on a subjective assessment (Chen et al. 2001). A variety of techniques are available to assign the weights. Widely applied weighting methods in MCE are; ranking, rating, trade-off analysis and pairwise comparison (Drobne & Lisec 2009; Malczewski 1999). Ranking is the simplest method of weighting, and involves analysis of decision criteria where each criterion is assigned a `rank' based on its perceived importance. The relative importance or weight can be calculated based on the ranks assigned to each criterion. In rating, weights are estimated explicitly for each criterion based on a predetermined scale. The scale is often between 0-1 or 0-100 (Drobne & Lisec 2009; Mendoza & Prabhu 2000). Trade-off analysis is based on direct trade-off assessments between pairs of alternatives (Drobne & Lisec 2009). Pairwise comparison provides a mathematical method for translating a pairwise comparison matrix of the criteria into a vector of relative weights for the criteria (Malczewski 1996). Amongst all of these methods, pairwise comparison has an added advantage of providing an organised structure for discussions, which helps to assign criterion weights during the decision-making process (Drobne & Lisec 2009). It also does not introduce 88 the uncertainty associated with the weighting process, which relies on the decision makers’ judgements (Chang et al. 2008). Therefore, pairwise comparison has become both popular and widely applied in multicriteria evaluation research. However, the selection of an appropriate weighting method is heavily dependent on the type of the decision rule utilised for the study.

The aggregation of criteria using a specific decision rule (aggregation procedure) is the most important step in the spatial multicriteria evaluation process. It allows decision- makers to make comparisons between the different alternatives on the basis of the values associated with the criteria (Chakhar & Martel 2003; Chakhar & Mousseau 2007). Different methods have been used to aggregate the criteria. The best type of aggregation to use is a context-specific choice because one method may be useful for some problems but not for others. For an example, aggregation method suitable for assessing land suitability may not be useful for assessing landslide risk. Therefore, the applicability of a given method depends on the way the problem is defined, on the data available and when these data are provided by the decision-maker (Chakhar & Martel 2003). In GIS this is defined as a spatial aggregation that combines different map layers (Chakhar & Mousseau 2007). Decision rules are regarded as the identities of the multicriteria methods (Chakhar & Mousseau 2007).

Widely accepted aggregation procedures include; Boolean Overlay, Weighted Linear Combination (WLC), Ordered Weighted Averaging (OWA), and Analytical Hierarchical Process (AHP). In Boolean Overlay, all criteria are assessed using suitability thresholds to produce Boolean maps, which are then combined using logical operators such as intersection (AND) and union (OR) (Jiang & Eastman 2000). However, Boolean overlay functions are extreme functions and often result in risk- taking or risk-averse solutions (Drobne & Lisec 2009).

WLC is a derivative of the Boolean overlay method but it softens the hard decisions of the Boolean overlay approach, avoiding the extremes (Drobne & Lisec 2009). It is a simple additive weighing method based on the values of standardised criteria and their assigned weights. The total score for each alternative is obtained by multiplying the weight by the standardised criterion value and summing the overall products. Decisions 89 are made based on the scores calculated for all of the alternatives (Drobne & Lisec 2009; Malczewski 2004). Boolean overlay and WLC are straightforward and are the most commonly employed aggregation techniques within MCE (Malczewski 2004; Malczewski 2006).

In WLC, weights assigned to each criterion also determine how they are traded off. Therefore, decision risk and uncertainty are associated with this method (Drobne & Lisec 2009). OWA has overcome this limitation by allowing decision-makers to determine the overall level of trade-offs (Drobne & Lisec 2009; Malczewski 2006). It provides an extension to conventional map aggregation methods such as the Boolean method and WLC (Malczewski 2004; Malczewski 2006). Unlike WLC, OWA involves two sets of weights: criterion importance weights and order weights. A criterion importance weight is assigned based on the layer’s relative importance. Order weights are assigned to a location’s attribute values in decreasing order without considering the criterion importance weights. The order weights are associated with the criterion values on a location-by-location basis. The OWA parameter is also associated with a trade-off measure indicating the degree of compensation between criteria (Drobne & Lisec 2009; Malczewski 2004). Although the above mentioned methods are widely used, they are also criticized because there is a risk of their incorrect use when they have been implemented without full understanding of the assumptions underlying these approaches (Thinh & Vogel 2007). Therefore, multicriteria evaluation should always be based on a strong theoretical framework.

Another widely accepted method in spatial MCE is AHP, introduced by Saaty (1980). AHP helps to minimise the risk of incorrectness discussed above by employing an underlying scale with values from 1 to 9. The scale uses pairwise comparison of alternatives/factors/indicators within a set of reciprocal matrices, and offers the advantage of a ratio scale, which can be effectively applied to quantify both quantitative and qualitative factors while treating the inconsistency in judging the relative importance of indicators (Thinh & Vogel 2007). The AHP process can be used in two distinctive ways within the GIS environment. It can be employed to derive the weights associated with map layers and the weights can be combined with the attribute map layers based on linear additive combination methods (Malczewski 2006). It also 90 involves structuring the problem from primary objectives to different levels of criteria and alternatives. The hierarchy is used to perform pairwise comparison of each element within each level of the hierarchy (Chang et al. 2008). The pairwise comparison allows group decision-making, which combines the experience and the knowledge of each member of the group (Chang et al. 2008).

Despite the popularity of AHP, it is also often criticized because it does not address the inherent uncertainty, vagueness and fuzziness associated with the process of transforming decision makers’ perceptions into real values. Furthermore the underlying assumption of precise input data are unrealistic and does not reflect typical styles of human thinking (Malczewski 2004; Vahidnia et al. 2008). Such issues can be addressed by introducing fuzzy set theory and fuzzy logic (Malczewski 2004). Therefore, some spatial multicriteria evaluation procedures have integrated fuzzy set theory with AHP to overcome the shortcomings of AHP (Chang et al. 2008).

“Fuzzy logic represents an extension of the classic binary logic, with the possibility of defining sets without clear boundaries or partial memberships of elements belonging to a given set” Zadeh (1965) in Malczewski (2004, p. 38). Fuzzy logic has been used to address problems such as vagueness, ambiguity and imprecision involved in defining constraints associated with the other conventional methods (Malczewski 2004). Jiang & Eastman (2000, p. 176) identified the reasons why the application of fuzzy set membership in criteria standardization is highly appealing. The reasons include: It provides a very strong logic for the process of standardization. Compared with linear scaling, standardization using fuzzy set membership represents a specific relation between the criterion and decision set. Fuzzy sets bridge a major gap between Boolean assessment and continuous scaling in weighted linear combination.

A fuzzy GIS approach consists of different fuzzy membership functions that are applied to data layers. This process is often called fuzzification. Fuzzy operations are then applied to fuzzified layers using either map algebra or a user-defined mathematical algorithm.

The above-mentioned multicriteria procedures have been incorporated in many GIS software packages such as ArcGIS and IDRISI. Some studies have developed their own MCE-GIS tools such as MCE-RISK (Chen et al. 2001) to address specific problems. However, IDRISI appears to be the most powerful currently available software tool in the context of MCE-GIS. It includes all the important tools such as AHP, fuzzy membership and OWA.

Multiple views about criteria and weights might exist in a multidisciplinary field like disaster management. “Different stakeholders in the design process can be anticipated to have different views about what is important, how that importance should be measured and how the various important factors should be combined” (Longley et al. 2010, p. 419). These decisions are based on experts’ own assessment of the importance and relevance of factors. Therefore, in this study, a GIS-based conceptual modelling approach that incorporates elements of risk and decision tasks that are characterised by multivariate factors is employed. This study utilises expert judgements of those who are knowledgeable about local conditions when evaluating decision criteria.

4.8 Materials and Methods

Methodological considerations for the bushfire hazard assessment are based on how best to explore the range of bushfire hazard mapping techniques with available models and data that are suitable to develop information on potential bushfire hazard levels at the UBI. The methodology was developed based on the conceptual framework shown in Figure 16. It consists of probability of occurrence and fire severity. Ignition probability was assessed using a kernel density estimation function to identify the probability of fire occurrence. Bushfire severity was assessed based on proximity to a fire event and the fire size (area burnt) using a GIS-based fuzzy MCE. Finally, the probability of occurrence and fire severity were combined using fuzzy overlay techniques to produce potential fire hazard maps for the study area. Considering the nature of the risk assessment, the distribution of census collection districts (CCDs) within the UBI, and socio-economic data availability, CCDs at the UBI were selected as the area of interest where interactions between residents and bushlands are present.

4.8.1 Ignition Probability

Traditional methods of assessing ignition probability are based on fire occurrence indices that capture relationships between causal factors. These indices are often calculated for an administrative unit (De la Riva et al. 2004). However, these methods do not explain the relationship between actual fires and their causes. This is because of a lack of appropriate fire history data sets in terms of their content and accuracy. Spatial distribution of fire occurrence and temporal sequences such as fire return intervals and seasons provide useful additional information for fire management, risk assessment and understanding the relationship between factors influencing fire regimes (Morgan et al. 2001). Probability of occurrence can be estimated using previous observations of fire occurrence. Therefore accessing accurate fire history data provides important information for any bushfire management activity (De la Riva et al. 2004).

Fire history databases often consist of fire occurrence records, which provide the location, area burnt, occurrence time and cause. Location is represented in spatial data as either point data or polygon data. Using data aggregation and spatial analysis methods, detailed information about bushfires can be generated. Temporal patterns of fire occurrence can be analysed using descriptive statistics. Furthermore, multivariate statistical techniques and geographically weighted averaging techniques can be used to understand the relationship between fire occurrence and causal factors (Koutsias et al. 2004). The goal of this fire ignition assessment is to model fire occurrence data spatially using available fire history data. Since the available fire history data does not include the cause of fires, this study does not attempt to identify any relationship between fire occurrence and causal factors.

Spatial interpolation is a common method that has been used to describe spatial fire occurrence patterns at the landscape scale (Amatulli et al. 2007; De la Riva et al. 2004; Koutsias et al. 2004). Interpolation techniques estimate values at unknown locations from observed values within the study area. They also convert point data to a continuous field (Burrough et al. 1998). Different interpolation techniques are widely used in many areas such as in estimating rainfall, temperature and other attributes where no direct measurements are available (Longley et al. 2010). Among them, inverse

93 distance weighting (IDW) and Ordinary Kriging are the most frequently used techniques (Li & Heap 2008). These models are based on the analysis of one variable as a function of the location of that variable. A value taken from sample locations is used to estimate values for all locations between sample sites (Amatulli et al. 2007; Levine 2002).

Nevertheless, interpolation may not be the most suitable method to calculate the spatial probability for individual point-level data. To calculate the probability of finite point locations such as point process variables, kernel density estimation is the best available method (Amatulli et al. 2007). It demonstrates the geographic variability of the density/probability by providing visual displays of ‘hot spot’ densities or probabilities while producing a smooth density or probability surface (Amatulli et al. 2007).

Kernel density estimation is a non-parametric statistical method that estimates probability densities. It places a symmetric probability density function (kernel) over each point observation, and results in a smooth surface function from the sums of all the kernels overlapping at the each intersection of a superimposed grid (Amatulli et al. 2007; De la Riva et al. 2004; Koutsias et al. 2004; Levine 2002). It has been extensively used in areas such as wildlife ecology and crime analysis. Kernel density estimation has also been used as a tool to identify high ignition risk areas within historical ignition patterns by producing a smooth, continuous surface that defines the level of ignition potential for that area (Amatulli et al. 2007; De la Riva et al. 2004; Koutsias et al. 2004). In addition, kernel density estimation represents a powerful way to conduct hot spot analysis and easily visualize trends over large areas (Levine, 2007). Its results allow users to identify hot spots quickly either by visually inspecting maps using expert knowledge or by defining hot spots based on statistical significance. It can be used in fire management to target areas of high concentrations of ignitions to develop scenarios in order to identify areas at higher fire risk for mitigation and prevention.

Kernel density estimation is sensitive to bandwidth size, which expresses the size of the kernel that controls the degree of smoothness of the density estimation (De la Riva et al. 2004). The problem of optimal bandwidth selection for kernel density estimation has been the subject of much research. Although some robust models are available, there is 94 no exact method for defining the size of the kernel. Therefore it is considered to be more of an art than a science (De Smith et al. 2009). Kernel density estimation mode varies depending on the method of selecting the bandwidth size (either a fixed value or multiple, adaptive, values) (De la Riva et al. 2004).

Based on the heterogeneous spatial distribution of the ignition point data, adaptive kernel density estimation was used for this study. The adaptive mode allows the adjustment of the bandwidth size in relation to the concentration of the ignition points. In this method bandwidth is adapted to be larger where the data are less dense and smaller where the data are denser. This approach also has the advantage of providing constant precision over the entire region for the estimated kernel density.

There are several kernel function types; normal distribution, triangular kernel, quadratic kernel, etc (De la Riva et al. 2004). Researchers have given less attention to understanding the effects of the type of kernel rather than bandwidth size because they believe that the kernel function is less important than bandwidth choice (Alegria et al. 2011; Danese et al. 2008). The normal distribution is the commonly used kernel function type (Levine 2002). In this study the normal distribution function was used for interpolation.

In order to get an idea about the size of the bandwidth to use in the kernel density estimation, a mean neighbourhood distance was calculated based on the inter-point distances of the ignition point pattern (Danese et al. 2008; Koutsias et al. 2004). Different smoothing parameters (bandwidth sizes) were tested (2, 5, 10 and 20 points) in order to identify the best one for the analysis. Among the different bandwidth sizes, the one with ten incidents was considered to be the most appropriate one based on visual examination, which shows more local variation. CrimeStat III (Levine 2002) spatial statistics programme was used for the analysis.

4.8.2 Fire Severity

In order to generate a fire severity surface, distance to past fire events and area burnt data were used. However, a limited amount of point data was available to perform the

95 task. Therefore, to reconstruct the underlying continuous layers of burnt area and distance from the available control point data, a point interpolation technique from ArcGIS 10.0’s Spatial Analyst tools was used. The objective of point interpolation is to produce a field of values to some satisfactory level of accuracy relative to the intended subsequent use of the data where the field itself has not actually been measured (Longley et al. 2010; O'Sullivan & Unwin 2010). The IDW method considers the distance between control points when predicting the unknown attribute values at certain locations. It is normally used when the set of control points is dense enough to capture the surface variation at the local extent (Childs 2004). Kriging is a statistical interpolation method used when assuming information about the distance and direction between control points, and it explains spatial variation in the surface (Childs 2004). To understand severity, an interpolated distance map of the proximity of dwellings to past bushfire events was created using the IDW method. Interpolated burnt area maps were developed using kriging. Both layers were then converted into fuzzy membership functions using linear functions to create standardised maps. Fuzzy maps of proximity and the area burnt were then combined using the Weighted Linear Combination (WLC) method to develop a bushfire severity map. Equal weights were assigned based on expert judgements derived from discussions with council members who are knowledgeable about local fires.

4.8.3 Exploratory Data Analysis

The temporal pattern of fires is another important variable at the landscape scale as the fire return interval determines the susceptibility of a given landscape. It is also an important factor that influences fire management strategies. Patterns of forest fires have often been analysed based on their frequency, burnt area and the fire season. In order to measure the temporal distribution of fires, descriptive statistical analysis was applied to fire history data. The statistical analysis includes fire seasonality, total number of fires, area burnt and fire frequency.

4.8.4 Recurrence Interval

The relationship between fire size and fire management is not clearly defined and is still an on-going debate (Ferreira Leite et al. 2010). Most larger fires are driven by the build

96 up of fuel (fuel load and an increased fraction of dead materials) and extreme weather conditions. Although fire suppression strategies would aim to minimise the impact of fires, heavy accumulation of fuel over a time period and weather driven factors may cause heavy impacts (Keeley & Fotheringham 2001). A return period is also known as a recurrence interval. It is an estimate of the interval of time between events of a certain intensity or size (Mays 2005). For bushfires, the recurrence interval can be calculated based on fire history data. It provides important information on extreme events that may occur in the future. A fire recurrence interval is a statistical measure of how often a hazard event of a given magnitude and intensity will occur. Extreme events have very low frequencies but very high magnitudes in terms of their destructive capacity. A bushfire event considered to be a twenty year fire event would cause more severe damage than a five year fire event (ADPC 2006).

4.9 Results

4.9.1 Fire History Database

The ignition points used in this analysis were extracted from the fire history database from the New South Wales Rural Fire Services (NSWFRS). Bushfire history data contains geographical areas as polygons and other information relevant to the fire (fire code, fire name, fire season, area burnt, etc.) for all fires that occurred in New South Wales from 1946 to 2006. However, complete information is available only for the most recent fire events. Therefore, in this study, data from 1980 to 2006 was used. As the database covers the entire state (New South Wales), fire history data was extracted only for the areas of interest. Bushfires that had some burnt area within three kilometres of each LGA boundary were selected using the Select by Location function in ArcGIS 10. However, some fields in the database are either incomplete or missing. The same fire has sometimes been recorded more than once, which causes data duplication. Therefore, the database was cleaned by removing duplications, records with inconsistent data, and records with missing data. After data cleaning, 136 records from the Ku-ring-gai area and 408 records from the Blue Mountains area remained in the database. The fire history database does not carry exact ignition point data and the polygons represent the burnt area of the fire. Therefore, fire perimeter polygons were converted to point data to extract the estimated ignition points for modelling by calculating centroids. This creates a new shapefile layer containing the points for the extracted centroid for each polygon 97 in the fire layer. For each ignition point, XY coordinates and area burnt were calculated (Figure 17 and Figure 18).

Figure 17: Spatial distribution of ignition points in Ku-ring-gai.

Two main regions can be identified in the map presented in Figure 17: bushland area and human settlements. The results in Figure 17 show that in Ku-ring-gai most events occurred either at the northern edge of the LGA where Ku-ring-gai Chase national park is located or near the southwestern edge of the LGA where Lane Cove national park is located. A few incidents have also been reported along the eastern edges of the LGA. The bushfire history data shows that the northwestern and northern (48 incidents) and southwestern and southern suburbs (62 incidents) can be considered to be high hazard

98 areas compared to other suburbs in the council area. In terms of area burnt, a larger area was burnt along the northern and northwestern edges (31km2), compared to the southwestern and southern edges (19km2). The map also reveals that most of the severe events occurred in the northeast. However, bushfires that occurred in the bushland along the northeastern side of the LGA did not encroach into the UBI because fire management authorities had enough time to suppress the fires before they reached the UBI. Structures in the southern and southwestern part of the LGA are much closer to the UBI and the nature of multiple ignition zone characteristics may escalate the difficulty of fire suppression.

Figure 18: Spatial distribution of ignition points in the Blue Mountains.

The results in Figure 18 show the ignition pattern in the Blue Mountains. It shows that most events occurred close to the human settlements in the Upper Mountains region. This may be due to high levels of human-environment interactions in that area. Ignitions are not as widespread in Lower Mountains region. Numerous ignitions can be found along the southern edge of the Lower Mountains region, but generally, the number of incidents in other areas is low compared to the Upper Mountains region. Most of the severe incidents reported occurred within bushland areas. However, large fires in the Upper Mountains have occurred close to Blackheath, Medlow Bath and Katoomba. One severe event can be found close to Springwood in the Lower Mountains region. This was the fire event that occurred in the Warrimoo, Valley Heights and Yellow Rock areas in 2001.

4.9.2 Exploratory Data Analysis

4.9.2.1 Fire History

10 NumberofIgnitions

0 1980 1982 1984 1986 1990 1992 1994 1996 2000 2002 2004 2006 Year

Figure 19: Number of ignitions in Ku-ring-gai (1980-2007).

In both areas the trend in the number of fires has increased over time (Figure 19 and Figure 20). In Ku-ring-gai, large numbers of fires were experienced in 1989, 1993 and 2000 (Figure 19). However, a decreasing tendency is evident after the 2001 fires. This may be because of the new policies on bushfire management, increased efforts of fire management authorities and initiatives such as fuel reduction burning and public education programmes. 100

45 40 35 30 25 20 15 NumberofIgnitions 10 5 0 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 Year

Figure 20: Number of ignitions in the Blue Mountains (1980-2006).

In the Blue Mountains the average number of ignitions per year is higher than in Ku- ring-gai (Figure 20). Over time, the number of ignitions has increased. Significant fire years were 1993, 1996, 2001 and 2002. Control of ignitions is a difficult task in the Blue Mountains. Unlike Ku-ring-gai, the number of ignitions has not been controlled over the years. This may be because of the length of the UBI and the higher level of human activities in the UBI.

120

100 ) 2

40 AnnualAreaburnt(km 20

0 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 Year

Figure 21: Annual area burnt in Ku-ring-gai (1986-2007).

101

The size of fires in terms of the area burnt has varied over the years (Figure 21 and Figure 22). In some years, the size of fires was very small compared to the maximum in the figure. In 1993, a large area was burnt by the historic Sydney bushfires. The results suggest that significant bushfires are likely to occur at certain intervals, such as every five to six years in Ku-ring-gai, driven by the accumulation of fuel and extreme weather conditions. Although the number of ignitions is high in the Blue Mountains, these incidents have not caused severe impacts when measured by the area burnt. The results show that in the Blue Mountains, significant fire events have occurred in four years since the 1993 fires. The most significant events occurred in 2001/2002, and the most recent event was in 2006.

1000 900 800 ) 2 700 600 500 400 300

AnnualAreaBurnt(km 200 100 0 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 Year

Figure 22: Annual area burnt in the Blue Mountains (1986-2006).

4.9.2.2 Recurrence Interval

In this research, the recurrence interval of bushfires in Ku-ring-gai and the Blue Mountains was calculated using fire history data from 1980 to 2006. The recurrence interval is the average time between fires that equal or exceed a given magnitude (Figure 23 and Figure 24). To calculate the fire recurrence intervals, burnt area was used as the measure of magnitude of the fires. Recurrence Interval (T) = (n + 1) / m Equation 6 (Mays 2005)

102 where, n = number of events in the record; m = rank of the event based on the size of the event (starting at 1 for the largest).

120

100 ) 2 80

40 AnnualAreaBurnt(km 20

0 0 5 10 15 20 25 30 RecurrenceInterval(Years)

Figure 23: Recurrence interval and the annual burnt area in Ku-ring-gai.

1000 900

) 800 2 700 600 500 400 300

AnnualAreaBurnt(km 200 100 0 0 5 10 15 20 25 30 RecurrenceInterval(Years)

Figure 24: Recurrence interval and the annual burnt area in the Blue Mountains.

Figure 23 and Figure 24 show the cumulative frequency curves for the observed bushfire data in the study areas. They show that in Ku-ring-gai a significant event that would burn more than 10km2 can occur about one in 10 to 15 years. In the Blue Mountains a significant event that could burn more than 150km2 might occur 103 approximately one in five years. The results also suggest that in the Blue Mountains the area burnt is much larger than in Ku-ring-gai. This is due to the large bushland extent in the area.

4.9.3 Bushfire Hazard in Ku-ring-gai

A bushfire hazard map for the study area was developed using the ignition probability and bushfire severity maps. The resulting map was then classified into five classes, which resulted in a hazard zonation map for the study area. This analysis was only performed for census collection districts (CCDs) that fall within three kilometres of the UBI because the focus of this research is on those locations. Other areas were excluded from the analysis.

Figure 25 shows the prepared input data layers and the intermediate results of the bushfire hazard assessment for Ku-ring-gai. These input maps show the spatial variability of factors influencing the bushfire hazard. For an example, it is evident that the ignition probability is high at the southwestern edge of the LGA whereas the burnt area is higher at the northern and western edges. The intermediate fire severity map is a combination of both the burnt area and distance maps.

104

Ignition Probability Distance to Fire Events

Area Burnt Fire Severity

Figure 25: Input maps for the bushfire hazard map (Ku-ring-gai).

The final bushfire hazard map for Ku-ring-gai is presented in Figure 26. It represents the relative level of bushfire hazard across the region based on spatial variation in ignition probability and fire severity. However, the resulting map itself cannot be used on its own for decision-making because it is hard to demarcate actual hazard zones. Therefore, a bushfire hazard zone map was derived using this map. 105

Figure 26: Bushfire hazard map for the Ku-ring-gai area.

The fire hazard map was classified into five hazard categories (Minimal, Low, Moderate, High and Extreme) using the quantile method to produce the hazard zonation map for the study area (Figure 27). The quantile method was selected for classification because it separates the hazard ratings into five intervals with an equal number of cells (area). This helps to compare the relative levels of hazard across the region. The classified map shows that the southwestern edge from Fox Valley to Lindfield and the northern edge from North Wahroonga to North Turramurra are extremely prone to bushfire. Relatively, the eastern and southeastern edges of the area are prone to moderate to minimal levels of bushfire hazard.

106

Figure 27: Hazard zonation map for the Ku-ring-gai area.

4.9.4 Bushfire Hazard in the Blue Mountains

Figure 28 shows the prepared input data layers and the intermediate results of the bushfire hazard assessment in the Blue Mountains. It is evident that the ignition probability is high in the Upper Mountains and at the southern edge of the Lower Mountains, whereas the burnt area is high across the area except at the southern edges of the Upper Mountains area. The intermediate fire severity map shows a high level of fire severity across most of the Lower Mountains region. The main reason for that is that much of the area consists of bushland rather than human settlements. In the Upper Mountains, the results suggest that the northern edge is more prone to bushfires than the southern edge. 107

Ignition Probability

Distance to Fire Events

Area Burnt

Figure 28: Input maps for the bushfire hazard map (Blue Mountains).

108

Fire Severity

Figure 28 continued.

Figure 29: Bushfire hazard map of the Blue Mountains area.

The bushfire hazard map is presented in

Figure 29. It represents the relative level of bushfire hazard across the UBI in the Blue Mountains based on spatial variation in ignition probability and fire severity. The hazard map was classified into five hazard categories (Minimal, Low, Moderate, High and Extreme) using the quantile method to produce a hazard zonation map for the study

109 area (Figure 30). The hazard zonation map shows that the northern edges of the Upper Mountains are extremely prone to bushfires. It also reveals that the area between Blackheath and Katoomba is more prone to bushfires in the Upper Mountains than other regions. In the Lower Mountains, the eastern and southern edges are extremely prone to bushfires.

Figure 30: Bushfire hazard zonation map of the Blue Mountains area.

4.10 Discussion

Bushfire hazard assessment at the LGA level is often hampered by various issues such as lack of capacity to run complicated fire dynamics models, limited data availability, and the absence of hazard assessment initiatives. Therefore, developing a bushfire hazard map is a challenge for many LGAs. Most LGAs lack precise information on bushfire hazard. The model presented here illustrates a simple approach to estimating bushfire hazard at the local level from data that is commonly available. Moreover, it is an improvement on the current method for delineating hazard risk zones used in most local councils: a simple buffer around the UBI boundary. It utilised a simple hazard assessment model that can be applied at the local council level using existing data, capabilities and resources. It integrates different criteria into a single hazard map that is easy to understand and easy to replicate in times of need. Bushfire hazard maps were developed using the characteristics of the past bushfire events in the study area. They 110 provide an insight into the bushfire hazard levels at the UBI in both areas and show areas that are prone to bushfires to different degrees. However, it is also possible to incorporate a range of other information such as cause of ignition and bushfire loss data into the model if it is available.

Despite being relatively simple and useful, the hazard assessment has its limitations. This model can be further modified by introducing additional variables (e.g., those that are not available for all LGAs such as fuel loads, temperature, wind and other weather factors), using different interpolation techniques and field-based, in-depth assessments. The assessment technique used here strongly depends on the point information. The input map layers are estimates derived from those point information. However other environmental properties and their relationships on the ground may influence fire ignition and severity. These conditions have not been taken into account in this model. Therefore, in reality the estimates may be more accurate in some locations than in others. Moreover, for this study, point of ignition was extracted using the burnt area polygon available in the bushfire history database. The extracted centroid may not represent the actual ignition point location. This may lead to underestimating/overestimating the actual scenario. However, this limitation can be overcome in the long run if the results can be compared with aerial images of actual fire events.

4.11 Conclusions

Hazard maps visualize the spatial distribution of potential bushfire ignition and severity. These maps frequently provide motivation for bushfire risk management decision- making to prioritise certain geographic locations, which would be difficult to obtain without visual interpretation. It will also be advantageous for many local level bushfire management and activities, ranging from educational programs to illustrate local hazards, to land use planning to reduce hazard impacts and to combine with other elements of risk to illustrate the level of risk.

In this method, only the ignition points and the area burnt were used to determine the level of bushfire severity. Other spatial data such as actual impact or loss data from

111 various sources can be added to improve the model. However the lack of availability of a complete database on bushfire impacts is a challenge for assessing bushfire hazard.

This model helps to estimate the level of hazard based on previous fire events. To understand fire severity only area and proximity measures were used. The findings can be further improved by coupling this model with existing fire behaviour models. This would result in the actual intensity of fires being based on parameters such as vegetation, weather and terrain features. These tools together would enable a more comprehensive approach to spatially explicit planning and development to minimise the level of hazard. However, fire behaviour models need more data and expert knowledge. Therefore, such models are often implemented at the state rather than local level.

4.12 Summary

This chapter discussed the bushfire hazard assessment and presented a bushfire hazard assessment model that can be applied at the local level using readily available resources and data. Bushfire hazard assessment was conducted for two LGAs using the bushfire hazard assessment model and the results of this assessment were discussed. The resulting hazard maps show spatial variation in bushfire hazard at the UBI. Such information is useful in bushfire risk management at the local level. Hazard assessment is one component of the bushfire risk assessment framework developed for this study. As discussed in Chapter 2, risk is considered to be a relationship between the hazard and vulnerable communities. Therefore, to understand the risk, vulnerability also needs to be understood. The risk assessment framework identifies three components of vulnerability; exposure and physical susceptibility, social vulnerability, and community response and coping capacity. The next chapter examines the social vulnerability of the population living at the UBI. It tries to develop social vulnerability profiles of the population in order to identify the most vulnerable social groups at the UBI.

112

Chapter Five: Social Vulnerability in the Context of Bushfire Risk at the Urban Bush Interface in Sydney

(A modified version of this chapter was accepted by Natural Hazards: Journal of the International Society for the Prevention and Mitigation of Natural Hazards, DOI: 10.1007/s11069-012-0334-y)

5.1 Overview

The concept of social vulnerability is widely used to understand social economic fragilities of the people who are vulnerable to natural disasters. Social vulnerability refers to population characteristics that influence the capacity of a community to prepare for, respond to and recover from disasters. Factors that contribute to social vulnerability are often hidden and difficult to capture. This chapter discuses the social vulnerability of people living at the UBI, mainly in the context of bushfire risk. Identification of vulnerable groups is an important first step in any preparedness and emergency management planning process. To understand the social vulnerability of people living at the UBI, a social vulnerability index was developed using Australian Bureau of Statistics (ABS) 2006 census data at the Census Collection District (CCD) level. The variables contributing to develop social vulnerability index and the components of social vulnerability were identified. Finally, the social vulnerability index was mapped at the CCD level. The results show different spatial patterns across the study areas, which provides useful information for identifying communities that are most likely to experience negative disaster impacts due to their socio-demographic characteristics. The resulting social vulnerability index map is used as an input for the overall vulnerability map. Therefore social vulnerability contributes to overall vulnerability as well as risk.

5.2 Introduction

Vulnerability arises from the relationship between human actions and hazard impacts, and has been discussed extensively since the 1970s (O'Keefe et al. 1976). The scientific community has put much effort to increasing our understanding of the concept of vulnerability and developing tools to assess the level of vulnerability within a population. Several conceptual frameworks have been developed to explain vulnerability to natural hazards. Widely recognized models include the Risk-Hazard

113

(RH) model (Davidson 1997), the Pressure and Release (PAR) and Access models (Blaikie et al. 1994; Wisner et al. 2004), the Comprehensive Vulnerability Analysis (CVA) model (Turner et al. 2003), the Vulnerability of Place model (Cutter 1996; Cutter et al. 2000), and several other integrated models (Dolan & Walker 2006; Füssel 2007; UNISDR 2004).

When translating different conceptual frameworks into practical vulnerability analysis tools, researchers have considered either individual dimensions of vulnerability such as physical (Douglas 2007), biophysical (Cutter et al. 2000) and socio-economic vulnerability (Clark et al. 1998; Cutter et al. 2003), or have taken a more holistic and integrated approach that addresses multiple dimensions of vulnerability (Dolan & Walker 2006; Preston et al. 2009). Socio-economic dimensions of vulnerability are a key element of many vulnerability frameworks (Füssel 2007; Hewitt 1997; UNISDR 2004; Wisner et al. 2004). The linkages between unsafe conditions and marginal socioeconomic conditions have been widely investigated in disaster management under the rubric of ‘Social Vulnerability’ (Clark et al. 1998; Cutter et al. 2003; Myers et al. 2008; Vincent 2004; Wei et al. 2004; Wood et al. 2010). In this research, I focus on social dimensions of vulnerability to bushfires at the UBI.

The literature on social vulnerability and bushfires is not well developed compared to that of other hazards. In this research I reviewed the literature on social vulnerability to other hazards to inform this work on bushfires. While several indicators of social vulnerability have been developed in Australia (Baum et al. 2008; Brunckhorst et al. 2011; Pink 2006), none of them have been applied in the context of bushfires. The lack of work on social vulnerability and bushfires was addressed by the Victorian Bushfires Royal Commission (2009), which highlighted the importance for effective emergency response of having demographic information on vulnerable populations. Such information can guide emergency managers to recognise the specific needs of vulnerable people such as those who might need evacuation assistance or separate consideration, particularly on code red (high fire risk) days (Victorian Bushfires Royal Commission 2009).

114

Constructing a social vulnerability index is a common approach to assessing social vulnerability to hazards. Such an index helps to identify where socially vulnerable groups are located (Fekete 2009). Vulnerability indices can help planners to prioritise locations of populations with high levels of social vulnerability when they make emergency and risk management decisions (Chakraborty et al. 2005), as well as to see variations in social vulnerability over time and across space (Cutter & Finch 2008).

5.3 Social Vulnerability

Social vulnerability in particular refers to demographic and socioeconomic characteristics of a population that influence its capacity to prepare for, respond to and recover from hazards and disasters (Cutter et al. 2009). This difference in capacity may be one reason why communities with similar levels of exposure may experience different impacts from a particular hazard event (Finch et al. 2010; Cutter et al. 2009). Over the last decade, the importance of social vulnerability has been increasingly recognized and it is now considered to be an important part of understanding disaster risk. It can be considered a by-product of social inequalities, which increase an individual or population’s level of susceptibility to a particular hazard by limiting access to opportunities to prepare, respond and recover (Cutter et al. 2003; Yeletaysi et al. 2009). Some people often struggle to cope with their everyday life due to social conditions such as low income, occupation, etc. For such populations, disaster preparedness is not the priority.

Social vulnerability can affect all phases of the disaster cycle. However, its effects are most visible in the response and recovery phases. For example, in Australia, people in non-English speaking households may be slower to receive and respond to early warning and evacuation orders (Goudie 2008). During the recovery phase, people who can afford to insure their properties and who have better savings will recover faster than people who are financially deprived. Socially vulnerable groups are often the least prepared for a hazard as their ability to prepare is hampered by their limited financial resources and/or physically constrained access to preparedness, response and recovery supplies (Tapsell et al. 2010). For example, low income groups are less likely to have insurance, less likely to receive and believe disaster management messages and less

115 likely to act on orders that will help them mitigate their experience of a disaster (Zahran et al. 2008).

5.3.1 Social Vulnerability Assessment

Measuring vulnerability is an essential pre-requisite to initiate efficient disaster reduction measures (Birkmann 2006a). It provides a more analytically rigorous and data driven approach to risk management decision making and also allows for comparison between geographic locations (Cardona 2005). This has been widely practiced at the global level (Dilley et al. 2005; UNU-EHS 2011), national level (Davidson & Lambert 2001) as well as local level (Davidson 1997; Granger 2003). Indices and indicators of disaster risk and vulnerability at national and sub-national scales are the most commonly used tools for identifying and measuring vulnerability to hazards (Birkmann 2006b). One common practice is constructing a Social Vulnerability Index (SoVI), which is an indicator-based approach that relies on socioeconomic and demographic data to model social vulnerability. SoVI methods reduce a large number of variables into a few statistically significant dimensions that describe different aspects of social vulnerability. Then a model (i.e., an index) is developed to quantify social vulnerability based on these dimensions (Adger et al. 2004; Schmidtlein et al. 2008).

In general terms, an index is defined as a composite representation of several numerical measurements within one single value (Simpson & Katirai 2006). The goal in building an index is to reduce the complexity of information without diminishing the information’s capacity to shed light on the problem (Bowen & Riley (2003) in Adger et al (2004, p. 65). Use of indices has gained wide popularity because they are easier for the public to interpret than identifying common trends across many individual variables (Saltelli 2007). They are also useful for benchmarking a given condition, and for identifying trends across space and through time (Nardo et al. 2005).

However, the effective use of indices presents some challenges. Composite indices may send misleading policy messages if they are poorly constructed or misinterpreted (Nardo et al. 2005; Saisana et al. 2005; Saltelli 2007). Therefore, when constructing an index, the selection of indicators, choice of model, the weighting of indicators and

116 treatment of missing values need to be addressed carefully. Choosing indicators requires extensive review of prior research, and the construction process should be transparent and based upon robust statistical and conceptual principles (Nardo et al. 2005; Saisana et al. 2005; Saltelli 2007). With respect to interpretation, index values only indicate relative differences. Therefore, no significance should be attached to the absolute values of the index, and index values can only be used to compare different groups, entities or geographic areas in order to perform ranking. Ranking is useful when targeting and in identifying priorities (Eakin & Luers 2006). Despite these challenges, vulnerability indices are widely applied in the context of disaster research (Clark et al. 1998; Cutter et al. 2003; Holand et al. 2011; Wood et al. 2010).

The concept of a social vulnerability index was introduced by Cutter et al. (1997), who presented a simple method of developing a single score to measure social vulnerability at the US census block level using identified social vulnerability indicators. In 2000, Cutter et al. operationalised this approach as a component of their ‘hazard of place’ model. In their method, eight indicators that describe characteristics such as population size and structure, differential access to resources and physical weaknesses, wealth or poverty, and level of physical or structural vulnerability were selected to quantify social vulnerability based on prior knowledge of socioeconomic and demographic characteristics and their relationships with hazard impact. Standardized values of those indicators were summed to develop an aggregate value of social vulnerability. Since then their method has been used as an illustrative tool to capture the societal conditions of populations living within a given geographic unit (Chakraborty et al. 2005; Cutter et al. 2000; Wu et al. 2002).

Clark et al. (1998) employed a more complex method of analysing social vulnerability based on factor analysis. Vulnerability factors (poverty, transience, disabilities, immigrants, young families) were identified based on census data. In 2003, Cutter et al. further developed the social vulnerability index approach using principle component analysis. In this study they identified components that describe different dimensions of social vulnerability. This method has been widely accepted and utilized in many studies (Fekete 2009; Mendes 2009; Rygel et al. 2006; Wood et al. 2010).

117

Despite being widely used, there is not yet agreement about how SoVI components should be aggregated to yield a composite index. The most common method of aggregation is a non-weighted additive model in which all components contribute equally to the final index (Cutter et al. 2003; Fekete 2009; Mendes 2009; Myers et al. 2008). Other researchers have used a weighting scheme that is based on the percent of variance explained by each component (Cox et al. 2007; Wood et al. 2010). Schmidtlein et al. (2008) used both non-weighted additive and weighted sum methods to analyse the sensitivity of the SoVI. They found substantially different results using the weighted sums and equal weights approaches. A third strategy assigns weights based on other criteria. For example, Rygel et al. (2006) used a Pareto ranking of factors, while Haki et al. (2004) assigned weights based on a pairwise comparison technique carried out by experts from different disciplines such as employment, housing, education and household demography. Because of the different results that these different methods produce, it is important to try to verify and validate the results of a SoVI analysis, insofar as possible.

Fekete (2009) validated the SoVI approach using an independent second dataset describing the socio-demographic characteristics of a population exposed to river floods in Germany. He first created a census-derived SoVI, that predicted who would be socially vulnerable to river flooding. He then used a survey that identified individuals who were vulnerable to a real flood event to test the predictive power of his model. To do this, he used a logistic regression of the independent variables in his vulnerability index to predict the probability of whether an individual with particular socio- demographic characteristics would be forced to leave their home or seek emergency shelter. The variables that were significant in this analysis were retained and he then repeated the factor analysis that generated his original SoVI with these retained variables and compared the two indices. His results revealed that the regression- captured variables yielded the same factors that his original SoVI index contained. He concluded that these results demonstrated the validity of the original index. Fekete’s approach, however, is difficult to replicate in the absence of a hazard event and data on who was actually vulnerable during that event.

118

Vulnerability is a product of processes occurring at different spatiotemporal scales. Measuring vulnerability at different scales helps to understand cross scalar dynamics in vulnerability, from global to local or vice versa (Chen et al. 2003). Selecting an appropriate scale is crucial and often challenging as the relevance of a specific variable may change with scale and degree of data aggregation (Eakin & Luers 2006; Fekete et al. 2010; Turner et al. 2003). The scale of analysis plays an important role in identifying vulnerability of people and places as the spatial distribution of vulnerability varies with the scale. While sub-national scales provide insight into regional processes and patterns, finer scales are beneficial for better understanding and capturing the root causes of vulnerabilities (Fekete et al. 2010). In the context of disaster management, vulnerability assessments are often conducted at scales that coincide with the geographic levels at which disaster management decisions are taken (Cutter et al. 2008; Eakin & Luers 2006). The selection of appropriate scale is often influenced by data availability and quality, construction and presentation of outputs, and the extent of the study area.

5.3.2 Social Vulnerability Indicators

Assessments of social vulnerability have frequently utilized different indicators and proxies that describe the socioeconomic and demographic characteristics of individuals and communities and that influence the level of social vulnerability in a particular geographic location. Indicators are selected based on either theoretical understanding of their relationships to vulnerability or based on statistical relationships (Adger et al. 2004). Such indicators can be selected from the wealth of information about the population normally readily available through a national census. Selection of indicators is problematic due to complex interactions between variables. Therefore social vulnerability indices include varying degrees of detail depending upon the context and the information, data and resources available.

Widely accepted characteristics that influence social vulnerability include income, gender, race/ethnicity, age and level of education (Cutter et al. 2003; Cutter et al. 2000; Rygel et al. 2006). In some studies characteristics of the built environment such as building density and building age have also been incorporated (Holand et al. 2009; Mendes 2009). Studies that have used these variables have defined social vulnerability as a function of socioeconomic, demographic and built environment characteristics. The 119 aim of this chapter is to capture the socio-demographic dimensions of the people who are living at the UBI. Therefore I only focus on the socio-demographic attributes of communities and exclude built environment variables.

5.4 Methodology

This study relies on the method developed by Cutter et al. (2003). To describe social vulnerability at the UBI, I developed a composite index based on 2006 Australian Bureau of Statistics (ABS) Census data; the most recent data available. This index provides a relative measure of socioeconomic and demographic conditions at the CCD level. It also provides profiles of who lives where to further help to understand the types of residents living in the UBI and their social vulnerability.

I initially selected 51 variables from the 2006 Australian Census data to capture indicators of socioeconomic status, gender, age and disabilities, family structure, housing, education, migrants and cultural values. The selection of variables was guided by those previously reported in SoVI methods in the literature (Cutter et al. 2003; Cutter et al. 2000; Fekete 2009; Flanagan et al. 2011; Rygel et al. 2006; Wu et al. 2002). The spatial scale of this study is the CCD level, because that is the smallest unit available with complete data. Often community level bushfire management initiatives are conducted at the street level and the CCD scale is best suited for such initiatives. I obtained data for the 239 CCDs that cover the UBIs in the Ku-ring-gai and Blue Mountains LGAs. Several CCDs contained no population. These CCDs are sites that have been recently developed and that contained no residents in 2006. Therefore, they were removed from the analysis and labelled as containing “no data”.

This SoVI employed Principle Component Analysis (PCA). PCA is a multivariate statistical technique that is frequently used as a data reduction method. It decomposes an original set of variables into a smaller number of linear variates (Field 2009). PCA works well when the distribution of variables varies across cases, or in this instance CCDs. The result of PCA is a set of uncorrelated components that represent a linear weighted combination of the initial set of variables (Vyas & Kumaranayake 2006). I then used the resulting components to create a social vulnerability index.

120

The original variables were tested for multicollinearity as PCA does not produce robust results when input variables are perfectly correlated or completely uncorrelated (Field 2009), retaining 20 variables from Ku-ring-gai and 26 variables from the Blue Mountains for the analysis Table 2. Density, per capita or percent functions were created for all remaining variables. This enabled comparison of CCD units of different sizes. These transformed variables were then standardized so that variables could be compared with each other. Because the PCA was performed separately for each LGA, the results can be compared within one LGA but not between LGAs. I deliberately made this choice to investigate whether the same variables would be important for describing social vulnerability in different parts of the Sydney basin.

121

Table 2: Social vulnerability indicators and variables used in the PCA

Indicator Variables used Percentage of persons over 65 Age Percentage of children under 52 Median age2 Gender Percentage of females Percentage of one parent families with a child under 152 Family structure Average household size Percentage of population living alone Median household income per week Median individual income per week2 Percentage of people with negative income2 Socioeconomic status Percentage of people with $1-$249 income per week Median housing loan repayment per month2 Median rent per week Percentage of population who speak a language other than English at home Percentage of migrants Race and Ethnicity Percentage of non-Australian citizens2 Percentage of people who do not speak English well or at all1 Percentage employed in service occupations2 Occupation Percentage employed as managers and professionals Percentage of people over 20 years who have education Education only up to 10 year or below Special needs Percentage of people needing special assistance Percentage of dwellings with no vehicle Household resources Percentage of housing units with no internet connection Percentage of rented housing units Renters Percentage of cooperative, community, or church housing units1 Percentage of volunteers2 Volunteer work Percentage of people who provided unpaid assistance2 Percentage of visitors Visitors and recent migrants Percentage of population who migrated < 1 year ago Percentage of people who did unpaid domestic work less Domestic work than 5hrs/wk 1= Variables retained only for Ku-ring-gai, 2= Variables retained only for the Blue Mountains

122

The PCA was performed using SPSS (SPSS 18.0) using varimax rotation to find the best possible fit to the dataset in order to calculate the highest variance for each component. I applied the Kaiser criterion (eigenvalues > 1) for component selection. I used the stepwise exclusion approach and repeated this process until the variables and components of the model were stable and statistically robust. To check the robustness of the model, I used the Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy and Bartlett’s test of sphericity and communalities. Finally, the sign of the components was adjusted according to their directional influence based on literature reports (Cutter et al. 2003). Components that increased vulnerability were considered to be positive and components that decreased vulnerability were considered to be negative (Wood et al. 2010). There were no variables that loaded both positively and negatively on a component, so I did not apply any absolute value transformations as did Cutter et al (2003).

Finally, I developed a composite index by adding weighted component scores. Each component score was weighted based on the percent of variance explained by the component divided by the total variance explained by all extracted components, in order to produce weights that ranged in value from 0 to 1. I classified the final SoVI using the standard deviation from the mean method, which provides a relative measure of deviation from the mean for each CCD (Wood et al. 2010). Finally, I mapped the results at the CCD level to illustrate their spatial variability. In order to compare the level of social vulnerability in the area, CCDs with a SoVI higher and lower than 1.5 positive or negative standard deviations from the mean were classified as high and low, respectively. Individual component scores were mapped using the quantile classification method to facilitate comparisons between variables. Individual component score maps illustrate spatial variability of each component score among CCDs. They also highlight the dimension(s) of social vulnerability that are most important in particular locations.

Vulnerability models can be validated by comparing their predictions with an independent dataset that includes measures of specific post-event outcomes (Fekete 2009). Such data sets are not often available and therefore validation of SoVIs is difficult. Hence, few authors have validated their SoVI models (Clark et al. 1998; Cutter et al. 2003; Mendes 2009; Wood et al. 2010). No household level post-event data

123 set was available for these study areas. Therefore validating the result empirically was not possible. Nevertheless, I used expert interviews with council employees, community fire volunteers and community members and observations on field trips to check that the SoVIs’ results were consistent with local knowledge about social vulnerability. This verification process led me to conclude that the SoVI components broadly captured the level of social vulnerability in the area.

5.5 Results

The PCA generated six components for both the Ku-ring-gai and Blue Mountains LGAs. In both locations, the analyses explained more than 70% of the total cumulative variance in social vulnerability. One component was common across areas: Immigrants, which explains 12% of the variance in Ku-ring-gai and 10% in the Blue Mountains (Table 3).

Table 3: Social Vulnerability Index comparison

Ku-ring-gai Blue Mountains Number of components 6 6 Number of variables 21 26 extracted Percent of variance 76.44 73.04 explained Index range -0.792 - 1.561 -0.97 - 1.29 Standard deviation 0.453 0.474 Special Needs, Elderly and Tenancy and Economy (27.57) Gender (25.79) Tenancy and Housing Age and Fragile (12.50) (12.54) Component Immigrants (12.28) Immigrants (10.33) interpretation (percent Low Education and Volunteers of variance explained) Wealth (10.03) (9.41) Occupation (7.92) Gender and Domestic Work (7.63) Visitors and Low Occupation and Personal Wealth Education (7.87) (5.58) Extraction method Principle Component Analysis; Rotation Method: Varimax with Kaiser Normalization.

124

5.5.1 Dimensions of Social Vulnerability in Ku-ring-gai

Results of the PCA (Table 4) identified six components of social vulnerability. A KMO of 0.754 indicates that the variables were suitable for analysis and the extracted commonalities were above 0.5, which indicates that the extracted components represent the variables well (Fekete 2009).

Table 4: Social vulnerability and its dimensions – Ku-ring-gai LGA

Correlation Name and Percent of between direction of Component variance Dominant variables dominant sign explained variable and adjustment component Percent of population living Special .868 alone Needs, 1 25.79 Percent of people who did Elderly and unpaid domestic work less than -.598 Gender (+) 5 hrs/wk Tenancy and 2 12.54 Percent of rented housing units .887 Housing (+) Percent of population speaking a Immigrants 3 12.28 language other than English at .903 (+) home Median individual income ($ / .758 week) 4 Wealth (-) 10.04 Percent of people with incomes -.718 of $1 - $249 per week Percent employed as managers Occupation -.771 5 7.92 and professionals (+) Median rent ($ / week) .479 Visitors and Low 6 7.87 Percent of visitors .746 Education (+) Extraction method: Principle Component Analysis; Rotation Method: Varimax with Kaiser Normalization, N = 102 SoVI = (0.337*Component1) + (0.164* Component2) + (0.161* Component3) - (0.131* Component4) + (0.104* Component5) + (0.103* Component6)

The first component, Special Needs, Elderly and Gender, explained 25.79 percent of the total variance, and included variables such as the percent living alone (+), percent needing special assistance (+), percent of housing units with no internet connection (+), percent over 65 (+), percent of dwellings with no vehicle (+), percent female (+) and

125 percent who did unpaid domestic work less than five hours per week (-). The direction of loading for each variable is indicated in brackets. People who live alone, the elderly, people who need special assistance, and females may be more likely to need assistance from others in the event of a disaster and are therefore more vulnerable than those who have family connections, are young, those who don’t need special assistance, and males. Access to information and ability to act upon that information during response and recovery is an important component that determines the level of social vulnerability. Households with limited independent mobility are also vulnerable during a time of evacuation (Chakraborty et al. 2005; Granger 2003). Households with no Internet connection or vehicle may have difficulties during a disaster event, increasing their social vulnerability. Some women may be more vulnerable to disasters. For example, a majority of single parents are women, and women from some cultural backgrounds, including those found among migrants to Australia’s cities, may be disempowered (Dwyer et al. 2004). Unpaid housework is also important because it is possible that well-maintained properties are somewhat better prepared (and hence less vulnerable) than more poorly maintained ones (Poudyal et al. 2012).

The second component, Tenancy and Housing, accounts for 12.54 percent of the total variance. It includes three variables: percent of housing units that are rented (+), percent who recently moved house in the past year (+), and percent of dwellings that are cooperative, community or church housing units (+). Ownership and quality of housing is an important factor in assessing social vulnerability. In general, rental houses are not as well-maintained as owner-occupied houses and renters are less likely to take actions to protect property from a disaster (Cutter et al. 2003; Kuhlicke et al. 2011). Often, low income levels limit the range of dwelling type available for renters’ families (Baum et al. 2008). People who have recently moved may be new to the area and their level of awareness of local risk conditions is likely to be low. Levels of available individual financial resources are also typically lower for both renters and transient populations.

The third component, Immigrants, is comprised of the percent of the population that speaks a language other than English at home (+), the percent who do not speak English well or at all (+) and the percent of migrants from overseas (+). This component represents 12.28 percent of the total variance. Immigrants are vulnerable to natural

126 hazards due to their different cultural backgrounds and racial and ethnic inequalities that often lead to discrimination against them (Wood et al. 2010). Often immigrants lack experience with local disasters and their awareness of risk is also low because they may have limited English proficiency.

The fourth component, Wealth, captures three variables and explains 10.04 percent of the total variance. The variables include median individual income (+), median household income (+) and percent of people with incomes of $1 - $249 per week (-). People with higher levels of income have better capacities to prepare for and recover from disasters. People with low levels of income often do not have any disaster recovery plans such as insurance and may have few savings or other assets. Therefore, low-income households are often disproportionately impacted by a disaster (Cox et al. 2007; Flanagan et al. 2011; Wood et al. 2010).

The fifth component is Occupation; accounting for 7.92 percent of the variance. It is comprised of the percent who are employed as managers and professionals (-) and median rent per week (+). Occupation contributes to social vulnerability (Holand et al. 2011; Kuhlicke et al. 2011). People employed as managers and professionals often have higher incomes, which influences their ability to prepare and recover from disasters.

The final component is Visitors and Low Education; it explains 7.87 percent of the variance. It includes two variables, namely percent of visitors (+), and percent of people over 20 years of age who have a year 10 education or less (+). Visitors’ lack of familiarity with the local environment and response planning, along with their relative isolation from the general community makes them more vulnerable during disasters (Granger 2003; King & MacGregor 2000). Education is typically associated with an individual’s level of income and poverty. Moreover, a low level of education may also negatively influence a person’s knowledge and awareness of bushfire management practices.

5.5.2 Dimensions of Social Vulnerability in the Blue Mountains

The same initial set of 51 variables was used for the Blue Mountains LGA analysis; 26 variables were retained for the PCA. These variables represent six components, which describe 73.04 percent of the total variance in the data (Table 5). A KMO of 0.829 127 indicates that the variables were suitable for analysis and the extracted commonalities were above 0.5.

Table 5: Social vulnerability and its dimensions – Blue Mountains LGA

Correlation Percent Name and between of Component direction of sign Dominant variables dominant variance adjustment variable and explained Component Percent of rented housing Tenancy and 0.933 units 1 Household Wealth 27.57 Median household income (+) -0.803 ($/week) Age and Special Median age of persons 0.845 2 12.51 Needs (+) Percent of children under 5 -0.723 Percent speaking a 3 Immigrants (+) 10.33 language other than 0.866 English at home Percent of people over 20 Low Education 4 9.42 years with a year 10 0.794 (+) education or below 0.769 Gender and Percent of females 5 Domestic Work 7.63 Percent who did unpaid (+) domestic work less than -0.596 5hrs/wk Percent employed as managers and 0.597 Occupation and professionals 6 Personal Wealth 5.58 Percent of people with (-) incomes of $1-$249 per -0.576 week Extraction method Principle Component Analysis; Rotation Method: Varimax with Keizer Normalization, N = 138 SoVI = (0.377*Component1) + (0.171*Component2) + (0.141*Component3) + (0.129*Component4) + (0.104*Component5) - (0.076*Component6)

The first component, Tenancy and Household Wealth, is comprised of ten variables and explains 27.57% of the variance. Percent of rented housing units (+) is the dominant variable. Other variables that contribute are percent of dwellings with no vehicle (+), percent living alone (+), percent of housing units with no internet connection (+), percent of one parent families with child under 15 (+), percent who moved house within the past year (+), percent of visitors (+), median household income per week (-), median rent per week (-) and median housing loan repayment per month (-). Negatively loading 128 variables explain the economic circumstances that contribute to a household’s level of preparedness and recovery. Household resources such as income, vehicle, and an internet connection are also very important during the emergency response phase because they support an effective response.

The second component contains four variables, which generally relate to Age and Special Needs. It explains 12.50 percent of the variance. It is comprised of median age of persons (+), percent of the population aged over 65 (+), percent who need special assistance (+) and percent of children under 5 (-). The elderly and people who need special assistance are considered to be vulnerable to disasters (King & MacGregor 2000). This is due to their low levels of mobility and high levels of isolation. Children from the youngest age group (below five years) are one of the most vulnerable groups in a disaster event. They are heavily dependent on others because they cannot protect themselves (Flanagan et al. 2011)

The third component represents the Immigrant population, accounting for 10.33 percent of the variance. It consists of four variables; percent speaking a language other than English at home (+), percent of migrants (+), percent of people with negative income (+) and percent of non-Australian citizens (+).

The fourth component describes Low Education. It explains 9.41 percent of the variance and includes three variables; percent over 20 years who have a year 10 education or below (+), percent of people provided unpaid assistance (+) and percent of volunteers (+).

The fifth component is Gender and Domestic Work, which explains 7.63 percent of the variance and includes two variables; percent of females (+) and percent of people who did unpaid domestic work less than five hours per week (-). The literature suggests that women are more vulnerable than men during disasters due to family care responsibilities and sector-specific employment such as personal services (Rygel et al. 2006; Tapsell et al. 2010).

The final component represents Occupation and Personal Wealth. It explains 5.58 percent of the variance. The variables include percent employed as managers and

129 professionals (+), percent of people with incomes of $1-$249 per week (-) and percent employed in service occupations (+). Occupation is an important factor that influences incomes. Low wage occupations, including many service occupations, are a group that might face slower recovery after a disaster (Cutter et al. 2003; Poudyal et al. 2012; Wood et al. 2010).

5.5.3 Social Vulnerability Index and Mapping

With the increased use of geographic information systems and mapping techniques, visualising the distribution of vulnerability has become a central tool for communicating the results of vulnerability research to decision makers and the general public (Eakin & Luers 2006; King 2001). It is being widely used to capture spatial and temporal changes in social vulnerability (Cutter et al. 2003; Cutter & Finch 2008; Fekete 2009; Schmidtlein et al. 2008). Preston et al (2011) have identified twofold benefits of vulnerability mapping; motivating policy responses and social learning.

For any disaster management activity, a relative measure of vulnerability gives useful insights into potential hotspots of high vulnerability. A SoVI provides a relative measure of social vulnerability across CCDs. A single composite index was calculated for each CCD using component scores and their variances. I believed that not all components contribute equally to social vulnerability. Therefore each component was weighted by the percent of variance it explained divided by the total variance explained by all components. CCDs with high index values represent high vulnerability whereas lower index values represent low vulnerability. In order to identify the spatial patterns of social vulnerability, the resulting SoVI scores were mapped at the CCD level (Figure 31). CCDs were classified using a five category mean-standard deviation classification method, with classes ranging from below -1.5 SD to more than +1.5 SD. Then each (unweighted) individual component was mapped to illustrate the spatial heterogeneity among the CCDs, using a quantile classification method to facilitate comparisons between components.

130

Ku-ring-gai Blue Mountains

Figure 31: Social vulnerability by CCD at the UBI based on SoVI.

In both LGAs, most CCDs show moderate levels of social vulnerability. In the Kur- ring-gai LGA, eight CCDs can be classified as highly vulnerable because their index values are higher than +1.5 SD. Social vulnerability is highly associated with the component Special Needs, Elderly and Gender in the communities in the area’s northern and eastern suburbs, such as North Turramurra and St Ives. In the southern part of the LGA, the level of social vulnerability is more closely related to the Tenancy and Housing component. The high vulnerability CCDs that are in close spatial proximity to the major highway in the area (it runs through the non-UBI region in the middle of the map) includes several CCDs with high numbers of immigrants (Component 3, Figure 32). The value of mapping individual component scores to aid in the understanding of the drivers of vulnerability in particular locations is well-illustrated by the high vulnerability CCD near Fox Valley, whose vulnerability derives from high values on components 3-6 rather than components 1 and 2.

In the Blue Mountains LGA, the results show that social vulnerability to bushfire is significantly higher in the suburbs of the Upper Mountains than in those of the Lower 131

Mountains, particularly in the areas in and around Katoomba. A total of 13 CCDs fall into the highest vulnerability category. Eleven of these highly vulnerable CCDs are in the Upper Blue Mountains and two are in Springwood, which is located in the Lower Blue Mountains. The highest levels of social vulnerability are driven primarily by the Tenancy and Economy, Age and Special Needs components (Figure 34). Immigrants is also an important component in the Upper Mountains. The Occupation and Personal Wealth component shows highly vulnerable CCDs largely concentrated around the Katoomba, Springwood and Blaxland town centres. Low education is evenly distributed across the study area. Most of the CCDs in the Mid-Mountains area show average or low levels of vulnerability for all the components except for Gender and Domestic Work.

132

Special Needs, Elderly and Gender Tenancy and Housing Immigrants

Figure 32: Maps showing distribution of individual component scores by CCD in the Ku-ring-gai LGA (components 1-3).

133

Visitors and Education Wealth Occupation

Figure 33: Maps showing distribution of individual component Scores by CCD in the Ku-ring-gai LGA (components 4-6).

134

Tenancy and Economy Age and Special Needs Age and Special Needs

Figure 34: Maps showing distribution of individual component scores by CCD in the Blue Mountains LGA (components 1-3).

135

Low Education Gender and Domestic Work Occupation and Personal Wealth

Figure 35: Maps showing distribution of individual component scores by CCD in the Blue Mountains LGA (components 4-6).

136

5.6 Discussion

When considering the implementation of any mitigation, preparedness, response and recovery measure, it is important to address the particular social dimensions that make communities more vulnerable. Therefore understanding the social dimensions of vulnerability in a community is a vital step in any efficient planning process. A SoVI summarises the complexity of social dimensions. It is multidimensional in nature and explains the sensitivity of a population to natural hazards and its ability to respond to and recover from the impacts of hazards (Cutter & Finch 2008). Although a SoVI provides a relative measure of vulnerability, the individual component scores can also be used to understand different dimensions that make a particular area socially vulnerable (Rygel et al. 2006). Mapping each component illustrates the spatial variability of that component among CCDs and can help emergency managers to plan and prioritise suitable strategies that target specific problems and specific populations at specific places.

SoVI is a relative measure. It only provides information on inequality between CCDs, and in our study, between CCDs within one LGA. SoVI does not provide information on absolute levels of socioeconomic and demographic characteristics within a CCD. Moreover, PCA is sensitive to variable selection (Fekete 2009). The choice of variables can have an impact on final index values. Therefore special consideration is required during variable selection process.

This research replicated the analysis for both the Ku-ring-gai and Blue Mountains LGAs. The components of social vulnerability and the contribution of each component to vulnerability levels were not exactly the same in both areas; an important point for local policy makers. Each community has its own social dimensions and variables that explain who is most vulnerable (Schmidtlein et al. 2008). This research identified several components that are broadly consistent with other social vulnerability studies (e.g., Cutter et al. 2003, Rygel et al. 2006).

The results reveal differences in the drivers of social vulnerability across space. The primary component that contributes to the overall level of social vulnerability was different among different CCDs. Social vulnerability maps, particularly those of the

137 index’s components, provide deeper insights into what is important in shaping a particular community’s social vulnerability, which will help fire managers to identify and prioritise specific fire management programmes to meet the specific needs of particular vulnerable communities.

For effective wildfire mitigation and adaptation, fire management authorities should focus on places where socially marginal populations intersect with higher wildfire risk because these populations have the added vulnerabilities of lower capacity to prepare for hazard events (Poudyal et al. 2012). In NSW several community-level bushfire management initiatives, such as Community Fire Units (CFU) and Street FireWise community education programmes are being conducted. However, currently these units are not necessarily being established in places where they are most needed: those with the most vulnerable populations. Rather, they are located in areas of relative affluence where residents are often already among the most prepared (Greg Buckley, 2010, personal communication).

While local response organizations (e.g., police, fire services) are typically aware of where socially disadvantaged populations can be found, social disadvantage is not the only correlate of vulnerability. Furthermore, although a standard index of socioeconomic disadvantage (SEIFA) is available in Australia, it excludes many variables that are very relevant to social vulnerability, such as people needing special assistance during fires and recent migrants. Therefore, the use of SEIFA on its own is insufficient to meet the needs of either a disaster planner or an emergency manager (King 2001). Moreover, a SoVI index provides a ranking of comparative levels of vulnerability that can be used to prioritize interventions when scarce resources do not allow for intervention in all vulnerable areas.

Most social vulnerability assessments have been conducted using census data and census geographies. In this research, variables were derived from the 2006 Australian Census. However, census data only provides some information about the communities and the final index is influenced by available data. King & MacGregor (2000) and (King 2001) have identified some of the limitations of this approach. For example, as administrative units, census geographies may not represent actual communities. Actual

138 communities may cross the boundaries of census geographies. Residents’ perceptions, norms, values and cultural attitudes as well as the complexity of social interactions within communities may not be represented by census data.

Available census data also may not represent the current sociodemographics of an area. The Australian Census is undertaken in winter, when the characteristics of the population in a given location may not be the same as during the summer and school holidays. The census also measures night-time population, which differs in important respects from daytime populations. On weekdays most of the nighttime population may not be at home and only retired persons, the elderly and young children (and their carers) are at home. These population characteristics are hard to capture with census data. Nevertheless, for many communities, the census remains the best available source of population data because the data are consistent, repeatable and readily available. Yet, these limitations mean that a critical examination of the map, its underlying data and assumptions is required before decisions are made based upon the map (Preston et al. 2011)

The assessment of social vulnerability using a SoVI is not an end product. Knowledge about the geographic and sociodemographic context are required to link the results of a SoVI analysis with the real world. A SoVI’s results need to be verified and validated insofar as possible. When possible, SoVIs need to be validated to verify the relationship between social dimensions and hazard impacts using an independent secondary data set (Fekete 2009).

This research used local knowledge and data on exposure to recent bushfire events in other locations in Australia to verify and interpret the results. I could not directly validate the SoVI generated in this research since a secondary dataset on socio- demographic characteristics of a bushfire-affected population for the study area is not available. Indeed, Adger et al (2004) have noted the difficulty in general of validating vulnerability indicators because hazard events in a given location are relatively rare. This is exacerbated by the fact that the characteristics of populations in a given location change over time, so validation requires relatively recent data on a population’s experiences of a hazard to identify who was vulnerable. It has been three years since the most recent bushfire in the Ku-ring-gai LGA and just one year since the most recent fire 139 in the Blue Mountains Council LGA. However, the lack of a validation data set does not mean that some method of assessing the levels of vulnerability within a community is not needed for hazard management. In the absence of a validation data set, it is possible to get some indication of whether the results are reasonable by examining populations who were vulnerable to recent bushfire events in other parts of Australia.

Recent studies on bushfire impacts in Australia have shown that social vulnerability and related processes that place specific segments of human populations at risk can create a disaster when a bushfire strikes. Reviews of fatalities in the Black Saturday bushfires in Victoria show that many of those who died in the event (i.e., those who were vulnerable; over 44%) were disabled, elderly or very young (i.e., children) (Handmer et al. 2010). In both LGAs, these variables were included in either the first or second principle component (i.e., those that explain a large proportion of variation in the data). Unfortunately the Black Saturday fatality report did not detail any other sociodemographic characteristics of the affected population that could be used for validation. The lack of access to a validation data set is a limitation of this research.

5.7 Conclusions

Prevailing policies and disaster management frameworks should address social inequalities and their impacts on disaster preparedness, and social vulnerability needs to be better integrated into current hazard management planning. Social vulnerability information can be used at any stage of the disaster management cycle. At the mitigation and preparedness phase SoVI can be used as a proactive tool to plan community based disaster mitigation activities that support more vulnerable areas in order to increase community resilience. At the response and recovery phases, it can be used as a reactive tool to plan warning, evacuation and recovery needs (Cox et al. 2007; Granger 2003).

Social vulnerability to natural hazards is dynamic (Cutter & Finch 2008). A SoVI can be used to capture the dynamics of the social demographic characteristics at the UBI. This type of work can provide critical insights into the changing nature of the UBI over time. The Australian census was recently undertaken (August 2011) and when the data are released, a comparative analysis could be completed. It would also be interesting to 140 investigate whether there is a relationship between UBI expansion and the level of social vulnerability of the people who move to the UBI. Knowledge about the spatiotemporal dynamics of social vulnerability can also be used to suggest future preparedness and response planning needs.

Nevertheless, social vulnerability is only one component of vulnerability. To have a complete picture, physical factors such as infrastructure and geographic features, community resources, and mitigation and preventive measures that may help to minimize impacts need to be combined. Social vulnerability is a vital component of the risk paradigm as it contributes to determining the level of resilience of particular segments of society and in this instance, particular geographical locations. Therefore a social vulnerability assessment is an important step in any in future risk and vulnerability studies.

The results of this chapter emphasize the importance of social vulnerability assessment in an Australian context: different population characteristics are related to vulnerability within populations in different places. This methodology can be incorporated into existing emergency management plans, and used in conjunction with other existing data on vulnerability and risk to understand the vulnerability of a particular community and area. Such assessments can help to develop more effective programs in order to utilize available resources in an efficient way by identifying the most vulnerable populations in a region. Once hazard managers know who is vulnerable, they can determine what specific actions might be best taken to decrease impacts of the hazard on that population.

5.8 Summary

This chapter discussed the socioeconomic fragility component of vulnerability. A social vulnerability index was developed using census data to understand the underlying causes of social vulnerability. Social vulnerability maps described the spatial variation of social vulnerability at the UBI. As discussed in Chapter 3, exposure and physical susceptibility, and emergency response and coping capacities also determine the vulnerability. The following chapter discusses the vulnerability assessment process that

141 integrates all three components of vulnerability. It further explores the integrated risk assessment process that combines the hazard and vulnerability assessment results.

142

Chapter Six: Bushfire Risk Assessment: An Integrated Modelling Approach

6.1 Overview

This chapter presents an integrated bushfire risk assessment based on the framework described in Chapter 3. Firstly, vulnerability to bushfires at the UBI was assessed using different indicators of vulnerability. The social vulnerability was presented in the previous chapter (Chapter 5). This chapter presents the assessment results of other two components; exposure and physical susceptibility, and preparedness and response capacity. All three components were then combined to develop the overall vulnerability map. The hazard map developed in Chapter 4 was then combined with the overall vulnerability map to develop the final bushfire risk map. Different data sources (spatial and non spatial) were used for this analysis and GIS data and spatial modelling techniques were used to develop indicators and output maps. The outputs of this analysis are presented as a series of maps. The resulting bushfire risk map presents high risk spots at the UBI in the study area. The input maps can be used to understand the underlying drivers of bushfire risk.

6.2 Introduction

In Australia, fire behaviour research dominates most of the bushfire hazard research activities. Fire behaviour models are mostly concerned with factors that determine the intensity and the behaviour of a fire (Whittaker 2008). A few researchers have done studies on buildings and structural vulnerabilities (Blanchi et al. 2011; Blanchi et al. 2006). Interactions between the physical event and the human system have been less commonly studied. According to Alexander (1993), people become vulnerable because they inhabit bushfire-prone areas. Collins (2005) has stated that fire problems have risen to prominence for three reasons: frequent occurrence of destructive fires, population increase in biophysically hazardous areas and inadequate mitigation and response measures. Therefore it is important to investigate the strengths and weaknesses that are induced by the social, economic and physical conditions of the human system. In the context of bushfires, risk assessment is one of the least investigated tasks, due partly to the lack of relevant detailed socioeconomic data and the difficulty of their effective 143 spatial representation for integration with physical environmental data on hazards (Chen et al. 2003).

6.3 Bushfire Risk and Conceptual Frameworks

The concept of holistic bushfire risk is not well recognized in the bushfire research community. Traditional fire terminology focuses more on fire dynamics than understanding the impacts caused by fires when they reach and enter the UBI (Chuvieco et al. 2010; Murnane 2006). Several conceptual models have been developed to address this issue. These models produce a realistic estimation of the actual magnitude of the risk posed in a given area. Unlike fire behaviour models, these models include a component that accounts for damage, impact or consequences.

Bachmann and Allgower (2000) defined fire risk as “the probability of a wild land fire occurring at a specific location under specific circumstances, together with its expected outcome as defined by its impact on the objects it affects” (p.28). They constructed scenarios to represent possible realizations of fires. These scenarios are based on the relevant preconditions and causes, which enables the quantitative determination of risk while emphasising that total risk is determined by the collective risk of a scenario (Bachmann & Allgower 2001). The structured wildfire risk model incorporates ‘risk’ in terms of probability and outcome based on wildfire occurrence, wildfire behaviour, and expected wildfire impact. Finney (2005) described fire risk as an expected net value change that is determined by the summation of losses and benefits from all potential fire behaviours (under all weather conditions from all ignition locations). Compared to Bachmann & Allgower (2005), in this framework, all losses and benefits are evaluated based on a monetary value. According to Finney (2005), prescribed burnings are considered to be beneficial.

In Australia, bushfire risk assessment frameworks (Atkinson et al. 2010; Shields & Tolhurst 2003) have been introduced based on the risk definition in the Australian and New Zealand Standard (AS/NZS 2004). According to these studies, risk is a product of likelihood and consequence:

144

Risk = Likelihood x Consequences Equation 7

Shields & Tolhurst (2003) developed a bushfire risk assessment based on the above risk equation. Likelihood is calculated based on the probability of an ignition in the landscape reaching the urban interface. The likelihood of a fire reaching each urban interface is, in turn, determined by the possible ignitions and varying fire weather scenarios. Based on these likelihoods, risk measures can then be obtained by incorporating consequences. Consequence represents the impact of a fire starting and spreading. It is assessed by estimating the physical impact upon specified objects. Atkinson et al. (2010) adapted this framework and estimated consequence by using dwelling density and land use.

The above mentioned models are mainly focused on consequences or the outcome of the bushfire events. They capture tangible effects of fire in order to estimate potential losses. In the context of bushfire management, understanding hazard levels as well as vulnerabilities and the capacities of the communities at the UBI is important. Therefore, integrated risk needs to be understood as the interaction between hazard and vulnerable communities. Blanchi et al. (2002) presented a bushfire risk model as a combination of hazard and vulnerability. In this model, hazard is characterised by the probability of fire ignition and its intensity. Vulnerability is conceptualised as the foreseeable consequences of a fire determined by the level of fire intensity, damage and prevention, and fire fighting resources. Although this model has addressed some components of vulnerability, it has ignored social vulnerability. Therefore, it does not present a robust measurement of overall vulnerability.

Chuvieco et al. (2010) have also proposed a fire risk model that includes a vulnerability component. It focuses on the assessment of potential damage caused by fire. However, this vulnerability component only captures socioeconomic values, degradation potential and landscape values. In this model, socioeconomic values consist of the monetary values of wood products and recreational and tourist resources. Important vulnerability characteristics such as social vulnerability and community capacities have not been considered. Therefore, this model does not present the actual level of risk posed to a

145 given UBI. It includes only a limited description of vulnerability and only measures losses in terms of monetary values.

Chen et al. (2003) introduced a conceptual framework that represents a holistic and multidisciplinary approach to assessing bushfire risk. In their model, risk is described using individual aspects of a hazard, a community’s vulnerability, and their interactions. According to their framework, the hazard is measured using fire occurrence and behaviour. Vulnerability is determined by physical, socioeconomic, environmental and management characteristics. Although this model fills the gap of having an integrated bushfire risk assessment model, the relationship between hazard vulnerability and each of the components of physical, environmental and socioeconomic characteristics and fire management activities that determine the level of impact are not explained well. The lack of relevant detail on each of the components makes this model difficult to apply. Therefore, this conceptual model is yet to be implemented.

It is evident that existing attempts at modelling risk for fire management have not been based on a common understanding of risk. Each has missed one or more components of risk. Although most of these models have addressed impacts or damages, none of these models has addressed the components of vulnerability such as social characteristics, response capacities and exposure to bushfires. Another key limitation that can be seen in current bushfire risk assessment frameworks is not having well-established measures for physical, environmental, social and bushfire management components of risk. This limitation reveals the complexities associated in understanding of these concepts. Therefore, it is also important to establish a means and mechanism though which to better understand the relationships between and measure these components.

In Australia, a generalized definition of ‘fire risk’ has been commonly used. Although the term ‘risk’ is used, most assessments have only focused on the hazard component and the consequences of the fire hazard. Factors affecting vulnerability at the UBI are often ignored. Traditional fire terminology only focuses on ignition and propagation potential and little emphasis has been given to potential damages of the fires (Chuvieco et al. 2010). Thus the potential impact is mainly evaluated only based on the monetary value of the properties. As discussed in Chapter 2, impact of a potential hazard is a 146 product of both hazard and vulnerability. Therefore, the monetary value of properties does not fully reflect the actual impacts. To understand the actual impacts, the level of hazard, vulnerability as well as their interactions need to be identified. In order to do that, hazard and vulnerability need to be addressed separately and then integrated into the risk model.

6.4 Integrated Approach

The integrated approach presented in this study was based on the integrated framework discussed in Chapter 3. The analytical framework is shown in Figure 36. The hazard element of this integrated risk assessment framework was addressed in Chapter 4. In the context of bushfire risk, vulnerability is determined by key three components; exposure and physical susceptibility, social vulnerability, and response and coping capacities. Social vulnerability at the UBI was discussed in Chapter 5. The combination of exposure and physical susceptibility, social vulnerability, and response and coping capacities in a given region are referred to as vulnerability. Figure 36 shows the components of vulnerability and the respective indicators selected to analyse individual vulnerability components. In total, 16 indicators were selected to assess exposure and physical susceptibility and response and coping capacities (Table 6 and Table 10). Overall vulnerability to bushfires at the UBI is then analysed based on individual vulnerability components. Finally, hazard and vulnerability are combined to understand the level of overall risk.

147

Risk

Hazard Vulnerability

Exposure and Social and Response and Physical Economic Coping Susceptibility Fragilities Capacities

Fire occurrence Proximity to Social Road Fire severity bushland vulnerability Connectivity Building density index Proximity to Type of land use evacuation point Type of Proximity to fire vegetation station Density of critical Proximity to infrastructure hospital Slope Dwellings per Aspect CFU Density of fire hydrants Density of static

Figure 36: Integrated bushfire risk assessment model.

6.5 Bushfire Risk Assessment - MCE Model

Figure 37 explains the multicriteria evaluation (MCE) model used in this study. It is driven by three components; constructive elements, conditions and constraints of the process, and the spatial multicriteria evaluation process. The process starts with the selection of indicator maps, the way criteria are going to be structured, and standardisation and weighting methods (Abella & Van Westen 2007). For this study, the constructive elements are vulnerability and hazard. The elements are subdivided into different criteria based on the indicators identified in the hazard and vulnerability assessment. Expert opinion was used to design the criteria tree based on the risk assessment framework proposed in Figure 36. Vulnerability and hazard indicators were standardised using an appropriate fuzzy membership function (Table 8 and Table 12). In the next step, weights for each indicator and criterion were decided. Finally, the criteria were aggregated using assigned weights with the selected algorithm. This process

148 results in hazard and vulnerability maps. The final risk map is derived as a function of the hazard and vulnerability maps.

Constructive elements Conditions and constraints Vulnerability Objective Socio-economic Hazard Scale conditions Ignition Expert Data availability Built environment probability conditions opinion Fire Community severity capacities

Criteria tree design

Vulnerability Hazard indicators indicators

n Standardisation Standardisation

Weighting Weighting

Aggregation Aggregation

atial Multicriteria Evaluatio Multicriteria atial Vulnerability Hazard p

S map map

Bushfire risk map

Figure 37: MCE model for bushfire risk assessment. Adapted from (Abella & Van Westen 2007, p. 312).

The MCE model depicted in Figure 37 was used in the MCE process. The MCE model suggests three sub-models that capture the components of vulnerability. Each sub-model consists of data that explains the relevant vulnerability component. Sub-models contribute to the overall model of vulnerability. A summary of data and indicators used to assess exposure and physical susceptibility and response and coping capacity can be viewed in Table 6 and Table 10. Social vulnerability was assessed based on the SoVI calculated in Chapter 5.

149

Some of the resulting indicators were made possible with the use of proximity analysis, 3D analyst and Spatial Analyst tools within ArcGIS 10 software (ESRI Inc. Redlands, USA). Density layers were generated using kernel density estimation with a fixed bandwidth size using CrimeStat 3.3 software. Various other tools such as geoprocessing, data editing and data conversion tools available with the ArcGIS 10 software were used to manage selected vector and raster data sets.

6.5.1 Standardisation, Weighting and Identifying Evaluation Rules

Because the indicator maps use different measurement scales, they need to be standardised from their original values to a common value scale (Abella & Van Westen 2007; Rashed et al. 2007). This enables comparisons between layers. This research uses the fuzzy approach (discussed in Chapter 4.7.1) because of the inherent uncertainty associated with other methods. A number of sigmoid fuzzy membership functions (monotonically increasing, monotonically decreasing and symmetric), shown in Figure 38, and user-defined fuzzy functions are used to convert original data into fuzzy sets. This process is known as fuzzification. A fuzzy membership function was directly applied to the indicators that are measured as ratio data. Nominal indicators such as land use zones and type of vegetation were ranked based on likelihood of occurrence of fires and the potential impact on the area before applying the relevant user-defined fuzzy membership function.

Fuzzy membership functions standardise indicators into either a 0-1 real number scale or 0-255 bytes scale. In this research a 0-1 real number scale was used. IDRISI Selva software (Clark Lab, Worcester, MA, USA) was used for this fuzzification process.

150

Figure 38: Different sigmoid fuzzy membership functions

Identifying evaluation rules is one of the most critical parts of the MCE process. In this process, the standardised indicators are identified based on their contribution towards the overall objective. The objective of this step is to identify areas of high vulnerability. Therefore, indicators were organized according to their contribution to vulnerability. Another important aspect of MCE is the use of constraint indicators. Constraint indicators act as a mask for the MCE process. In this research, human settlement area is used as a constraint layer as the research is only focused on locations where humans live. This mask helps to identify the areas where human activities are present.

After deriving appropriate indicators, and defining their standardisation and evaluation rules, weights were assigned to each indicator and to intermediate results. This research utilised the Pairwise Comparison Method to determine the weights. This method allows the comparison of only two indicators at once. In this research, information gathered from discussions with key personnel from local government councils, NSW Fire and Rescue, and researchers was used when defining the pairwise comparison matrix using a standard pairwise comparison tool (Appendix III). The pairwise comparison matrix takes the pairwise comparisons as an input and produces relative weights as an output. The output is used as a set of relative weights for the indicators.

The purpose of this study is to identify areas with higher levels of vulnerability at the UBI. In order to achieve that, the fuzzy criteria and their weights need to be aggregated. In this study, two aggregation methods were used. Weighted Linear Combination

151

(WLC) was used to combine indicators to produce vulnerability components. In this step only the exposure and physical susceptibility component and the response and coping capacity component were assessed as the social vulnerability component was assessed in Chapter 5. The resulting vulnerability component as well as the social vulnerability component were then standardised and aggregated using the Ordered Weighted Average (OWA) method (see Chapter 4.7.1 for descriptions of WLC and OWA). In this method, order weights are used in addition to component weights. Order weights allow for direct control over the level of trade-offs and uncertainty.

6.6 Vulnerability Assessment

6.6.1 Exposure and Physical Susceptibility

In vulnerability research, exposure has a number of different meanings. Exposure is related to the location of the community with respect to the hazard and environmental surroundings (Villagrán de León 2006). It also refers to the concentration of elements at risk, and includes those elements within a given area that have been, or could be, subject to the impact of a particular hazard. Elements at risk might include people, property, systems, or other elements present in hazard zones that are thereby subject to potential losses. Exposure is influenced by the number of people, buildings and other infrastructure within an area (Geoscience Australia 2011). The concept of exposure is utilised by some disciplines to refer particularly to the physical aspects of vulnerability (UNISDR 2004). Depending on the context of the study it is defined as “exposure” or “placed in harm’s way” or “being in the wrong place at the wrong time” (UNISDR 2004). Exposure may be measured using population density levels, remoteness of a settlement, the site, and the design of buildings and other infrastructure within an area (Geoscience Australia 2011; UNISDR 2004).

The objective of exposure and physical susceptibility analysis is to identify probable areas where direct and indirect consequences may occur because of a severe bushfire event due to its physical and environmental condition. It also estimates the expected effects of a potential fire hazard at the UBI. An area that has conditions that may enhance the impact of the hazard is considered to be an area with high exposure and areas where minimum disruption is predicted are considered to be areas with low

152 exposure. Analysis of exposure requires identification and development of extensive datasets that define the most exposed components of communities.

Data that provides the basis for assessing exposure can be classified into two groups. Physical environmental conditions describe human settlement locations, infrastructure and facilities, and the distribution of people. Natural environmental factors capture the natural features of a particular location such as slope, aspect, type of vegetation, etc. Risk assessment based only on potential fire hazard levels at the UBI will generally be misleading, because different areas of a given community can have substantially different levels of physical and environmental conditions that influence the fire’s impacts. Therefore, spatial variability of physical and environmental conditions should be taken into account. Exposure or physical susceptibility describes the quantity and the distribution of physical and environmental elements. According to the holistic approach and framework discussed in Chapters 2 and 3, it is a key component of vulnerability. Regardless of the hazard’s intensity, there would be no damage or harm if the affected area is free from exposed elements with zero susceptibility. It also determines the level of actual risk (Davidson 1997). The larger the exposure, the greater the risk, and the lower the exposure, the lower the risk. In this analysis, exposure is addressed with respect to physical environmental and natural environmental features in the UBI.

Based on the literature and data availability, natural and physical environmental features that determine the level of exposure and physical susceptibility were identified. Identified features were used to derive indicators for the MCE analysis. Physical features describe properties, infrastructure and other assets that could be subjected to potential future harm whereas natural environment features describe the place-based natural conditions that increase severity and distribution of a potential hazard and the disruption in the area from the hazard (Lavell et al. 2012). Terrain features, fuel characteristics and weather conditions largely determine bushfire severity and intensity at the UBI. Natural environmental conditions vary across the area. Therefore, a uniform distribution of hazard intensity may not be observed and damage to physical environmental properties may also vary.

153

The selected indicators are given in Table 6. Data was obtained from the Land and Property Management Authority (LPMA) and Local Councils in the study area. While weather conditions can vary considerably over a localised scale, the limited number of 24 hour weather stations (n = 2) within the two study areas does not allow for a valid and robust construction of weather-related indicators. Therefore, weather related indicators have not been included in this study. To a limited extent, some variability in exposure caused by spatially variable weather conditions will be captured through other factors (slope, aspect and vegetation communities).

Table 6: Indicators of exposure and physical susceptibility

Available Data Layer Derived Indicator Data Source Dwellings Dwelling density LPMA Critical infrastructure facilities Density of critical infrastructure LPMA Dwelling/Bushland Distance to bushland LPMA Land use Land use zones at the UBI Local Council Vegetation community Flammable vegetation Local Council Digital Elevation Model Slope LPMA Digital Elevation Model Aspect LPMA

Variables that capture natural environmental features were identified as factors that are generally beyond the control of humans. They include elevation, aspect, slope and type of vegetation. Elevation, slope and aspect were derived using a 5m resolution Digital Elevation Model (DEM). Elevation is an important variable that is used to derive slope and aspect, which influence fire intensity and propagation. Fire travels faster up slopes than down slopes and across flat ground (Jaiswal et al. 2002). Slopes with west, northwest or north aspects are exposed to longer periods of sun during the day and are warmer than flat areas or those facing other directions. Because of this, drier soil is more prone to ignition (Zeng et al. 2003). Northwest aspect slopes have higher temperatures, more robust winds, lower humidity and lower fuel moisture compared to north, northeast and west aspects. The type of vegetation and its fuel content affect the potential ignition and spread of fires (Jaiswal et al. 2002; Zeng et al. 2003).

Availability of flammable vegetation provides an important indicator of exposure in terms of potential fire intensity. Key vegetation communities in the vegetation classification layer in the Ku-ring-gai area are Blue Gum High Forest, Sydney

154

Turpentine Ironbark Forest, Coastal Shale Sandstone Forest, Duffys Forest, Sydney Sandstone Ridgetop Woodland, Sydney Sandstone Gully Forest and Gully Rainforest (Colyer 2010). In the Blue Mountains area, the vegetation layer is classified into a different set of vegetation communities. Key vegetation communities that can be found in the Blue Mountains vegetation layer are; heathlands and scrub, dry sclerophyll forest, wet sclerophyll forest, forest wetlands and swamps, escarpment complex, rainforest, transition forest complex and modified bushland (vegetation thinned or removed) (Hammill & Tasker 2010).

Physical environmental factors describe the characteristics of the built environment. They also include the factors that illustrate the interaction between humans and the UBI. UBI areas are more fire prone where they are located adjacent to high building densities, which increases activities that result in interactions between humans and bushland (Vadrevu et al. 2010). Forested regions near to settlements and houses are more prone to fire ignitions because fire can rapidly propagate through structure-to- structure ignition. Densely populated areas require more time and effort to evacuate and a greater impact can be expected since more structures are impacted (Haight et al. 2004). Distance between the UBI and bushland is a critical parameter that influences the level of exposure and physical susceptibility. The probability of home destruction decreases almost linearly with increasing distance from the bushland boundary. The probability of home destruction within the first 50 m of the forest edge is about 60%, which provides an empirical basis for pricing and differentiating bushfire risk to homes at the bushfire-prone urban boundary (Chen 2005).

Land use is another important factor considered for the analysis. It describes the area where human activities happen as well as activity types. Common land use classifications are residential, commercial and industrial, infrastructure, recreational and open space and vegetation. Critical infrastructure includes individual lifelines, resources and facilities within the UBI. It is important to identify their locations and geographical distribution to determine the level of exposure of the area. The people in critical locations infrastructure often seek outside help during a time of disaster (Boyd et al. 2002). In this study, critical infrastructure locations include hospitals, primary schools, day care centres, nursing homes and other aged care facilities. 155

The vulnerability analysis was performed in a GIS environment by processing, preparing and transforming available raster and vector data into 50m resolution raster grids. In order to perform the MCE process, the prepared raster grids were standardised using the relevant fuzzy membership function (Table 8). A fuzzy GIS approach also helps to evaluate the spatial distribution of identified indicators and the vulnerability they would cause. Each pixel value in the raster data set is determined by the association between each indicator and a particular fuzzy membership function, ranging from 0 to 1 (Malczewski 1999). In this case, 0 represents low vulnerability whereas 1 represents high vulnerability.

The evaluation was done in terms of the numbers, sizes and geographical distribution of indicators within the UBI. To assign weights for the indicators, the pairwise comparison method was used. In this study three different groups were considered to be decision- makers: council members, fire managers and researchers. These experts are knowledgeable about local conditions and therefore they could directly compare the relative importance of two indicators using a nine-point scale. The pairwise comparison matrix shown in Table 7 was developed based on their judgements. Based on these expert judgements, distance to bushland and flammable vegetation around the UBI area were considered key factors that determine the level of exposure and physical susceptibility. The least important factor was density of critical infrastructure. The resulting matrix was analysed for consistency to ensure consistency of judgements in the pairwise comparison (Saaty 1980). The consistency ratio (CR) was 0.07, which is less than the recommended 0.1. If CR <= 0.1, consistency is acceptable.

Table 7: Pairwise comparison matrix of indicators of exposure and physical susceptibility DisBL VEG SLP ASP LUSE DwelDEN DenCI Weights DisBL 1 2.3 5.6 3.3 5.3 5.6 6.8 0.41 VEG 3/7 1 1.7 1.7 4.0 2.9 5.1 0.18 SLP 1/6 3/5 1 1.6 4.3 2.8 4.0 0.12 ASP 1/3 3/5 5/8 1 5.7 4.0 5.7 0.13 LUSE 1/5 1/4 1/4 1/6 1 0.4 0.6 0.05 DwelDEN 1/6 1/3 1/3 1/4 2 1/2 1 1.0 0.06 DenCI 1/7 1/5 1/4 1/6 1 5/7 1 1 0.05

156

Consistency Ratio (CR) = 0.07 (< 0.1), DisBL= Distance to bushland, VEG = Flammable vegetation, SLP = Slope, ASP = Aspect, LUSE = Land use zones, DwelDEN = Dwelling Density, DenCI = Density of critical infrastructure. Table 8 shows the fuzzy memberships, evaluation criteria and the weights used for each indicator in the study.

Table 8: Indicators of exposure and physical Susceptibility - Fuzzy membership functions, evaluation criteria and weights Derived Indicator Fuzzy Membership Weight Sigmoidal Monotonically Distance to bushland 0.41 Decreasing Flammable vegetation User Defined 1 0.18 Slope User Defined 0-10 0 10-20 0.25 0.12 20-30 0.5 30-40 0.75 Above 40 1 Aspect User Defined N 0.75 NE 0.5 E 0 SE 0 0.13 S 0.5 SW 0.5 W 1 NW 1 Land use zones at the UBI User Defined Residential 1 Industrial / Business 0.75 Development / Infrastructure 0.5 0.05 Open space / Recreational 0.25 Environmental protection / National 0 Parks Sigmoidal Monotonically Dwelling density 0.06 Increasing Sigmoidal Monotonically Density of critical infrastructure 0.05 Increasing 1: See Table 9 for more details

157

Table 9: Vegetation communities in Ku-ring-gai and the Blue Mountains

LGA Vegetation Community Fuzzy Criteria Sydney Sandstone Ridgetop Woodland 1 Sydney Sandstone Ridgetop Woodland (wet heath) 1 Duffys Forest 0.8 Sydney Sandstone Ridgetop Woodland 0.6 Ku-ring- Coastal Shale Sandstone Forest 0.4 gai Sydney Turpentine - Ironbark Forest 0.4 Sydney Sandstone Gully Forest 0.2 Blue Gum High Forest 0.2 Gully Rainforest 0 Heathlands and scrub 1 Dry sclerophyll 0.8 Granite slope forest 0.8 Escarpment complex 0.6 Transition forest complex 0.6 Blue Mixed (dry/wet) sclerophyll 0.4 Mountains Modified bushland (vegetation thinned or 0.4 removed) Forest wetlands 0.2 Swamps 0.2 Wet Sclerophyll 0.2 Rainforest 0

Figure 39 and Figure 40 show the fuzzy criterion maps of exposure and physical susceptibility in the Ku-ring-gai and Blue Mountains areas, respectively. These criterion maps were aggregated in an additive fashion using Weighted Linear Combination. The final maps (Figure 41 and Figure 42) show the vulnerability of each area based on exposure and physical susceptibility. These raster maps were classified into five vulnerability groups; extreme, high, moderate, low and minimal using the quantile method. The result of the exposure and physical susceptibility analysis is used to assess the integrated vulnerability of the area (Chapter 6.6.4).

158

Proximity to bushland Vegetation

Slope Aspect

Figure 39: Fuzzy criterion maps of the indicators of exposure and physical susceptibility in Ku-ring-gai.

159

Land use zones Dwelling density

Density of critical infrastructure

Figure 39 continued.

160

Proximity to bushland

Vegetation

Slope

Figure 40: Fuzzy criterion maps of the indicators of exposure and physical susceptibility in the Blue Mountains. 161

Aspect

Land use zones

Dwelling density

Figure 40 continued.

162

Density of Critical Infrastructure

Figure 40 continued.

Figure 41 shows the exposure and physical vulnerability of Ku-ring-gai. It is a combination of both physical environmental and natural environmental characteristics. As discussed in Chapter 2, risk is a result of interaction between a vulnerable community and the external hazard. Therefore, to understand the vulnerability of communities, only the human settled area within 3 km of the UBI was considered to be an area of interest when developing the integrated vulnerability map.

163

a) Exposure and physical susceptibility map b) Exposure and physical susceptibility zonation map

Figure 41: Exposure and physical susceptibility maps - Ku-ring-gai.

The results suggest that most of the UBI in Ku-ring-gai is exposed and physically susceptible to bushfires. Extremely vulnerable areas are found close to bushland. This reflects the fact that the highest weights were assigned to distance to bushland and type of flammable vegetation. According to the results, the eastern part of the area shows low levels of exposure and physical susceptibility compared to the other areas.

164

Figure 42: Exposure and physical susceptibility map - Blue Mountains.

Figure 43: Exposure and physical susceptibility zonation map - Blue Mountains.

Figure 42 and Figure 43 show the exposure and physical susceptibility in the Blue Mountains area. The results suggest that most of the area in the Lower Mountains is more exposed than that of the Upper Mountains area, mostly due to higher proximity to bushland. The areas that have a minimal level of exposure and physical susceptibility are located away from bushland and are not adjacent to flammable vegetation.

165

6.6.2 Social and Economic Fragilities

The conceptual framework identifies this component as social vulnerability. The indicators of social vulnerability and the process of developing a social vulnerability index were discussed in Chapter 5. The calculated social vulnerability index for the study area was used to understand the socioeconomic fragilities in each area. The resulting SoVI map was standardised and used as an input into the MCE model.

6.6.3 Response and Coping Capacities

It has been recognised that improving disaster preparedness and emergency response and coping capacities at various levels reduces the level of vulnerability and risk from a potential hazard event. Response and coping capacity describe how effectively and efficiently a community can respond to a particular disaster by suppressing fires at the UBI, thereby reducing vulnerability, and recovering from the impact more quickly. This component includes actions that are taken before, during and after the disaster. Therefore emergency response and coping capacity is considered to be a key component of a vulnerability assessment (see Chapters 2 and 3).

As the focus here is on vulnerability in the context of bushfires, short-term actions taken during and soon after a disaster are considered. In the context of bushfires, preventing spot fires at the UBI during a fire is an action that should be taken into account. Fire fighting support from fire management authorities and community volunteers are critical to protecting the UBI from spot fires. The selection of emergency response assessment indicators was based upon an analysis of the literature and consultation with experts in the field of bushfire management. The selected indicators include access to resources, fire suppression capacity and evacuation potential. The identified data layers and derived indicators are shown in Table 10. Data was obtained from the Land and Property Management Authority and New South Wales Fire and Rescue.

166

Table 10: Indicators of response and coping capacity

Available Data Layer Derived Indicator Data Source Road network Road density LMPA Road network Distance to main roads LPMA Distance to natural water Natural water sources LPMA bodies Locations of fire hydrants and static water Density of fire NSW Fire and supply (SWS) points hydrants/SWS points Rescue Location of hospitals Distance to hospital LPMA NSW Fire and Community Fire Unit (CFU) locations CFU density Rescue Location of fire station Distance to fire station LPMA Distance to evacuation LPMA Location of evacuation centre centre

There are many forms of resources. In this analysis, fire suppression resources and facilities that minimise the level of impact were considered. The identified resources include water, health care and transport facilities. Availability of water is an important factor when it comes to fire suppression. Access to fire hydrants and other static water sources helps to minimise the risk of the fire spreading and causing a significant impact (Zeng et al. 2003). Health care facilities are another important resource that minimise the impact of a fire on residents at the UBI. During a bushfire on-site health care facilities must be provided and injured people should be transported quickly to healthcare facilities as necessary (Davidson 1997). This study included public and private health care facilities within and around the study area. The road network is another key resource that determines people’s movements and their level of access to other resources. It also important to maintain the performance of emergency response functions during a bushfire event. The road network is not evenly developed across the UBI. Only one access road can be seen in some suburbs in the area. It is also obvious that a transport network in a compact UBI operates closer to its capacity and therefore has greater difficulty to cope with unscheduled events. In this study, road density and proximity to main roads are considered to be key indicators that describe the characteristics of the road network in the area.

Community response capacity plays a critical role during bushfire events (Lowe et al. 2008). It describes community level preparedness and mitigation activities. It is more 167 vital when the fire intensity exceeds fire-fighting capacity. In the study area, the Community Fire Unit (CFU) approach operates under the direct supervision of Fire and Rescue New South Wales (FRNSW). CFUs consist of trained fire fighters and the necessary fire-fighting equipment. Proximity to fire stations is also an important factor when suppressing fires. In New South Wales, FRNSW is the responsible agency for fire management at the UBI. Therefore, the location of their fire stations and their proximity to the dwellings were used as indicators.

Successful evacuation requires individual decisions and marked exit routes (Paton 2003). Although evacuation decision-making is a complex process, the distance to an identified evacuation point and connectivity that facilitates the evacuation process during the time of evacuation are both important. Evacuation potential is largely determined by the road network, number of vehicles and potential traffic incidents in the area. In this study, road density was used as an indicator to measure the connectivity in the study area. Proximity to the nearest evacuation centre is an important element of emergency response capacity. Proximity to evacuation points was derived based on the point distance from dwellings to the closest evacuation point.

The same methodology discussed in section 6.7.1 was applied to assess emergency response and coping capacities. In this assessment, evaluation criteria were applied to identify areas that lacked emergency response and coping capacity. Therefore, in the fuzzified layers, 0 represents a higher level of emergency response and coping capacity whereas 1 represents a low level of emergency response and coping capacity. To assign weights for the indicators, the pairwise comparison method was used with the same decision-makers. The pairwise comparison matrix shown in Table 11 was developed based on their judgements. The pairwise matrix was obtained with a satisfactory level of consistency ratio (0.09) (Table 11).

168

Table 11: Pairwise comparison matrix of indicators of emergency response and coping capacity

DenWH DisMRD RDNet DenCFU DisFS DisEP DisHOS DisNW Weight /SWS

DisMRD 1 0.2 1.1 0.4 0.3 1.2 1.3 1.0 0.10 RDNet 5 3/5 1 0.4 0.8 1.2 1.7 0.7 0.7 0.13 DenCFU 1 2 1/2 1 0.8 0.8 1.7 0.7 1.2 0.13 DisFS 2 1/2 1 1/5 1 1/5 1 0.6 1.4 0.6 0.8 0.12 DisEP 3 5/9 5/6 1 2/7 1 3/5 1 2.5 0.5 0.5 0.13 DisHOS 5/6 3/5 3/5 2/3 2/5 1 0.4 0.5 0.07 DenWH/ 3/4 1 4/9 1 4/9 1 5/7 2 1/7 2 1/3 1 1.7 0.18 SWS

DisNW 1 1 4/9 5/6 1 1/5 2 2 1/7 3/5 1 0.14 Consistency ratio = 0.09 (<0.1), DisMRD = Distance to main road, RDNet = Road network (Road density), DenCFU = CFU density, DisFS = Distance to fire station, DisEP = Distance to an evacuation point, DisHOS = Distance to hospital, DENWH/SWS = Density of water hydrants/static water supply, DisNW = Distance to natural water source

Table 12 shows the indicators used to assess the level of response and coping capacity in the area, fuzzy evaluation criteria, and the criterion weights derived from the pairwise comparison matrix. Based on expert judgements, the highest weight is assigned to the density of fire hydrants and static water supply, and distance to natural water sources. Road density, CFU density, and distance to evacuation points were considered to be equally important factors. The least important factor is distance to a hospital. The resulting fuzzy criterion maps are shown in Figure 44 and Figure 45.

Table 12: Indicators of exposure and physical susceptibility - Fuzzy membership functions, evaluation criteria and weights

Derived Indicator Fuzzy membership Weight Density of fire hydrants/SWS Sigmoidal Monotonically Decreasing 0.18 Distance to natural water Sigmoidal Monotonically Increasing 0.14 Road density Sigmoidal Monotonically Decreasing 0.13 CFU density Sigmoidal Monotonically Decreasing 0.13 Distance to an evacuation point Sigmoidal Monotonically Increasing 0.13 Distance to fire station Sigmoidal Monotonically Increasing 0.12 Distance to main roads Sigmoidal Monotonically Increasing 0.10 Distance to hospital Sigmoidal Monotonically Increasing 0.07

169

Fuzzy criterion maps of emergency response and coping capacity are shown in Figure 44 and Figure 45. The maps were then aggregated using the Weighted Linear Combination method (Figure 46(a) and Figure 47(a)). The aggregated maps of the emergency response and recovery component of vulnerability were classified into five vulnerability groups based on the level of emergency and response capacity; extreme, high, moderate, low, and minimal using the quantile method (Figure 46(b) and Figure 47(b)). The result of the emergency response and coping capacity analysis was used as an input to assess the overall vulnerability of the area.

170

Density of fire hydrants/SWS Distance to natural water sources

Road density Density of CFUs

Figure 44: Fuzzy criterion maps of the indicators of emergency response and coping capacity in Ku-ring-gai. 171

Distance to an evacuation point Distance to the fire station

Distance to the main road Distance to the hospital

Figure 44 continued.

172

Distance to natural water bodies

Density of Fire hydrant and static water supply

Density of CFUs

Figure 45: Fuzzy criterion maps of the indicators of emergency response and coping capacity in the Blue Mountains. 173

Distance to the main road

Proximity to the hospital

Proximity to the fire station

Figure 45 continued.

174

Road density

Proximity to the evacuation point

Figure 45 continued.

175

Figure 46 shows the level of vulnerability based on emergency response and capacity in the Ku-ring-gai area. It shows the UBI from the north to northeast and the western edges is extremely vulnerable due to lack of emergency response and recovery capacities. The main reasons are lack of water hydrants and alternative water sources, an inadequate road network, lack of community level fire management activities and longer distances to emergency fire services.

a) Emergency Response and Coping b) Emergency Response and Coping Capacity Capacity Zonation

Figure 46: Emergency response and coping capacity maps - Ku-ring-gai.

Figure 47 depicts emergency response and coping capacity in the Blue Mountains. The results suggest that the eastern part of the Lower Mountains area and the Mid Mountains area are the most vulnerable. The potential causes are lack of fire hydrants and insufficient static water supply, the lack of a well-connected road network, and distance to other fire response services. It is also evident that vulnerability is minimal in the town areas compared to other residential areas. The Upper Mountains region shows a higher level of response and coping capacity. This reflects the local level fire management efforts and infrastructure available in the Upper Mountains region due to high concern about bushfire risk. 176

a) Emergency Response and Coping Capacity

b) Emergency Response and Coping Capacity Zonation

Figure 47: Emergency response and coping capacity maps - Blue Mountains.

6.6.4 Integrated Vulnerability

A key objective of this study is to develop an integrated vulnerability map combining the components of vulnerability. As discussed in the analytical framework, overall vulnerability is a combination of exposure and physical susceptibility, social vulnerability, and emergency response and recovery capacity. To integrate those 177 elements, OWA aggregation was applied. OWA allows for control of the trade-offs among the vulnerability components (Kandilioti & Makropoulos 2012). It includes factor weights and order weights. The component maps were first standardised using fuzzy criteria given in Table 14. In this aggregation 1 is considered to be high vulnerability whereas 0 is assigned to low vulnerability. To calculate the factor weights the same pairwise comparison approach was applied. A pairwise comparison matrix was obtained with a consistency ratio of 0.03 (Table 13). In this study an intermediate approach OWA (0.5,0.25,0.25), which assigns more weight to the highest rank and equal weight to the lower ones for each location in the study area, was applied (Kandilioti & Makropoulos 2012). The OWA module in IDRISI Selva was used to perform the analysis. In the MCE process, only the human settled areas were considered by using a constraint mask. This helps to analyse vulnerability in areas where human activities are present.

Table 13: Pairwise comparison matrix of vulnerability factors EXPHS SOC ERCC FW OW EXPHS 1 5.6 5.0 0.72 0.5 SOC 1/6 1 1.4 0.14 0.25 ERCC 1/5 2/3 1 0.14 0.25 Consistency Ratio = 0.03 (<0.1), FW = Factor Weights, OW= Order Weights EXPHS = Exposure and physical susceptibility, SOC = Social vulnerability, ERCC = Emergency response and coping capacity

Table 14: Factors of vulnerability - Fuzzy membership functions, evaluation criteria and weights Factor Order Vulnerability Factor Fuzzy Criteria Weight Weight Sigmoidal Monotonically Exposure and Physical Susceptibility 0.72 0.5 Increasing Sigmoidal Monotonically Social Vulnerability 0.14 0.25 Increasing Emergency Response and Coping Sigmoidal Monotonically 0.14 0.25 Capacity Increasing

178

The integrated vulnerability maps resulting from the MCE process are presented in Figure 48 and Figure 50. The maps were then classified into five categories; extreme, high, moderate, low and minimal using the quantile method. Classified vulnerability zonation maps are presented in Figure 49 and Figure 51.

Figure 48: Integrated vulnerability map - Ku-ring-gai.

179

Figure 49 shows the vulnerability zones for the UBI in the Ku-ring-gai area. It suggests that extremely vulnerable areas can be found at the northern, southeastern and western edges. Vulnerability component maps suggest that all components of vulnerability drive these areas into an extremely vulnerable situation. In other words, those edges are highly exposed and physically susceptible, socially vulnerable and emergency response and coping capacity in the area is not adequate.

Figure 49: Integrated vulnerability zonation map - Ku-ring-gai.

180

Figure 51 shows the classified overall vulnerability map for the Blue Mountains area. It reveals that the edges of the Upper Mountains and the Mid Mountains region are extremely vulnerable compared to the Lower Mountains region. In the Upper Mountains region, vulnerability is mainly driven by the high level of social vulnerability and exposure and physical susceptibility in the area. However, in the Mid Mountains region, the overall vulnerability is driven by exposure and physical susceptibility and the level of emergency response and coping capacities.

Figure 50: Integrated vulnerability map - Blue Mountains.

181

Figure 51: Integrated vulnerability zonation map - Blue Mountains.

6.7 Integrated Risk

In this research, risk is defined as a product of hazard and vulnerability. The equation Risk = Hazard x Vulnerability elaborates the relationship between hazard and vulnerability. The results of the bushfire hazard assessment and the integrated vulnerability assessment were used to generate the integrated risk map using the equation given above. First both the hazard and the integrated vulnerability maps were standardised using a sigmoidal monotonically increasing fuzzy membership function. Then the standardised layers were combined using a mathematical operation with map algebra. The resulting integrated risk maps are presented in Figure 52 and Figure 54. It shows the spatial distribution of bushfire risk in the area. Integrated risk maps were then classified using the quantile method to develop risk zonation maps (Figure 53 and Figure 55).

182

Figure 52: Integrated risk map - Ku-ring-gai.

The risk zonation map for the Ku-ring-gai area shown in Figure 53 demonstrates that the northern edge and the western, southwestern and southern edges are at high risk. Although the eastern edge showed a high level of vulnerability, the area becomes moderate/high risk because the area is less prone to bushfire hazard. The southeastern edge poses the lowest risk compared to the other edges in the area. Both vulnerability and the level of hazard are relatively low in the area.

183

Figure 53: Integrated risk zonation map - Ku-ring-gai.

184

Figure 54: Integrated risk map - Blue Mountains.

Figure 55: Integrated risk zonation map - Blue Mountains.

185

The risk zonation map of the Blue Mountains is presented in Figure 55. It reveals that the eastern and southern edges of the Lower Mountains, the southern edges of the Mid Mountains and the northern and the northwestern edges of the Upper Mountains regions are at high risk. The Lower Mountains area shows high risk mainly due to a high level of bushfire hazard. In the Mid and Upper mountains, the level of risk is mainly driven by the level of vulnerability as well as the level of hazard. It is evident that the main town centre area in the Upper Mountains (Katoomba) appears to be at minimal risk.

6.8 Conclusions

This study presents a GIS-based fuzzy MCE approach to assessing bushfire risk at the UBI based on the holistic risk framework discussed in Chapter 3. The proposed integrated framework characterises and quantifies the level of risk. In this study, bushfire risk was conceived as a product of bushfire hazard and vulnerability. An integrated approach was used to assess the level of vulnerability at the UBI. The components of vulnerability considered were exposure and physical susceptibility, social vulnerability, and response and coping capacity. Vulnerability factors were assessed by combining the vulnerability indicators using the WLC method. Integrated vulnerability assessment was performed by combining the components of vulnerability using the OWA method.

The findings of this study help to understand spatial variability of bushfire risk across the study areas. The resulting maps present relative levels of risk and vulnerability in the area. They also help to understand the factors affecting bushfire risk and the underlying causes of those factors.

The integrated vulnerability and risk maps can give fire managers, planners, insurers and emergency services a valuable tool to understand the level of bushfire risk in the area. Each of them needs to assess risk, vulnerability, and their underlying causes in order to make decisions to minimise potential losses. These vulnerability and risk maps are also useful for land use planning and local government councils who propose development projects in the area.

186

This study also reviewed the role of GIS-based MCE and then outlined the different MCE approaches available for decision-making. The MCE model used fuzzy evaluation techniques followed by WLC and OWA to combine the identified indicators and components of vulnerability. This study also used pairwise comparison to calculate weights. These techniques allow vulnerability indicators and components to be integrated more effectively while including subjective judgements into the modelling process.

The results revealed that the fuzzy MCE and OWA approach used in this study is well suited to understanding the spatial variation of bushfire vulnerability and risk at the local council level. The model provides a cost effective and comprehensive method for assessing bushfire risk with limited data, time, knowledge and experts.

Although the proposed model provides an avenue for developing bushfire risk profiles at the local level, there are a number of challenges for its implementation. One of the challenges is the lack of grid-based population and socio-economic data. In this research, CCD boundaries were used to map social vulnerability. Recently a number of initiatives have been undertaken to develop global gridded population datasets such as the LandScan population grid. However, the resolution of this population grid dataset is considerably coarser. High-resolution grid data are crucial for spatial vulnerability assessment at the local or community level.

In this study, the heterogeneity of bushfire events was not taken into account. The bushfire hazard assessment was mainly based on previous bushfire occurrence data. However, in reality the level of bushfire hazard varies according to fire behaviour. This could be overcome by incorporating bushfire simulation models. Bushfire hazard maps can be generated using previous bushfire simulation results. Such initiatives also help fire managers to predict the level of potential risk.

Vulnerability analysis requires input from and on-going collaboration with a range of bushfire management stakeholders, and it may be challenging to engage and maintain the proper individuals to sustain the momentum of the process. There might not have been adequate expertise or resources to understand the process of collecting and 187 integrating data. Data are often not freely available and there is not always proper spatial data infrastructure available. Data belongs to different agencies and different agreements need to be processed in order to access data. Although other data belongs to national agencies, it is also necessary to compile and ground truth this information with local stakeholders. It is highly recommended to have a framework for Spatial Data Infrastructure (SDI) for spatial data, metadata, users and tools that are interactively connected in order to use spatial data for bushfire management activities in an efficient and flexible way.

6.9 Summary

This chapter highlighted the importance of understanding levels of bushfire risk in the bushfire management process. The results revealed different zones of bushfire vulnerability and risk based on the selected criteria. It provided insight into the interaction between hazard and vulnerability in the area. It further provides information about the underlying causes of vulnerability such as exposure and physical susceptibility, social vulnerability and emergency response and coping capacities. However, the proposed model does not capture the decision-making processes of residents. Individual decision-making is mainly determined by personal judgements. For effective bushfire management a combination of risk estimation as well as understanding the perceptions and decision-making of residents is important (Cardona et al. 2012). The following chapter analyses the perceptions of people living at the UBI in the context of bushfire preparedness, response, and recovery.

188

Chapter Seven: Understanding Communities at the Urban Bush Interface for Bushfire Preparation, Response and Recovery

7.1 Introduction

The Australian UBI has been under bushfire threat since settlements were established. Bushfires are considered an inevitable part of Australian landscapes that result in substantial loss of properties and lives (McGee & Russell 2003). As the population living at the UBI has increased with trends in population growth and migration, UBIs are becoming more complex, with increased human-environment interaction (Cottrell & King 2007). Climate change may be causing more fire weather days, producing frequent bushfires that have a devastating impact on the UBI (Hennessy et al. 2005). Such fire events are increasingly overwhelming emergency responders (e.g., the Black Saturday fires in 2009).

It has been widely recognized that fire management in the landscape is not the only solution to minimizing the impact of fires, and that preparedness and emergency response activities need to be considered to be equally important (Beringer 2000). The interaction between fire management authorities and communities also remains a crucial factor that successfully minimises substantial losses to infrastructure and property. There have been numerous efforts to encourage household and community level preparedness. Given that much effort and many resources are being utilized for fire management, it is important to understand whether these initiatives have made any impact on residents’ levels of preparedness. This will also help to identify deficiencies in the current bushfire management framework, levels of household and community preparedness and factors influencing household preparedness decisions. A clear understanding of household and community level preparedness and related issues is an important step in any bushfire management practice. Such information helps to develop strategies to encourage residents who are more vulnerable and less prepared to take actions to reduce their risk of suffering severe impacts.

In Australia research on public perceptions of bushfire management strategies to reduce bushfire risk is rare. Available studies are generally focused on a particular region or 189 particular bushfire management strategy (Bushnell et al. 2007; Cottrell et al. 2008; Lowe et al. 2008). Household and community level risk management and preparedness activities are often influenced by social, cultural and institutional processes (Bushnell et al. 2007; Miceli et al. 2008; Paton et al. 2000). Risk reduction behaviour is a function of local level characteristics such as past experience, coping capacity and community norms (Kruger et al. 2002; Paton et al. 2000). Socio-demographic and economic characteristics of the household may also influence risk perception and preparedness decisions (Miceli et al. 2008; Paton et al. 2000; Sjöberg 2000). Therefore, people from one locality may not perceive risk in the same way as those of others, and even within the same community, people perceive risk differently (Paton et al. 2000). Simple extrapolation of results from previous studies into the present is difficult and sometimes wrong because of the differences in physical environmental context, the type of risk that is posed, different personal characteristics of the residents, and risk reduction measures (Miceli et al. 2008). Understanding local residents living at the UBI, the factors affecting their decisions and their perceptions of bushfire risk, and their attitudes towards current bushfire management strategies is an important step towards effective bushfire risk management. Such knowledge will facilitate improvements in existing risk communication and risk reduction strategies (Cottrell et al. 2008; Paveglio et al. 2009).

This chapter investigates household perceptions and views of bushfire risk and attitudes towards current bushfire management strategies in two bushfire-prone UBIs; Blue Mountains City Council area and Ku-ring-gai local government council area in New South Wales. It further investigates the association between levels of preparedness and selected variables that explain the socioeconomic status of residents, community characteristics, bushfire management activities and previous bushfire experience. Overall, this chapter provides insight into community level bushfire management issues at the UBI. It discusses community perceptions of current bushfire management practices, what the public expects from bushfire management authorities and household and community-level engagement in bushfire management activities in their communities. It further identifies factors that can influence perceptions of bushfire risk and other local level bushfire management issues. The results can be used to understand the relationship between bushfire preparedness and social vulnerability components discussed in Chapter 5. It further helps to verify whether current bushfire management 190 practices have met the actual expectations of residents (Cortner et al. 1990). Finally, this chapter discusses how understanding bushfire risk perceptions and other community level issues can be used to improve specific strategies to minimise bushfire risk.

7.2 Treating Bushfire Risk at the UBI

Treating risk is an important component of the risk management process (AS/NZS 2009; Asian Disaster Reduction Center 2005). As discussed in Chapter 3, risk treatments are essential when the level of risk remains intolerable. According to Durham (2003), the main purpose of treating risk is to reduce the likelihood of harm to the community and the environment. This can be done by selecting and implementing appropriate risk treatment options.

Risk treatment involves identifying risk reduction measures for treating priority risks. It includes prevention, preparedness, response, and recovery, selecting options, planning and implementing risk reduction strategies (Asian Disaster Reduction Center 2005; De Guzman 2003). It should address a wide range of issues such as social, economic, institutional, political and legal options and the expectations of the people living at the UBI (Cortner et al. 1990; Durham 2003). Community level risk management is well recognized as the most effective approach to disaster risk management as it involves all level of governments, decision makers and the community (Allen 2006). It is important to have community engagement and support in order to make the risk treatment options more effective (Durham 2003). Although the disaster risk management paradigm has shifted from post-disaster response and recovery measures to pre-disaster prevention and preparedness measures through proper risk identification and treatment, the approach to disaster management in Australia has still been oriented more towards response and reaction (Cronstedt 2002; Gabriel 2009). Community level prevention and preparedness measures are still at the initial stages and are not as developed as they could be.

In Australian risk management, bushfire risk is understood as a product of probability of occurrence and consequences (Bush Fire Coordinating Committee 2008; Shields &

191

Tolhurst 2003; Tolhurst et al. 2008). Probability is a difficult concept for the public to understand. It is technical in nature and is a tool utilized by experts for risk identification. Public judgments of risk are often based on consequences as they are easy to understand and less complex (Bushnell et al. 2007; Sjoberg 1999; Smith 2001). As discussed in Chapter 2, researchers are yet to agree upon a single definition of risk and people differ in how they use the term risk. This ambiguous nature of risk can influence risk perception and demands for risk reduction.

Understanding these different perceptions, including public views and underlying issues, beliefs, access to resources, informal social interactions, fire safety knowledge, previous experience and attitudes towards community level fire prevention and maintenance programs can illustrate differences in the social construction of bushfire risk within the UBI, and can provide a great deal of information to bushfire management authorities in order to take actions to enhance current bushfire risk management activities.

7.3 Social Construction of Bushfire Risk

Social constructionism is a useful concept that has been applied to various social research projects (Hacking 1999). It explains real environmental and social problems and the way that people deal with those problems (Cottrell et al. 2008). Often the social adjustments that help to solve problems result from individual and community judgements, beliefs and attitudes (Tobin & Montz 1997). Cannon (2008) has stated that disasters are socially constructed because they are largely a product of processes involving economic, political and social factors. The relationships between these factors make the adjustment process complex. People understand risk based on their past experiences, beliefs and their relationships with others, agency involvement, human capacity and social capacity (McGee & Russell 2003; Prior 2010). Their social responses to prepare for hazards are influenced by these individual, community and historical factors (Paton et al. 2006; Prior 2010). Social response refers to “the actions that improve decision making, organization, management and planning that help communities to assess, support and choose among different approaches to wildfire management” (Paveglio et al. 2009, p. 1087). The level of social response is determined

192 by different social constructions of risk, and results in differences in the effectiveness of different social responses within the UBI (Paveglio et al. 2009).

In an emergency people often act based upon their own evaluation of risk, which is different from that of researchers and experts (Cottrell et al. 2008; Slovic 2000). Public views, presumptions and attitudes may put communities at great risk if the risk is misunderstood. Some people underestimate risk and are over-confident about their safety. “People who underestimate the level of risk will be likely to expose themselves to the risk to a greater extent than a person with more realistic perceptions thereby becoming more vulnerable” (Bushnell & Cottrell 2007b, p. 7). Bushfire hazard researchers have identified individual (perceived responsibility), social (sense of community and trust) and situational (timing) factors that influence the risk reduction decision making process (Cottrell et al. 2008; Rhodes 2011). Often those factors are locally based issues and unique to a particular community (Cottrell et al. 2008). Community level risk reduction measures can be built through understanding and addressing these local issues while promoting community based initiatives.

The term community is used frequently yet has a wide range of meanings. A community is usually referred to as a group of people living in one geographical area (Schellong 2007). A community has number of things in common, which may be defined by location, interests, or function (Bushnell & Cottrell 2007b). Within a community people often interact together, share things and hold values, norms and goals in common (Fitzgerald & Fitzgerald 2005). According to Bushnell & Cottrell (2007b), a person can belong to a number of different communities, and there can be a number of different overlapping communities within a single location. Common bases for community formation and cohesion are physical proximity, lifestyle and values, social class, work, ethnicity, age, gender, length of residence and religion (Fitzgerald & Fitzgerald 2005). Based on a survey in Victoria, Australia, Pope (2010, p. 4) has noted that “an ideal community would have good local facilities and services, friendly and helpful people, a pleasant environment, opportunities to participate, and government that was working with communities to build a secure future (safety, jobs, responsiveness to local needs).” Unfortunately, in reality it is hard to find such ideal communities in the context of bushfire management. Local governments are not able to meet the demands that the 193 community needs to prepare for bushfires. Community participation is often weakened by the busy lifestyles of community members. With the increasing heterogeneity among populations, community networks and relationships are becoming more complex and there can be people within a community who do not work well together, get on, or who do not share similar values (Buckle et al. 2001).

Although many bushfire management activities have been adopted at the state and local council level, only a few studies have examined social complexities and their influence on bushfire risk reduction measures at the community level. Most bushfire management interventions are designed to address risk in terms of potential loss and the value of the assets at risk, but they often fail to acknowledge the social construction of risk. This chapter provides insight into the attitudes and perceptions of residents living at the UBI. It also explores the household and community characteristics that influence household and community level preparedness and community perceptions of bushfire preparedness. To achieve the research aims, a household survey and in-depth semi- structured interviews were undertaken with residents. The results of this survey shed light on community and household level behaviours and perceptions of bushfire management. They can be then used to improve current management policies and preparedness programs.

7.4 What Makes Communities More Vulnerable and Less Resilient?

The degree of hazard impact in a given community at risk is mainly shaped by its level of vulnerability and resilience (Buckle et al. 2001). In general, vulnerability refers to situations where people experience susceptibility to hazards with inadequate capacity to cope and adapt to possible impacts (Adger 2006; Birkmann 2006b; Whittaker 2008). According to Wisner et al. (2004, p. 11) vulnerability is “the characteristics of a person or group in terms of their capacity to anticipate, cope with, resist and recover from the impact of a natural hazard.” Resilience is often seen as the capacity of a system to absorb disturbance and re-organize while undergoing changes, yet retain the same function, structure, identity and feedbacks (Berkes 2007; UNISDR 2004; Walker et al. 2004). According to Fitzgerald & Fitzgerald (2005), resilience has four components: reduction (mitigation), readiness (preparedness), response, and recovery. Resilience and

194 vulnerability have been defined as different sides of the same coin. Cannon (2000), cites Granger’s (1997) view as being that the “vulnerability of each element at risk within the community can be measured along a continuum from total resilience at one end to total susceptibility at the other.” Buckle et al. (2001), however, suggested that they need not necessarily be conceptualized as opposite ends of a continuum. Buckle (2006, p. 91) further explained that “resilience and vulnerability do not cancel each other out to arrive at a neutral state. They complement rather than confront each other.” Finally, Paton & Johnston (2006) conceptualized vulnerability as losses and resilience as gains. They further explained that these complementary factors have actions that are independent from each other and that can enhance adaptive capacity. This chapter does not attempt to address the issues of resilience, but tries to understand the factors affecting the level of preparedness that may also be related to resilience.

Certain individual, household and social characteristics influence both levels of vulnerability and resilience. These characteristics represent the heterogeneous nature of a community. Paton & Johnston (2006) have explained these relationships in the figure given below (Figure 56).

Figure 56: Hazard and risk relationship (Paton & Johnston 2006).

195

According to Figure 56, personal, community, institutional and environmental factors influence vulnerability and resilience. On the other hand, vulnerability and resilience influence the level of preparedness, response and recovery. Therefore individual, community, institutional and environmental factors and the relationships between those factors and preparedness, response and recovery need to be better understood to facilitate better risk management (Paton & Johnston 2001; Paton & Johnston 2006). Risk management initiatives should focus on promoting resilience and preparedness through various strategies including communication, managing vulnerability and facilitating resilience and growth (Paton & Johnston 2001; Paton & Johnston 2006).

Cannon (2000) identified five components of vulnerability; social protection, social capital, initial well-being, self-protection, and livelihood resilience. Initial well-being, self-protection, and livelihood resilience are mostly determined by socio-demographic characteristics whereas social protection and social capital are determined by the socio- political structure and community characteristics that enable community level relationships, such as social networks, trust and bonds (Whittaker 2008). In the Pressure and Release (PAR) model (Blaikie et al. 1994), social relations, public actions and institutions are identified as important components that determine the level of vulnerability. An appropriate approach to understanding community vulnerability is to consider individual, household and social characteristics that influence community vulnerability and resilience so that strategies can be recommended to decrease the vulnerability of a given community.

In order to achieve that, researchers have employed various methods to identify these individual, household and social characteristics. Individual factors that influence preparedness for natural hazards have been identified by various research activities (Tierney 2001). The concept of social vulnerability is widely used as a measure of identifying vulnerability and resilience based on the individual socio-demographic characteristics of populations (Clark et al. 1998; Cutter et al. 2008; Cutter et al. 2003; Wood et al. 2010). It describes the specific social inequalities and unequal exposure of different groups to natural hazards (Kuhlicke et al. 2011). Research on social vulnerability indicates that the elderly, females, disabled people, single parent families, renters, low-income people, unemployed people and people with low education are 196 associated with high levels of social vulnerability (Clark et al. 1998; Cox et al. 2007; Cutter et al. 2003; Fekete 2009; Wood et al. 2010). The concept of social vulnerability is mostly operationalized using indicator-based approaches (Clark et al. 1998; Cutter et al. 2003; Fekete 2009; Wood et al. 2010). Those studies rely heavily on available census data and quantitative statistical techniques. In Chapter 5, social vulnerability at the UBI and related issues were discussed. However, while social vulnerability analysis gives an insight into a socio-demographic profile based on the individual characteristics of residents, social vulnerability often does not focus on the social cognitive process of vulnerability because those characteristics are rather difficult to capture using only census data. Only in-depth household and community level analysis will provide such information.

While individual characteristics have been identified as a key factor that determines vulnerable communities, the social component of vulnerability and resilience also plays a key role in disaster preparedness and emergency management. Social components are often qualitative in nature and hard to capture. At the UBI differences can be seen in both quantitative demographic elements and less tangible qualities related to social context, such as informal knowledge networks, place-based knowledge and experience, and scientific or technical knowledge (Paveglio et al. 2009).

7.5 Social Factors Shape the Level of Vulnerability and Resilience to Bushfires

In the context of disaster management, many factors and processes contribute to vulnerability and resilience and their relationship with disaster preparedness. Some of these factors are intrinsic aspects of culture and demography whereas others reflect personal dispositions, societal and institutional characteristics and processes (Jakes et al. 2002; McGee & Russell 2003; Miceli et al. 2008; Paton & Johnston 2006; Paveglio et al. 2009).

There have been a few attempts to identify the vulnerability of people and communities to bushfires in terms of their level of preparedness (Bushnell & Cottrell 2007a; Cottrell & King 2007; McGee & Russell 2003). Within the same community some people, some groups and some areas are at greater risk than others, and some groups may be more

197 vulnerable and less resilient than others. A number of studies have demonstrated that variability in knowledge, demographics and the nature of social networks make UBI communities different from each other (Paveglio et al. 2009). Understanding this issue requires knowledge of the embedded and often unrecognised drivers of contemporary social, economic, political and environmental existence (Buckle et al. 2001). Understanding the relationship between these drivers and preparedness, response and recovery measures is an area that is under-researched. In order to develop sustainable, fire resilient communities and community self-sufficiency for fire safety, it is important to recognize what makes communities vulnerable to bushfires, and how socioeconomic, social and institutional factors like economic well-being, access to resources and participation in social networks contribute to bushfire resilience.

Paton et al. (2006) described preparedness as a process that involves people making judgments regarding the relationship between themselves, the hazards present in their environment, and action available to mitigate the attendant risk. According to the literature, there are two major types of factors that contribute to household and community level bushfire preparedness. Those factors can be identified as cognitive factors that drive the individual or household decision-making process and the social factors that determine the household level as well as community level collective decision-making process. In this chapter the main focus was to identify these social factors.

Over the years, researchers have identified different social factors that affect a community’s level of disaster preparedness. Jakes et al. (2002) stated that community level resources and community level decision-making enhance bushfire preparedness. In their capital-based approach, they identified that a high level of community preparedness is supported by human capital, social capital, cultural capital, and agency involvement with a strong social foundation that enables community networks to share social responsibilities in order to have a more resilient community (Jakes et al. 2002; Winkworth et al. 2009). Social capital is defined as community characteristics such as leadership, trust, norms, networks and flows of information and mobilization of resources that facilitate collective social actions (Adger 2003; Jakes et al. 2002; Mayunga 2007). Human capital includes knowledge and skill among individuals about 198 hazards, hazard history, and hazard risk that is obtained through education and training. Cultural capital includes knowledge and experience that is accumulated through disaster experience, heritage and place attachment (Adger 2003; Jakes et al. 2002; Mayunga 2007). Agency involvement is identified as the process of agencies working towards a common goal (e.g., bushfire management in the area) (Adger 2003; Jakes et al. 2002; Mayunga 2007).

Paveglio et al (2009) describe these social factors as a mosaic of complex settlement patterns with overlapping resource problems and capacities. The identified elements are; demographic and structural characteristics, place-based knowledge and experience, informal interactions or relationships among residents and access to scientific and technical knowledge networks. Fire experience may be a collective memory; it may also explain behaviour and attitudes towards others during a time of disaster (Cottrell 2009).

7.5.1 Perception of Bushfire Risk

The perception of risk is understood as “the judgments people make when they are asked to characterize and evaluate hazardous activities” (Slovic 1987, p. 236). Different individuals, households and groups perceive risk and related issues differently from the scientific community (Bushnell & Cottrell 2007b; Paton 2006c; Sjöberg 1998; Sjöberg 2000). This individual and social perception of risk is identified as an important component in risk management as it triggers the social cognitive process of risk reduction (Li 2009; Paton et al. 2000; Slovic 2000; Tobin & Montz 1997). Risk perception is also a dynamic process (Paton et al. 2000; Sjoberg 1999; Sjöberg 2000). Previous risk perception research has only explained a small fraction of its variation (Sjöberg 2000).

People often have varying perceptions of the risks they face and the degree to which they are vulnerable to bushfire risk. Researchers have identified several factors that influence people’s perceptions. These include past experience of bushfires; demographic differences including gender, age, education, occupation, ethnic background (including ability to speak English); level of literacy, which determines whether they can understand messages communicated to them; whether people belong

199 to local organisations; whether residents have a realistic view of their situation and degree of isolation from other people, services and communications, within a local area (Buckle et al. 2001; Pearce 2000).

In the context of bushfires, risk perception can be extremely subjective (McCaffrey 2004). Within the bushfire research community, several individual and community characteristics that influence how people interpret the hazardous circumstances that could prevail in their community have been identified (Heath et al. 2011; Paton 2006b). Paton et al. (2006) have demonstrated that preparedness is a function of how people interpret their relationship with the hazardous aspects of their environment. Risk perception is linked to preparedness as risk reduction actions are determined by people’s perception of the relative importance of hazard and people’s interpretation of their ability to take action to increase their safety (McCaffrey 2004; Paton et al. 2006). Perceptions of risks are sometimes culturally and socially constructed. Social groups, their behaviour and attitudes are complex, and often they construct different meanings than individuals (McIvor & Paton 2007; Tobin & Montz 1997). For example, individuals who migrated recently may evaluate the level of bushfire risk at the UBI totally differently than community fire unit members who are working on local level fire management initiatives. Therefore, it is important to identify these social and cultural influences before developing a risk management strategy. Then these influences can be accommodated to create an effective risk management strategy (McIvor & Paton 2007).

7.5.2 Risk Knowledge, Awareness

Knowledge and awareness of risks that people are exposed to are important influences on individual and community perceptions, attitudes and actions towards bushfire preparedness and mitigation (Enders 2001; Ryan & Wamsley 2008). They are also essential for developing and maintaining capacity to minimise fire impacts (Buckle 2006). A high level of knowledge and awareness about the bushfire management process, including the physical processes involved in a bushfire event, increases the use of fire prevention measures and faith in survival strategies (Beringer 2000). Knowledge

200 and awareness are also identified as individual and community characteristics that affect the level of resilience and vulnerability to bushfires (Paton & Johnston 2006).

People differ in their prior knowledge, beliefs and awareness and also the way that acquire information about hazard activities (Paton et al. 2000; Ryan & Wamsley 2008). This knowledge can be gained through their own experience, media, and materials from disaster management authorities, and community level education programmes. Many people moving into the UBI are urban residents who have not been in a bushfire prone area before. Therefore they have limited awareness of bushfire risk (Cottrell & King 2007). Poor understanding of the physical environment in which people are living and low awareness of bushfire hazards and their consequences make people more vulnerable (Cortner et al. 1990). Awareness and knowledge about hazards are related to different household level characteristics such as length of residence, age and proximity to the hazard area (Hohenemser et al. 1983). Bushnell & Cottrell (2007b) have explained that good knowledge of bushfire is related to past bushfire experience, close association with local fire brigades and community cohesiveness. Being in the same neighbourhood for a long time will help people to understand the characteristics of the physical and social environment, previous hazards, as well as emergency communication channels.

Although occupation is not often recognised as a vital component of awareness, it impacts the level of bushfire awareness. People who engage with services that directly deal with fire management issues are more aware of bushfire preparedness issues than others. Community level volunteer activities are widely recognized as a valid model that helps to increase people’s knowledge and awareness about their local level hazards. Close association with bushfire management voluntary groups such as the local fire brigade increases the level of knowledge and awareness about fire history, preparedness actions and response measures (Bushnell & Cottrell 2007b). Knowledge and awareness are interlinked with other factors in the disaster preparedness process and can influence the effectiveness of risk communication.

201

7.5.3 Risk Communication

Risk communication involves translation of scientific knowledge into useful constructs or concepts that non-technical audiences can understand (Glik 2007). Risk communication is an important component of risk management that should take place during all stages of the risk management process (AS/NZS 2009). It is a non-structural risk reduction process that enables risk experts and concerned citizens to participate in discussion, share information and opinions, and then build partnerships (Okada & Matsuda 2005). During the process of risk communication, issues such as the risk itself, its causes, its consequences and the measures being taken to treat it need to be addressed.

Risk communication should be a two-way exchange process. Information regarding local hazards, risk and hazard management issues needs to be shared with the community and the results of risk assessments need to be shared with decision-makers and policy makers. Feedback in both directions is important to monitoring and evaluating the risk communication process (Pearce 2000). Thus, multi-lateral communication between stakeholders, researchers and communities is required to understand the basis on which decisions are made, and the reasons why particular actions are required (AS/NZS 2009; Okada & Matsuda 2005).

Trust is an important component of an effective communication process. Level of trust can be affected by factors such as inadequate information sharing and dissemination, beliefs, and incomplete and inconsistent information (Paton 2006c). Therefore providing the right information to the right people at the right time will encourage community preparedness and response actions.

In the context of disaster management, most of the research has focused on disaster warnings that are delivered during an event (Pearce 2000). Effective communication at every phase of the disaster management cycle is critical in order to take sustainable risk reduction measures. Risk knowledge and awareness is one parameter that explains the effectiveness of risk communication, providing people with information on hazards and their potential consequences. It includes pre-event preparedness and safety messages,

202 emergency warnings, post-event communications and advice that minimizes potential losses (Bushfire CRC 2011). It enhances the sense of partnership between residents, neighbours, local governments and fire management authorities in terms of bushfire management and response (Lowe 2008).

Risk communication plays an important role in making individuals or households aware of the level of bushfire risk as well as mitigation and preparedness decisions in order to take actions well in advance of the bushfire season (Eriksen & Prior 2011). It should provide meaningful and understandable information about the tools and skills that can be used by an individual, family, or community to better prepare, given their particular circumstances. Effective risk communication can be achieved through community engagement, promoting positive preparedness outcomes, expectancy, attitudes and information dissemination and sharing (Prior et al. 2008).

Communication during fire emergencies is recognised as a major problem facing communities (McLeod 2003; Victorian Bushfires Royal Commission 2009). Information about the nature and propagation of the fire and safe evacuation routes is essential in order for community members to prepare themselves for a safe evacuation (Beringer 2000). Frequency of information dissemination is another component that is important in the context of risk communication and preparedness (Pearce 2000). Yamamoto & Quarantelli (1982) (cited in Pearce 2000) have stated that the more often people obtain information; (1) the more they trust disaster predictions; (2) the more they prepare for potential disaster situations; (3) the stronger their anxieties are; (4) the stronger their desires to move are; and (5) the more severe damage they predict.

It is been widely accepted that two-way communication between fire agencies and communities is much more effective for bushfire mitigation and management than a top-down approach (Okada & Matsuda 2005; Prior 2010). Some communities have been actively engaged with such management programmes. In Australia several community level bushfire management programs have adopted a collaborative approach where residents are actively involved in bushfire mitigation and preparedness in their neighbourhoods. These programs bring community members together to learn about local bushfire risk, develop strategies for reducing risk, and to implement these 203 measures. Importantly, such programs have the potential to build resilience through building the capacity of communities to anticipate and respond to a disaster and thereby lead to sustainable bushfire mitigation (McGee 2011).

Poor risk communication and lack of knowledge and awareness results in a significant proportion of the population in many communities failing to respond appropriately or adequately to fire weather and fire emergency warnings and ultimately leads to significant loss (Bushfire CRC 2011). Communities at the UBI continue to demonstrate lack of knowledge and preparedness for bushfire hazards despite the fact that fire management authorities have made efforts to encourage people to take adequate preparedness measure before bushfires (Paton 2006b). Failure to prepare before a bushfire strikes may result in a catastrophic experience. The main reason that people give for not preparing for potential fire is that they simply do not perceive they are at risk (Beringer 2000). The role of risk communication is important in reinforcing people’s perceptions and attitudes about bushfire management activities in their communities. Therefore, it is important to understand issues related to risk communication and bushfire preparedness.

7.5.4 Experience and Local Environment

Often direct or indirect experience with a natural hazard is seen as an important element in influencing a person’s perception of a hazard, and the mitigation, response and recovery actions they undertake (McCaffrey 2004; Tobin & Montz 1997). Personal experience of bushfire activity in the local environment as a resident, a fire fighter or other type of fire responder is a factor that influences perceptions of bushfire risk and attitudes towards the risk management process (McGee & Russell 2003; Paton 2006b; Ryan & Wamsley 2008). Past experiences often keep people motivated and provide greater understanding of the need to prepare. However, sometimes decisions that are made based on a single experience can cause misunderstanding of the real potential impacts. Therefore more frequent experience generally increases a person’s capacity to make a realistic assessment of potential impacts and the appropriate mitigation measures to take to deal with that level of risk (McCaffrey 2004). Physical location also remains as an important component of vulnerability to bushfire as it defines the level of

204 exposure to bushfire risk (Cottrell 2005). People with little understanding of the fire hazard in the local environment are vulnerable compared to those who have experience (Cortner et al. 1990). A lack of experience in the local area may also make it more difficult to implement risk reduction strategies as well as response measures (Harris et al. 2011).

Research has also revealed that prior experience or lack thereof has been linked to both increased and decreased levels of preparedness (Paton 2006c). People who live in the area for a long time often become a central part of the community, and this may make them more inclined to believe that they have a responsibility to prepare their neighbourhood. Beringer (2000) identified differences in bushfire perception and preparedness between residents who recently moved to the area and long-term residents. He further explained that residents who recently moved do not understand their actual level of exposure until they assess the level of bushfires in the area.

The existence of community level bushfire management activities such as experience sharing also increases individual and community awareness of fires. It also helps to identify community level leadership, which is an important factor when initiating, coordinating and organizing community level preparedness activities with the local authorities. Greater experience of fires means people are more likely to be prepared for the next event by acquiring the resources they need to be prepared (Goodman & Proudley 2008). Tobin & Montz (1997, p. 157) identified recency as a characteristic of experience; “more recent the experience, the greater the awareness of the hazard”. The magnitude and frequency of an experience also affects how people perceive and respond (Tobin & Montz 1997). This further highlights the importance of risk communication to keep the memories of those who have experienced bushfires alive in order to motivate them on bushfire management activities. They also noted that more memorable and more severe the experience, the more accurate the perception of risk, the more adequate of response to the hazard and the better the ability to recover from that hazard.

205

7.5.5 Community Strength

In the context of disaster preparedness, a community’s strength is defined by its social structure and function. Some researchers have identified this characteristic as community capacity (Goodman et al. 1998). It is characterised by high levels of social capital, sense of community, leadership and community participation (Cottrell 2009; Goodman et al. 1998; McGee & Russell 2003; Tobin & Montz 1997). Disasters are socially constructed events that can be observed in time and space. These events have impacts on both communities and individuals (Cannon 2008; Schellong 2007; Weichselgartner 2001). Community strength is an important component of local level bushfire management, and facilitates productivity and coordinated actions that contribute to minimising potential losses through preparedness, emergency response and recovery measures (McGee & Russell 2003; Schellong 2007). Community capacity is required for effective community-based fire management initiatives.

The concept of social capital has been gaining a wide interest among disaster risk researchers and policy makers as it helps to understand the role of important community characteristics in the production of disaster management. Definitions of social capital have a common emphasis on social structure, trust, norms and social networks that facilitate collective decision-making (Mayunga 2007). Putnam (2001) (cited inWestern et al. 2005, p. 1097) described social capital as “connections among individuals-social networks and the norms of reciprocity and trustworthiness that arise from them.” Cohesive and strong communities are characterised by high levels of social capital that reflect the quality of social cooperation (Mayunga 2007; Western et al. 2005). Pelling & High (2005, p. 317) have stated that “multi-layered and multi-faceted social ties of everyday social interaction may be a community’s best resource in maintaining a capacity to change collective direction.” Maintaining associations with wider society through close groups encourages social trust and interaction, enhances the flow of information, building equality and collective actions that reduce exposure and increases the capacity of group members to deal with hazards (Adger 2003; Pelling & High 2005). These cohesive networks and interactions, which are governed by norms, trust and reciprocity, are beneficial in collective decision making processes that allow people to resolve their problems in disaster planning and the adjustment process (Adger 2003).

206

Therefore enhancing social capital is seen as an important task in risk management (Lowe 2008).

A sense of community is characterized by high concern for community issues, respect for and service to others, and a sense of connection (Cutter et al. 2010). It is also identified as societal commitment (Bushnell & Cottrell 2007b). Sense of community is directly related to social cohesion, including mutual concerns and shared values (Cutter et al. 2010; Norris et al. 2008). “Strong sense of community tends to be well networked and communicate well on issues such as bushfire preparedness. Similarly, a well- connected community, which shares and discusses concerns about bushfires, tends to be better prepared. Some communities also have a ‘culture of preparedness’ which also contributes to capacity to deal with bushfires” (Cottrell 2009, p. 15).

Bushnell & Cottrell (2007b) found that people who have a strong sense of community are generally well prepared and are in a better position than others to deal with fires. Sense of community is often displayed by providing social support, tangible assistance, information, or care for community members when needed (Goodman et al. 1998). People feel connected to their community through various means such as church, schools, community organizations and cultural organizations (Goodman et al. 1998). McMillan & Chavis (1986) (cited in Goodman et al. 1998, p. 269) identified four elements that characterise a strong sense of community (1) membership, or a feeling of belonging; (2) influence, or a feeling that the individual and community matters; (3) fulfilment of needs, or a feeling that members' needs will be met by resources received through membership; and (4) emotional connection, or the belief that members share common experiences and history. It is evident that some of these variables are described by the construct of social capital. It is important to identify the influence of these variables on a community’s preparedness to take constructive actions to mitigate the effects of bushfires.

Leadership is one of the most effective elements of community capacity in enhancing collective actions in the disaster management process (Goodman et al. 1998; Nakagawa & Shaw 2004). Lang et al. (2006) identified the importance of community level leadership in the context of bushfire management. Leaders are important because they 207 understand the community and are able to encourage mitigation and preparedness in a number of ways including: Helping to identify important local issues and create a vision for action. Developing a preparedness strategy that takes community members’ goals into account. Obtaining commitment to act by communicating with other residents and building one-on-one relationships. Mobilizing financial and material resources.

In bushfire management, the credibility of community leaders is often derived from their local knowledge, experience and their ability to harmonise bushfire mitigation actions with the community’s needs and concerns (Paton 2006b). Community leadership is an important first link in establishing ties between community groups and local fire management agencies, and identifying key preparedness and mitigation issues to implement in community level fire management programmes. Therefore it is important to identify community level leadership, provide training to improve leaders’ skills, and reward commitment by sharing ownership or providing funding for future efforts (Lang et al. 2006).

For better results, leadership needs to have a strong base of actively involved residents (Goodman et al. 1998). This community participation ensures people’s active involvement in their community. It will be more effective in planning for a sensible and sustainable community fire management plan suitable for the community. Paton (2006b) identified the benefits of community participation, including acquiring new information from community discussions, learning new skills (e.g., fire fighting skills), being involved with important issues, making contacts, personal recognition, and developing a sense of community by contributing to the community. All these benefits help people to identify their own issues in undertaking actions for better preparedness.

7.5.6 Available Resources

To make disaster management successful, adequate resources need to be available at the household and community levels. Access to resources is an important factor that determines the impacts of a natural event (Morrow 1999). Allen (2006) found that 208 community members clearly experience different degrees of access to community resources, depending on social status and particularly the social capital provided by their household networks. Resources can be financial, natural, human or technical.

Financial resources play an important role in household level hazard mitigation. They include household income, savings, credit, insurance and other financial investments, and improve a household’s ability and capacity for preparedness, response and recovery (Mayunga 2007). Before implementing any community level fire management programme, funding sources need to be identified. For community level bushfire management, funds come through local governments and other fire management authorities. According to Ku-ring-gai Council (2012), local councils allocate substantial resources to minimize the risk to residents through programs such as: Fire trail and walking track maintenance Fire break maintenance in high fire risk areas Community education Support for the Ku-ring-gai Bushfire Brigade Development controls

Natural resources such as water, land, and vegetation are also important factors in fire management and preparedness. Natural water sources act as natural fire breaks and provide water supplies for fire fighting. Land and vegetation play an important role in determining fire severity based on the fuel load and other terrain features.

Human resources are an important resource in disaster preparedness. For example, single householders often face difficulties when preparing compared to households with many people. Available human resources are important to implement any community level fire management plan (Harris et al. 2011). Communities can be seen as a human resource, especially when a fire exceeds local fire fighting capacity (Cottrell et al. 2008). Communities can use their human resources to reduce the impacts from fires through sharing responsibilities with fire management authorities for the development or revision of emergency response plans, disaster recovery, hazard mitigation strategies, and comprehensive land use plans that incorporate sustainable development practices (Flax et al. 2002).

209

Access to technology can provide a community with innovative ideas and resources that make their preparedness more effective (Goodman et al. 1998). In the context of bushfire management, fire fighting equipment, communication channels, and early warning systems can be considered to be innovative ideas and resources.

7.5.7 Institutional Arrangements

To achieve any risk reduction goals, multi-dimensional approaches and innovative institutional arrangements are required (Yodmani 2001). These include organizations, information channels, laws, regulatory and enforcement arrangements, which often influence people’s risk reduction attitudes and actions (Winter & Fried 2000). These arrangements are often made by a large number of governmental, non-governmental, public and private bodies involved in disaster management. These institutions typically include disaster management experts and policymakers. Each body’s organizational structure is usually different, and they each have a different orientation and priorities and use different strategies. Institutions are responsible for disaster management initiatives that have been designed to respond to different needs at different phases of the disaster management cycle. To deal with complex risk reduction issues, national decision-making requires strong, sustainable and accepted institutional structures, and a population and civil society educated about risk reduction issues and alternatives (O'Brien et al. 2006). Therefore good links between disaster management institutions and communities deliver a successful disaster management structure.

In New South Wales (NSW), three bodies are primarily responsible for bushfire fire fighting and mitigation services: the NSW Rural Fire Service; the National Parks and Wildlife Service; and the State Forests of NSW. There are other administrative bodies that also take part in the Bushfire Coordinating Committee. This committee provides a forum for government and non-government organisations with an interest in prevention, mitigation and suppression of bushfires. It also plays a key role in coordinating the work of Bushfire Management Committees through its policy on Bushfire Risk Management. Each Bushfire Management Committee is responsible for developing and implementing a Bushfire Risk Management Plan for its area under section 52 of the Rural Fires Act 1997. Emergency NSW is responsible for emergency preparedness, response and

210 recovery arrangements for NSW. Subsidiary plans including State Plans, District Plans, Supporting Plans, and Sub Plans, are available "to set out the arrangements for preventing, preparing for, responding to and initially recovering from bush fire events by combat, participating and support agencies in NSW" (Montoya 2010).

7.5.8 Shared Responsibility

In the context of risk, the concept of responsibility refers to the fact that “certain parties have a prospective obligation to undertake actions to manage risk” (McLennan & Handmer 2012, p. 1). Within a risk management framework different parties are responsible for different aspects of risk management. These judgments and obligations of different parties to manage risk are often supported by particular ways of framing responsibility sharing (McLennan & Handmer 2012).

Residents’ preparation for bushfires is often linked to perceptions of responsibilities. People are more likely to prepare when they perceive themselves to be responsible for risk reduction (Bushnell & Cottrell 2007a). Paton (2003) has argued that it is important that people share responsibilities rather than be fully dependent on the government and local bodies. In order to minimize the impact on people and property, communities need to understand that the risk, and the responsibility for bushfire mitigation and management, are shared by individuals, landholders, communities, fire and land management agencies, researchers, and governments (Ellis et al. 2004; Victorian Bushfires Royal Commission 2009) Within the shared responsibility framework, “well- informed and well-prepared individuals and communities complement the roles of land managers and fire agencies” (Ellis et al. 2004, p. 39).

Within a community, different opinions about responsibilities are visible. McGee & Russell (2003) and Winter & Fried (2000) identified these different opinions about shared responsibility. Some residents believe that they are responsible for preparing for and responding to bushfires. People with strong feelings of belonging to the community and a place are more likely to accept personal responsibility for their safety (Paton 2003). They continue to take initiative on preparedness and response measures while hoping that the fire services would be there to assist them if their resources are overrun.

211

Other people think that initial responsibility for fire suppression needs to be taken by fire services and their support can extend afterwards. There are also residents who are not involved in any preparedness or response measures, nor do they accept any responsibility for doing so. They believe that preparedness would not make any difference to their safety (Paton et al. 2008). They also state that the fire services are responsible for protecting the community against bushfires and are less likely to take mitigation actions (Paton 2003). These different opinions need to be identified in developing strategies to facilitate community level initiatives to share responsibilities in order to minimise potential losses.

Researchers have employed different frameworks that help to understand the relationship between elements of social context and levels of preparedness (Jakes et al. 2007; Jakes et al. 2002; Paton 2003; Paveglio et al. 2009). These frameworks can be divided into two types: the community wildfire preparedness models (Jakes et al. 2007; Jakes et al. 2002) and a social-cognitive preparation model (Paton 2003). Jakes et al. (2002) and Jakes et al. (2007) describe individual and community level actions and resources and their relationships with the decision making process, which increases community level preparedness (Figure 57).

Collective Readiness Response Impacts Community Restoration/ Community wildfire recovery actions preparedness Improved forest Organisational health decisions

Improved Individual/ Individual individual wildfire Homeowner actions preparedness

Figure 57: Framework of community wildfire preparedness (Jakes et al. 2007).

212

In this framework, while the importance of community and household level decision- making processes for better preparedness is discussed, factors affecting the preparedness process are not discussed. Furthermore, the relationship between household level preparedness and the community level preparedness is not well defined in this framework and is rather difficult to understand.

In Paton’s (2003) social-cognitive preparation model (Figure 58), social-cognitive processes and the factors and the relationships that affect adjustment and preparation are discussed.

Figure 58: Social cognitive preparation model (Paton 2003).

In the social-cognitive preparation model, Paton identified three phases that are influenced by a specific set of variables; the motivation phase, intention formation phase and the linking intentions and preparedness phase. According to this model, the motivation phase is comprised of factors that motivate people such as critical awareness of hazards, risk perception and hazard anxiety. Factors such as problem focused coping, response efficacy, outcome expectancy and self-efficacy link this initial motivation with the formation of intentions. The final relationship between preparatory intentions and 213 actual preparation is explained by factors such as sense of community, response efficiency, timing of the hazard activity, perceived responsibility, and trust and empowerment. Although this model provides a good platform to understand the social- cognitive process and the variables influencing the decision making about adjustment and preparation, it describes individual level decision making rather than household level and community level processes. Furthermore it only discusses factors affecting the social-cognitive process, which only describes the level of vulnerability, resilience and preparedness up to some point. To have a better overall understanding, socio-cognitive factors need to be combined with other factors influencing the level of preparedness in an overall bushfire preparedness model.

This chapter focuses only on the effects of household and community level decision making processes on bushfire preparedness, mainly concentrating on social factors rather than cognitive factors. However, it is also evident that there is some overlap between both cognitive and social factors and those factors are closely interlinked. These relationships are further explained by developing a model of bushfire preparedness. In order to address the limitations of previous models discussed above, both social and cognitive factors are included in this model. It explains household level and community level decision-making processes separately while summarizing their common features. Therefore, this research attempted to identify several factors that determine household and community level bushfire preparedness processes. Figure 59 shows the relationship between household and community level preparedness processes and the identified factors. This research utilised the model in Figure 59 to investigate how household and community level factors and interactions affected bushfire preparedness.

214

Risk Behaviour Socio Communication Attitudes demographics

Household Knowledge level and preparedness awareness Individual / HH level perception Experience

Sharing Bushfire responsibilities Preparedness

Community Sense of Experience perception community

Community level preparednes

Community Institutional Access to based arrangements resources initiatives

Figure 59: Bushfire preparedness model, adapted from Paton (2003) and Jakes et al. (2007)

Figure 59 explains the overall bushfire preparedness process of the community. According to this framework, overall levels of bushfire preparedness are determined by household level preparedness, community level preparedness and the perception of shared responsibilities. Perceptions of shared responsibilities are also shaped by household and community level preparedness. Household level preparedness is largely influenced by individual/household level perception. Several factors shape individual and household level perception. They include socio-demographics, behaviour and attitudes towards preparedness, previous experience with bushfires and knowledge and awareness of fire impacts and preparation. Risk communication plays an important role in bushfire management. It helps to improve the level of knowledge and awareness, as well as attitudes and behaviour towards fire management and preparedness, which helps to improve the individual and household perceptions of bushfire preparedness.

215

Community perception is the driving force that determines the level of community preparedness. Community perception is shaped by factors such as sense of community, community based initiatives, and whether community has had previous bushfire experience or not. Institutional involvement and available community level resources (human, financial, etc.) play an important role in community level initiatives. This framework also shows a relationship between household and community level preparedness. Factors that influence household level preparedness affect the level of community preparedness and vice versa.

7.6 Community Based Bushfire Management in the Study Area

Promoting community level activities to motivate people to prepare for bushfires is an important step towards sustainable bushfire management (Newport & Jawahar 2003; Paton et al. 2008; Pearce 2003). Godschalk et al.(1998), (cited in Pearce 2003) also recognised the importance of local communities in initiating and implementing disaster mitigation policies that will lead to the adoption of mitigation strategies, while state and federal governments and agencies have a role to play in establishing policies. Newport & Jawahar (2003) recognized community level initiatives as a social process, in which vulnerable groups organize themselves for their common needs and problems with support from various sources. Communities can participate in resource identification, and have capabilities and coping mechanisms which can be effective in the decision- making process (Newport & Jawahar 2003).

One of the key objectives of bushfire management authorities is to empower individuals and communities to maintain their functioning at normal levels before, during and after a significant disruption by a bushfire event. People’s level of preparedness to deal with their exposure to adverse hazard consequences is an important component of their normal functioning (Paton et al. 2006). Thus community fire preparedness and response capacity is a crucial component in minimizing losses in any emergency situation. Encouraging people to prepare for bushfires has been identified as a significant factor and has already been given attention as a public policy issue in Australia (McLeod

2003).

216

The sustainability of any risk management programme depends on how people are involved and transform risk management processes as a function of their community (Pearce 2003). Community level activities such as Community Fire Units (CFU) contribute significantly to bushfire safety and preparedness. Such programmes help people to understand risk and what actions need to be taken to manage that risk. Hence they can motivate people to act as an individual or a group. They also help to share responsibilities between the community, fire management authorities and government as recommended by the Victorian Royal Bushfire Commission (Teague et al. 2009).

In New South Wales, several community level fire management operations exist at the UBI. The Community Fire Unit (CFU) Program is one of the main community level activities that enhances community safety. It was initiated by the NSW Fire Brigade (now Fire and Rescue NSW). A CFU consists of a team of local residents who live in urban areas close to bushland in NSW. The members of the CFU are given fire fighting training by the NSWFB and are equipped with physical resources (e.g., fire fighting equipment). The main task of the CFU is to prepare for and protect their properties from spot fires and ember attack in the event of a bushfire, until the fire services arrive (Fire and Rescue NSW 2009). CFUs can be inspired by reasons such as an increased number of fire incidents, awareness raising programmes, community values, other community level activities, and place attachments (Lowe 2008). CFUs minimise risk and vulnerability in fire-prone communities by empowering communities to develop survival strategies and disseminate vital information (Lowe 2008).

The New South Wales Rural Fire Service volunteer programme is better resourced than CFUs and focuses on fire fighting in bushlands. However, they often organize education campaigns for the residents living near bushlands. Both groups are essential when firefighting resources overrun the capacities of the fire management authorities. They also help to increase community resilience by organizing community level training, awareness campaigns and community level discussions to help people understand bushfire risk. Community bush care groups are another initiative. They are run by the local government council and focus on enhancing sustainable agriculture and biodiversity. Bush care groups are actively involved in fuel management in the area and help to minimise the potential bushfire threats. Bush care groups also organize a variety 217 of events such as meetings, workshops, field days and educational programmes to promote better natural resource management. These events also encourage people to build better social networks, which will help to increase social capital (Lowe 2008). Finally, Street FireWise (SFW) is another community level initiative that has been developed by the Blue Mountains branch of the NSW Rural Fire Services. The main objective of SFW is to increase knowledge and awareness of bushfire risk within bushfire prone areas. This is coordinated through frequent meetings and discussions at the street level during the bushfire season (Lowe 2008).

7.7 Household Level Activities in the Study Area

Household level bushfire preparedness activities are another important component of community fire management strategies. Fire impacts can be minimized if people are adequately prepared for bushfires. These preparedness actions need to be undertaken before an event occurs. They increase response capacity when disaster does strike (Tierney 2001). Paton et al. (2006) have noted that personal actions can mitigate a problem or result in favourable outcomes for the person.

Householders have their own responsibilities within a household. Household preparedness activities include minimizing the amount of fine fuel by creating a defensible space around the home, cleaning leaves from guttering, placing metal fly screens on windows, screening eaves, ensuring access to resources for extinguishing spot fires, and determining householders’ ‘stay or go’ positions (McGee & Russell 2003; Paton 2006b; Paton et al. 2008). These issues have been discussed in many public policy documents.

Household level responses are driven by several factors such as socioeconomic conditions, experience, risk communication, awareness and social capital and institutional arrangements (Collins 2005; McGee & Russell 2003; Paton 2003; Paveglio et al. 2009). Some householders are reluctant to trust advice from fire management authorities while others are heavily dependent on fire management authorities (McGee & Russell 2003). These factors and household level behaviour and preparedness are important components that need to be explored in order to address issues of household

218 level preparedness and risk communication. Strengthening household level preparedness is important to meet the ever-increasing demand for fire management resources (Lowe 2008). This strategy involves “the encouragement of residents to decide, prior to the start of each fire season, whether they will prepare to stay and actively defend their property from bushfires or leave well before a fire arrives” (Lowe 2008, p. 3).

The lack of household level efforts highlights a need for risk management strategies to include a focus on increasing household preparedness. To address that problem, it is important to understand household level preparedness, response and recovery strategies. This research explores some major features of household level preparedness, response and recovery measures. It also reviews social and other factors that influence whether people decide to prepare for bushfires, and can respond quickly and recover rapidly after a bushfire. It further describes what people expect from fire management authorities, their level of trust in them and whether they have received what they expect over time.

7.8 Role of the Fire Management Authorities in the Study Area

Traditionally, government and other fire management authorities are responsible for bushfire risk reduction strategies such as fire suppression, fire prevention through reducing fuel loads and building firebreaks, and building codes and zoning. In NSW, reducing fuel loads is a primary task of the local government council and the NSW Rural Fire Services. NSW Fire and Rescue maintain responsibility for fire prevention within the urban bush interface. Local councils operate a Bushfire Coordination Committee (BFCC). The main responsibility of the BFCC is to develop a Bushfire Risk Management Plan for the council area. Bushfire management zoning is based on the outcome of the Bushfire Risk Management Plan.

More recently there has been a redistribution of rights and obligations and/or changes in the allocation of costs and benefits from the government to the public (Winter & Fried 2000). Local councils and other fire management agencies promote bushfire education in order to share some of the responsibility with households and communities, with a

219 particular emphasis on encouraging people to become members of community level organizations. In addition to individual fire preparation efforts (e.g., creating a defensible space around the home by clearing vegetation), regulations have been imposed to make residents’ homes more fireproof (e.g., retrofitting roofs) or to pay tax for additional fire protection, motivate residents to take preventative measures and engage in community programs to reduce the fire risk to themselves and their neighbourhood (Bushnell & Cottrell 2007b).

To implement bushfire risk management programmes, community support is essential. None of these agency initiatives has been recognized as ‘the’ best practice. Rather it is suggested that a combination of strategies is more successful (Bushnell & Cottrell 2007b). Policies and bushfire management strategies should address community needs and expectations (Bushnell & Cottrell 2007b). Community views are important to understanding what people expect from fire management authorities. They will also help to change policies and strategies to fulfil the expectations of the people living at the UBI.

In order to encourage greater community level preparedness at the UBI, it is important to understand how people’s experiences, attitudes and beliefs contribute to their decisions about bushfire preparation and how people perceive bushfire risk, use and respond to risk information, and strengths and weaknesses of their communities. To understand the importance of shared responsibilities for bushfire management and mitigation, it is also extremely important to investigate how people view their individual responsibilities and institutional responsibilities and interact within their communities and within society, and how these interactions shape their understanding and behaviour.

7.9 Methodology

Numerous studies have discussed disaster preparedness issues in the context of different natural hazards such as floods, earthquakes, volcanoes and bushfires or wildfires (Jakes et al. 2007; McGee & Russell 2003; Miceli et al. 2008; Paton et al. 2000). It is also widely accepted that it is necessary to understand community level issues with respect to disaster preparedness and mitigation. While scholars have widely examined

220 community safety and related community level issues, only a small body of research has looked at household level preparedness (Beringer 2000; Collins 2005; Winter & Fried 2000). Paton (2006a) stated that disaster loss reflects how the characteristics of a hazard interact with individual, community and societal elements. Therefore, to have a better understanding about the level of preparedness in a given area, it is important to identify linkages and interactions between individual households, communities and other societal elements.

In order to examine household and community level preparedness, perceptions, attitudes, and beliefs and factors influencing their decisions, two data collection techniques were employed; a household survey and community level discussions. This mixed-methods approach was utilized to gather a wide range of views and perspectives about community level bushfire preparedness. It satisfies both the need for deep understanding of issues related to bushfire preparedness and their interaction and processes. The data-gathering techniques utilised in this survey were household surveys (postal survey and face-to face discussions), semi-structured interviews and number of follow up discussions. This was done in three phases: development of data collection tools, a pilot study and final data collection. To identify important issues to be addressed by this survey, previous literature on community surveys about bushfire issues was reviewed. Furthermore, a number of focus groups were initially undertaken with community, fire brigade groups, the two local government councils and researchers. Finally, the survey tools were formulated and piloted; the final survey included some questions proposed by local government councils, the NSW Rural Fire Services and NSW Fire and Rescue.

7.9.1 Household Survey

The household survey was designed to collect primary data from householders living at the UBIs in the Ku-ring-gai LGA and the Blue Mountains LGA. To ensure that householders were sampled randomly yet from across the geographic area, a stratified random sample by geographical unit was employed. First, samples of census collection districts within a 30m buffer of the UBI were selected and then a sample was made of householders within those census collection districts.

221

For the household survey, an eight page self-completion questionnaire was designed (Appendix II) to collect data on a wide range of social factors including: demographic information and property/lifestyle factors; bushfire risk knowledge and awareness; bushfire experience; perception of local hazard risks; participation in community level bushfire preparation activities; preferences for bushfire information; views on responsibility for bushfire-related activities; views on service providers and services provided; involvement in community organisations and views on shared responsibility. The questionnaire was trialled in a pilot survey and appropriate changes made before the final version was deployed. The questionnaire consisted of both open and closed questions, including subjective continuum scale questions. For the purpose of statistical analysis, responses to questions were treated as variables. Both continuous and discrete variables were captured in this study and coding was done prior to the analysis.

A detailed introduction letter explaining the purpose of the survey with ethics approval from UNSW (Appendix I), a formal invitation from the council/New South Wales Fire and Rescue, as well as a postage-paid return envelope was sent together with the questionnaire. In total, 1400 questionnaires were delivered; 600 in Ku-ring-gai and 800 in the Blue Mountains council area. The procedure involved hand-delivery to mailboxes (in Ku-ring-gai) and mailing (in the Blue Mountains). Respondents returned surveys by mail using the postage-paid return envelopes. Addresses were not requested to keep respondents anonymous and confidential.

7.9.2 Interviews with Community Members and Key Personnel

The key objective of these interviews was to further develop understanding of community characteristics and fire management issues in the area. In-depth, semi- structured interviews were conducted with community members and community fire unit members in Ku-ring-gai and the Blue Mountains area before the start of the 2011/2012 bushfire season. With the help of local councils, NSW Rural Fire Services and NSW Fire and Rescue, several bushfire prone streets that were suitable for interviews were identified using maps of the Ku-ring-gai and the Blue Mountains local government areas. Interviews were conducted with randomly selected households living in properties within the 30m buffer zone and in areas close to bushlands. Face to face

222 household surveys were conducted in selected streets by door knocking, introducing and explaining the purpose of the survey and asking residents to take part in the interview. Before the interview all participants were informed about the ethics approval, that the interview was completely voluntary and confidential and that no identifying information would be used in any publication from the study.

Figure 60: Interviewing community members and key personnel (Author’s Photograph).

Sixty-two household level interviews were undertaken using a semi-structured questionnaire. In each street an adequate number of interviewees was recruited randomly. In addition to that, group discussions were conducted with New South Wales Rural Fire Service volunteers to understand previous bushfire impacts and on-going community level fire management activities in the area. General weekly meeting days were used to meet NSWRFS volunteers at their community fire brigade. These discussions were also used to collect information about community level fire management issues. Topics were highlighted, focusing more on the qualitative aspects of community risk perceptions and motivations. These included community strengths including social capital, access to resources, awareness and risk knowledge, preparedness, response, warning and evacuation and recovery. In addition various other community level issues such as local council involvement that influenced preparedness were discussed.

223

7.9.3 Data Analysis

Out of the 1400 surveys that were deployed (600 in Ku-ring-gai and 800 in the Blue Mountains), 365 surveys were returned, giving a 26% response rate. A higher response rate (29%) was observed in Ku-ring-gai, compared to 24% in the Blue Mountains. Although previous studies have shown a slightly higher response rate (28%) in other parts of the Australia (Bushnell & Cottrell 2007a; Paton et al. 2006), past studies in NSW had response rates as low as 18.5% (Prior 2010). Therefore, it is concluded that the response rate for this survey was pretty good, especially given the length of the survey.

Exploratory data analysis techniques were used to summarise the findings using descriptive statistics. Chi-square tests were performed to understand statistically significant associations between variables of interest. Chi-square statistics 2 were used to define the association between variables at p <0.05 (Field 2009). In order to facilitate Chi-square tests, composite variables were derived to summarise household levels of preparedness, community levels of preparedness, perceptions of shared responsibility and available resources by aggregating categorical variables (Table 15). The advantage of having one composite indicator is that a wider range of variables can be incorporated, ideally leading to a more comprehensive model of reality (Vincent 2004).

The categorical variables are measured at different scales, which makes their aggregation complex and requires a mechanism to identify optimal scaling (Benedetti et al. 2010). This was achieved by using a scoring algorithm to construct a scale (Hays et al. 1995). In this study, a scale ranging from 0-1 was used. The final aggregated value was transformed linearly to a 0-1 scale using the scale function given below (Hahn et al. 2009).

224

Finally, respondents were classified based on the standardised value (z score) of the transformed scale values.

Data were analysed using SPSS 20 software under several themes, demographics of respondents, bushfire awareness and knowledge; bushfire preparedness (household and community level), perceptions of shared responsibility and response and recovery measures. Each theme has implications for bushfire management.

Table 15: Variables and derived composite variables

Composite Variable Variables When do you normally start to prepare for bushfires? In which month do you normally start bushfire preparedness activities? How do you prepare? Do you have insurance for your residence? Have you discussed what to do in the event of a fire with your family? Household level Have you discussed the following assessment tools with your preparedness family? Household bushfire risk assessment by NSW Rural Fire Services (NSW RFS) Bushfire survival plan by NSW RFS Asset protection zone construction tool by NSWRFS Recover after a fire event by NSW RFS Have you identified a potential evacuation route? Sources of the connection you feel to your community How do people in your community help each other in an emergency? Do people in your community discuss bushfire preparedness? Does your community work together on bushfire preparedness activities? My community is protected against potential bushfire events. Community level preparedness I am happy with current bushfire management practices in my community Bushfire management activities have improved in my community since 2005 How involved are you in community level bushfire prevention activities? Would you like to be involved in a community fire guard /Community Fire Unit programme? 225

Rate your satisfaction with the following infrastructure/service facilities in your community Water supply Available resources Road network / transport Fire fighting resources Emergency warning Emergency healthcare Who do you believe is the most responsible party/parties for following bushfire preparedness and mitigation activities in your area? Safe Prevention, Preparation and Suppression Management of Fire in the Landscape Community Self-Sufficiency for Fire Safety Protection of People and Property Shared responsibilities Education, Training and Communication (Household) Who is responsible for the following emergency response and recovery activities in your area? Early warning Fire fighting Assisting people to evacuate from the area Providing relief and recovery needs Remove debris after an event

7.10 Results

7.10.1 Household Survey

7.10.1.1 Demographics of Respondents

Background information on the respondents is useful to determine the sample’s representativeness. Data collected from the household surveys was compared with available statistics from the Australian Bureau of Statistics (ABS) (2006) (Table 16). However, there are some limitations to the procedure used to establish the extent to which the sample was representative of the UBI community. As the survey was only administered within the 30m buffer zone of the UBI, the sample did not include respondents across the whole population in the LGA. ABS data was obtained from the Collection Districts (CDs) that intersect the 30m buffer zone. Unfortunately these CDs do not uniformly conform to the shape of the UBI. Furthermore, in this study, data from the 2006 census was used as it was the most recent census data available at the time when the analysis was conducted. Over the years, demographics may have changed in 226 the buffer zone and they may not be the same as in 2006. Therefore, census data may be slightly different, but a comparison does provide at least some information about the representativeness of the sample. The first section of the questionnaire asked key questions on socio demographic characteristics. It provides a socio-demographic profile of the respondents and can be used in decision-making.

227

Table 16: Demographic profile of the respondents

Ku-ring-gai Blue Mountains Census Census Collection Survey Collection Survey Data District, ABS, Data District, 2006 ABS, 2006 n=176 (HH) n= 14231 (HH) n=189 (HH) n= 15542 (HH) Gender Male 49.6% 48.4% 49.7% 48.1% Female 50.4% 51.6% 50.3% 51.8% Age (Years) < 5 6.6% 5.4% 3.3% 6.4% 5-17 15.7% Not available 16.6% Not available 18-65 57.4% Not available 62.5% Not available 65+ 22.1% 17.4% 18.3% 13.9% Household income

(Per Week) Less than $500 6.8% 8.7% 14.6% 18.5% $500-$999 21.4% 14.3% 33.3% 24.5% Above $1000 71.8% 77.0% 52.1% 57.0% Labour Force Full time 45.1% 61.7% 48.5% 60.7% Part time 15.9% 34.7% 18.7% 34.2% Casual 4.5% Not available 6.1% Not available Retired 30.1% Not available 22.6% Not available Unemployed 3.3% 3.6% 3.6% 5.1% Year of residence >1 Year 94.2% 85.4% 93% 74.6% >5 Year 82% 61.0% 81% 51.0% Type of household Couple with no children 38.7% 27.1% 37.2% 23.7% Couple with children 39.9% 48.4% 29.8% 29.4% Single parent 4.0% 8.2% 2.7% 9.4% Other 8.7% 16.3% 8.0% 37.5% Tenancy Own property 92.6% 89.3% 94.5% 80.3% Renting 7.4% 9.9% 4.9% 18.7% Other 0% 0.8% 0.5% 1.0%

228

Approximately equal numbers of males and females responded to the survey in both areas. Gender statistics from the household survey and the ABS show that the sample collected through the household questionnaire is quite representative of the general population (Figure 61).

100%

80% 50.3% 51.8% 50.4% 51.6%

60%

40%

49.7% 48.1% 49.6% 48.4%

PercentageofRespondents 20%

0% HHSurvey ABS HHSurvey ABS BlueMountains Kuringgai

Male Female

Figure 61: Gender demographics comparison with ABS data.

Age groups were defined based on the level of vulnerability (Figure 62). It is well documented that age below 5 years and above 65 are highly vulnerable compared to other age groups (see Chapter 5). Therefore, information about the residents whose age was above 65 or below 5 years was collected specifically. The other age groups defined in the questionnaire were the same as the age groups defined by the ABS, and therefore could not be compared. The age group of above 65 is over-represented in both samples, which is probably due to higher participation of the elderly households because they have time to participate. In the Blue Mountains, the age group of below 5 years is under-represented. This may be due to the variation within the CCD boundary problem that was discussed above.

229

100% 13.9% 18.3% 22.1% 17.4% 80%

78% 60% 62.5% 80% 57.4% 40%

PercentageofRespondents 20% 16.6% 15.7% 6.4% 6.6% 0% 3.3% 5.4% HHSurvey ABS HHSurvey ABS BlueMountains Kuringgai

<5 517 1865 65+

Figure 62: Age demographics comparison with ABS data.

Figure 63 illustrates levels of income in Ku-ring-gai and the Blue Mountains. It shows that in Ku-ring-gai, the percentage of high-income earners is higher and percentage of low-income earners is lower than the Blue Mountains. ABS data indicates that middle- income earners are under-represented in the samples for both study areas. This may in part be explained by the difference in income distribution between the buffer zone and the whole CD. It is also evident that higher numbers of middle-income earners live closer to public transport services and town centres and not within the 30m zone.

230

100%

80% 52.1% 57% 71.8% 60% 77%

40% 33.3% 24.5%

PercentageofRespondents 20% 21.4% 14.3% 14.6% 18.5% 6.8% 8.7% 0% HHSurvey ABS HHSurvey ABS BlueMountains Kuringgai

Lessthan$500 $500$999 Above$1000

Figure 63: Level of household income/week among respondents, in comparison with ABS data.

To compare the surveyed households’ statistics with the ABS data, family household statistics were used. I assumed that in the UBI one family could be considered to be one household. Single-person household data was not readily available with the ABS and we could not compare those responses. Therefore, single person households in the survey data were considered as the ‘other’ household type. Surveyed samples and ABS data show some differences, which may arise from differences in the mix of household types within CCDs and in the buffer zone (Figure 64). The primary household type in the area is an important factor for bushfire management practices. Residents who live alone and single parent householders are more vulnerable and may need to seek external assistance in times of emergency. In the Blue Mountains, the major group of households in the survey is couples with no children and in Ku-ring-gai the major group is couples with children. Single person households make up 22.3/33.3% of the households in the other category in the Blue Mountains, and 8.7/17.4% in Ku-ring-gai.

231

100% 17.4% 16.3% 30.3% 80% 37.5% 4% 8.2%

2.7%

60% 9.4% 39.9% 29.8% 48.4%

40% 29.4%

PercentageofRespondents 20% 37.2% 38.7% 23.7% 27.1%

0% HHSurvey ABS HHSurvey ABS BlueMountains Kuringgai

Couplewithnochildren Couplewithchildren Singleparent Other

Figure 64: Household type comparison with ABS data.

In both areas, there is a high level of home ownership (Figure 65). In Ku-ring-gai, the household survey and ABS proportions are quite similar. In the Blue Mountains sample, rented properties are underrepresented. This difference may indicate that rental properties are located within the CD but not within the buffer zone.

232

100% 1.0% 0.8% 4.9% 7.4% 9.9% 18.7% 80%

60%

94.5% 92.6% 89.3% 40% 80.3%

PercentageofRespondents 20%

0% HHSurvey ABS HHSurvey ABS BlueMountains Kuringgai

Ownproperty Renting Other

Figure 65: Home ownership comparison with ABS data.

Figure 66 illustrates the employment status of the respondents. This information on employment status could not be compared with the ABS as the groups defined in the household survey are not comparable with ABS information. More than 45% of the responding households in both areas work fulltime (i.e., at least 35 hours a week). The elderly, retired population is critical when determining the level of vulnerability at the UBI. A significant proportion of respondents is retired in both areas (30% in Ku-ring- gai and 23% in the Blue Mountains) (Figure 66). Low unemployment rates among respondents can be seen in both Ku-ring-gai and the Blue Mountains.

233

100% 3.6% 3.3%

22.6% 80% 30.1%

6.1% 4.5% 60% 18.7% 15.9%

40%

PercentageofRespondents 48.5% 20% 45%

0% BlueMountains Kuringgai

Fulltime Parttime Casual Retired Unemployed

Figure 66: Employment status comparison.

Level of education is an important socio-economic indicator, and is often highly correlated with income. Level of income increases the level of resilience of the household as it increases the available opportunities for bushfire preparedness and coping measures. High-income households tend to have access to more response and recovery measures in the context of natural hazards. A high proportion of the residents of Ku-ring-gai and the Blue Mountains hold a tertiary qualification (Figure 67). Eighteen percent hold a trade/technical qualification in Ku-ring-gai, while 22.3% hold trade/technical qualifications in the Blue Mountains. People who have education levels lower than secondary school are rare in Ku-ring-gai (2.6%) compared to the Blue Mountains (6.4%).

234

100% 17.2% 24.3% 80%

31.3% 60% 42.1%

40% 22.3%

17.8% 20% 22.8% 13.1% 6.4% 0% 2.6% BlueMountains Kuringgai

Lessthansecondaryschool Secondaryschool(year12) Trade/technicalqualification Universitydegree Postgraduate

Figure 67: Level of education of adults.

Differences in daytime and nighttime populations can complicate planning for and response to bushfires. An unattended property at a time of bushfire is a critical issue. People who live alone and are geographically and socially isolated with no co-resident carer or family are a population that needs to be considered for emergency management and planning. In both areas, many residents travel outside of the area for employment. It is evident that on a typical working day, significant numbers of household properties are unattended (25% in Blue Mountains; 18% in Ku-ring-gai). Only one person can be found at home during the daytime in a significant proportion of households both in the Blue Mountains (43%) and Ku-ring-gai (39%) (Figure 68).

235

100% BlueMountains 90% Kuringgai 80%

70%

60% 48% 50% 43% 41% 43% 40% 39% 33% 34% 30% 25% 25% 25% Percentageofrespondents 20% 18% 10% 9% 10% 7%

0% 012>2012>2 Daytime Nighttime

Figure 68: Number of people at home during day and nighttime.

7.10.1.2 Knowledge and Bushfire Awareness

Bushfire preparedness is often influenced by knowledge and awareness about bushfires (Paton 2006b). Understanding of physical properties of the fire and what measures need to be taken to minimise fire impacts is vital for effective preparedness (Beringer 2000; McGee & Russell 2003). It also enhances enthusiasm for fire prevention and willingness to initiate community level preparedness activities (Beringer 2000; Newport & Jawahar 2003; Prior 2010). Several questions were asked to measure respondents’ levels of bushfire awareness and knowledge (see questionnaire in Appendix II). Each response was individually assessed and a composite variable was derived using the questions asked.

Length of residence in the area influences many important aspects of knowledge about bushfire risk such as awareness, attitudes and preparedness (Beringer 2000; Prior 2010). It also influences the level of community engagement (McGee & Russell 2003). A certain amount of knowledge about bushfire risk and fire management activities can be gained over the years from previous bushfire seasons. Recent migrants to the area often do not understand their level of exposure until they assess the level of bushfire hazard in 236 the area. A question was asked about how long respondents have been living at their current suburb (Q9). Ku-ring-gai residents have lived in their current suburb on average somewhat longer (19.3 years) than Blue Mountains residents (15.6 years). Figure 69 illustrates the length of residence of the respondents.

Years in the current suburb was classified into four classes; fewer than two years, two to five years, five to ten years and more than ten years. Most households in both areas have resided in the area for more than 10 years (above 60%). Many of these residents would be well aware of the surrounding natural environment. Thirteen percent of respondents in the Blue Mountains and eleven percent in Ku-ring-gai have lived in the area for 2-5 years. In the Blue Mountains, seventeen percent of respondents moved to the area between five and ten years ago whereas in Ku-ring-gai it is fifteen percent. In both areas, a minority of residents are recent migrants to the area (Blue Mountains, 10%; Ku-ring-gai 11%). Compared with the ABS figures, years of residence discussed in all four classes in the survey samples from both areas are over represented. This may be due to the change of movement of population over the time and the differences between the buffer zone and the whole CCD.

100% BlueMountains 90% Kuringgai 80%

70% 64% 60% 60%

50%

40%

30% Percentageofrespondents 20% 17% 13% 15% 10% 11% 11% 10%

0% <2Years 25Years 510Years >10Years

Figure 69: Years of residence in their current suburb.

237

People move to the UBI for various reasons. Their decision to move is primarily influenced by the amenity of the UBI and its associated landscape aesthetics. People value the peace and quiet, space, and trees and bushland afforded by their property (Figure 70). Other reasons mentioned in the literature include affordable housing, less crime, pollution and crowding (Cottrell & King 2007). This locational preference may suppress individuals’ willingness to recognize bushfire hazards in the area (Cortner et al. 1990). A question was asked about the reason for living in their residential location, with the ability to specify multiple options because that provides an indication of sources of attachment to the area (Q10).

100%

90% BlueMountains 80% Kuringgai 69% 70% 61% 59% 60% 52% 50% 47% 44% 40%

30% 27% 23% 18% 18% Percentageofrespondents 20% 17% 17% 14% 15% 10%

0% Affordable Proximityto Qualityof Qualityof Lifestyle Igrewup Closetomy housing work thenatural facilities reasons there family environment

Figure 70: Reason for living in Ku-ring-gai or the Blue Mountains.

A majority of households in Blue Mountains (61%) indicate that they moved to the Blue Mountains for lifestyle reasons whereas 59% valued the quality of natural environment. In Ku-ring-gai, a majority (69%) indicated that they value the quality of the natural environment in the area. People have also moved to the Ku-ring-gai area expecting a different lifestyle. Forty two percent suggested lifestyle reasons as a motivation in Ku- ring-gai. The quality of facilities such as education, health, or security is another important factor driving residential choice, among 47% of households in Ku-ring-gai. The quality of facilities is not a major driver for residents living in the Blue Mountains (17%). The other prominent reasons in Ku-ring-gai are proximity to work and 238 affordable housing (27% and 23%, respectively). In the Blue Mountains affordable housing was another important reason (47%); only seventeen percent have mentioned proximity to work as a reason. Eighteen percent live in Ku-ring-gai and fifteen percent live in the Blue Mountains because it is close to their family.

People are often less accurate in evaluating their level of actual bushfire risk if they have not experienced a bushfire event. Various types of previous experience of bushfires can be important for residents to perceive levels of risk accurately and prepare for fires (Bushnell & Cottrell 2007a; Goodman & Proudley 2008). Respondents were asked whether they had experienced fire in their current neighbourhood or elsewhere (Q11). This question was included to see whether previous fire experience had influenced preparedness at the UBI. In Ku-ring-gai only thirty one percent have experienced bushfire in their neighbourhood while a large portion (42%) of the respondents have not yet experienced a fire in their neighbourhood (Figure 71). Nineteen percent have experienced fire somewhere else and eight percent of residents have experienced fire in Ku-ring-gai and somewhere else. Altogether, 58% of the responding households have experienced fire either in their neighbourhood or somewhere else.

Compared to Ku-ring-gai, fifty percent of respondents have experienced fire in the Blue Mountains and sixteen percent have experienced fire somewhere else. Fourteen percent have experienced fire in the Blue Mountains and somewhere else. Only twenty percent are yet to experience bushfires. Results revealed that in the Blue Mountains, 80% of respondents have experience fire in some location.

239

100%

90% 20%

80% 42% 14% 70%

60% 16% 8% 50%

40% 19%

30%

Percentageofrespondents 50% 20% 31% 10%

0% BlueMountains Kuringgai

Inthisneighbourhood Somewhereelse Inthisneighbourhoodandsomewhereelse None

Figure 71: Previous bushfire experience.

Providing bushfire prevention information is an important component of public policy in bushfire education programmes. Access to specific bushfire prevention information increases levels of awareness and encourages preparedness of residents living in the UBI (Paton 2006b). Fire management agencies have identified the importance of engaging with communities in order to improve risk communication strategies that will enhance levels of preparedness while emphasising shared responsibilities. The obvious choices of sources of information are fire management agencies, the local council in the area and the media (Paton 2006b). However, in Ku-ring-gai most respondents indicated that they have not received any information about bushfires in last six months (64%). Only 36% of respondents have received such information. In the Blue Mountains a higher proportion of respondents has received information (73%), and only 27% have not received any (Q13).

240

Figure 72 illustrates the percentages of households who received bushfire information from various sources (Q14). In both areas, respondents received information through the mass media (TV, radio and newspapers) and local council newsletters. Fifty five percent of the Blue Mountains respondents have received information through the media, while 57% have in Ku-ring-gai. Similar percentages of respondents have received information from the local council in the two study areas (34% in the Blue Mountains and 36% in Ku-ring-gai). Compared to Ku-ring-gai (31%), a large percentage (52%) of Blue Mountains respondents has received information by mail. Community level discussions and community newsletters were important sources of information in both locations. In the Blue Mountains 27% of respondents received information from their community newsletter, as did 30% in Ku-ring-gai. The other information sources that people mentioned include the NSW RFS, neighbours or friends, and the Internet. In both areas very few people reported receiving information through the Internet (3% in Ku-ring-gai and 12% in the Blue Mountains). Although there are many sources of fire management information or tools online, residents are reluctant to seek out such information themselves. Traditional methods of communication seem to be much more efficient than online promotions.

241

100% 3% 12% 11% 90% 13% 8% 15% 80% 36% 34% 70%

60%

50% 55% 57%

40%

30% 27% 30% 20%

52% 10% 31%

0% BlueMountains Kuringgai Bymail Communityactivities Massmedia Localcouncilnewsletter NSWRFS Neighbours/friends Internet Figure 72: Sources of information.

Despite the availability of information, providing information for the residents in an appropriate delivery method (relevant information to the right people at the right time) is important to make significant impact on behaviour. A question was asked to understand whether the received information was useful or not (Q15). A majority of the respondents in both areas had received bushfire information from any source (79% in Ku-ring-gai and 88% in the Blue Mountains) indicated that the information that they received was useful, while 21% respondents in Ku-ring-gai and 12% respondents in the Blue Mountains believed the information was not useful or they had not read it.

On a day of extreme weather conditions, it is important to communicate fire ban information to those who are most at risk. Identifying proper channels of communication would help to improve dissemination of fire ban information. A

242 majority of respondents rely on either television/radio advertisements or roadside fire danger signs to receive information about fire danger warnings (Table 17) (Q17).

Table 17: Communication of total fire ban information

Blue Mountains Ku-ring-gai Local government newsletter 4% 8% Community newsletter 5% 6% From the neighbours 5% 8% Newspaper advertisement 14% 19% Television/Radio advertisement 51% 75% Roadside fire danger signs 82% 35% Internet 28% 14% Other 14% 10%

Fire bans are a critical policy issue throughout the bushfire season. Some people are unaware of fire ban rules and regulations, which may result in accidental fires. It is important to educate people about fire bans in order to prevent those accidental fire events. Respondents were asked to rate their level of awareness of fire bans on a five point scale with 5= fully aware, and 1= not aware (Q18). The overall results revealed a good level of awareness about fire bans among respondents in both areas, with a mean value of 3.75 in Ku-ring-gai and 4.1 in the Blue Mountains. Figure 73 illustrates the respondents’ self-reported levels of awareness about fire bans. According to the Likert scale, 77% in the Blue Mountains and 64% in Ku-ring-gai were well aware of the fire bans (scores of 4 or 5). Three percent of respondents in the Blue Mountains and 7% in Ku-ring-gai were not aware of fire ban rules and regulations.

243

100%

90%

80% 37% 47% 70%

60%

50% 27%

40% 30% 30% 17% 20% 14% 12% 10% 6% 7% 0% 3% BlueMountains Kuringgai

Notaware 2 3 4 Fullyaware

Figure 73: Awareness of fire ban rules and regulations.

A survey conducted by Bushnell & Cottrell (2007a) found a relationship between respondents’ hazard perceptions and their preparedness. Risk perception is a well recognised factor that also influences the social construction of risk and community level initiatives (Bushnell et al. 2007). Risk perception is often unique to communities, individuals and households (Cottrell et al. 2008). Although it is difficult for the public to understand the actual level of risk, the perceived level of risk influences the level of bushfire management at both household and neighbourhood levels.

To measure perception of bushfire risk, respondents were asked to rate how significant they think the threat of bushfire to life and property was in their neighbourhood using a Likert scale (1= No risk to 5= Very high risk) (Q19). Residents in the Blue Mountains rated the threat of bushfire as substantially higher (mean 4.75) than residents in Ku- ring-gai (mean 3.71). The mean responses in both areas are above 3.5, which means that residents in both areas generally considered that bushfires are likely to occur in their area. Figure 74 shows that 83% of the Blue Mountains respondents perceive risk to be at either a high or very high level, compared with 60% in Ku-ring-gai. Only 3% of

244 respondents in the Blue Mountains think there is no or very low bushfire risk while in Ku-ring-gai 13% believe that they live in an area which has no or very low bushfire risk.

100%

90% 25% 80% 46% 70%

60% 35% 50%

40% 37% 30% 27% 20%

10% 14% 12% 1% 0% 2% 1% BlueMountains Kuringgai

Norisk Lowrisk Moderaterisk Highrisk Veryhighrisk

Figure 74: Perceived level of bushfire risk.

7.10.1.3 Household Level of Preparedness

Household level bushfire preparedness is the initial starting point for any preparedness activity. A series of questions was asked to identify factors affecting household level preparedness.

Knowledge about the bushfire season influences when bushfire preparedness activities might occur. Most people do not carry out preparedness activities throughout the year. They only start to prepare when the fire season comes. Therefore it is important to know whether people are aware of when the bushfire season begins and ends.

245

1% Idon'tknow 2%

1% March April 1%

8% January February 6%

35% November December 30%

55% September October 62%

0% 10% 20% 30% 40% 50% 60% 70%

Kuringgai BlueMountains

Figure 75: Knowledge of the month when the bushfire season starts.

A question was asked to understand whether people know when the bushfire season starts (Q20). Figure 75 presents respondents’ knowledge of the bushfire season. In NSW, bushfire season officially starts on 1 October and it ends on 31 March. However, in some bushfire seasons it starts in September due to fire weather conditions. Respondents demonstrated an overall understanding of the bushfire season. One percent of respondents in Ku-ring-gai and 2% in the Blue Mountains indicated that they don’t know the answer. More than fifty-five percent of respondents in both areas (62% in the Blue Mountains and 55% in Ku-ring-gai) answered that the bushfire season starts in either September or October. Thirty five percent of respondents in Ku-ring-gai and 30% in the Blue Mountains thought it starts in the months of November or December. Nine percent of respondents in Ku-ring-gai think the bushfire season starts in January or February, while only 6% do in the Blue Mountains. One percent in both areas think it starts in either March or April.

Having knowledge of the bushfire season, and general timeframe of preparedness, it is also important to understand the exact time when preparation is carried out and what motivates household level bushfire preparedness activities. There are different factors 246 that encourage people to initiate and start their bushfire preparedness activities (Hughes & White 2005). They include risk communication, community activities, personal experience, and risk perception (Paton 2006b; Prior 2010). The ideal time to start preparedness activities would be either before or at the beginning of the bushfire season. A question was asked about the time when people generally start their bushfire management activities (Q21). Responses are given in Table 18. Interestingly, in Ku- ring-gai, 32% of respondents normally don’t prepare, while 14% of respondents in the Blue Mountains don’t take any preparedness actions. It is evident that most of the residents in the Blue Mountains (63%) start their bushfire preparedness activities before bushfire season starts. In Ku-ring-gai only 37% of residents start their preparedness activities ahead of the bushfire season.

Table 18: When do you normally start to prepare for bushfires?

Blue Time of bushfire preparedness Ku-ring-gai Mountains After a bushfire ignites in the Sydney metropolitan 7% 14% area After a bushfire ignites elsewhere in Australia 1% 1% When the authorities tell me to prepare 11% 11% When I see neighbours or others in the community 4% 5% preparing Before the bushfire season starts 63% 37% I don’t normally prepare 14% 32%

Some people do not understand bushfire risk and do not prepare at all. These actions may be due to under-estimation of risk, overconfidence because of a perceived preparedness for bushfire or perception of responsibility (Bushnell et al. 2006). A question was asked to gain insight into the reasons respondents choose to not prepare (Q25). Figure 76 indicates that most of the non-preparers have issues other than the options provided in the questionnaire. In that ‘other’ option, many (38%) non-preparers noted that they are too old to take any preparedness actions on their own, while others stated that although they live in a bushfire prone area (within 30m of the UBI), bushfire is highly unlikely to have an impact on their property. Surprisingly, an equal number of respondents from both areas (33%) stated that they don’t know how to prepare. Twenty

247 one percent of non-preparers in the Blue Mountains and 8% of non-preparers in Ku- ring-gai stated that they don’t have enough resources to prepare. In the Blue Mountains no respondents believed that they do not live in a bushfire prone area and all non- preparers believed that it is their responsibility to prepare. In Ku-ring-gai 24% of non- preparers believe they do not live in a bushfire prone area and only 2% believed that bushfire preparedness is not their responsibility.

100.0%

90.0%

80.0%

70.0%

60.0%

50.0% 42% 40.0% 33% 34% 36% 30.0% 24% 25% 21% 20.0% 8% 10% 10.0% 2% 0.0% Idon’tliveina It’snotmy Idon’thave Idon’thave Idon’tknow Other bushfireprone responsibility resources time howtoprepare areasoit’snot necessary

BlueMountains Kuringgai

Figure 76: Reasons for not conducting bushfire preparedness activities.

There are several circumstances that encourage people to initiate and start their bushfire preparedness activities. Receiving information from various sources triggers both awareness as well as preparedness activities. A question was asked to identify what type of information encourages preparedness among respondents (Figure 77) (Q23). In Ku- ring-gai, bushfires in the Sydney metropolitan area (42%) and previous experience of bushfire events (38%) are the most important factors that encourage household level bushfire preparedness. In the Blue Mountains, previous experience of bushfires (64%) and information received from the local council (28%) made the greatest impacts on the level of bushfire preparedness of respondents. Information received through media 248 advertisements and bushfires elsewhere in Australia have similar impacts on preparedness in both areas. Respondents also identified other circumstances that encouraged them to start preparedness activities. They included information received from the NSW RFS, education programmes and information received from their workplace.

100% BlueMountains 90% Kuringgai 80% 70% 64% 60% 50% 42% 38% 40% 28% 26% 30% 28% 24% 21% 20% 15% 12% 10% 0% Media Information Bushfiresinthe Bushfires Previous advertisements receivedfrom Sydney elsewherein experiencewith localauthorities metropolitan Australia bushfires area

Figure 77: Information that encourages bushfire preparedness activities.

Household level preparedness is a critical step that minimizes the potential impacts of fires on a household. Respondents were asked to describe the activities they had undertaken to prepare for bushfires (Q24) (Figure 78). Eighty-seven percent of respondents in the Blue Mountains stated that they conduct bushfire preparedness activities. In Ku-ring-gai, 69% of respondents had undertaken bushfire preparedness actions. The actions most often undertaken in both areas are clearing gutters of leaves (above 80% of those undertaking preparedness actions) and removing leaf litter and undergrowth around the house (above 88%). Other important actions that respondents take include moving combustible materials such as firewood or fuel away from the house and clearing vegetation away from the house. The lowest priorities are given to house modifications such as installing sprinklers and replacing wooden materials, which increases the house’s level of fire resistance, and other factors such as checking the water tank/pump, the emergency kit and replacing hoses. This may indicate that

249 residents are quite reluctant to spend money on bushfire preparation. They rather focus on preparatory actions that have a low financial burden of preparation.

100% 93% 89% 90% 83%86% BlueMountains 80% Kuringgai 70% 61% 60% 57%

50% 46% 40% 40% 38% 37%

30% 22% 18% 20% 15% 15% 15% 10% 8%

0% Removeleaf Cleargutters Maintaina Move Get Clear House Other litter, ofleaves firebreak combustible equipment vegetation modification undergrowth, materials suchasa awayfrom toincrease etcaround suchas ladder, thehouse bushfire thehouse firewoodor bucketand resistance fuelaway mopsfor (install fromthe spotfires sprinklers, house replace wooden materials)

Figure 78: Bushfire preparedness - household level activities.

Property insurance is a common method of transferring risk, and increases the recovery capacity of the household after a disaster. Almost all residents at the UBI in the sample have insured their properties (Q27, Q28). In Ku-ring-gai and the Blue Mountains most property owners (above 95%) have insurance on their property. Eighty four percent of insured properties have coverage for both rebuilding and refurnishing. Some renters either don’t have any insurance or they don’t know if their landlords have insurance that covers their property (5% in Ku-ring-gai and 3% in the Blue Mountains).

In order to understand the intentions of residents to seek bushfire prevention information and the relationship between residents and bushfire management

250 authorities, several questions were asked about whether respondents discussed bushfire prevention methods with any agency they recognize as having responsibility for bushfire management (Q30, Q31) (Figure 79). Thirty eight percent of the Blue Mountains respondents have discussed their situation with local authorities while only 17% of Ku-ring-gai respondents have done so. In both areas the majority of those who have consulted a local authority have discussed their issues with the NSW RFS (81% in the Blue Mountains and 59% in Ku-ring-gai). In the Blue Mountains amongst those who have consulted a local authority, 25% stated that they have also discussed preparedness issues with NSW Fire and Rescue, while a higher percentage has in Ku- ring-gai (36%). Compared to the Blue Mountains, more respondents in Ku-ring-gai consulted the local council (34% versus 14%). Some respondents also consulted with NSW Parks and Wildlife as well.

100% BlueMountains 90% Kuringgai 81% 80%

70% 59% 60%

50%

40% 34% 34% 30% 25% 21% 20% 14% 10% 7% 3% 6% 0% Localcouncil NSWruralfire NSWfirebrigade NSWparksand Other services wildlife

Figure 79: Discussion of preparedness issues with fire management authorities.

Fire management agencies have developed different bushfire risk assessment materials to help residents to perform self-assessments. Available materials include the Household Bushfire Risk Assessment document developed by the NSW RFS, the

251

Bushfire Survival Plan by the NSW RFS, the Asset Protection Zone Construction Tool by the NSW RFS and the document Recover after a Fire Event developed by NSW RFS. It is residents’ responsibility to discuss their household’s circumstances with family members and implement these plans in order to minimise losses. Despite media campaigns and other community education programmes, a majority of respondents have not discussed at least one of these action plans with their family members (Q33) (Table 19). Blue Mountains respondents are more likely to use these tools than Ku-ring-gai respondents. The Bushfire Survival Plan is the most commonly used tool in both areas (14% in the Blue Mountains and 15% in Ku-ring-gai). The Household Risk Assessment Tool had been used by 31% of Blue Mountains respondents, but only 12% of Ku-ring- gai respondents. Use of the Asset Protection Zone Development Tool is not common in either area. Respondents in both areas have not have not focused on recovery measures and only six percent of respondents in the Blue Mountains and 4% of respondents in Ku-ring-gai have a recovery plan.

Table 19: Use of bushfire assessment tools Percentage of LGA Assessment tools Respondents Household bushfire risk assessment NSWRFS 30% Blue Bushfire survival plan NSWRFS 41% Mountains Asset protection zone construction tool NSWRFS 10% Recover after fire NSWRFS 6% Household bushfire risk assessment NSWRFS 12% Bushfire survival plan NSWRFS 15% Ku-ring-gai Asset protection zone construction tool NSWRFS 7% Recover after fire NSWRFS 4%

Individual perceptions of levels of preparedness are an important factor that influences actual levels of preparedness. Some people are overconfident about their level of preparedness and important implications are often therefore ignored. To understand the respondents’ perceived levels of preparedness, they were asked to rate their perceptions of how prepared their households were for future bushfire events using a Likert scale question (1= Not prepared; 5= Fully prepared) (Q26). The perceived level of preparedness in both areas is moderate. It is slightly higher in the Blue Mountains (Mean=3.38) than Ku-ring-gai (Mean=3). Figure 80 illustrates that only 47% of 252 respondents in the Blue Mountains and 35% in Ku-ring-gai believe that they are prepared. Thirty five percent of respondents in both areas indicated that that they are somewhat prepared. Seventeen percent of respondents in the Blue Mountains and 29% of respondents in Ku-ring-gai believe that they are not well prepared.

100% 12% 8% 90% 27% 35% 80%

70% Fullyprepared 60% 36% 4 50% 35% 3 40% 2 30% Notprepared 17% 20% 13% 10% 12% 0% 4% BlueMountains Kuringgai

Figure 80: Perceived levels of preparedness.

7.10.1.4 Household Level Preparedness and Related Issues

The household level of preparedness is related to many household and community characteristics and to fire management activities in the area. Results show that household level of preparedness is associated with household type (2 = 15.24, df = 6, p =.018). People who live alone and other family types, which include single parent and non-related group households, tend to prepare less compared with couple only households and couple households with children. Household levels of preparedness are also influenced by the number of people who stay at home during the day (2 = 12.28, df = 4, p = .015). This may be a measure of the time that residents can allocate for their preparedness activities. The results revealed that households with residents staying at home during the daytime are more likely to be prepared than households with no residents at home during the daytime.

253

Respondents who had received information about bushfire preparedness were more likely to prepare than those who had not (2 = 42.59, df = 2, p < 0.001). Residents who had lived in the area for fewer than two years were not well-prepared. The longer the period of residency, the higher the level of preparedness (2 = 10.16, df = 4, p = .038). Respondents must be able to assess the level of threat to which they are exposed before they are likely to undertake preparedness measures. It often takes some time to develop this understanding and new residents may be less likely to perceive a threat. They may need to directly experience severe fire weather conditions to realise the importance of preparedness.

Respondents with past bushfire experience were generally more likely to prepare for bushfires (2 = 31.99, df = 2, p < 0.001). Past experiences help to perceive the actual level of risk and those who had an experience where their lives and property had been threatened by bushfire were more likely to consider the importance of household level bushfire preparedness. Residents are often reluctant to believe their actual level of threat and often underestimate the level of risk. Residents who had identified their level of exposure to risk accurately and believed that they lived in a high bushfire risk area were more likely to be better prepared (2 = 18.79, df = 4, p = .001).

To implement preparedness strategies effectively, an understanding of actual levels of preparedness is also required. Residents often overestimate their level of preparedness, and they may therefore not be adequate to withstand a severe fire. Increased understanding of level of preparedness results in increased enthusiasm for prevention and greater faith in survival strategies. Results show an association between perceptions and the level of preparedness. The lower the level of residents’ perceived preparedness, the lower their actual level of preparedness (2 = 106.43, df = 4, p<0.001). A higher level of perceived preparedness also indicated higher preparedness. So people seem to understand their preparedness accurately.

Community interactions also help residents to share their experiences, knowledge, and resources. The results suggest that respondents who were more concerned about their community and who feel like they are part of the community were more likely to 254 prepare than those who were unconcerned or who did not feel part of the community (2 = 14.32, df = 4, p = 0.006). It was also weakly evident that there is an association between level of education and household preparedness (approaching significance 2 = 9.09, df = 4, p = 0.059). This suggests that respondents with higher levels of education may be more likely to be prepared for bushfires.

7.10.1.5 Community Level Preparedness

Social interactions and shared identity help to increase the bond between residents and their community. Tightened social bonds may lead to greater community stability, which facilitates community level bushfire preparedness activities (Lowe et al. 2008). To examine how people feel about their community’s cohesion, respondents were asked a Likert scale question (1-5) about whether they felt they were a part of the community (Q34). A majority of respondents felt like they belong to their local community (70% in the Blue Mountains and 67% in Ku-ring-gai). Twenty-five percent in both areas neither agreed nor disagreed, while seven percent of respondents did not agree with the statement and didn’t feel like they belong to their community (Figure 81).

100% 1% 1% 6% 6% 90% Stronglydisagree Disagree 80% 24% 25% Neitheragreenordisagree 70% Agree 60% Stronglyagree

50% 46% 48% 40%

30%

20%

24% 10% 19%

0% BlueMountains Kuringgai

Figure 81: “I feel like I belong to the local community”

255

There are certain locations and events that help residents to make connections with other community members. Respondents were asked to describe their points of social connections (Q35). Neighbours are the most common point of social connection in both areas. In the Blue Mountains and Ku-ring-gai more than 80% of respondents stated that they make community connections through their neighbours (Figure 82). Community organizations are the next best option in both areas (>25%). In the Blue Mountains, twenty-one percent of respondents use school to make social connections while 27% mentioned that their workplace allows them to build social connections. Clubs were also an important source of connections (20%) in the Blue Mountains, as was church (16%). In Ku-ring-gai, twenty-eight percent of respondents use school to make social connections and 23% make social connections at church. In Ku-ring-gai, thirteen percent of respondents think clubs help to increase social connections. The ‘other’ source of connections includes local shops and common recreational places such as parks.

100% BlueMountains 90% 83% Kuringgai 80% 80%

70%

60%

50%

40%

29% 28% 30% 26% 27% 23% 20% 21% 20% 13% 16% 15% 11% 12% 10%

0% Club Church Community Neighbours Work School Other organization

Figure 82: Source of social connection.

256

To further understand the characteristics of their local neighbourhood, a question was asked about whether people help each other in an emergency in their community (Q36) (Figure 83). Forty-five percent of Blue Mountains respondents stated that community members always help each other when there is an emergency whereas in Ku-ring-gai 34% mentioned that they always help each other. Seventeen percent of respondents in the Blue Mountains and Ku-ring-gai indicated that their community members sometimes help each other. Only one percent of Blue Mountains respondents stated that they never help each other and in Ku-ring-gai no one mentioned that. Thirty-two percent of respondents in the Blue Mountains either have not experienced an emergency or they don’t know whether community members help each other during an emergency.

9% Idon’tknow 7%

Havenotexperienced 29% 25%

Neverhelpeachother 1%

11% Occasionallyhelpeachother 5%

17% Sometimeshelpeachother 17%

34% Alwayshelpeachother 45%

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Kuringgai BlueMountains

Figure 83: Community help each other in an emergency

When examining the level of participation in bushfire preparedness and mitigation issues (Q37), 55% of respondents in the Blue Mountains and 31% of respondents in Ku- ring-gai stated that community discussions were conducted about bushfire preparedness activities. Forty-five percent of respondents in the Blue Mountains and 69% in Ku-ring- gai stated that either their community does not discuss bushfire preparedness issues or they do not know about such discussions (Figure 84).

257

100% BlueMountains 90% Kuringgai 80%

70%

60%

50% 45% 40% 33% 34% 33% 28% 27% 30%

20%

10%

0% Yes No Idon'tknow

Figure 84: Community discussions about bushfire preparedness activities.

Three Likert scale questions (1=Strongly agree, 5= Strongly disagree) were asked about respondents’ perceptions of the bushfire safety, the protection of their community and the improvement of fire management activities in their communities over time (Q39, Q40, Q41). Table 20 illustrates the results. A majority of the respondents in the Blue Mountains and Ku-ring-gai neither agree nor disagree with each of the statements. On average 5% of respondents in the Blue Mountains and 3% of respondents in Ku-ring-gai strongly agreed with all three statements.

In both areas a majority of respondents are unsure about their community’s level of preparedness (Q39). In the Blue Mountains the mean value is below the midpoint of the scale (2.91) and in Ku-ring-gai it is right at the midpoint (3.09). In the Blue Mountains 36 % of respondents believed that their community is protected against bushfires while only 24% of respondents agreed in Ku-ring-gai. Thirty-eight percent of respondents in the Blue Mountains and 47% of respondents in Ku-ring-gai are unsure of their community’s safety. Twenty-seven percent of respondents in the Blue Mountains and 29% of respondents in Ku-ring-gai think that their community is not protected against bushfires.

258

When examining levels of satisfaction with community level bushfire management activities (Q40), the Blue Mountains respondents showed a level of satisfaction below the midpoint of the scale (Mean=2.84). Forty-two percent of respondents indicated (including 7% of responded who strongly agreed) that they are happy with the current bushfire management activities in their community. Thirty-one percent of respondents don’t know about community level bushfire management activities and 27% of respondents are not satisfied with current community level fire management activities. In Ku-ring-gai 28% of respondents (including 4% who strongly agreed) are happy with current bushfire management practices while 24% of the respondents are unhappy. Approximately half of respondents (49%) gave a neutral response, which may be explained by their level of awareness of current fire management practices. However, Ku-ring-gai respondents also showed a mean level of satisfaction that is lower than midpoint scale (2.95).

In both areas a majority of respondents were not able to judge the level of improvement of bushfire management activities in their community (Q41). Forty-three percent of respondents in the Blue Mountains and 57% in Ku-ring-gai were unable to provide a direct answer. However, 39% of respondents in the Blue Mountains and 25% of respondents in Ku-ring-gai believe that bushfire management activities in their community have improved since 2005 while 18% of residents in the Blue Mountains and 19% of residents in Ku-ring-gai do not agree with that statement.

Table 20: Community protection and bushfire management activities

Neither Strongly agree Strongly Agree Disagree agree nor disagree disagree My community is Blue 4% 32% 38% 22% 5% protected against Mountains bushfire events Ku-ring-gai 2% 22% 47% 21% 8% I am happy with current Blue 7% 35% 31% 23% 4% bushfire management Mountains practices Ku-ring-gai 4% 24% 49% 19% 5% Bushfire management Blue 7% 32% 43% 13% 5% activities have improved Mountains in my community since Ku-ring-gai 4% 21% 57% 13% 6% 2005

259

In the event of bushfires, community fireguard programmes that promise to reduce the vulnerability of residents are a key to community safety, preparedness and response by residents Boura (1998). When respondents were asked whether they would like to be involved in a community level fire guard programme (Q44), 33% of respondents in the Blue Mountains and 28% percent of respondents in Ku-ring-gai have stated that either they have already been involved in community level activities or would like to be a part of community level fire guard programmes. In the Blue Mountains, 21% of respondents indicated that they are not interested while 29% of respondents in Ku-ring-gai are not willing to take a part in community level fire guard programmes (Figure 85). In both areas, around 20% of respondents stated that although they would like to participate, they are unable to get involved as they are busy with other work. Twenty-three percent of respondents in the Blue Mountains and Ku-ring-gai mentioned other reasons for not becoming involved. The most commonly cited other reasons included health issues and age.

100%

90% 23% 23% Other 80% I'dlikebutIamtoobusy 70% No/Notinterested 23% 20% 60% Yes/Alreadyinvolved

50%

21% 29% 40%

30%

20% 33% 28% 10%

0% BlueMountains Kuringgai

Figure 85: Participation in CFU/fireguard programmes.

260

Community participation is an important factor that influences the level of community safety (McGee & Russell 2003; Pearce 2003). Community group approaches to bushfire preparedness are more likely to be responded to than individual approaches (Prior 2010). Community involvement provides motivation and enthusiasm, thus increasing the likelihood of creating a culture of safety in a locality (Boura 1998). Ellis et al. (2004) also recognised the importance of enhancing community participation in the fire management process. In both areas, a majority of respondents report never participating in bushfire prevention activities in their community (53% in the Blue Mountains and 66% in Ku-ring-gai; Q42, Figure 86). Eleven percent of respondents in the Blue Mountains and 6% of respondents in Ku-ring-gai participate frequently in bushfire prevention activities in their community. Although thirty-seven percent of respondents in the Blue Mountains indicated that they participate in such activities, they are not frequently involved, while 27% of those in Ku-ring-gai are also infrequently involved.

100%

90% Ineverparticipate 80% 53% Irarelyparticipate 70% 66% Isometimesparticipate 60% Ialwaysparticipate 50%

40% 21% 30% 16% 20% 16% 10% 11% 11% 6% 0% BlueMountains Kuringgai

Figure 86: Participating in community level activities.

To examine residents’ perceptions of best practices for soliciting community involvement in bushfire preparedness and mitigation, they were asked to select the method they thought would be the best practice for their area. In the Blue Mountains a majority of respondents would prefer to have small group activities within the

261 community (Figure 87) (Q46). In Ku-ring-gai, a majority of respondents (42%) think distribution of information through local letterboxes is a better practice than others. The second most commonly selected among Blue Mountains respondents is distribution of information through local letterboxes (27%) while the second best option for Ku-ring- gai respondents (22%) was considered to be small group activities within their community. A significant proportion of respondents from both areas (Blue Mountains: 17%, Ku-ring-gai 15%) think larger community level activities would be effective.

100% 4% 8% Other 8% 90% 5% 9% 8% 80% Generaladvertisingandmedia coverage(television,radio,etc)

70% Informationinlocalmedia(local 27% newsletters,noticeboardsetc.) 60% 42% Distributionofinformation 50% 2% throughlocalletterboxes

40% 17% Clubsandotherorganizational 2% levelactivities

30% 15% Largercommunitylevelactivities withallresidentsinapublic 20% 33% location Smallresidentgroupactivities 10% 22% withinthecommunity

0% BlueMounttains Kuringgai

Figure 87: Best practice for bushfire preparedness and mitigation involvement.

Residents were also asked about their views on bushfire management activities that need to be better developed in the future (Q47). A majority of respondents in both areas believe that education, training and communication, and safe prevention, preparation and suppression should be given priority (Table 21). The next most important area that needs to be developed is management of fire in the landscape. Equal importance was given to community self sufficiency for fire safety in both areas. The lowest priority

262 was allocated to the protection of people and property. Other important activities that people mentioned were the use of modern technology for risk commutation, individual risk assessments, and building community level profiles.

Table 21: Areas of community preparedness that respondents believe need further development

Blue Mountains Ku-ring-gai Safety, Prevention, Preparation and Suppression 55% 49% Education, Training and Communication 53% 51% Management of Fire in the Landscape 41% 38% Community Self-Sufficiency for Fire Safety 39% 39% Protection of People and Property 24% 34% Other 10% 7%

Availability of resources is a factor that facilitates levels of bushfire preparedness. A question was asked to examine respondents’ perceptions of available resources in their neighbourhood (Q48). A majority of respondents in both areas were satisfied with the available water supply in their neighbourhood (Figure 88). Twenty-seven percent of respondents in the Blue Mountains were dissatisfied with the road network while 20% were neither satisfied nor unsatisfied. In Ku-ring-gai 20% of respondents were dissatisfied with the road network and 18% were unable to make a clear judgment. While a majority of Blue Mountains respondents were satisfied with their level of fire fighting resources, in Ku-ring-gai only 43% were satisfied, but 45% of respondents did not rate their level of satisfaction with fire fighting resources. A majority of respondents believed that emergency warning and emergency response remains under-resourced in their community, though a significant proportion of respondents were unable to judge the quality of available resources.

263

Kuringgai 5% 36% 47% 7% 5%

BlueMountains 6% 32% 42% 19% 2% healthcare Emergency

Kuringgai 4% 22% 55% 11% 7%

warning BlueMountains 9% 33% 37% 18% 3% Emergency

Kuringgai 7% 36% 45% 9% 4%

BlueMountains 1%

resources 14% 52% 28% 7% Firefighting

Kuringgai 11% 50% 18% 16% 4% port Road BlueMountains 9% 43% 20% 23% 4% network/trans

Kuringgai 28% 51% 15% 5% 2%

BlueMountains 25% 56% 11% 7% 2% Watersupply 0% 20% 40% 60% 80% 100%

Highlysatisfied Satisfied Neithersatisfiednorunsatisfied Unsatisfied Highlyunsatisfied

Figure 88: Access to resources.

7.10.1.6 Community Preparedness and Related Issues

Unlike household level preparedness, community level preparedness is driven by many social factors such as social capital and sense of community. Respondents’ community level bushfire preparedness was linked with a number of factors. The results suggest an association between community level preparedness and female male ratio of households (2 = 13.13, df = 6, p = .041). More preparedness is evident if the ratio is one. This may mean that the involvement of couples with no children or elderly couples who live alone. Residents are more likely to be prepared as a community if they had received literature on bushfire preparedness (2 = 35.82, df = 2, p<0.001). In particular, bushfire preparedness information that highlights the importance of community level activities may influence community level preparedness.

Communities are better prepared if they consist of the residents who have previous bushfire experience (2 = 25.89, df = 2, p<0.001). People with previous bushfire 264 experience may work with other members of the community in order to minimise the bushfire threat. Risk perception is an important influence on community level bushfire preparedness. Respondents who perceive higher levels of risk were more likely to engage in community level preparedness activities (2 = 22.98, df = 4, p<0.001).

Results also revealed an association between community level bushfire preparedness and respondents’ perceived level of bushfire preparedness (2 = 47.6, df = 4, p<0.001). Community level preparedness is higher when respondents’ perceived level of preparedness is high. Availability of community level resources may influence the level of community level preparedness. The results suggest that perceived availability of resources is associated with a higher level of community level preparedness (2 = 20.64, df = 4, p<0.001). Respondents who believed that the availability of resources is satisfactory were more likely to participate in community level activities. Often community level bushfire management practices can be seen in places where social interactions are high. Respondents’ participation in community level bushfire preparedness activities is associated with their sense of community. Community level preparedness is high if residents’ sense of community is high (2 = 54.73, df = 4, p<0.001).

7.10.1.7 Shared Responsibilities

The roles and responsibilities of various bushfire management authorities and the sharing of responsibility between bushfire management authorities, residents and communities are topical issues, which are also being discussed at the policy level. The Victorian Bushfires Royal Commission (2009, p. 6) advocated shared responsibility, recognising the role of government agencies but also noting that: “communities, individuals and households need to take greater responsibility for their own safety and to act on advice and other cues given to them before and on the day of a bushfire.” Residents often tend to transfer their responsibilities to fire management authorities. With a lack of resources, fire management authorities may not be able to ensure the protection of all people living in areas at risk (Prior 2010). Therefore, in order to have effective bushfire mitigation, preparedness and response, fire management and

265 emergency management responsibilities need to be shared between relevant agencies and residents.

To understand residents’ willingness to share responsibilities for bushfire preparedness and mitigation activities and emergency response activities with relevant organizations, respondents were asked to identify the most responsible parties for bushfire preparedness and mitigation activities using a matrix provided in the questionnaire (Q45 and Q56). Figure 89 illustrates respondents’ willingness to share responsibilities for bushfire preparedness and mitigation activities. Interestingly a majority of respondents in both areas indicated that they are themselves responsible for prevention, preparation and suppression as well as community self-sufficiency for fire safety. The results clearly show that respondents believe that three key parties are responsible for safe prevention, preparation and suppression, residents, local councils and NSW RFS. Respondents perceived both NSW RFS and NSW Parks and Wildlife as being the most responsible for management of fire in the landscape. Most respondents believe that residents as well as CFUs and the local council are responsible for community self-sufficiency for fire safety. Both NSW RFS and NSW Fire and Rescue were perceived to be key parties to protecting people and property during a fire by a majority of respondents. A significant proportion of respondents has indicated that they themselves are also responsible for protecting people and property. Both local councils and NSW RFS were perceived to be responsible for education, training and communication. However, some respondents believed that seeking information is their own responsibility as well.

266

13% Kuringgai 31% 76% 56% 51% 27% 13%

Education, BlueMountains 31% 70% 81% 46% 34% Trainingand Communication

Kuringgai 41% 30% 40% 56% 76% 20%

Property BlueMountains 38% 37% 29% 80% 66% 19% Peopleand Protectionof

Kuringgai 54% 48% 58% 21% 23% 10%

FireSafety BlueMountains 63% 54% 46% 46% 16% Sufficiencyfor

CommunitySelf 8% 13% Kuringgai 19% 48% 60% 48% 45% 10% Fireinthe Landscape BlueMountains 19% 31% 87% 36% 52% Managementof

Kuringgai 63% 33% 64% 43% 38% 40%

BlueMountains 72% 34% 52% 67% 36% 43% Suppression Preparationand SafePrevention,

Residents CFU LocalCouncil NSWRFS NSWFire&Rescue NSWParks&Wildlife

Figure 89: Perceptions of responsibility for community protection and bushfire management activities.

To understand residents’ perceptions of the responsibilities for response and recovery activities during and after bushfires, they were asked to select the most responsible parties for each activity (Q56). Figure 90 illustrates the results for both areas. A majority of respondents had transferred their responsibilities to a fire management or emergency management authority. Respondents clearly perceived the NSW police, NSW RFS and NSW Fire and Rescue as being most responsible for disseminating early warning messages and assisting people to evacuate from the area. Some respondents also believed that it’s their responsibility. NSW Fire and Rescue and NSW RFS were perceived to have responsibility for fire fighting during a bushfire event. About 40% of

267 respondents believe that CFUs should share responsibility with fire management authorities. The local council was mostly perceived as having responsibility for providing relief and recovery needs and removing debris after a bushfire event. Thirty six percent of respondents believe that relief and recovery needs are their responsibility and 57% of respondents are willing to share responsibility for removing debris after an event.

3% Kuringgai 49% 23% 81% 21% 14% 5% Remove

anevent BlueMountains 65% 18% 66% 32% 16% debrisafter

Kuringgai 33% 19% 84% 14% 13% 15%

needs BlueMountains 39% 21% 82% 20% 10% 22% recovery Providing reliefand

Kuringgai 29% 22% 25% 29% 37% 83%

area BlueMountains 32% 23% 18% 44% 28% 88% Assisting fromthe evacuate peopleto 9% 6% Kuringgai 10% 36% 69% 89% 6% 10% BlueMountains 20% 43% 94% 76% Firefighting Kuringgai 33% 28% 33% 47% 53% 59% Early

warning BlueMountains 22% 23% 20% 76% 42% 51%

Residents CFU LocalCouncil NSWRFS NSWFire&Rescue NSWPolice

Figure 90: Community protection and bushfire management activities.

The roles and responsibilities of various bushfire service agencies have been discussed at various levels. Allocating responsibility among multiple parties can help to overcome risk management conflict issues during a fire (McLennan & Handmer 2012). Results suggest that the views on shared responsibility are associated with both community level preparedness (2 = 12.80, df = 4, p = .012) and household level preparedness (2 = 10.87, df = 4, p = .028). In other words, respondents stated that they are willing to take more responsibility for bushfire preparedness issues if their level of household and 268 community preparedness is high. Households comprised only of people above age 65 are unlikely to take much responsibility as these respondents generally consider themselves to be too old to take action (2 = 7.27, df = 2, p = .026). Those respondents who had received bushfire preparedness information were more willing to share a greater portion of the responsibility (2 = 15.76, df = 2, p<0.001). Respondents with previous bushfire experience were more likely to state that they are willing to take responsibility for bushfire preparedness (2 = 9.22, df = 2, p = .01). It is also evident that sharing responsibilities is also associated with respondents’ perceived level of risk. Respondents with higher levels of perceived risk tended to view themselves are responsible for bushfire preparedness measures (2 = 9.79, df = 4, p = .04).

7.10.1.8 Response and Recovery Actions

Individual decisions about preparedness and emergency response often minimise potential impacts (Paton 2003; Paton 2006b). If a bushfire threatens, with the ‘stay or go’ policy, residents have to take appropriate decisions while not putting their lives in danger (Victorian Bushfires Royal Commission 2009). A series of questions was asked to predict respondents’ behaviour during a time of emergency. Almost all respondents identified at least one immediate contact point to inform authorities about a fire event when they see an incident in their area (Q49) (Figure 91). For a majority of respondents (94%) in both areas, the first contact point would be “000”2, followed by their family, friends or neighbours. In the Blue Mountains people put faith in the NSW RFS, and 39% of respondents would contact them, while in Ku-ring-gai it is the NSW Fire and Rescue with 32%. In both areas the lowest priority was given to NSW Parks and Wildlife and their local councils. This is understandable because neither agency undertakes emergency response activities.

2 Emergency call service in Australia. 269

Family,friendsorneighbors 48% 51%

NSWparksandwildlife 7% 6%

Communityfireunit 13% 13%

NSWfirebrigade 32% 23%

NSWruralfireservices 18% 39%

Localcouncil 8% 4%

000 94% 94%

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Kuringgai BlueMountains

Figure 91: Immediate contact point.

People who are not confident about their level of available resources and hence the thought of defending their property, consider evacuation during life-threatening bushfire events. Often many people take the “wait and see” approach, hoping that no action will be needed after all (Prior 2010). Residents’ decisions to leave their property may be influenced by a variety of factors. Previous studies have highlighted that the most dangerous option and the cause of most fatalities during a bushfire emergency is last minute evacuation (Handmer & Tibbits 2005). Knowledge of residents’ perceptions of evacuation, their evacuation potential and intended evacuation process provides insight into their capacity to evacuate safely. Such information can be useful for evacuation planning at any given point. A Likert scale question was asked to investigate whether respondents would evacuate in the event of a life-threatening bushfire (1=strongly disagree, 5=strongly agree) (Q50). Results show that in both areas people state that they would be likely to evacuate during a fire event, with a mean value of 4.3. However in the Blue Mountains, 9% of respondents would stay while in Ku-ring-gai only 3% have decided to stay even during a life-threatening fire. Interestingly, 8% of respondents in the Blue Mountains and 9% respondents in Ku-ring-gai have yet to take a decision 270

(Figure 92). The results further highlight the importance of emergency evacuation planning as the majority of people state they have decided they would evacuate in a life- threatening bushfire event.

100% 1% 3% 2% 6% 9% 90% Stronglydisagree 8% Disagree 80% Neitheragreenordisagree 20% 70% 34% Agree Stronglyagree 60%

50%

40%

30% 62% 53% 20%

10%

0% BlueMountains Kuringgai

Figure 92: My family would evacuate in an event of life threatening bushfire.

Often evacuation efforts take place too late, placing people at extreme risk. During the evacuation process people may be advised to evacuate, or may be forcefully removed from their homes under threat without any precautionary measures (Goudie 2007). Bushfire awareness, including about physical aspects of the fire, may help people to decide the right time to evacuate. The best time to evacuate is to go when the fire is so mild that there is little or no danger. This means that any evacuation should be carried out when you know there are bushfires in nearby areas, and there is any possibility of them spreading to your area (Goudie 2007). “When bushfires are burning on days when ‘extreme’ or ‘catastrophic’ fire danger ratings are expected, leaving early may be the only safe option, even for people planning to defend well-prepared buildings” (Killalea & Llewellyn 2010, p. 12). When a question was asked about when they think is the right

271 time to evacuate (Q51), a majority of respondents in the Blue Mountains (54%) and in Ku-ring-gai (58%) indicated that they would wait until the authorities ask them to leave (Figure 93). This highlights the importance of early warning and communication during a time of emergency. Twenty-four percent of respondents in the Blue Mountains wait until fire approaches the edge of their suburbs and 6% would evacuate when they feel smoke and heat. In Ku-ring-gai, 17% would evacuate when fire reaches the edge of their suburb while 10% would evacuate when they feel heat and smoke. The results do not indicate that evacuation decisions are influenced by neighbours’ actions. During the interviews, a number of people also mentioned that this decision is made based on their personal judgment, from experience and given the circumstances.

10% Other 13%

58% Iwaituntilauthoritiestellmetoleave 54%

5% Whenneighboursstarttoevacuate 3%

10% WhenIfeelheatandseesmoke 6%

Assoonasfireapproachestheedgeofthe 17% suburbs 24%

0% 10% 20% 30% 40% 50% 60% 70% 80% 90%100%

Kuringgai BlueMountains

Figure 93: When is the right time to evacuate?

It is important to have identified the best evacuation route as well as alternatives for safe evacuation before evacuation is required. A majority of respondents in both areas have identified evacuation routes (Q54). In the Blue Mountains, 83% of respondents have identified a potential evacuation route. However, in Ku-ring-gai, only 57% of respondents have identified their evacuation route. In both areas, 96% of respondents 272 would leave using their own car. However, in both the Blue Mountains and Ku-ring-gai some respondents would seek assistance to evacuate.

It is also important to identify a point of evacuation before any fire event. This point of evacuation may vary depending on fire conditions on the day of evacuation. However, knowing a place of evacuation would help people to make their decision quickly during a time of evacuation. A majority of the Blue Mountains respondents would evacuate to the identified evacuation point (42%), while 34% of respondents would evacuate to a friend or a relative’s house (Figure 94). In both areas, 15% of respondents are uncertain about to where they would evacuate. In the ‘other’ option, respondents mentioned that they couldn’t provide an exact location because it all depends on circumstances on the given day.

100% BlueMountains 90% Kuringgai 80%

70%

60%

50% 47% 42% 40% 34% 30% 24% 20% 15% 16% 10% 5% 6% 2% 3% 3% 4% 0% Hotel/motel Uncertain Identified Community Friendor Other evacuation centre relative’s point house

Figure 94: Point of evacuation.

During a bushfire people usually have two safe options; leaving early or staying and defending their well-prepared properties. Defending their own property is the choice for many physically fit and emotionally prepared people, although it will still pose some risk. If they do not evacuate, adequate resources and preparedness are necessary to 273 defend successfully (Killalea & Llewellyn 2010). A question was asked to understand what reasons would trigger a stay and defend decision (Q55) (Table 22). A significant portion of respondents noted that they would stay to save their property (43% in the Blue Mountains and 49% in Ku-ring-gai). Respondents also indicated that they would stay because they have faith in bushfire management activities (20% of Blue Mountains respondents and 24% in Ku-ring-gai). Seventeen percent of respondents in the Blue Mountains believe that it is safer to stay than to evacuate while in Ku-ring-gai 13% believed this. Surprisingly, results indicate that only small number of respondents have a fire bunker at their properties despite the fact that these residents live in a high fire risk area.

Table 22: Reason not to evacuate Blue Mountains Ku-ring-gai To save the property 43% 49% We have a fire bunker 4% 1% It is safer to stay than to evacuate 17% 13% I have faith in bushfire management 20% 24% activities

7.10.2 Household Interviews and Focus Group Discussions

A number of household interviews were undertaken to identify whether there are any other important issues that were not addressed by the mail survey. It also helped to further develop deeper understanding of issues addressed by the questionnaire. These interviews were undertaken with both CFU members, and non-CFU members in each community. Issues identified by community members included a lack of community leadership, lack of community knowledge about evacuation planning, insufficient resources and lack of community participation in bushfire preparation. Community members also demonstrated some conflicting views on the latest developments in emergency management and methods of information dissemination during a time of emergency. In both areas community members raised the issue of the involvement of their local councils in bushfire prevention. They generally believed that the councils should do more. They identified issues such as fuel management, green waste collection and waste management, which are mainly controlled by their local councils. 274

Although respondents in both areas agreed that the council was doing a good job in relation to bushfire management, they believe that more could have been done. Respondents highlighted that they are having problems with disposing of green waste once they clear vegetation around their properties. Accumulation of debris in the UBI increases the level of fuel and bushfire risk, and was considered to be a critical issue. This was demonstrated by following responses:

“We do what we can do for cleaning the property, but the council should collect green waste. Otherwise waste around the property would increase our risk” (Respondent 5, non CFU member, Ku-ring-gai).

“More council cleanup days and frequent council green waste collection is a must before bushfire season starts. It will also encourage people to clean their properties” (Respondent 2, CFU member, Blue Mountains).

Therefore, respondents believed that council should consider this issue in order to take action to provide adequate services for green waste disposal and debris removal.

Some residents highlighted that regulations imposed by the local council made it difficult for them to clear overgrown vegetation around their property in order to minimise fuel loads and reduce the potential impacts of fire. This was mentioned by a majority of residents and is demonstrated in the following quotes from the interviews demonstrate:

“Council regulations are too strict on clearing vegetation around properties. It seems they are more concerned about vegetation rather than properties and lives” (Respondent 4, CFU member, Blue Mountains).

“All services need to be improved when it comes to realistic tree pruning for the protection of the property. Council needs to upgrade the poor judgement on allowing trees to be protected, especially Turpentine trees, which grow to a hedge height and explode during a bushfire” (Respondent 6, CFU member, Blue Mountains).

275

“There needs to be a balance between a safe environment and preservation” (Respondent 3, non CFU member, Ku-ring-gai).

On the other hand, there were also residents who prefer to live in a natural environment amongst trees and plants. They believed that bushland is one of the best aspects of the area and do not want to clear vegetation. These respondents also stated that they would evacuate with only their valuable documents if there is a fire. They do not worry about their property as it has been insured, and hence they have transferred the risk.

Some respondents showed support for controlled burning as a bushfire management strategy. However, there were other respondents who strongly believed that controlled burning is not the best solution. They are concerned about wildlife and believed that fire fighting and emergency response activities need to be strengthened to minimise losses instead. Some respondents also mentioned health issues such as asthma related to controlled burning. Some positive and negative responses are given below:

“Back burning needs to be increased and maintained” (Respondent 5, non CFU member, Ku-ring-gai).

“I believe more fuel reduction burning would be done in the areas of high bushfire risk” (Respondent 1, CFU member, Blue Mountains).

“I think controlled burning is the least beneficial to the community” (Respondent 7, non CFU member, Blue Mountains).

Respondents in the some parts of the urban edge were not happy about their water supply. Although most residents have an alternative water source (water tank), many of them believed that it is not sufficient in a severe fire event. Therefore, they highlighted the lack of water hydrants in their area as a critical issue for councils to address. For example:

“What if we run out of water during a severe bushfire?” (Respondent 7, non CFU member, Blue Mountains). 276

“I am pretty sure the water pressure won’t be good enough and do not expect an uninterrupted water supply” (Respondent 1, CFU member, Blue Mountains).

Respondents also criticized council regulations that do not enforce the maintenance of properties. Some people raised the issue of unprepared neighbours, whether they are other residents or public land managers. One respondent noted:

“Although we undertake bushfire preparedness measures to be well prepared if the neighbours do not clear their waste and vegetation then we are all stuffed” (Respondent 2, non CFU member, Blue Mountains).

“There should be an asset protection zone in all bush areas adjoining homes on crown land” (Respondent 5, non CFU member, Ku-ring-gai).

In the Blue Mountains almost all respondents living at the UBI were well aware of the threat from bushfire in their area. They also perceive that they are at high risk. The Blue Mountains residents highlighted many issues throughout the interviews. Their main worry was the limited connectivity of and access to the road network. The Blue Mountains City Council area is connected to other points by one highway (the Great Western Highway) and for many residents it is the only accessible evacuation route during a fire event. This limited access between and across suburbs reduces connectivity while creating obstacles to movement during a bushfire event. It is often closed during bushfires, which makes conditions worse and many households fear that they would be trapped during a time of evacuation: a critical issue.

“Just one way, the Great Western Highway. I would expect more traffic and might get stuck in the middle of the highway during an evacuation” (Respondent 3, non CFU member, Blue Mountains).

Many residents highlighted the importance of evacuation planning. However, a majority of residents in the Blue Mountains are not aware of any evacuation planning procedures done by the fire management authorities. They are also not sure about how they would receive an early warning message. 277

“We have no idea from where we would get an early warning message” (Respondent 5, non CFU member, Ku-ring-gai).

“I would rather switch on my radio or TV to get latest information about warnings” (Respondent 4, non CFU member, Blue Mountains).

Therefore, it is important to have a mechanism to disseminate early warning messages that reach every resident in the community.

During the survey, it became clear that residents in both areas had not considered recovery planning as a preparedness measure. Most respondents have not yet discussed recovery plans as they believe that it is highly unlikely their properties would be burnt out. In case of any recovery needs, a majority of residents have a faith in their insurance and would expect some assistance from the council to clear debris, find temporary shelter, etc. during early the stages of relief and recovery phases of an event.

“I always keep my personal effects somewhere else and I have fully insured the property. During a fire I would simply evacuate and would not worry about the property” (Respondent 8, non CFU member, Blue Mountains).

“Insurance does not cover clearing debris. Residents need some external help to clear debris soon after a fire. I think it’s council responsibility to help people” (Respondent 3, non CFU member, Ku-ring-gai).

The Blue Mountains is a perfect holiday location. Therefore holiday houses are common in the area. These properties are often not attended and are only occupied during vacations or school holidays. Such properties are easy to identify, with rubbish/leaves all over their front gardens, overgrown vegetation around the house and gutters that are visibly full of leaves. Full-time residents believe that these properties put their properties in danger and it is council’s responsibility to carry out inspections to clean those properties. A lack of community interaction and community level preparedness activities was observable in the streets with high numbers of holiday homes. 278

“The trees from the holiday home next door grow over my roof. It’s not well maintained. Council should enforce this” (Respondent 2, non CFU member, Blue Mountains).

Respondents who live in a rented property stated that preparation is their landlord’s responsibility. Some respondents reported that they couldn’t clear vegetation around their house because it is beyond their control. Renters were also reluctant to spend time or money on household preparedness. They indicated that their priority is their valuable possessions rather than the property itself. This issue was common in both study areas.

“It is not my responsibility to clean vegetation around the property and I am not allowed to. The owner is responsible for clearing the vegetation around this property” (Respondent 2, non CFU member, Blue Mountains).

“I would not worry about this property because it does not belong to me. I would worry more about the safety of my family” (Respondent 5, non CFU member, Ku-ring-gai).

In the Blue Mountains, communities are actively involved in bushfire management issues and often they work at the street level. Different community level initiatives were identified during the survey. The CFU is a prominent community level body that is responsible for community bushfire preparedness. Overall, perceptions of the CFU programme were very positive. Respondents appreciated the availability of fire fighting equipment and the training they were offered. Both member and non-member residents had some faith in their local CFU. For example, one respondent noted:

“CFU here is excellent. Great training and education” (Respondent 10, non CFU member, Blue Mountains). CFUs serve as an access point to fire expertise for a community. Residents also indicated that becoming involved with a CFU allowed them to improve their connection to other community members. However, some respondents did not see the CFU as a part of their community. They rather described it as a separate group of people. Therefore, strategies need to be taken to strengthen CFUs as a strong community body.

279

“I have no idea what a CFU is. This is a closed community and people are on their own. I can’t find any community leader role here” (Respondent 15, non CFU member, Ku-ring-gai).

In the Blue Mountains in the streets where there is no CFU, respondents showed interest in community level preparedness activities. However, the presence of community level leadership was identified as a critical factor that triggers community level preparedness activities.

“We have a CFU in this area. Mr. XXX is the leader of our CFU. He is very capable of organising community activities. Although I am not a member of CFU I feel safe here because of the work that they have done” (Respondent 12, non CFU member, Blue Mountains).

Often community leaders initiate and coordinate activities with the fire management agencies while disseminating information to other community members. A community with a lack of leadership is unable to organise a community group that connects with fire management authorities. During the survey no community level bushfire preparedness activity was noticed if a leader was absent.

In the Blue Mountains a majority of residents maintained a good relationship with their neighbours. Residents of some streets (especially in cul-de-sacs) interacted well with each other. Residents in those streets stated that they used to organize street level activities such as Australia Day BBQs and Christmas lunch. These events create a sense of community amongst individuals and will help residents to share their experiences and knowledge of bushfire events. This closer association between residents will increase their willingness to work together as a community to minimise bushfire risk. CFUs were more likely to be found in the streets with high level of social interaction. CFUs were not formed in streets where isolated residents live since it was hard to find interest and initiative.

Although CFUs function in Ku-ring-gai, they do not operate as a broader community unit. Ku-ring-gai is much closer to the Sydney CBD and the UBI is much more urban in 280 nature. Therefore, community level activities do not function well throughout the UBI compared to the Blue Mountains. Most of the residents keep close relationships only with their neighbours. This has made community level initiatives much more difficult to initiate in the Ku-ring-gai area.

Residents who had experienced bushfires were actively involved in community level activities. Often people who had fire experience had become community leaders in fire management. It was also noticed that service occupants (e.g., State Emergency Services, NSW Fire and Rescue) who are involved in fire management or emergency response also work actively on household and community fire management issues. These residents often check whether their resources, such as pumps, hoses, and water tanks are in sufficient working order to combat fires.

“I have experienced many fires and have seen the impact. I do not want to see another disaster here. It is always better to be prepared for the next fire event”(Respondent 1, CFU member, Blue Mountains).

Despite the importance of integrating vulnerability and risk management into the development process, some development projects are being continued without considering bushfire management issues. A retirement village that was built in close proximity to bushland in Ku-ring-gai demonstrates this problem. Lack of access to evacuation routes and increased need for external help may put the residents of that retirement village at high risk if there is a bushfire threat.

During the survey, some duplication of community level activities was identified. Some examples include: CFU (NSW Fire and Rescue) and StreetWise (NSW RFS) Static water supply (SWS) (NSW Fire and Rescue) and SWS (NSW RFS)

These programs may replicate the same function and thereby confuse residents. Liaisons between the NSW Fire and Rescue, NSWRFS, council, community and other groups would promote more efficient service with positive impacts. During the survey, it became evident that different sources of information are available and published by

281 different organizations. Residents are sometimes confused about which one to choose. Residents suggested having one common portal rather than having different communication channels.

“There is a lot of information coming from a lot of sources. Coordinated one stop information assistance would be more useful” (Respondent 15, non CFU member, Ku- ring-gai).

7.11 Discussion and Limitations

The survey was executed among a random sample of households living within a 30m buffer zone of the UBI. Because of limited resources and time constraints, the expected sample size and response rates were considered before conducting the mail survey and guided the number of survey that were distributed.

A demographic profile of the residents living in the buffer zones was not available in order to check the representativeness of the sample. To overcome this issue a demographic profile was generated using census collection districts (ABS Census 2006) that overlap with the buffer zone, assuming the CD represents the buffer zone. However, the demographics of the buffer zone may not be representative of the entire census collection district area. Therefore it is possible that the estimates generated by the sample may deviate from the true population parameter and therefore generalizations should be done with care.

Although an adequate sample size was obtained overall, for some groups, there were few responses (e.g., renters). For these groups, sample size was not sufficient to perform chi square tests. Therefore these associations could not be investigated.

Levels of household preparedness, community preparedness and views of shared responsibilities were determined by aggregating selected variables Table 15 into a composite variable. Although aggregation gives a useful summary, there is a chance of losing some information during the process because it may not allow reflect all of the properties of the original variables.

282

In some areas residents have not experienced a severe bushfire for quite a long time. In those areas bushfire is often less prominent to people due to a lower frequency and impact. This may have affected their responses and answers.

These limitations should be taken into consideration when viewing and using the results of this study.

7.12 Conclusions

This study was undertaken to identify selected individual and social factors that influence household and community level disaster preparedness at the UBI based on the bushfire preparedness model discussed in Figure 59. It also identified factors affecting perceptions of shared responsibilities. The results of this study provide a snapshot of household and community level bushfire preparedness. Data was collected through household surveys using a mailed survey questionnaire and face-to-face discussions. Statistical analysis (descriptive statistics, cross-tabulations and chi square tests) was applied to explore its results and identify associations between variables. This approach can help fire management authorities to have an insight into household and community level decision-making in bushfire management. It also investigated issues with current bushfire management activities and identified areas where bushfire management might be improved.

As identified in Figure 59, the findings of this study suggest that there are multiple determinants of household level preparedness, community level preparedness and sharing bushfire management responsibilities. These factors contribute to overall bushfire preparedness in the area. These factors are often hidden and hard to capture using census data. It is important to understand these characteristics prior to any bushfire management programme in order to address the actual needs of the residents. It found associations between household level preparedness, community preparedness and shared responsibilities. Household level bushfire preparedness activities are associated with; household type, number of people staying at home during the day, information on bushfire preparedness, years of residency, previous experience with bushfires, perception of risk, perception of preparedness, sense of community and the

283 level of education. These results explain the household level factors identified in the proposed bushfire preparedness model. Community level bushfire preparedness activities are associated with; ratio of females to males, information on bushfire preparedness, bushfire education, previous experience with bushfires, perception of risk, perception of preparedness, availability of community resources and sense of community. These associations explain the community level factors identified in the proposed model. Views on shared responsibility are associated with; household level preparedness, community level preparedness, elderly people, information on bushfire preparedness, years of residency, previous experience with bushfires, and perception of risk.

The complex set of social, economic and environmental development at the urban bush interface has set dynamic interactions between the residents, bushfire management agencies and emergency services with regard to bushfire prevention and preparation. According to the proposed model, perception is a key driver of preparedness. The results also revealed that the perception of bushfire risk is an important component that influences household level preparedness, community level preparedness and views of shared responsibilities. Perception of bushfire risk varies between residents even when they live in the same area. These perceptions need to be identified in order to address local bushfire management issues. Bushfire management authorities need to consider appropriate and effective risk communication strategies in order to educate people about the actual level of risk to which residents are being exposed.

These results suggest that some bushfire management issues are related to local issues such as availability of local resources. Knowledge of such local issues is necessary to support in order to have an effective bushfire management practices these local issues need to be understood. Communities are ready to share responsibility with fire management authorities. Therefore, community based initiatives can be implemented through community based bushfire management activities. Residents also would like to have clear guidelines on what to do and what should not be done including fire management resources prior to a bushfire event. At present, it is bushfire management authorities who brief residents about these responsibilities and protocols. However understanding community willingness to share responsibilities will help to promote 284 community level bushfire management initiatives. Therefore, it needs to be looked at as a bi-directional rather than a top down approach.

Bushfire prevention information has a direct impact on household level preparedness, community level preparedness and views on shared responsibilities. Residents have accepted that they are expected to take household level initiative and not heavily depend on fire management authorities. Many residents also would like to take part in community level activities. However, people are not very aware of bushfire prevention guidelines as they had little information about them. Bushfire prevention messages have not reached every corner of the UBI. Therefore bushfire management authorities should identify strategies to transfer bushfire prevention information more effectively.

Previous bushfire experience motivates residents to prepare for bushfires, and become a member of a CFU, hence participating in community level fire management activities. People with experience need be considered to be community resource and sharing their knowledge and experience with other community members would help other people to understand what to expect during a fire and its potential impact. Bushfire management authorities should consider initiating strategies to transfer previous bushfire knowledge and lessons learnt activities to increase community education.

Formation of community fire units has increased the faith of the UBI residents in their ability to respond to fires themselves. Community leadership is critical in initiating community level bushfire mitigation activities, including the formation of CFUs. Long periods without any significant fire activity may decrease the level of interest within the community. It is fire management authorities’ responsibility to keep residents motivated for bushfire management.

Most residents did not have an evacuation plan. People do not have any idea about modes of communication during a bushfire event and the enforcement of evacuation orders. Residents have also not focused much on recovery planning. Bushfire management authorities should focus on enhancing residents’ emergency response and recovery capacities. Early warning issues and community level evacuation drills need to be practised at least once in a while. 285

The results of this study highlighted gaps in the bushfire management activities coordinated by local government councils. Knowledge of these gaps will help councils to take additional steps forward to improve fire management in their areas. It also highlighted the importance of integrating bushfire management into the development approvals process. Local government councils have to play a key role in developing new policy that would help to minimise the bushfire risk within future development projects.

The results of this study demonstrate the importance of collaborative approaches to managing bushfires at the UBI. This can be implemented at the community level involving all responsible parties; local council, NSW RFS and NSW Fire and Rescue. Such activities will minimize duplication and enhance the effectiveness of programmes.

286

Chapter Eight: Conclusions of the Research

8.1 Introduction

Bushfires are unpredictable events. The magnitude of catastrophic impacts in terms of loss of life and livelihoods, destruction of property and environmental damage on local communities is significant. To minimise the potential impact of future events, and thus effectively manage bushfire risk, there is a high demand for understanding bushfire risk in both holistically while incorporating scientific knowledge. This research presented an integrated bushfire risk assessment model to understand risk at the UBI and a bushfire preparedness model to understand the decision-making processes of residents. Both the models and other results of this study can help to facilitate effective bushfire risk management decision-making. In the previous chapters, the process of developing an integrated risk assessment, a bushfire preparedness model and their results were discussed. It explored the importance of each model, limitations and the refinements that might improve the results of each step. This final chapter concludes with a summary of the overall process, contributions and potential future work related to this study.

8.2 Summary of Results

It is now evident that the cause of bushfire disasters is not simply the severity of the fires themselves. Disasters are also shaped by the level of vulnerability of the population (Whittaker 2008). Based on the extensive literature review done in Chapter 2, bushfire risk was identified as an interaction between a hazardous bushfire event and vulnerable social conditions. Vulnerability was defined as pre-disaster conditions of a population that amplified its level of susceptibility to bushfire impacts. This research delivers an integrated bushfire risk assessment framework that can be applied within the bushfire risk management process (Figure 11). The framework highlights the fact that bushfire risk management is not simply about suppressing fires. It suggests that there are other deeper causes such as social and economic fragilities, community capacities and social cognitive processes, expectations and actions. Clearly, a deeper

287 understanding of those elements is required for effective bushfire risk mitigation and response.

To understand the elements of risk, an integrated bushfire risk assessment model was developed (Figure 36). In this model, bushfire risk was considered to be a product of bushfire hazard and vulnerability. Bushfire hazard consists of two components; ignition probability and fire severity. Vulnerability is determined by exposure and physical susceptibility, social vulnerability and emergency response and coping capacity. Each element was analysed separately and then the elements were combined using a fuzzy multi-criteria evaluation method. The results demonstrate the spatial variation of hazard, vulnerability and risk at the UBI. The criteria maps clearly explain the underlying causes of risk and vulnerability. It is evident that the bushfire hazard vulnerability and risk are not equally distributed even within the buffer zone. They also show that bushfire risk varies not only with the level of hazard but also with the other dimensions of vulnerability such as exposure and physical susceptibility, social vulnerability and emergency response and coping capacity. Recognition of these dimensions and their contribution toward overall risk could be used for bushfire risk management, social improvement, and mitigation for exposure to potential bushfires.

This research has highlighted the importance of GIS for assessing and visualising of bushfire risk, and for facilitating the decision making for bushfire risk management at the LGA level. The model was developed using a standard set of indicators based on available data. The spatial model was applied in two LGA areas, which shows applicability in the local context and shows promising results. Therefore, the model can be repeated at LGA level in other locations. The model can also be applied at the state level. However further research is needed to identify whether the model will generalize to that scale, or needs to alter the variables used to fit state level goals and resources.

Hazard assessment in this model was based on a bushfire history database. Therefore, the quality of records in that bushfire history database influences the risk assessment result. Existing bushfire inventory databases do not provide comprehensive information on particular fire events. They only provide the area burnt and date of ignition. These existing bushfire history databases could be enhanced by adding information such as the 288 ignition point, cause, and damage or loss. Thus, an attached incident report would help users to understand the nature and impact of the fire. Therefore more research is needed to validate, enhance and update existing bushfire history databases.

In the method used in this research, GIS based fuzzy multi-criteria evaluation techniques were extensively utilised. Although conventional spatial multi-criteria evaluation methods have been used for many other natural hazard risk assessments (Abella & Van Westen 2007), fuzzy multi-criteria modelling is a rather new approach in disaster risk assessment. It is based on interfacing fuzzy and crisp modelling with GIS, which enables soft risk decision-making. So far it is been more widely used in land suitability assessment (Gbanie et al. 2012; Nisar Ahamed et al. 2000). It consists of several key steps such as standardising, criteria evaluation, weighting and integration, which generates different intermediate outputs and a single final output. This process allows for the incorporation of expert judgment into the bushfire risk assessment using tools such as pairwise comparison techniques and fuzzy models. Fuzzy models are capable of describing both subjective (knowledge-driven) and deterministic (data- driven) information. This thesis has demonstrated the application of GIS based fuzzy multi-criteria modelling in disaster risk assessment and its advantages. Therefore, a GIS based fuzzy multi-criteria evaluation technique can be considered to be a suitable solution for understanding the spatial variation of risk at the local level when other suitable tools to support bushfire risk management decision making process are not available due to lack of data, resources and/or expertise.

Weights used when aggregating the data layers can affect the outcome of the analysis. Expert judgements were used to calculate weights in this study. The expert judgments of this study were made with the intention of minimising potential impacts on people and properties at the UBI. Those judgments play a key role in all analytical steps. Most of these decisions were taken after discussion with relevant experts who have prior knowledge about bushfire management. Thus, it is a case-specific process mainly determined by the perception and acceptance by the decision makers (Fekete 2011). Therefore, results of this study are mainly suitable for bushfire risk management. However, the study is replicable for other decision-making contexts such as habitat risk

289 management and environmental risk management. In such cases, expert judgments need to be made by decision-makers with expertise in those domains.

The bushfire risk assessment model used in this thesis was designed to work with the minimal resources and data available at the local level. It can be further expanded by exploring more variables that are relevant. However, the mathematical combination and the relative importance of variables may vary with the inclusion of new variables. Therefore, the model would need to be refined if more variables are added.

Difficulty of validation in risk assessment results is an issue raised in the literature (Abella & Van Westen 2007; Grünthal et al. 2006). Validation requires certain data sources that better account for the spatial variation of disaster loss. However lack of bushfire loss inventories and information on past bushfire incidents hinders the validation process for this study. Moreover, the results of this study estimate the existing level of bushfire risk across the area. Validation based on previous events may not reflect the present situation. As conditions changed over time and new housing developments are built, the validation of such models become more challenging (Abella & Van Westen 2007).

The findings of this research are delivered in a map form that shows the spatial variation of hazard, vulnerability and risk across the study area. There is no nationally or internationally recognised standard format for risk mapping (Abella & Van Westen 2007). In this study, hazard, vulnerability and risk zonation maps were generated using the quantile classification method. It assigns the same number of pixels into each zone. Maps are unique and powerful tools for visualization and communicating results (National Research Council - Committee on Planning for Catastrophe a Blueprint for Improving Geospatial Data Infrastructure 2007). Risk reduction decisions are often based on policymakers’ interpretations of the map. Therefore mapping risk assessment results is an important step in risk assessment.

The resulting maps only provide a relative measure of bushfire risk in the area. Maps can only be used to compare the areas within a LGA. Although the same methodology is applied for another LGA, areas across the LGAs are not comparable. For example, 290 maps of Ku-ring-gai LGA cannot be directly compared with the maps of the Blue Mountains LGA.

The risk assessment presented in Chapter 6 is a scientific method for understanding bushfire risk at the UBI. As discussed in Chapter 7, residents often view risk differently and their estimation of risk is socially constructed. For effective bushfire risk management, the results of the scientific risk assessment process and residents’ decision-making processes need to be linked. Household and community level decision- making processes are influenced by a number of social factors. The third objective of this thesis was to understand household and community characteristics that influence household and community decisions about bushfire management activities. By conducting an extensive literature review, a bushfire preparedness model that explains household and community level bushfire preparedness was developed (Figure 59). The model identified key factors and their relationships that influence local level bushfire preparedness.

The results of the primary survey that was conducted to understand household and community characteristics that influence decisions about bushfire management activities at the local level identify the factors that affect household and community level decision-making process. It clearly illustrates an association between identified social characteristics of the household and community level bushfire preparedness. In other words, the survey results helped to validate the proposed bushfire preparedness model. The bushfire preparedness model makes an important contribution to the field of bushfire risk management at the local level. The local level bushfire preparedness process should ensure that it addresses specific local level issues such as risk communication, local level education and training in order to increase bushfire preparedness and minimise bushfire risk at the local level.

Building upon the social vulnerability factors identified in Chapter 5, the methodology in Chapter 7 enabled the testing of associations between identified social vulnerability factors and household and community level preparedness. It further identified that some characteristics that influence levels of preparedness are not identified by social vulnerability factors. Variables such as disaster experience, risk perception and 291 available information also influence the level of preparedness. However these factors are not captured in traditional vulnerability indices. The results also confirmed that some identified social vulnerability factors influence levels of preparedness. This study only focused on preparedness issues. But social vulnerability also influences response and recovery. Further research is needed to understand social vulnerability factors and underlying causal relationships between vulnerability, response and recovery.

The results of both the risk assessment and the understanding of the decision-making process of people living at the UBI provide two different perspectives on managing bushfire risk. Both results provide necessary information for bushfire risk managers. Successful bushfire risk mitigation programmes must deal with both measurable risk as well as decision-making processes. They should encompass a wide range of risk reduction options, targeting various locations, social groups and specific priorities. Therefore, the mixed methods approach used in this research provided more useful information for an effective bushfire risk management than typical risk assessment models.

Interagency relationships are essential components for reducing vulnerability and risk (Pelling 2003). Gillen (2005) stated that politics and the silo-mentality of NSW government institutions leads to their resistance and general reluctance to engage in collaborative efforts. Unfortunately this is valid for bushfire management practices as well. During the study the importance of collaborative efforts between Local Government, the NSW RFS and NSW Fire and Rescue to implement more effective bushfire management programmes such as risk communication and community level fire mitigation in the future became evident.

8.3 Future Research

The key research need emerging from this research is that of a deeper examination and understanding of risk at the UBI. The study is one of the first steps towards local level bushfire risk assessment using an integrated approach in Australia. Although the research illustrates spatial variations in hazard, vulnerability, and risk and underlying

292 causes, the relationships between these causes are not well established. Understanding such relationships also helps in selecting indicators.

The population dynamics of vulnerability were not touched upon in this research. The daytime/nighttime population dynamics and seasonal changes in population were not considered. Behaviour of the population and the movement of population during an emergency provide important information for understanding risk. Incorporating these variations into the interaction between vulnerable communities and potential bushfire scenarios would create a more realistic representation of bushfire risk. The model presented in this study can be further developed to capture this dynamic nature of risk. Bushfire simulation models are widely used to explain variability across spatial and/or temporal units of analysis (Mercer & Prestemon 2005). These simulation models produce more than just fire propagation. They also deliver the fire intensities at each point, spot fires, etc. Agent-based models are used to model human and systems behaviour during or after disaster events (Fiedrich & Burghardt 2007). A combination of both bushfire simulation models and agent-based models based on the developed framework will provide an ideal platform for understanding these dynamics. However, developing such models is both time and resource consuming. Therefore, the application of such models at the local level is at best a long-term objective.

The survey results could have been related to socio-demographic characteristics and integrated into the spatial data. For example, address information could have been collected from survey respondents and their home locations geocoded. In that way, the household results could have been fed back into the integrated vulnerability assessment. So, greater integration of the multi-method elements could potentially be achieved. However sample size, integration method, and ethical issues would remain challenging.

The present study represents bushfire risk based on prevailing conditions at the UBI. Various indicators of exposure and physical susceptibility, social vulnerability and emergency response and coping capacity were used in this study. Those indicators are not static in nature and may increase or decrease over time. Thus, the relative importance of these indicators also may vary with time. This may affect the pairwise comparison results and final weights. Transforming vulnerability and risk assessment 293 into a continuous monitoring system would facilitate understanding these temporal variations. Disaster managers and emergency planners must therefore develop and apply appropriate monitoring systems that enhance effective bushfire risk management plans and measures.

8.4 Use of this Study

There are many potential uses for this research including bushfire risk management, emergency planning and disaster response. The estimation and assessment of vulnerability and risk will remain key issues in the future, especially with regard to future development at the UBI and the implementation and expansion of community level fire management initiatives. It could be used for environmental justice analysis to show how different vulnerable communities are exposed to potential bushfire hazard. It could also be used for urban planning to limit development in very hazardous areas, to require the use of bushfire building codes and standards for future construction in those areas, or to require additional insurance in those areas and encourage vulnerable communities to enhance their response and coping capacities. Tax incentives could be used to encourage bushfire preparedness activities in the vulnerable communities at the UBI. In addition, because of the density and importance of vulnerable locations, it would be unrealistic for local and national governments to prevent or even limit building and occupation of the UBI. It would also be unrealistic in terms of economic costs, to reinforce every building within the bushfire zone. Therefore, it is of vital importance that disaster managers and emergency planners have detailed information on which buildings, structures, infrastructural units and groups of people are particularly vulnerable to bushfires. When such data are available, cost-effective mitigation measures may be developed and applied.

Engaging with relevant fire management stakeholders to understand the geography of the local level risk is an important issue when planning towards safer communities. The results of this research offer a useful starting point for stakeholder engagement at the local level. Visual representation of results reflects the local context and therefore stimulates the recognition and a sense of responsibility among stakeholders (Preston et

294 al. 2009). The acceptance of the concept of risk by a diversity of stakeholders will provide a platform to receive constructive feedback that may help to improve the model.

Prevailing bushfire risk management policies are more focused on the physical aspects of bushfires. However, vulnerability and its dimensions such as social vulnerability, social inequalities and their impacts on disaster preparedness, and local level emergency response and coping capacities need to be better integrated into current bushfire risk management planning. For example, maintaining community level inventories with information such as people who need special assistance during an emergency, or people who can provide resources during an emergency can be very useful during a bushfire emergency. Findings such as the SoVI and factors affecting household level preparedness can be used at the mitigation and preparedness phases; to promote community based disaster mitigation activities to enhance community capacity, which may reduce the level of vulnerability and potential impacts.

295

296

REFERENCES

Abella, E.A.C. & Van Westen, C., 2007, Generation of a landslide risk index map for Cuba using spatial multi-criteria evaluation, Landslides, 4(4), 311-325.

Adger, W.N., 2000, Social and ecological resilience: Are they related?, Progress in Human Geography, 24(3), 347-364, doi: 10.1111/j.1944-8287.2003.tb00220.x.

Adger, W.N., 2003, Social capital, collective action, and adaptation to climate change, Economic Geography, 79(4), 387-404, doi: 10.1111/j.1944- 8287.2003.tb00220.x.

Adger, W.N., 2006, Vulnerability, Global Environmental Change, 16(3), 268-281, doi: 10.1016/j.gloenvcha.2006.02.006.

Adger, W.N., Brooks, N., Bentham, G., Agnew, M. & Eriksen, S., 2004, New Indicators of Vulnerability and Adaptive Capacity, Technical Report 7, Tyndall Centre for Climate Change Research, Norwich.

ADPC, 2006, Capacity Building in Asia Using Information Technology Applications (CASITA) Asian Disaster Preparedness Centre, Bangkok. Available from: http://www.adpc.net/casita/course-materials/Mod-2-Hazards.pdf [10 March 2012].

Al-Adamat, R.A.N., Foster, I.D.L. & Baban, S.M.J., 2003, Groundwater vulnerability and risk mapping for the Basaltic aquifer of the Azraq basin of Jordan using GIS, Remote sensing and DRASTIC, Applied Geography, 23(4), 303-324, doi: 10.1016/j.apgeog.2003.08.007.

Alcántara-Ayala, I., 2002, Geomorphology, natural hazards, vulnerability and prevention of natural disasters in developing countries, Geomorphology, 47(2- 4), 107-124, doi: 10.1016/S0169-555X(02)00083-1.

Alegria, A.C., Sahli, H. & Zimanyi, E., 2011, Application of density analysis for landmine risk mapping, Spatial Data Mining and Geographical Knowledge Services (ICSDM), 2011 IEEE International Conference, 29 June -1 July 223- 228 pp.

Alexander, D., 1993, Natural Disasters, UCL Press, London.

Allen, K.M., 2006, Community based disaster preparedness and climate adaptation: local capacityVbuilding in the Philippines, Disasters, 30(1), 81-101, doi: 10.1111/j.1467-9523.2006.00308.x

297

Alwang, J., Siegel, P. & Jorgensen, S., 2001, Vulnerability: A view from different disciplines, No 23304, Social Protection Discussion Papers from The World Bank. Available from: http://EconPapers.repec.org/RePEc:wbk:hdnspu:23304 [20 June 2011].

Amatulli, G., Perez-Cabello, F. & de la Riva, J., 2007, Mapping lightning/human- caused wildfires occurrence under ignition point location uncertainty, Ecological Modelling, 200(3-4), 321-333, doi: 10.1016/j.ecolmodel.2006.08.001.

American Geological Institute, 1984, Glossary of geology, American Geological Institute Falls Church, Virginia.

AS/NZS, 2004, AS/NZS 4360: Risk Management, Standards/Australia/Standards New Zealand, Sydney and Wellington.

AS/NZS, 2009, AS/NZS ISO 31000: Risk Management Principles and Guidelines Standerds Australia/Standards New Zealand

Asian Disaster Reduction Center, 2005, Total Disaster Risk Management, Good Practices, Asian Disaster Reduction Center, Kobe, Japan.

Atkinson, D., Chladil, M., Janssen, V. & Lucieer, A., 2010, Implementation of quantitative bushfire risk analysis in a GIS environment, International Journal of Wildland Fire, 19(5), 649-658, doi: 10.1071/WF08185.

Atkinson, D., Janssen, V., Lucieer, A. & Chladil, M., 2007, Bushfire risk assessment-an integrated approach using GIS, Spatial Science Institute Biennial International Conference SSC2007, Hobart, Tasmania, 14 -18 May

Australian Bureau of Statistics, 2009, Regional population growth, Australia, 2007-08 Australian Bureau of Statistics, viewed 26 October 2011, .

Baas, S., Ramasamy, S., DePryck, J.D. & Battista, F., 2008, Disaster Risk Management Systems Analysis: A Guide Book, Food and Agricultural Organisation (FAO), Rome

Bachmann, A., 2001, GIS-based Wildland Fire Risk Analysis, PhD thesis, University of Zurich,Zurich.

Bachmann, A. & Allgower, B., 2001, A consistent wildland fire risk terminology is needed!, Fire Management Today, 61(4), 28-33.

Bankoff, G., Frerks, G. & Hilhorst, D., 2004, Mapping Vulnerability: Disasters, Development, and People, Earthscan, London.

Barnett, J., Lambert, S. & Fry, I., 2008, The hazards of indicators: insights from the environmental vulnerability index, Annals of the Association of American Geographers, 98(1), 102-119, doi: 10.1080/00045600701734315.

298

Barredo, J. & Bosque-Sendra, J., 1998, Comparison of Multi-Criteria Evaluation Methods Integrated in Geographical Information Systems to Allocate Urban Areas, Department of Geography, Universidad de Alcalá de Henares, Spain.

Batty, M., Xie, Y. & Sun, Z., 1999, Modeling urban dynamics through GIS-based cellular automata, Computers, Environment and Urban Systems, 23(3), 205-233, doi: 10.1016/S0198-9715(99)00015-0.

Baum, S., Horton, S. & Choy, D., 2008, Local urban communities and extreme weather events: Mapping social vulnerability to flood Australasian Journal of Regional Studies, 14(3), 251-273.

Benedetti, R., Piersimoni, F., Bee, M. & Espa, G., 2010, Agricultural Survey Methods, Wiley Online Library.

Beringer, J., 2000, Community fire safety at the urban/rural interface: the bushfire risk, Fire Safety Journal, 35(1), 1-23, doi: 10.1016/S0379-7112(00)00014-X.

Berkes, F., 2007, Understanding uncertainty and reducing vulnerability: lessons from resilience thinking, Natural Hazards, 41(2), 283-295, doi: 10.1007/s11069-006- 9036-7.

Birkmann, J., 2006a, Indicators and criteria for measuring vulnerability: Theoretical bases and requirements, in Measuring Vulnerability To Natural Hazards: Towards Disaster Resilient Societies, J. Birkmann (ed.), United Nations University Press, Tokyo, pp. 55-73.

Birkmann, J., 2006b, Measuring vulnerability to promote disaster-resilient societies: Conceptual frameworks and definitions, in Measuring Vulnerability To Natural Hazards: Towards Disaster Resilient Societies, J. Birkmann (ed.), United Nations University Press, Tokyo, pp. 9-54.

Blaikie, P., Wisner, B., Cannon, T. & Davis, I., 1994, At Risk: Natural Hazards, People's Vulnerability and Disasters, Routledge, London.

Blanchi, R., Jappiot, M. & Alexandrian, D., 2002, Forest fire risk assessment and cartography. A methodological approach, Forest Fire Research & Wildland Fire Safety. Available from: http://www.incendies-de-foret.org/recherche/1998- 3/coimbra-2002.pdf. [29 June 2011].

Blanchi, R., Leonard, J., Leicester, R., Lipkin, F., Boulaire, F. & McNamara, C., 2011, Assessing vulnerability at the urban interface, The 5th International Wildland Fire Conference, Sun City, South Africa. Available from: ftp://ftp2.fs.fed.us/incoming/wo_fam/Wildfire2011%20Conference.SouthAfrica/ Wildfire2011%20e-programme/Programme%20contents /Wildfire2011%20 Con ference%20Papers/Wildfire2011%20Conference%20Papers/Conference%20Pap ers/Raphaele%20Blanchi.pdf [1 August 2012].

299

Blanchi, R., Leonard, J.E. & Leicester, R.H., 2006, Lessons learnt from post-bushfire surveys at the urban interface in Australia, Forest Ecology and Management, 234(Supplement 1), S139-S139, doi: 10.1016/j.foreco.2006.08.184.

Blue Mountains Bush Fire Coordinating Committee, 2008, Bushfire Risk Management Plan, Blue Mountains City Council Katoomba.

Blue Mountains City Council, 2012, Community Atlas:Index of Relative Socio- economic Disadvantage, the Blue Mountains City 2006, viewed 14 May 2012, .

Bogardi, J. & Birkmann, J., 2004, Vulnerability assessment: The first step towards sustainable risk reduction, Disaster and Society–From Hazard Assessment to Risk Reduction. Logos Verlag Berlin, Berlin, 75-82.

Bohle, H.G., Downing, T.E. & Watts, M.J., 1994, Climate change and social vulnerability: Toward a sociology and geography of food insecurity., Global Environmental Change, 4, 37-48, doi: 10.1016/0959-3780(94)90020-5.

Bollin, C., Cardenas, C., Hahn, H. & Vatsa, K., 2004, Disaster Risk Management by Communities and Local Governments, Inter-American Development Bank, Washington, DC.

Bonham-Carter, G., 1994, Geographic Information Systems for Geoscientists: Modelling with GIS, Pergamon Press, Oxford; New York.

Boura, J., 1998, Community Fireguard: Creating partnerships with the community to minimise the impact of bushfire, Australian Journal of Emergency Management, 13(3), 59-64.

Bowen, R.E. & Riley, C., 2003, Socio-economic indicators and integrated coastal management, Ocean & Coastal Management, 46(3-4), 299-312.

Boyd, K., Hervey, R. & Stradtner, J., 2002, Assessing the vulnerability of the Mississippi Gulf Coast to coastal storms using an on-line GIS-based coastal risk atlas, Oceans 2002: Reflections of the Past, Visions of the Future, Biloxi, Missisippi, 23-31 October 2002, 1127-1133 pp. Available from: http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=1192124 [1 August 2012].

Bradstock, R., Davies, I., Price, O. & Cary, G., 2008, Effects of climate change on bushfire threats to biodiversity, ecosystem processes and people in the Sydney region, Final report to the New South Wales department of environment and climate change: climate change impacts and adaptation research project, New South Wales Department of Environment and Climate Change, Sydney, 65 pp. Available from: https://wikis.utas.edu.au/download/attachments/ 12852129/ BushfireReport2008Carey.pdf [10 August 2012].

300

Bradstock, R.A., Bedward, M., Kenny, B.J. & Scott, J., 1998, Spatially-explicit simulation of the effect of prescribed burning on fire regimes and plant extinctions in shrublands typical of south-eastern Australia, Biological Conservation, 86(1), 83-95, doi: 10.1016/S0006-3207(97)00170-5.

Brooks, N., 2003, Vulnerability, Risk and Adaptation: A Conceptual Framework, Tyndall Centre for Climate Change Research, and Centre for Social and Economic Research on the Global Environment (CSERGE), School of Environmental Sciences, University of East Anglia, Norwich. Available from: http://tyndall.ac.uk/sites/default/files/wp38.pdf [20 June 2011].

Brunckhorst, D., Reeve, I., Morley, P., Coleman, M., Barclay, E., McNeill, J., Stayner, R., Glencross-Grant, R., Thompson, J. & Thompson, L., 2011, Hunter & Central Coasts New South Wales – Vulnerability to climate change impacts, Department of Climate Change and Energy Efficiency, Australia.

Bründl, M., Romang, H., Bischof, N. & Rheinberger, C., 2009, The risk concept and its application in natural hazard risk management in Switzerland, Natural Hazards and Earth System Sciences, 9, 801-813.

Bruneau, M., Chang, S., Eguchi, R., Lee, G., O’Rourke, T., Reinhorn, A., Shinozuka, M., Tierney, K., Wallace, W. & von Winterfeldt, D., 2003, A framework to quantitatively assess and enhance the seismic resilience of communities, Earthquake Spectra, 19(4), 733-752.

Buckle, P., 2006, Assessing social resilience, in Disaster Resilience: An Integrated Approach, D. Paton & D. Johnston (eds), Charles C Thomas, Springfield, pp. 88-104.

Buckle, P., Marsh, G. & Smale, S., 2000, New approaches to assessing vulnerability and resilience, Australian Journal of Emergency Management, 15(2), 8-14.

Buckle, P., Marsh, G., Smale, S. & Australia, E.M., 2001, Assessing Resilience & Vulnerability: Principles, Strategies & Actions, Emergency Management Australia.

Burg, J., 2008, Measuring populations' vulnerabilities for famine and food security interventions: The case of Ethiopia's chronic vulnerability Index, Disasters, 32(4), 609-630, doi: 10.1111/j.0361-3666.2008.01057.x

Burrough, P.A., McDonnell, R.A. & McDonnell, R., 1998, Principles of Geographical Information Systems, Oxford University Press Oxford.

Burton, D. & Laurie, E., 2009, Blue Mountains City Council Climate Change Risk Assessment, Climate Risk Pty Limited (Australia), Sydney.

Burton, I., Kates, R.W. & White, G.F., 1978, The Environment as Hazard, Oxford University Press, New York.

301

Burton, I., Kates, R.W. & White, G.F., 1993, The environment as hazard, Second edition edn, The Guilford Press, New York.

Bush Fire Coordinating Committee, 2008, Bushfire Risk Management Planning Guidelines for Bushfire Management Committees (Policy No. 1/2008), New South Wales Rural Fire Services. Available from: http://www.rfs.nsw.gov.au/file_system/attachments/State08/Attachment_200810 03_C07A52CD.pdf [1 August 2009].

Bushfire CRC, 2011, Effective Risk Communication, Bushfire CRC, viewed 06 January .

Bushnell, S., Balcombe, L. & Cottrell, A., 2007, Community and fire service perceptions of bushfire issues in Tamborine Mountain: What’s the difference?, Australian Journal of Emergency Management, 22(3), 3-9.

Bushnell, S. & Cottrell, A., 2007a, Increasing community resilience to bushfire— implications from a north Queensland community case study, Australian Journal of Emergency Management, 22(2), 3-9.

Bushnell, S. & Cottrell, A., 2007b, Living with bushfire: What do people expect, Communities living with hazards. , Centre for Disaster Studies, James Cook University, Townsville, 215-253 pp. Available from: http://www- public.jcu.edu.au/public/groups/everyone/documents/conference_paper/jcutst_0 56266.pdf [11 Sepetember 2011].

Bushnell, S., Cottrell, A., Spillman, M. & Lowe, D., 2006, Thuringowa Bushfire Case Study–Technical Report, James Cook University, Townsville.

Buxton, M., Haynes, R., Mercer, D. & Butt, A., 2011, Vulnerability to Bushfire Risk at Melbourne's Urban Fringe: The Failure of Regulatory Land Use Planning, Geographical Research, 49(1), 1-12.

Cannon, T., 1994, Vulnerability analysis and the explanation of ‘natural’disasters, in Disasters, Development and the Environment, A. Varley (ed.), John Wiley & Sons, Chichester, pp. 13-30.

Cannon, T., 2000, Vulnerability analysis and disasters, in Floods, D.J. Parker (ed.), Routledge, pp. 45-55.

Cannon, T., 2008, Vulnerability,“innocent” disasters and the imperative of cultural understanding, Disaster Prevention and Management, 17(3), 350-357, doi: 10.1108/09653560810887275.

Cardona, O.D., 2004, The need for rethinking the concepts of vulnerability and risk from a holistic perspective: A necessary review and criticism for effective risk management, in Mapping Vulnerability: Disasters Development and People G. Bankoff, G. Frerks & H. Dorothea. (eds), Earthscan, London, pp. 37-51.

302

Cardona, O.D., 2005, Indicators of disaster risk and risk management. Summary report. IDB/IDEA Program on Indicators for Disaster Risk Management, Inter- American Development Bank, Washington DC.

Cardona, O.D. & Barbat, A.H., 2000, The Seismic Risk and its Prevention, Calidad Siderúrgica, Madrid, Spain.

Cardona, O.D. & Hurtado, J.E., 2000, Holistic Approach for Urban Seismic Risk Evaluation and Management, Proceedings of the Sixth International Conference on Seismic Zonation: Managing Earthquake Risk in the 21st Century, Palms Springs, CA, 12-15 November 2000.

Cardona, O.D., van Aalst, M.K., Birkmann, J., Fordham, M., McGregor, G., Perez, R., Pulwarty, R.S., Schipper, E.L.F. & Sinh, B.T., 2012, Determinants of Risk: Exposure and Vulnerability, A Special Report of Working Groups I and II of the Intergovernmental Panel on Climate Change (IPCC), Cambridge University Press, Cambridge, UK, and New York, NY, USA.

Carreño, M., Cardona, O. & Barbat, A., 2007, Urban seismic risk evaluation: A holistic approach, Natural Hazards, 40(1), 137-172, doi: 10.1007/s11069-006-0008-8.

Carver, S.J., 1991, Integrating multi-criteria evaluation with geographical information systems, International Journal of Geographical Information Science, 5(3), 321- 339, doi: 10.1080/02693799108927858

Chakhar, S. & Martel, J.M., 2003, Enhancing geographical information systems capabilities with multi-criteria evaluation functions, Journal of Geographic Information and Decision Analysis, 7(2), 47-71.

Chakhar, S. & Mousseau, V., 2007, An algebra for multicriteria spatial modeling, Computers, Environment and Urban Systems, 31(5), 572-596, doi: 10.1016/j.compenvurbsys.2007.08.007.

Chakraborty, J., Tobin, G.A. & Montz, B.E., 2005, Population evacuation: Assessing spatial variability in geophysical risk and social vulnerability to natural hazards, Natural Hazards Review, 6(1), 23-33, doi: 10.1061/(ASCE)1527- 6988(2005)6:1(23).

Chambers, R., 1989, Editorial Introduction: Vulnerability, coping, and policy, IDS Bulletin, 20(2), 1-7.

Chang, N.B., Parvathinathan, G. & Breeden, J.B., 2008, Combining GIS with fuzzy multicriteria decision-making for landfill siting in a fast-growing urban region, Journal of Environmental Management, 87(1), 139-153, doi: 10.1016/j.jenvman.2007.01.011.

Chaudhuri, S., Jalan, J. & Suryahadi, A., 2002, Assessing household vulnerability to poverty from cross-sectional data: A methodology and estimates from Indonesia, Department of Economics Discussion Paper Series, 102, 52.

303

Chen, K., 2005, Counting bushfire-prone addresses in the Greater Sydney region, Planning for Natural Hazards - How Can We Mitigate the Impacts?, University of Wollongong, 2-5 February 10 pp.

Chen, K., Blong, R. & Jacobson, C., 2001, MCE-RISK: integrating multicriteria evaluation and GIS for risk decision-making in natural hazards, Environmental Modelling & Software, 16(4), 387-397, doi: 10.1016/S1364-8152(01)00006-8.

Chen, K., Blong, R. & Jacobson, C., 2003, Towards an integrated approach to natural hazards risk assessment using GIS: With reference to bushfires, Environmental Management, 31(4), 546-560, doi: 10.1007/s00267-002-2747-y.

Childs, C., 2004, Interpolating surfaces in ArcGIS spatial analyst, ArcUser, July- September, 32-35.

Chou, Y., Minnich, R. & Chase, R., 1993, Mapping probability of fire occurrence in San Jacinto Mountains, California, USA, Environmental Management, 17(1), 129-140, doi: 10.1007/BF02393801.

Chuvieco, E., Aguado, I., Yebra, M., Nieto, H., Salas, J., Martín, M.P., Vilar, L., Martínez, J., Martín, S. & Ibarra, P., 2010, Development of a framework for fire risk assessment using remote sensing and geographic information system technologies, Ecological Modelling, 221(1), 46-58, doi: 10.1016/j.ecolmodel.2008.11.017.

Chuvieco, E. & Salas, J., 1996, Mapping the spatial distribution of forest fire danger using GIS, International Journal of Geographical Information Science, 10(3), 333-345, doi: 10.1080/02693799608902082.

City of Blue Mountains, 2009, State of city report - summary findings Blue Mountains City Council, Katoomba. Available from: http://www.bmcc.nsw.gov.au/index. cfm? cx= 018081763861296252136%3An 58dywkggym&cof= FORID% 3A10&ie=UTF-8&q= State+of+city+report+-+summary+findings +&s=C57 6E07E-3048-1075-63533E928 C097BE8 [26 July 2011].

Clark, G., Moser, S., Ratick, S., Dow, K., Meyer, W., Emani, S., Jin, W., Kasperson, J., Kasperson, R. & Schwarz, H., 1998, Assessing the vulnerability of coastal communities to extreme storms: the case of Revere, MA., USA, Mitigation and Adaptation Strategies for Global Change, 3(1), 59-82, doi: 10.1023/A:1009609710795.

Collins, T.W., 2005, Households, forests, and fire hazard vulnerability in the American West: A case study of a California community, Global Environmental Change Part B: Environmental Hazards, 6(1), 23-37, doi: 10.1016/j.hazards.2004.12.003.

304

Colyer, P., 2010, Mapping and Assessment of Key Vegetation Communities Across the Ku-ring-gai Local Government Area. , Ku-ring-gai Council, Gordon, NSW. Available from: http://www.kmc.nsw.gov.au/resources/ documents/ attomc22 June2010GB.05-A.pdf [16th January 2012].

Cortner, H.J., Gardner, P.D. & Taylor, J.G., 1990, Fire hazards at the urban-wildland interface: What the public expects, Environmental Management, 14(1), 57-62, doi: 10.1007/BF02394019.

Cottrell, A., 2005, Communities and bushfire hazard in Australia: More questions than answers, Global Environmental Change Part B: Environmental Hazards, 6(2), 109-114, doi: 10.1016/j.hazards.2005.10.002.

Cottrell, A., 2009, Know Your Patch to Grow Your Patch, Bushfire CRC Briefing Paper, Understanding Communities Project Bushfire CRC & Centre for Disaster Studies James Cook University. Available from: http://www.bushfirecrc.com/ managed/resource/108524_know_your_patch.pdf [10 May 2011].

Cottrell, A., Bushnell, S., Spillman, M., Newton, J., Lowe, D. & Balcombe, L., 2008, Community perceptions of bushfire risk, in Community Bushfire Safety, J. Handmer & K. Haynes (eds), CSIRO, Melbourne, Australia, pp. 11-20.

Cottrell, A. & King, D., 2007, Planning for more bushfires: implications of urban growth and climate change, Queensland Planner, 47(4), 23-26.

Cox, J.R., Rosenzweig, C., Solecki, W.D., Goldberg, R. & Kinney, P.L., 2007, Social Vulnerability to Climate Change: A Neighbourhood Analysis of the Northeast U.S. Megaregion, Union of Concerned Scientists, viewed 26 October, .

Crichton, D., 1999, The risk triangle, in Natural Disaster Management, J. Ingleton (ed.), Tudor Rose London, pp. 102-103.

Cronstedt, M., 2002, Prevention, preparedness, response, recovery-an outdated concept?, Australian Journal of Emergency Management, The, 17(2), 10-13.

Cutter, S.L., 1996, Vulnerability to environmental hazards, Progress in Human Geography, 20(4), 529-539, doi: 10.1177/030913259602000407

Cutter, S.L., Barnes, L., Berry, M., Burton, C., Evans, E., Tate, E. & Webb, J., 2008, A place-based model for understanding community resilience to natural disasters, Global Environmental Change, 18(4), 598-606, doi: 10.1016/j.gloenvcha.2008.07.013.

Cutter, S.L., Boruff, B.J. & Shirley, W.L., 2003, Social vulnerability to environmental hazards, Social Science Quarterly, 84(2), 242-261, doi: 10.1111/1540- 6237.8402002

305

Cutter, S.L., Burton, C.G. & Emrich, C.T., 2010, Disaster resilience indicators for benchmarking baseline conditions Journal of Homeland Security and Emergency Management 7(1), doi: 10.2202/1547-7355.1732

Cutter, S.L., Emrich, C.T., Webb, J.J. & Morath, D., 2009, Social Vulnerability to Climate Variability Hazards: A Review of the Literature: Final Report to Oxfam America, Hazards and Vulnerability Research Institute, University of South Carolina.

Cutter, S.L. & Finch, C., 2008, Temporal and spatial changes in social vulnerability to natural hazards, Proceedings of the National Academy of Sciences, 105(7), 2301-2306.

Cutter, S.L., Mitchell, J.T. & Scott, M.S., 1997, Handbook for Conducting a GIS-based Hazards Assessment at the County Level, University of South Carolina, Columbia.

Cutter, S.L., Mitchell, J.T. & Scott, M.S., 2000, Revealing the vulnerability of people and places: a case study of Georgetown County, South Carolina, Annals of the Association of American Geographers, 90(4), 713-737, doi: 10.1111/0004- 5608.00219.

Danese, M., Lazzari, M. & Murgante, B., 2008, Kernel density estimation methods for a geostatistical approach in seismic risk analysis: The case study of Potenza hilltop town (southern Italy), International Conference on Computational Science and its Applications, Part I, Springer-Verlag, Perugia, Italy, 415-429 pp.

Davidson, R., 1997, An urban earthquake disaster risk index, John A. Blume Earthquake Engineering Center Report, Department of Civil and Environmental Engineering, Stanford University, Stanford, CA Available from: https://blume.stanford.edu/sites/default/files/TR121_Davidson.pdf [20 March 2011].

Davidson, R. & Lambert, K., 2001, Comparing the hurricane disaster risk of U. S. coastal counties, Natural Hazards Review, 2(3), 132-142.

De Guzman, E.M., 2003, Towards total disaster risk management approach. Available from: http://nirapad.org/admin/soft_archive/1308222759_Towards%20Total%20Disas ter%20Risk%20Manangement%20Approach.pdf [4th April 2012].

De la Riva, J., Perez-Cabello, F., Lana-Renault, N. & Koutsias, N., 2004, Mapping wildfire occurrence at regional scale, Remote Sensing of Environment, 92(3), 363-369, doi: 10.1016/j.rse.2004.06.022.

De Smith, M.J., Goodchild, M.F. & Longley, P., 2009, Geospatial Analysis: A Comprehensive Guide to Principles, Techniques and Software Tools, 3rd edn, Matador, Leicester.

306

Denzin, N. & Lincoln, Y., 2000, Handbook of Qualitative Research, Sage Publications, Inc, Thousand Oaks.

Deodhar, V., 2004, Does the Housing Market Value Heritage?: Some Empirical Evidence, 1864089202, Dept. of Economics, Macquarie University.

Dhakal, A.S., Amada, T. & Aniya, M., 2000, Landslide hazard mapping and its evaluation using GIS: An investigation of sampling schemes for a grid-cell based quantitative method, Photogrammetric Engineering and Remote Sensing, 66(8), 981-989.

Dilley, M., Chen, R.S., Deichmann, U., Lerner-Lam, A.L. & Arnold, M., 2005, Natural Disaster Hotspots.A Global Risk Analysis, World Bank, Washington, DC.

Dixon, D., 2005, Needs of an Actual Community Post Disaster-Hornsby Ku-ring-gai, The Australian Journal of Emergency Management, 20(3), 33-38.

Dolan, A. & Walker, I., 2006, Understanding vulnerability of coastal communities to climate change related risks, Journal of Coastal Research, SI 39, 1316-1323.

Douglas, J., 2007, Physical vulnerability modelling in natural hazard risk assessment.

Drobne, S. & Lisec, A., 2009, Multi-attribute decision analysis in GIS: weighted linear combination and ordered weighted averaging, Informatica: An International Journal of Computing and Informatics, 33(4), 459-474.

Durham, K., 2003, Treating the risks in Cairns, Natural Hazards, 30(2), 251-261, doi: 10.1023/A:1026174602731.

Dwyer, A., Zoppou, C., Nielsen, O., Day, S. & Roberts, S., 2004, Quantifying Social Vulnerability: A methodology for identifying those at risk to natural hazards, Geoscience Australia Record, Geoscience Australia, Canberra, 101 pp. Available from: http://www.ga.gov.au/webtemp/1241759/Rec2004_014.pdf [4th April 2011].

Eakin, H. & Luers, A.L., 2006, Assessing the vulnerability of social-environmental systems, Annual Review of Environmental Resources, 31, 365-394, doi: 10.1146/annurev.energy.30.050504.144352.

Eastman, J.R., Jin, W., Kyem, P.A.K. & Toledano, J., 1995, Raster procedures for multi-criteria/multi-obiective decisions, Photogrammetric Engineering & Remote Sensing, 61(5), 539-547.

El Morjani, Z.E.A., Ebener, S., Boos, J., Ghaffar, E.A. & Musani, A., 2007, Modelling the spatial distribution of five natural hazards in the context of the WHO/EMRO Atlas of Disaster Risk as a step towards the reduction of the health impact related to disasters, International Journal of Health Geographics, 6(1), 8-35, doi: 10.1186/1476-072X-6-8.

307

Ellis, S., Kanowski, P. & Whelan, R., 2004, National Inquiry into Bushfire Mitigation and Management, Council of Australian Governments, Canberra.

EMA Disaster Database, 2012, Australian Emergency Management Institute, annually updated, viewed 20 January 2012, .

Emmi, P.C. & Horton, C.A., 1995, A Monte Carlo simulation of error propagation in a GIS-based assessment of seismic risk, International Journal of Geographical Information Systems, 9(4), 447-461.

Enders, J., 2001, Measuring community awareness and preparedness for emergencies, Australian Journal of Emergency Management, 16(3), 52-58.

Eriksen, C. & Prior, T., 2011, The art of learning: wildfire, amenity migration and local environmental knowledge, International Journal of Wildland Fire, 20(4), 612- 624.

Fedra, K., 1998, Integrated risk assessment and management: Overview and state of the art, Journal of Hazardous Materials, 61(1), 5-22, doi: 10.1016/S0304- 3894(98)00102-2.

Fekete, A., 2009, Validation of a social vulnerability index in context to river-floods in Germany, Natural Hazards and Earth System Sciences, 9(2), 393-403, doi: 10.5194/nhess-9-393-2009.

Fekete, A., 2011, Spatial disaster vulnerability and risk assessments: challenges in their quality and acceptance, Natural Hazards, 1-18.

Fekete, A., Damm, M. & Birkmann, J., 2010, Scales as a challenge for vulnerability assessment, Natural Hazards, 55(3), 729-747, doi: 10.1007/s11069-009-9445-5.

Ferreira Leite, F., Bento Gonçalves, A. & Vieira, A., 2010, The recurrence interval of forest fires in Cabeço da Vaca (Cabreira Mountain--northwest of Portugal), Environmental Research, 111(2), 215-221, doi: 10.1016/j.envres.2010.05.007.

Fiedrich, F. & Burghardt, P., 2007, Agent-based systems for disaster management, Communications of the ACM, 50(3), 41-42.

Field, A.P., 2009, Discovering Statistics using SPSS, 3rd edn, Sage Publications Ltd, London.

Finney, M.A., 2005, The challenge of quantitative risk analysis for wildland fire, Forest Ecology and Management, 211(1-2), 97-108, doi: 10.1016/j.foreco.2005.02.010.

Finney, M.A. & Station, R.M.R., 1998, FARSITE, Fire Area Simulator--model development and evaluation, US Department of Agriculture, Forest Service, Rocky Mountain Research Station.

Fire and Rescue NSW, 2009, CFU Frequently Asked Questions, viewed 25 April 2011, .

308

Fitzgerald, G. & Fitzgerald, N., 2005, Assessing Community Resilience to Wildfires: Concepts and Approach, Paper prepared for SCION Research. Available from: http://www.fitzgerald.co.nz/documents/Reports/Assessing-Community- resilience-to-wildfires.pdf [3 May 2009].

Flanagan, B.E., Gregory, E.W., Hallisey, E.J., Heitgerd, J.L. & Lewis, B., 2011, A social vulnerability index for disaster management, Journal of Homeland Security and Emergency Management, 8(1), Article 3, doi: 10.2202/1547- 7355.1792.

Flax, L.K., Jackson, R.W. & Stein, D.N., 2002, Community vulnerability assessment tool methodology, Natural Hazards Review, 3(4), 163-176, doi: 10.1061/(ASCE)1527-6988(2002)3:4(163).

Fleming, G., Merwe, M. & McFerren, G., 2007, Fuzzy expert systems and GIS for cholera health risk prediction in southern Africa, Environmental Modelling & Software, 22(4), 442-448, doi: 10.1016/j.envsoft.2005.12.008.

Fuchs, S., Kuhlicke, C. & Meyer, V., 2011, Editorial for the special issue: Vulnerability to natural hazards—the challenge of integration, Natural Hazards, 58(2), 609- 619, doi: 10.1007/s11069-011-9825-5

Füssel, H.M., 2007, Vulnerability: A generally applicable conceptual framework for climate change research, Global Environmental Change, 17(2), 155-167, doi: 10.1016/j.gloenvcha.2006.05.002.

Gabriel, P., 2009, Victoria's state-level emergency risk assessment method, The Australian Journal of Emergency Management, 24(1), 5-10.

Gall, M., 2007, Indices of social vulnerability to natural hazards: A comparative evaluation, PhD Thesis, University of South Carolina,South Carolina.

Gallopín, G., 2006, Linkages between vulnerability, resilience, and adaptive capacity, Global Environmental Change, 16(3), 293-303, doi: 10.1016/j.gloenvcha.2006.02.004.

Gbanie, S.P., Tengbe, P.B., Momoh, J.S., Medo, J. & Kabba, V.T.S., 2012, Modelling landfill location using Geographic Information Systems (GIS) and Multi-Criteria Decision Analysis (MCDA): Case study Bo, Southern Sierra Leone, Applied Geography.

Geoscience Australia, 2011, What is Exposure?, Geoscience Australia, viewed 16 January 2011, .

Gillen, M., 2005, Urban governance and vulnerability: exploring the tensions and contradictions in Sydney's response to bushfire threat, Cities, 22(1), 55-64, doi: 10.1016/j.cities.2004.10.006.

309

Gillian, P. & Derek, C., 2011, Implementing the Phoenix fire spread model for operational use, Surveying and Spatial Sciences Biennial Conference Wellington, New Zealand, 21-25 November 2011.

Glik, D.C., 2007, Risk communication for public health emergencies, Annual Review of Public Health, 28, 33-54, doi: 10.1146/annurev.publhealth.28.021406.144123.

Godschalk, D., Kaiser, E. & Berke, P., 1998, Hazard assessment: The factual basis for planning and mitigation, in Cooperating with Nature: Confronting Natural Hazards with Land-Use Planning for Sustainable Communities, R. Burby (ed.), Joseph Henry, Washington, DC, pp. 85-118.

Gómez-Fernández, F., 2000, Contribution of geographical information systems to the management of volcanic crises, Natural Hazards, 21(2), 347-360, doi: 10.1023/A:1008150816028.

Goodman, H. & Proudley, M., 2008, Social contexts of responses to bushfire threat: A case study of the Wangary fire, in Community Bushfire Safety, J. Handmer & K. Haynes (eds), CSIRO Publishing, Melbourne, Australia, pp. 47-56.

Goodman, R.M., Speers, M.A., McLeroy, K., Fawcett, S., Kegler, M., Parker, E., Smith, S.R., Sterling, T.D. & Wallerstein, N., 1998, Identifying and defining the dimensions of community capacity to provide a basis for measurement, Health Education & Behavior, 25(3), 258-278, doi: 1177/109019819802500303.

Goudie, D., 2007, Bushfire Web Review, Bushfire CRC, Centre for Disaster Studies, James Cook University, Townsville. Available from: http://www.jcu.edu.au/cds/public/groups/everyone/documents/technical_report/j cutst_056257.pdf [30 September 2011].

Granger, K., 1997, UN-IDNDR & QUIPUNET Internet conference on Floods, Drought:Issues for the 21st Century. Available from: www.quipu.net:1997/English/Welcome.htm [14th April 2012].

Granger, K., 1998, Developing an understanding of urban geohazard risk, Australian Journal of Emergency Management, 13(4), 13-17.

Granger, K., 1999, Community Risk in Cairns: A multi hazard risk assessment Australian Geological Survey Organisation, Canberra. Available from: http://www.ga.gov.au/image_cache/GA4176.pdf [30 September 2011].

Granger, K., 2003, Quantifying storm tide risk in Cairns, Natural Hazards, 30(2), 165- 185, doi: 10.1007/s11069-004-2291-6.

Granger, K. & Hayne, M., 2000, Natural Hazards and Risk they Pose to South East Queensland, Australian Geological Survey Organisation, Canberra. Available from: http://www.ga.gov.au/image_cache/GA4200.pdf [30 September 2011].

Greene, R., 2002, Confronting Catastrophe: A GIS Handbook, Esri Press, Redlands, California.

310

Greiving, S., Fleischhauer, M. & Lückenkötter, J., 2006, A methodology for an integrated risk assessment of spatially relevant hazards, Journal of Environmental Planning and Management, 49(1), 1-19, doi: 10.1080/09640560500372800.

Grünthal, G., Thieken, A., Schwarz, J., Radtke, K., Smolka, A. & Merz, B., 2006, Comparative risk assessments for the city of Cologne–storms, floods, earthquakes, Natural Hazards, 38(1), 21-44.

GTZ, 2004, Risk Analysis - a Basis for Disaster Risk Management Deutsche Gesellschaft für Technische Zusammenarbeit (GTZ) GmbH, Eschborn. Available from: http://www2.gtz.de/dokumente/bib/05-0038.pdf [20 May 2010].

Gunasekera, R.C., 2009, Framework for a Methodology to Integrate Vulnerability to Develop Natural Hazard Risk Profiles for Sri Lanka, Preventionweb. Available from: http://www.preventionweb.net/files/15417_frameworkfordevleomentofriskprofil e.doc [15 March 2012].

Hacking, I., 1999, The social construction of what?, Harvard Univ Pr, Cambridge, Mass.

Hagenlocher, M., 2012, Towards a spatial assessment of disaster risk hot spots in the Sahel: Integration of concepts from climate change and disaster risk reduction, Z_GIS – Centre for Geoinformatics, University of Salzburg, Salzburg.

Hahn, H., 2003, Indicators and Other Instruments for Local Risk Management for Communities and Local Governments. , The German Technical Cooperation Agency, GTZ, for IADB. Available from: http://idbdocs.iadb.org/wsdocs/ getdocument.aspx?docnum=561184 [1 August 2012].

Hahn, M.B., Riederer, A.M. & Foster, S.O., 2009, The Livelihood Vulnerability Index: A pragmatic approach to assessing risks from climate variability and change: A case study in Mozambique, Global Environmental Change, 19(1), 74-88, doi: 10.1016/j.gloenvcha.2008.11.002.

Haight, R.G., Cleland, D.T., Hammer, R.B., Radeloff, V.C. & Rupp, T.S., 2004, Assessing fire risk in the wildland-urban interface, Journal of Forestry, 102(7), 41-48.

Hajkowicz, S. & Collins, K., 2007, A review of multiple criteria analysis for water resource planning and management, Water Resources Management, 21(9), 1553-1566, doi: 10.1007/s11269-006-9112-5.

Haki, Z., Akyürek, Z. & Düzgün, ., 2004, Assessment of social vulnerability using geographic information systems: Pendik, stanbul case study, 7th AGILE Conference on Geographic Information Science, Parallel Session 4.3- Environmental / Social Modelling, Heraklion, Greece.

311

Hammill, K. & Tasker, E., 2010, Vegetation, Fire and Climate Change in the Greater Blue Mountains World Heritage Area Department of Enviorenment, Climate Change and Water (NSW). Available from: http://www.environment.nsw.gov.au/resources/protectedareas/DECCW2010094 1.pdf [30 July 2012].

Handmer, J., O’Neil, S. & Killalea, D., 2010, Review of fatalities in the February 7, 2009, bushfires: Report Prepared for the Victorian Bushfires Royal Commission April 2010, Bushfire CRC, Centre for Risk and Community Safety, RMIT University, Melbourne.

Handmer, J. & Tibbits, A., 2005, Is staying at home the safest option during bushfires? Historical evidence for an Australian approach, Global Environmental Change Part B: Environmental Hazards, 6(2), 81-91, doi: 10.1016/j.hazards.2005.10.006.

Hardy, C.C., 2005, Wildland fire hazard and risk: Problems, definitions, and context, Forest Ecology and Management, 211(1-2), 73-82, doi: 10.1016/j.foreco.2005.01.029.

Harris, L.M., McGee, T.K. & McFarlane, B.L., 2011, Implementation of wildfire risk management by local governments in Alberta, Canada, Journal of Environmental Planning and Management, 54(4), 457-475, doi: 10.1080/09640568.2010.515881.

Hawkes, B., Beck, J. & Sahle, W., 1997, A wildfire threat rating system for the McGregor Model Forest, 13th Fire and Meteorology conference, Lorne, Australia, 27-31 pp.

Hays, R.D., Sherbourne, C.D. & Mazel, R., 1995, User's Manual for the Medical Outcomes study (MOS) Core Measures of Health-related Quality of Life, Rand Corporation, Santa Monica, CA.

Heath, J., Nulsen, C., Dunlop, P., Clarke, P., Bürgelt, P. & Morrison, D., 2011, The February 2011 Fires in Roleystone, Kelmscott and Redhill School of Psychology, University of Western Australia. Available from: http://www.bushfirecrc.com/managed/resource/bushfire_final_report_0.pdf [20 September 2011].

Hennessy, K., Lucas, C., Nicholls, N., Bathols, J., Suppiah, R. & Ricketts, J., 2005, Climate Change Impacts on Fire-weather in South-east Australia, Climate Impacts Group, CSIRO Atmospheric Research and the Australian Government Bureau of Meteorology, Aspendale, Melbourne.

Hewitt, K., 1997, Regions of risk: A geographical introduction to disasters, Longman London.

Hohenemser, C., Kates, R.W. & Slovic, P., 1983, The nature of technological hazard, Science, 220(4595), 378-384.

312

Holand, I.S., Lujala, P. & Rød, J.K., 2011, Social vulnerability assessment for Norway: A quantitative approach, Norsk Geografisk Tidsskrift-Norwegian Journal of Geography, 65(1), 1-17, doi: 10.1080/00291951.2010.550167.

Holand, I.S., Rød, J. & Lujala, P., 2009, A social vulnerability index for Norway: a viable approach in analyses of social consequences of climate change?, IOP Conference Series: Earth and Environmental Science: Climate Change: Global Risks, Challenges and Decissions, IOP Publishing Ltd, Copenhagen, 10–12 March.

Hornsby/Ku-ring-gai Bushfire Management Committee, 2008, Bushfire Risk Management Plan. Available from: http://www.hkbfmc.org.au/files /BFRMP _Final_ Lowres_ 04032010.pdf [26 October 2011].

Hufschmidt, G., 2011, A comparative analysis of several vulnerability concepts, Natural Hazards, 58(2), 621-643, doi: 10.1007/s11069-011-9823-7.

Hughes, P. & White, N., 2005, Living with bushfire risk: Residents accounts of their bushfire preparedness behaviour, AFAC/Bushfire CRC Conference, Auckland.

ICG, 2006, Risk Assessment Framework, International Centre for Geohazards, viewed 2 February 2010, .

InConsult, 2009, Risk Management Update: ISO 31000 Overview and Implications for Managers Available from: http://www.inconsult.com.au/Articles /ISO%2031000 %20Overview.pdf [15 March 2012 ].

Jaiswal, R.K., Mukherjee, S., Raju, K.D. & Saxena, R., 2002, Forest fire risk zone mapping from satellite imagery and GIS, International Journal of Applied Earth Observation and Geoinformation, 4(1), 1-10.

Jakes, P., Kruger, L., Monroe, M., Nelson, K. & Sturtevant, V., 2007, Improving wildfire preparedness: Lessons from communities across the US, Human Ecology Review, 14(2), 188-197.

Jakes, P.J., Nelson, K., Lang, E., Monroe, M., Agrawal, S., Kruger, L. & Sturtevant, V., 2002, A model for improving community preparedness for wildfire, Gen. Tech. Rep. NC-231. , U.S. Department of Agriculture, Forest Service, North Central Research Station, St. Paul, MN., pp. 4-9.

Jerrett, M., Burnett, R.T., Kanaroglou, P., Eyles, J., Finkelstein, N., Giovis, C. & Brook, J.R., 2001, A GIS-environmental justice analysis of particulate air pollution in Hamilton, Canada, Environment and Planning A, 33(6), 955-974, doi: 10.1068/a33137.

Jiang, H. & Eastman, J.R., 2000, Application of fuzzy measures in multi-criteria evaluation in GIS, International Journal of Geographical Information Science, 14(2), 173-184, doi: 10.1080/136588100240903.

313

Kaiser, G., 2006, Risk and vulnerability analysis to coastal hazards–an approach to integrated assessment, PhD Thesis, The Christian-Albrechts-University, Kiel, Germany.

Kandilioti, G. & Makropoulos, C., 2012, Preliminary flood risk assessment: the case of Athens, Natural Hazards, 1-28.

Kates, R., 1971, Natural hazard in human ecological perspective: hypotheses and models, Economic Geography, 47(3), 438-451.

Keeley, J.E. & Fotheringham, C., 2001, Historic fire regime in southern California shrublands, Conservation Biology, 15(6), 1536-1548.

Killalea, D. & Llewellyn, R., 2010, Position Paper on Bushfires and Community Safety, Australasian Fire & Emergency Service Authorities Council, East Melbourne.

King, D., 2001, Uses and limitations of socioeconomic indicators of community vulnerability to natural hazards: Data and disasters in Northern Australia, Natural Hazards, 24(2), 147-156.

King, D. & MacGregor, C., 2000, Using social indicators to measure community vulnerability to natural hazards, Australian Journal of Emergency Management, 15(3), 52-57.

Kirkwood, A.S., 1994, Why do we worry when scientists say there is no risk?, Disaster Prevention and Management, 3(2), 15-22.

Kohler, A., Jülich, S. & Bloemertz, L., 2004, Guidelines Risk Analysis - a Basis for Disaster Risk Management, Deutsche Gesellschaft für Technische Zusammenarbeit (GTZ) GmbH. Available from: http://www.gtz.de/de/dokumente/es-analisis-riesgo-base-para-la-gestion-de- riesgo.pdf [20 July 2011].

Koutsias, N., Kalabokidis, K.D. & Allgöwer, B., 2004, Fire occurrence patterns at landscape level: Beyond positional accuracy of ignition points with kernel density estimation methods, Natural Resource Modeling, 17(4), 359-375, doi: 10.1111/j.1939-7445.2004.tb00141.x.

Kruger, L.E., Agrawal, S., Monroe, M., Lang, E., Nelson, K., Jakes, P., Sturtevant, V., McCaffrey, S. & Everett, Y., 2002, Keys to community preparedness for wildfire, United States Department Of Agriculture Forest Service General Technical Report 10-17.

314

Ku-ring-gai Council, 2005, Ku-ring-gai Community Environmental Research Report, A joint project between Ku-ring-gai Council, The Institute of Environmental Studies at the University of NSW, and the Department of Geography and Environmental Science, Monash University., Gordon, NSW. Available from: http://www.kmc.nsw.gov.au/resources/documents/Kuringgai_Community_Eviro nmental_Research_Report_Feb_2005_Sections_1_2_3.pdf [10 June 2010].

Ku-ring-gai Council, 2006, Biodiversity Strategy, Ku-ring-gai Council, Gordon, NSW. Available from: http://www.kmc.nsw.gov.au/resources/ documents/ Biodiversity _Strategy_May_2006_final_for_adoption1[1].pdf [10 June 2010].

Ku-ring-gai Council, 2012, Bushfire Management and Prevention, viewed 1 August 2012, .

Kuhlicke, C., Scolobig, A., Tapsell, S., Steinführer, A. & De Marchi, B., 2011, Contextualizing social vulnerability: Findings from case studies across Europe, Natural Hazards, 58(2), 789-810, doi: 10.1007/s11069-011-9751-6.

Lang, E.A., Nelson, K.C. & Jakes, P., 2006, Working with community leadership to promote wildfire preparedness, Science Findings for Managers from National Fire Plan Research USDA Forest Service, Northern Research Station,General Technical Report NRS-1, US Depoartment of Agriculture, Forest Service, Northern Research Station, 137-149 pp.

Lavell, A., Oppenheime, M., Diop, C., Hess, J., Lempert, R., Li, J., Muir-Wood, R. & Myeong, S., 2012, Climate change: New dimensions in disaster risk, exposure, vulnerability, and resilience. in: Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation A Special Report of Working Groups I and II of the Intergovernmental Panel on Climate Change (IPCC), Cambridge University Press, Cambridge, UK, and New York, NY, USA, 25-64 pp.

Lee, E. & Jones, D., 2004, Landslide Risk Assessment, Thomas Telford Publishing, Heron Quay, London.

Lein, J. & Stump, N., 2009, Assessing wildfire potential within the wildland-urban interface: a southeastern Ohio example, Applied Geography, 29(1), 21-34, doi: 10.1016/j.apgeog.2008.06.002.

Levine, N., 2002, CrimeStat III: a spatial statistics program for the analysis of crime incident locations, Ned Levine & Associates, Houston, TX, and the National Institute of Justice, Washington, DC, doi: 10.3886/ICPSR02824.

Li, G.M., 2009, Tropical cyclone risk perceptions in Darwin, Australia: A comparison of different residential groups, Natural Hazards, 48(3), 365-382, doi: 10.1007/s11069-008-9269-8.

315

Li, J. & Heap, A.D., 2008, A Review of Spatial Interpolation Methods for Environmental Scientists, Geoscience Australia, Record 2008/23, Canberra, Australia. Available from: http://www.ga.gov.au/image_cache/GA12526.pdf [15 August 2012].

Lindell, M.K., 2011, Disaster studies, ISA eSymposium for Sociology. Available from: http://www.sagepub.net/isa/resources/pdf/Disaster%20Studies.pdf [20 March 2012].

Lindell, M.K. & Prater, C.S., 2003, Assessing community impacts of natural disasters, Natural Hazards Review, 4(4), 176-185, doi: 10.1061/(ASCE)1527- 6988(2003)4:4(176).

Lindell, M.K., Prater, C.S., Perry, R.W. & Nicholson, W.C., 2006, Fundamentals of Emergency Management, Federal Emergency Management Agency (FEMA), Washington DC, United States.

Longley, P.A., Goodchild, M.F., Maguire, D.J. & Rhind, D.W., 2010, Geographic Information Systems and Science, 3rd edn, John Wiley & Sons.

Lowe, T., 2008, New South Wales Fire Brigades Community Fire Unit Approach: A Report on the Background, Key issues and Future Directions Bushfire CRC, Melbourne.

Lowe, T., Haynes, K. & Byrne, G., 2008, Resilience at the urban interface: The community fire unit approach, in Community Bushfire Safety, J. Handmer & K. Haynes (eds), CSIRO, Melbourne, pp. 21-34.

Malczewski, J., 1996, A GIS-based approach to multiple criteria group decision- making, International Journal of Geographical Information Systems, 10(8), 955- 971, doi: 10.1080/02693799608902119.

Malczewski, J., 1999, GIS and Multicriteria Decision Analysis, John Wiley & Sons, New York.

Malczewski, J., 2004, GIS-based land-use suitability analysis: A critical overview, Progress in Planning, 62(1), 3-65, doi: 10.1016/j.progress.2003.09.002.

Malczewski, J., 2006, GIS based multicriteria decision analysis: A survey of the literature, International Journal of Geographical Information Science, 20(7), 703-726, doi: 10.1080/13658810600661508.

Mays, L.W., 2005, Water Resources Engineering, John Wiley & Sons, Inc.

Mayunga, J., 2007, Understanding and applying the concept of community disaster resilience: A capital-based approach, Draft working paper prepared for the Summer Academy for Social Vulnerability and Resilience UNU-EHS, Munich. Available from: http://www.ehs.unu.edu/file/get/3761 [20 June 2012].

316

Mazareanu, V., 2007, Risk management and analysis: Risk assessment (qualitative and quantitative), Analele Stiintifice ale Universitatii, 54, 42-46, doi: 10.2139/ssrn.1549186

McCaffrey, S., 2004, Thinking of wildfire as a natural hazard, Society & Natural Resources, 17(6), 509-516, doi: 10.1080/08941920490452445.

McCarthy, J.J., 2001, Climate change 2001: Impacts, Adaptation, and Vulnerability: Contribution of Working Group II to the Third Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge.

McGee, T., 2011, Public engagement in neighbourhood level wildfire mitigation and preparedness: Case studies from Canada, the US and Australia, Journal of Environmental Management, 90(10), 2524-2532, doi: 10.1016/j.jenvman.2011.05.017.

McGee, T.K. & Russell, S., 2003, “It's just a natural way of life…” an investigation of wildfire preparedness in rural Australia, Global Environmental Change B: Environmental Hazards, 5(1-2), 1-12, doi: 10.1016/j.hazards.2003.04.001.

McIvor, D. & Paton, D., 2007, Preparing for natural hazards: Normative and attitudinal influences, Disaster Prevention and Management, 16(1), 79-88, doi: 10.1108/09653560710729839

McLennan, B.J. & Handmer, J., 2012, Reframing responsibility-sharing for bushfire risk management in Australia after Black Saturday, Environmental Hazards, 11(1), 1-15, doi: 10.1080/17477891.2011.608835.

McLeod, R., 2003, Inquiry into the Operational Response to the January 2003 Bushfires in the ACT, Publication number 03/0537, ACT Government, Canberra. Available from: http://www.cmd.act.gov.au/ data/assets/pdf file/0008/113939/ McLeodInquiry.pdf [20 August 2012].

McMillan, D.W. & Chavis, D.M., 1986, Sense of community: A definition and theory, Journal of Community Psychology, 14(1), 6-23.

Mendes, J., 2009, Social vulnerability indexes as planning tools: Beyond the preparedness paradigm, Journal of Risk Research, 12(1), 43-58, doi: 10.1080/13669870802447962.

Mendoza, G.A. & Prabhu, R., 2000, Multiple criteria decision making approaches to assessing forest sustainability using criteria and indicators: A case study, Forest Ecology and Management, 131(1-3), 107-126, doi: 10.1016/S0378- 1127(99)00204-2.

Mercer, D.E. & Prestemon, J.P., 2005, Comparing production function models for wildfire risk analysis in the wildland-urban interface, Forest Policy and Economics, 7(5), 782-795, doi: 10.1016/j.forpol.2005.03.003.

317

Metzger, M.J. & Schröter, D., 2006, Towards a spatially explicit and quantitative vulnerability assessment of environmental change in Europe, Regional Environmental Change, 6(4), 201-216, doi: 10.1007/s10113-006-0020-2.

Miceli, R., Sotgiu, I. & Settanni, M., 2008, Disaster preparedness and perception of flood risk: A study in an alpine valley in Italy, Journal of Environmental Psychology, 28(2), 164-173, doi: 10.1016/j.jenvp.2007.10.006.

Middelmann, M. & Granger, K., 2000, Community risk in Mackay: a multi-hazard risk assessment, Australian Geological Survey Organisation, Canberra. Available from: http://www.ga.gov.au/image_cache/GA4177.pdf [10 July 2012].

Middelmann, M.H., 2007, Natural Hazards in Australia. Identifying Risk Analysis Requirements., Geoscience Australia, Canberra.

Montoya, D., 2010, Bushfires in NSW: An Update - Briefing Paper No 10/2010New South Wales Parliamentary Library Research Service, Sydney.

Morgan, P., Hardy, C.C., Swetnam, T.W., Rollins, M.G. & Long, D.G., 2001, Mapping fire regimes across time and space: Understanding coarse and fine-scale fire patterns, International Journal of Wildland Fire, 10(4), 329-342, doi: 10.1071/WF01032.

Morrow, B.H., 1999, Identifying and mapping community vulnerability, Disasters, 23(1), 1-18.

Murnane, R.J., 2006, Catastrophe risk models for wildfires in the wildland–urban interface: What insurers need, Natural Hazards Review, 7, 150-156, doi: 10.1061/ASCE1527-698820067:4150.

Myers, C., Slack, T. & Singelmann, J., 2008, Social vulnerability and migration in the wake of disaster: the case of Hurricanes Katrina and Rita, Population & Environment, 29(6), 271-291, doi: 10.1007/s11111-008-0072-y.

Nakagawa, Y. & Shaw, R., 2004, Social capital: A missing link to disaster recovery, International Journal of Mass Emergencies and Disasters, 22(1), 5-34.

Nardo, M., Saisana, M., Saltelli, A., Tarantola, S., Hoffman, A. & Giovannini, E., 2005, Handbook on constructing composite indicators: methodology and user guide, OECD Publishing.

National Research Council - Committee on Planning for Catastrophe a Blueprint for Improving Geospatial Data Infrastructure, 2007, Successful Response Starts with a Map: Improving Geospatial Support for Disaster Management, The National Academies Press Washington DC.

Newport, J.K. & Jawahar, G.G.P., 2003, Community participation and public awareness in disaster mitigation, Disaster Prevention and Management, 12(1), 33-36, doi: 10.1108/09653560310463838.

318

Nisar Ahamed, T., Gopal Rao, K. & Murthy, J., 2000, GIS-based fuzzy membership model for crop-land suitability analysis, Agricultural Systems, 63(2), 75-95.

Norris, F.H., Stevens, S.P., Pfefferbaum, B., Wyche, K.F. & Pfefferbaum, R.L., 2008, Community resilience as a metaphor, theory, set of capacities, and strategy for disaster readiness, American Journal of Community Psychology, 41(1), 127-150, doi: 10.1007/s10464-007-9156-6.

NSW Rural Fire Service, 2006, Guideline for Bush Fire Prone Land Mapping, Sydney. Available from: http://www.rfs.nsw.gov.au/file_system/attachments /State08 /At tachment2007022806EF9BB7.pdf [1 August 2012].

O'Brien, G., O'Keefe, P., Rose, J. & Wisner, B., 2006, Climate change and disaster management, Disasters, 30(1), 64-80, doi: 10.1111/j.1467-9523.2006.00307.x

O'Keefe, P., Westgate, K. & Wisner, B., 1976, Taking the naturalness out of natural disasters, Nature, 260, 566-567.

O'Sullivan, D. & Unwin, D.J., 2010, Geographic Information Analysis, 2nd edn, John Wiley & Sons Inc, Hoboken, NJ

O’Brien, K. & Leichenko, R., 2007, Human security, vulnerability and sustainable adaptation, Background Paper Commissioned for the Human Development Report, 2008.

Okada, N. & Matsuda, Y., 2005, Risk communication strategy for disaster preparedness viewed as multilateral knowledge development, IEEE International Conference on Systems, Man and Cybernetics, IEEE, Hawaii,USA, 640-647 pp.

Oliver-Smith, A., 1996, Anthropological research on hazards and disasters, Annual Review of Anthropology, 303-328.

Pastor, E., Planas, E. & Arnaldos, J., 2003, Mathematical models and calculation systems for the study of wildland fire behavior, Progress in Energy and Combustion Science, 29(2), 139-153, doi: 10.1016/S0360-1285(03)00017-0.

Paton, D., 2003, Disaster preparedness: A social-cognitive perspective, Disaster Prevention and Management, 12(3), 210-216, doi: 10.1108/09653560310480686.

Paton, D., 2006a, Disaster resilience: Integrating individual, community, institutional and environmental perspectives, in Disaster Resilience: An Integrated Approach, D. Paton & D. Johnston (eds), Charles C Thomas, Springfield, Illinois, pp. 305-317.

Paton, D., 2006b, Promoting Household and Community Preparedness for Bushfires: A review of issues that inform the development and delivery of risk communication strategies, Bushfire Co-operative Research Centre.

319

Paton, D., 2006c, Promoting Household and Community Preparedness for Bushfires: A Review of Issues that Inform the Development and Delivery of Risk Communication Strategies, Bushfire Co-operative Research Centre.

Paton, D., Bürgelt, P.T. & Prior, T., 2008, Living with bushfire risk: Social and environmental influences on preparedness, Australian Journal of Emergency Management, 23(3), 41-48.

Paton, D. & Johnston, D., 2001, Disasters and communities: Vulnerability, resilience and preparedness, Disaster Prevention and Management, 10(4), 270-277, doi: 10.1108/EUM0000000005930.

Paton, D. & Johnston, D., 2006, Disaster Resilience: An Integrated Approach, Charles C Thomas, Springfield.

Paton, D., Kelly, G., Burgelt, P.T. & Doherty, M., 2006, Preparing for bushfires: Understanding intentions, Disaster Prevention and Management, 15(4), 566- 575, doi: 10.1108/09653560610685893.

Paton, D., Smith, L.M. & Johnston, D., 2000, Volcanic hazards: Risk perception and preparedness, New Zealand Journal of Psychology, 29(2), 86-91.

Paveglio, T.B., Jakes, P.J., Carroll, M.S. & Williams, D.R., 2009, Understanding social complexity within the wildland–urban interface: A new species of human habitation?, Environmental Management, 43(6), 1085-1095, doi: 10.1007/s00267-009-9282-z.

Pearce, L., 2003, Disaster management and community planning, and public participation: How to achieve sustainable hazard mitigation, Natural Hazards, 28(2), 211-228, doi: 10.1023/A:1022917721797.

Pearce, L.D.R., 2000, An integrated approach for community hazard, impact, risk and vulnerability analysis: HIRV, Doctoral Dissertation, The University of British Columbia Vancouver.

Pelling, M., 2003, The Vulnerability of Cities. Natural Disasters and Social Resilience, Earthscan Publications, London.

Pelling, M. & High, C., 2005, Understanding adaptation: What can social capital offer assessments of adaptive capacity?, Global Environmental Change Part A, 15(4), 308-319, doi: 10.1016/j.gloenvcha.2005.02.001.

Petak, W. & Atkisson, A., 1982, Natural hazard risk assessment and public policy: anticipating the unexpected, Springer-Verlag, New York.

Pink, B., 2006, Socio-Economic Indexes for Areas (SEIFA):Technical Paper, Australian Bureau of Statistics, Canberra.

Pitman, A., Narisma, G. & McAneney, J., 2007, The impact of climate change on the risk of forest and grassland fires in Australia, Climatic Change, 84(3), 383-401.

320

Pope, J., 2010, Indicators of Community Strength at the Local Government Area Level in Victoria 2008, Department of Planning and Community Development, Melbourne. Available from: http://www.dpcd.vic.gov.au/ __data/assets/pdf_file/ 0014/30641/Indicators_of_Community_Strength_at_LGA_level_2008_FINAL_ low-res.pdf [10 July 2011].

Poudyal, N.C., Johnson-Gaither, C., Goodrick, S., Bowker, J. & Gan, J., 2012, Locating Spatial Variation in the Association Between Wildland Fire Risk and Social Vulnerability Across Six Southern States, Environmental Management, 9(3), 623-635, doi: 10.1007/s00267-011-9796-z.

Preston, B., Brooke, C., Measham, T., Smith, T. & Gorddard, R., 2009, Igniting change in local government: Lessons learned from a bushfire vulnerability assessment, Mitigation and Adaptation Strategies for Global Change, 14(3), 251-283, doi: 10.1007/s11027-008-9163-4.

Preston, B.L., Yuen, E.J. & Westaway, R.M., 2011, Putting vulnerability to climate change on the map: a review of approaches, benefits, and risks, Sustainability Science, 6(2), 177-202, doi: 10.1007/s11625-011-0129-1.

Prior, T., Paton, D. & Cottrell, A., 2008, Household bushfire preparation: decision- making and the implications for risk communication, The International Bushfire Research Conference 2008 - incorporating The 15th annual AFAC Conference, Adelaide, Australia, 1-3 September 2008.

Prior, T.D., 2010, Householder bushfire preparation: decision-making and the implications for risk communication, PhD Thesis, University of Tasmania.

Putnam, R.D., 2001, Bowling Alone: The Collapse and Revival of American Community, Simon and Schuster, New York.

Rashed, T., Weeks, J., Couclelis, H. & Herold, M., 2007, An integrative GIS and remote sensing model for place-based urban vulnerability analysis, in Integration of GIS and remote sensing, V. Mesev (ed.), John Wiley & Sons, New York, pp. 199-224.

Renn, O., 1992, Concepts of risk: A classification, in Social Theories of Risk, S. Krimsky & D. Golding (eds), Praeger Publishers, Westport, CT, USA, pp. 53- 82.

Rhodes, A., 2011, Opinion: Ready or not?: can community education increase householder preparedness for bushfire?, The Australian Journal of Emergency Management, 26(2), 6-10.

Rigina, O. & Baklanov, A., 2002, Regional radiation risk and vulnerability assessment by integration of mathematical modelling and GIS analysis, Environment International, 27(7), 527-540, doi: 10.1016/S0160-4120(01)00104-0.

321

Rushton, G., 2003, Public health, GIS, and spatial analytic tools, Annual Review of Public Health, 24(1), 43-56, doi: 10.1146/annurev.publhealth. 24.012902.14084 3.

Ryan, R.L. & Wamsley, M.B., 2008, Public perceptions of wildfire risk and forest management in the central pine barrens of Long Island(USA), Australasian Journal of Disaster and Trauma Studies, 2008(2).

Rygel, L., O’sullivan, D. & Yarnal, B., 2006, A method for constructing a social vulnerability index: An application to hurricane storm surges in a developed country, Mitigation and Adaptation Strategies for Global Change, 11(3), 741- 764, doi: 10.1007/s11027-006-0265-6.

Saaty, T., 1980, The analytical Hierarchy Process, McGraw-Hill, New York.

Saisana, M., Saltelli, A. & Tarantola, S., 2005, Uncertainty and sensitivity analysis techniques as tools for the quality assessment of composite indicators, Journal of the Royal Statistical Society: Series A (Statistics in Society), 168(2), 307-323.

Saltelli, A., 2007, Composite indicators between analysis and advocacy, Social Indicators Research, 81(1), 65-77.

Salter, J., 1997, Risk management in a disaster management context, Journal of Contingencies and Crisis Management, 5(1), 60-65, doi: 10.1111/1468- 5973.00037.

Sands, G.R. & Podmore, T.H., 2000, A generalized environmental sustainability index for agricultural systems, Agriculture, Ecosystems & Environment, 79(1), 29-41, doi: 10.1016/S0167-8809(99)00147-4.

Schellong, A., 2007, Increasing Social Capital for Disaster Response through Social Networking Services (SNS) in Japanese local governments, NCDG Working Paper No. 07-005, National Centre for Digital Government

Schmidtlein, M.C., Deutsch, R.C., Piegorsch, W.W. & Cutter, S.L., 2008, A sensitivity analysis of the social vulnerability index, Risk Analysis, 28(4), 1099-1114, doi: 10.1111/j.1539-6924.2008.01072.x.

Schneiderbauer, S. & Ehrlich, D., 2004, Risk, Hazard and People’s Vulnerability to Natural Hazards: A Review of Definitions, Concepts and Data, Brussels: European Commission–Joint Research Centre (EC-JRC), Brussels.

Secunda, S., Collin, M. & Melloul, A.J., 1998, Groundwater vulnerability assessment using a composite model combining DRASTIC with extensive agricultural land use in Israel's Sharon region, Journal of Environmental Management, 54(1), 39- 57, doi: 10.1006/jema.1998.0221.

Sharples, J., McRae, R., Weber, R. & Gill, A., 2009, A simple index for assessing fire danger rating, Environmental Modelling & Software, 24(6), 764-774, doi: 10.1016/j.envsoft.2008.11.004.

322

Shields, B. & Tolhurst, K., 2003, A theoretical framework for wildfire risk assessment, 3rd International Wildland Fire Conference and 10th Annual AFAC Conference, Sydney. Available from: http://www.fire.uni-freiburg.de/summit- 2003/3-IWFC/Papers/3-IWFC-130-Shields.pdf [20 June 2011].

Simpson, D. & Human, R., 2008, Large-scale vulnerability assessments for natural hazards, Natural Hazards, 47(2), 143-155, doi: 10.1007/s11069-007-9202-6.

Simpson, D. & Katirai, M., 2006, Indicator Issues and Proposed Framework for a Disaster Preparedness Index (DPi), University of Louisville, 49 pp.

Sjoberg, L., 1999, Consequences of perceived risk: Demand for mitigation, Journal of Risk Research, 2(2), 129-149, doi: 10.1080/136698799376899.

Sjöberg, L., 1998, Risk perception: Experts and the public, European Psychologist, 3(1), 1, doi: 10.1027//1016-9040.3.1.1.

Sjöberg, L., 2000, Factors in risk perception, Risk Analysis, 20(1), 1-12.

Slovic, P., 1987, Perception of risk, Science, 236(4799), 280-285, doi: 10.1126/science.3563507.

Slovic, P.E., 2000, The Perception of Risk, Earthscan Publications, London.

Smith, K., 2001, Environmental Hazards : Assessing Risk and Reducing Disaster, Routledge, New York.

Smith, K. & Petley, D.N., 2009, Environmental hazards: assessing risk and reducing disaster, Taylor & Francis, Bodmin.

Smolka, A., 2006, Natural disasters and the challenge of extreme events: risk management from an insurance perspective, Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 364(1845), 2147-2165, doi: 10.1098/rsta.2006.1818.

Tapsell, S., McCarthy, S., Faulkner, H. & Alexander, M., 2010, Social Vulnerability to Natural Hazards, CapHaz-Net’s WP4 Report, Flood Hazard Research Centre (FHRC), Middlesex University, London. Available from: http://caphaz-net.org /outcomes-results/CapHaz-Net_WP4_Social-Vulnerability2.pdf [26 October 2011].

Teague, B., McLeod, R. & Pascoe, S., 2009, Victorian Bushfires Royal Commission Final Report, Parliament of Victoria, Government Printer for the State of Victoria.

Thinh, N.X. & Vogel, R., 2007, Application of the analytic hierarchy process in the multiple criteria decision analysis of retention areas for flood risk management, Environmental Informatics and Systems Research. EnviroInfo Warsaw, 675-682.

323

Thywissen, K., 2006, Components of Risk: A Comparative Glossary., UNU-EHS, Bonn. Available from: http://www.ehs.unu.edu/file/get/8335 [10 June 2010].

Tierney, K.J., 2001, Facing the Unexpected: Disaster Preparedness and Response in the United States, Joseph Henry Press, Washington DC.

Timmerman, P., 1981, Vulnerability, resilience, and the collapse of society, Environmental Monograph, 1.

Tobin, G.A. & Montz, B.E., 1997, Natural Hazards: Explanation and Integration, Guilgord Publications, New York.

Tolhurst, K., Shields, B. & Chong, D., 2008, Phoenix: Development and application of a bushfire risk management tool, The Australian Journal of Emergency Management, 23(4), 47-54.

Tran, P., Shaw, R., Chantry, G. & Norton, J., 2009, GIS and local knowledge in disaster management: A case study of flood risk mapping in Viet Nam, Disasters, 33(1), 152-169, doi: 10.1111/j.0361-3666.2008.01067.x.

Turner, B.L., Kasperson, R.E., Matson, P.A., McCarthy, J.J., Corell, R.W., Christensen, L., Eckley, N., Kasperson, J.X., Luers, A., Martello, M.L., Polsky, C., Pulsipher, A. & Schiller, A., 2003, A framework for vulnerability analysis in sustainability science, Proceedings of the National Academy of Sciences of the United States of America, 100(14), 8074-8079, doi: 10.1073/pnas.1231335100.

UNDRO, 1980, Natural Disasters and Vulnerability Analysis in Report of Expert Group Meeting UNDRO, Geneva.

UNISDR, 2004, Living with Risk. A Global Review of Disaster Reduction Initiatives, United Nations, Geneva.

UNISDR, 2005, Hyogo framework for action 2005–2015: Building the resilience of nations and communities to disasters, World Conference on Disaster Reduction, Kobe, Hyogo, Japan.

UNISDR, 2009a, Global Assessment Report on Disaster Risk Reduction: Risk and Poverty in a Changing Climate, Manama.

UNISDR, 2009b, UNISDR Terminology for Disaster Risk Reduction Available from: http://www.unisdr.org/we/inform/publications/7817 [12 March 2012].

UNISDR, 2011, Global Assessment Report on Disaster Risk Reduction: Revealing Risk,Redefining Development, United Nations Available from: http://www.preventionweb.net/english/hyogo/gar/ [31 July 2012].

324

United Nations, 1999, Final report of the Scientific and Technical Committee of the International Decade for Natural Disaster Reduction, General Assembly Economic and Social Council, Geneva, 1-16 pp. Available from: http://www.preventionweb.net/files/8151_8151IDNDRSTCfinalreport819991.p df [30 July 2012].

UNU-EHS, 2011, World Risk Report 2011, The Alliance Development Works Bonn. Available from: http://www.ehs.unu.edu/file/get/9018 [25 March 2012].

Vadrevu, K.P., Eaturu, A. & Badarinath, K.V.S., 2010, Fire risk evaluation using multicriteria analysis—a case study, Environmental monitoring and assessment, 166(1), 223-239.

Vahidnia, M., Alesheikh, A., Alimohammadi, A. & Bassiri, A., 2008, Fuzzy analytical hierarchy process in GIS application, The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 37.

Van Westen, C.J., Rengers, N., Terlien, M. & Soeters, R., 1997, Prediction of the occurrence of slope instability phenomena through GIS-based hazard zonation, Geologische Rundschau, 86(2), 404-414, doi: 10.1007/s005310050149.

Vasilakos, C., Kalabokidis, K., Hatzopoulos, J., Kallos, G. & Matsinos, Y., 2007, Integrating new methods and tools in fire danger rating, International Journal of Wildland Fire, 16(3), 306-316, doi: 10.1071/WF05091.

Verde, J. & Zêzere, J., 2010, Assessment and validation of wildfire susceptibility and hazard in Portugal, Natural Hazards and Earth System Sciences, 10(3), 485-497, doi: 10.5194/nhess-10-485-2010.

Victorian Bushfires Royal Commission, 2009, The Interim Report of the 2009 Victorian Bushfires Commission, State Government of Victoria, Melbourne.

Villagrán de León, J.C., 2006, Vulnerability: A Conceptual and Methodological Review, UNU-EHS, Bonn.

Villagrán de León, J.C., 2008, GIRO: The Integral Risk Management Framework, An Overview, UNU-EHS Working Paper, Bonn.

Vincent, K., 2004, Creating an index of social vulnerability to climate change for Africa, Tyndall Center for Climate Change Research. Working Paper, 50 pp.

Vyas, S. & Kumaranayake, L., 2006, Constructing socio-economic status indices: how to use principal components analysis, Health Policy and Planning, 21(6), 459- 468, doi: 10.1093/heapol/czl029.

Wainwright, J. & Mulligan, M., 2004, Environmental Modelling: Finding Simplicity in Complexity, John Wiley & Sons Inc, Chichester.

325

Walker, B., Holling, C., Carpenter, S. & Kinzig, A., 2004, Resilience, adaptability and transformability in social--ecological systems, Ecology and Society, 9(2), article 5.

Watts, M.J. & Bohle, H.G., 1993, The space of vulnerability: the causal structure of hunger and famine, Progress in Human Geography, 17(1), 43-67, doi: 10.1177/030913259301700103.

Wei, Y.M., Fan, Y., Lu, C. & Tsai, H.T., 2004, The assessment of vulnerability to natural disasters in China by using the DEA method, Environmental Impact Assessment Review, 24(4), 427-439.

Weichselgartner, J., 2001, Disaster mitigation: the concept of vulnerability revisited, Disaster Prevention and Management, 10(2), 85-95, doi: 10.1108/09653560110388609.

Western, J., Stimson, R., Baum, S. & Van Gellecum, Y., 2005, Measuring community strength and social capital, Regional Studies, 39(8), 1095-1109, doi: 10.1080/00343400500328222.

White, G., 1964, Choice of adjustment to floods, Department of Geography Research Papers, No. 93, University of Chicago, Chicago.

White, G.F., 1973, Natural hazards research, in Directions in Geography, R.J. Chorley (ed.), Methuen, London.

White, R. & Engelen, G., 1993, A cellular automata and fractal urban form: a cellular modelling approach to the evolution of urban land-use patterns, Environment and Planning 25(8), 1175-1199.

Whitman, R., 1973, Damage probability matrices for prototype buildings, Massachusetts Institute of Technology, Department of Civil Engineering Research Report R73-57, Cambridge, MA.

Whittaker, J., 2008, Vulnerability to bushfires in south-eastern Australia: a case study from East Gippsland, Victoria., PhD Thesis, RMIT University,Melbourne, 282 pp.

Winkworth, G., Healy, C., Woodward, M. & Camilleri, P., 2009, Community capacity building: Learning from the 2003 Canberra bushfires, The Australian Journal of Emergency Management, The, 24(2), 5-12.

Winter, G. & Fried, J.S., 2000, Homeowner perspectives on fire hazard, responsibility, and management strategies at the wildland-urban interface, Society & Natural Resources, 13(1), 33-49, doi: 10.1080/089419200279225.

Wisner, B., Blakie, P., Cannon, T. & Davis, I., 2004, At Risk: Natural Hazards, Peoples Vulnerability and Disasters, 2nd edn, Routledge, London.

326

Wood, L.J. & Dragicevic, S., 2007, GIS-based multicriteria evaluation and fuzzy sets to identify priority sites for marine protection, Biodiversity and Conservation, 16(9), 2539-2558, doi: 10.1007/s10531-006-9035-8.

Wood, N., Burton, C. & Cutter, S., 2010, Community variations in social vulnerability to Cascadia-related tsunamis in the US Pacific Northwest, Natural Hazards, 52(2), 369-389, doi: 10.1007/s11069-009-9376-1.

Wu, S., Yarnal, B. & Fisher, A., 2002, Vulnerability of coastal communities to sealevel rise: A case study of Cape May County, New Jersey, USA, Climate Research, 22(3), 255-270, doi: 10.3354/cr022255.

Yalcin, G. & Akyurek, Z., 2004, Analysing flood vulnerable areas with multicriteria evaluation, ISPRS Congress Istanbul, 12-23 July 2004 12-23 pp. Available from: http://www.isprs.org/proceedings/XXXV/congress/comm2/papers/154.pdf [20 June 2011].

Yamamoto, Y. & Quarantelli, E.L., 1982, Inventory of the Japanese Disaster Research Literature in the Social and Behavioral Sciences, Disaster Research Center, Ohio State University, Columbus.

Yeletaysi, S., Ozceylan, D., Fiedrich, F., Harrald, J. & Jefferson, T., 2009, A framework to integrate social vulnerability into catastrophic natural disaster preparedness planning, TIEMS Annual Conference, Istanbul, 380-389 pp.

Yodmani, S., 2001, Disaster Risk Management and Vulnerability Reduction: Protecting the Poor, paper presented at The Asian and Pacific Forum on Poverty, Manila, Philippines.

Zadeh, L.A., 1965, Fuzzy sets, Information and Control, 8(3), 338-353, doi: 10.1016/S0019-9958(65)90241-X.

Zahran, S., Brody, S.D., Peacock, W.G., Vedlitz, A. & Grover, H., 2008, Social vulnerability and the natural and built environment: A model of flood casualties in Texas, Disasters, 32(4), 537-560, doi: 10.1111/j.1467-7717.2008.01054.x.

Zeng, T., Hudson, J., Kay, S., Laginestra, E. & Authority, S.O.P., 2003, A fuzzy GIS approach to fire risk assessment: a case study of Sydney Olympic Park, Australia, Spatial Sciences Conferences 2003.

Zerger, A., 2002, Examining GIS decision utility for natural hazard risk modelling, Environmental Modelling & Software, 17(3), 287-294, doi: 10.1016/S1364- 8152(01)00071-8.

Zerger, A.Z., 1998, Cyclone inundation risk mapping, PhD Thesis, Centre for Resource and Environmental Studies, Australian National University, Canberra.

327

Zhijun, T., Jiquan, Z. & Xingpeng, L., 2009, GIS-based risk assessment of grassland fire disaster in western Jilin province, China, Stochastic Environmental Research and Risk Assessment, 23(4), 463-471, doi: 10.1007/s00477-008-0233- 7.

328

APPENDIX I: Participant Information Statement

Ethics approval No: A-10-54

01st December 2011

Dear Respondent,

Bushfire Risk Assessment Survey - 2011: Participant Information Statement

You are invited to participate in a study about bushfire risk at the urban-bush interface in Sydney. This survey is being conducted by researchers from the University of New South Wales @ ADFA, Canberra in conjunction with the Blue Mountains City Council. Participation in the study is voluntary. You were chosen to participate in this study because you live in Blue Mountains City Council area.

8.4.1.1.1 Study Purpose This study is attempting to identify characteristics of communities that lead to community-level resilience to bushfires. This information can help managers more effectively direct resources to places that need support to become more resilient to bushfires. 8.4.1.1.2 Study Details The survey will take approximately 20 minutes to complete. It contains general demographic questions and questions about the things that may affect your risk of experiencing a bushfire and actions that you have taken to minimize the impact of bushfire in your community.

Please look over the questionnaire and, if you choose to do so, complete the questionnaire and send it back in the enclosed postage-paid envelope.

Recompense to participants There is no cost to you to for participating in this research, nor is there any remuneration available for your participation.

8.4.1.2 Confidentiality

You are not asked to provide your name or address and your response will not be identifiable. In other words, your participation will be anonymous.

If you give your permission to participate in this research, the results will be used for a PhD thesis and articles in scientific journals.

Completion and return of this survey indicates your consent to participate in the research. Since it will not be possible to identify any particular person’s survey after the surveys have been returned, it is not possible for you to withdraw from the research after returning the survey. This should be borne in mind when making your decision about participation.

329

Your decision on whether or not to participate in this research will not prejudice your future relations with the University of New South Wales or your local council. You may keep this form.

If you have any questions regarding this survey, please contact Daminda Solangaarachchi (Phone: 0421566108, E-mail: [email protected]) or Dr. Amy Griffin (Phone : 02 6268 8949, email : [email protected]).

Complaints may be directed to Dr Stephen Coleman; Convenor, Human Research Ethics Advisory Panel, UNSW@ADFA, CANBERRA 2600 (phone (02) 6268-8812, fax (02) 6268-8899, email [email protected]). Any complaint you make will be investigated promptly and you will be informed of the outcome.

330

APPENDIX II: Household Survey Questionnaire

Household Survey Questionnaire

Bushfire Risk Assessment Survey –2011

This survey is being conducted by researchers from the University of New South Wales ADFA in conjunction with Local Council and the NSW Fire Brigade. The purpose of this questionnaire is to collect information on awareness of bushfire risk and bushfire preparedness activities in the Ku-ring-gai area. The questionnaire will take approximately 20 - 25 minutes to complete and all the information provided will be kept confidential. Please mark the correct answers to the questions by circling the option that best represent your opinion and provide more detailed information when appropriate.

Demographics

This selection of the survey contains questions on basic aspects of your household demographics. We are interesting in finding out this information so that we can compare your level of bushfire awareness, preparedness and community level participation with other similar respondents.

1. What is your household type? 1. I live alone 2. Couple only 3. Couple with children 4. Single parent 5. Group (unrelated adults) 6. Extended family (related adults and children) 7. Other (specify: ______)

2. Specify the number of people in your household falling into the categories given below

Age Males Females Below 5 years 5-17 years 18 – 65 years 65 + years

3. How many adults in your household work?

Type # of adults Full time (35 hours per week or more) Part time (less than 35 hours per week) Casual Temporary worker Retired Unemployed

331

4. What is your household income? 1. Less than $300/pw 2. $300 - $499/pw 3. $500 - $699/pw 4. $700 - $999/pw 5. Above $1000/pw 6. Prefer not to answer

5. What is the level of education of the adults in your household? Level of education # of adults Less than secondary school Secondary school (year 12) Trade / technical qualification University degree Postgraduate

6. On a typical week day how many people are in your house? During day time: ______During night time: ______

7. What type of residence do you live in? 1. Single detached house 2. Semi-detached, row or terrace house or townhouse 3. Flat / Unit /Apartment 4. Other (specify:______)

8. Which of the following statements applies to your residence? 1. I own the residence and have no mortgage. 2. I own the residence but have a mortgage. 3. I am the sole renter. 4. I rent the property with others. 5. I live in public housing 6. Other (specify:______)

Risk knowledge and bushfire awareness

In this section, we are interested in finding out your experience with past bushfire events, risk knowledge and awareness. This information will help to improve the future bushfire awareness activities in your area.

9. How many years have you lived in your current residence? ______Years Suburb: ______

332

10. Why did you choose to live in Ku-ring-gai/Blue Mountains? (Select all that apply) 1. Affordable housing 2. Proximity to work 3. Quality of the natural environment 4. Quality of facilities (Education, Health, Security, etc.) 5. For lifestyle 6. I grew up there 7. Close to my family 8. Other (specify:______)

11. Has anyone in your household experienced a bushfire before? (Select all that apply)

a. Yes in this neighbourhood b. Yes, in somewhere else c. No

12. Rank the potential risk (1 to 4) of the following hazards in your suburb. (1- lowest risk ; 4 – highest risk)

Hazard Risk Bushfire Earthquake Severe storms Floods

13. Have you received information about bushfire prevention methods in the last six months? 1. Yes 2. No (Go to question 16)

14. How did you receive that information? (Select all that apply) 1. By mail 2. Community discussions 3. TV / Radio 4. Local council newsletter 5. Newspaper 6. Neighbours/friends 7. Community newsletter 8. (Source: ______) 9. Internet 10. Other (Specify:______)

15. If so, did you think that information was useful? 1. Yes, useful 2. Not sure 3. Not useful (explain why )______4. I didn’t have time to read it

333

16. How many total fire bans were there in your suburb in the last bushfire season? ______

17. How do you find out about total fire bans about your area? (Select all that apply) 1. Local government newsletter 2. Community newsletter (Source: ______) 3. From the neighbours 4. Newspaper advertisement 5. Television/Radio advertisement 6. Roadside fire danger signs 7. Internet 8. Other (Specify: ______)

18. On a scale from 1 to 5, where 1 is not aware and 5 is fully aware, how do you evaluate your level of awareness on total fire ban rules and penalties?

Not aware 1 2 3 4 5 Fully aware

19. On a scale from 1 to 5, where 1 is no risk and 5 is very high risk, evaluate the level of bushfire risk in your area by circling the number that best represents your opinion.

Very low risk 1 2 3 4 5 Very high risk

Preparedness to bushfires (Household level)

In this part of the survey, we want to find out about your household’s bushfire preparedness activities. This information will help to develop programmes to encourage household bushfire preparedness activities.

20. When do you think the bushfire season starts? 1. September – October 2. November – December 3. January- February 4. March – April 5. I don’t know

21. When do you normally start to prepare for bushfires? 1. After a bushfire ignites in the Sydney metropolitan area 2. After a bushfire ignites elsewhere in Australia 3. When the authorities tell me to prepare 4. When I see neighbours or others in the community preparing 5. Before the bushfire season starts 6. I don’t normally prepare (Go to question 25)

22. In which month do you normally start bushfire preparedness activities? 1. July - August 2. September - October 3. November - December 4. January - February

334

23. What encourages you to prepare for bushfires? (Select all that apply) 1. Media advertisements 2. Information received from local authorities 3. Bushfires in the Sydney metropolitan area 4. Bushfires elsewhere in Australia 5. Previous experience with bushfires 6. Other (specify: ______)

24. How do you prepare? (Select all that apply) 1. Remove leaf litter, undergrowth, etc around the house 2. Clear gutters of leaves 3. Maintain a fire break 4. Move combustible materials such as firewood or fuel away from the house 5. Get equipment such as a ladder, bucket and mops for spot fires 6. Clear vegetation away from the house 7. House modification to increase bushfire resistance (install sprinklers, replace wooden materials) 8. Other (specify: ______)

25. If you don’t prepare, why not? (Select all that apply) 1. I don’t live in a bushfire-prone area so it’s not necessary 2. It’s not my responsibility 3. I don’t have resources 4. I don’t have time 5. I don’t know how to prepare 6. Other (specify: ______)

26. On a scale from 1 to 5, where 1 is not prepared and 5 is fully prepared, how do you assess your household level preparedness for a bushfire? Circle the number that best represents your opinion.

Not prepared 1 2 3 4 5 Fully prepared

27. Do you have insurance for your residence? 1. Yes 2. No (Go to Question 29) 3. Don’t know (Go to Question 29)

28. Which of the following costs will your insurance fully cover? 1. Rebuilding the house only 2. Refurnishing the house only 3. Both rebuilding and refurnishing 4. Don’t know

29. Please indicate the roofing and wall materials of your property 1. Roofing: ______2. Wall: ______

335

30. Have you discussed bushfire safety and prevention methods with a local authority? 1. Yes 2. No (Go to question 32)

31. If yes, with which authority? (Select all that apply) 1. Local council 2. NSW rural fire services 3. NSW fire brigade 4. NSW parks and wildlife 5. Other (specify: ______)

32. Have you discussed what to do in the event of a fire with your family? 1. Yes 2. No

33. Have you discussed the following assessment tools with your family?

I am not Tool Yes No aware of Household bushfire risk assessment by NSW Rural fire services (NSW RFS) Bushfire survival plan by NSW RFS Asset protection zone construction tool by RFS Recover after a fire event by NSW RFS

Preparedness to bushfires – Community level

In this section, we are interested in finding out what your community is like and how it is prepared for bushfires. This information will help to improve bushfire preparedness activities in your area.

34. Please indicate how strongly you agree or disagree with this statement “I feel like I belong to the local community”.

1. Strongly agree 2. Agree 3. Neither agree nor disagree 4. Disagree 5. Strongly disagree

35. What are the sources of the connection you feel to your community? (Select all that apply)

1. Club 2. Church 3. Community organization 4. Neighbours 5. Work 6. School 7. Other (specify: ______)

336

36. How do people in your community help each other in an emergency? 1. They always help each other 2. They sometimes help each other 3. They occasionally help each other 4. They never help each other 5. There hasn’t been an emergency since I have lived here 6. I don’t know

37. Do people in your community discuss bushfire preparedness? 1. Yes 2. No 3. I don’t know

38. Does your community work together on bushfire preparedness activities? 1. Yes, always 2. Yes, sometimes 3. Yes, rarely 4. No 5. I don’t know

Please indicate how strongly you agree or disagree with the following statements.

39. My community is protected against potential bushfire events.

1 2 3 4 5 Strongly disagree Disagree Neither agree nor disagree Agree Strongly agree

40. I am happy with current bushfire management practices in my community.

1 2 3 4 5 Strongly disagree Disagree Neither agree nor disagree Agree Strongly agree

41. Bushfire management activities have improved in my community since 2005.

1 2 3 4 5 Strongly disagree Disagree Neither agree nor disagree Agree Strongly agree

42. How involved are you in community level bushfire prevention activities? 1. I always participate 2. I sometimes participate 3. I rarely participate 4. I never participate

43. If you never participate, why? 1. I don’t live in a bushfire-prone area so it’s not necessary 2. It’s not my responsibility 3. I don’t know how to get involved 4. I don’t have time 5. I don’t care 6. Other (specify: ______)

337

44. Would you like to be involved in a community fire guard /Community Fire Unit programme?

1. Yes 2. No / Not interested 3. I’d like to, but I am too busy 4. Other (specify: ______)

45. Who do you believe is the most responsible parties for bushfire preparedness and mitigation in your area? Select all parties for each activity

NSW Rural NSW Local NSW Fire Fire management activity Residents CFU Fire Parks and Council Brigade Services Wildlife Safe Prevention, Preparation

and Suppression Management of Fire in the

Landscape Community Self-Sufficiency

for Fire Safety Protection of People and

Property Education, Training and

Communication

46. What do you believe would be the best practice for bushfire preparedness and mitigation in your area?

1. Small resident group activities within the community 2. Larger community level activities with all residents in a public location 3. Clubs and other organizational level activities 4. Distribution of information through local letter boxes 5. Information in local media (local newsletters, notice boards etc.) 6. General advertising and media coverage (television, radio, etc) 7. Other (Specify: ______)

47. What areas do you think needs to be better developed? (Select all that apply) 1. Safe Prevention, Preparation and Suppression 2. Management of Fire in the Landscape 3. Community Self-Sufficiency for Fire Safety 4. Protection of People and Property 5. Education, Training and Communication 6. Other (Specify ______)

338

48. Rate your satisfaction with the following infrastructure/service facilities in your community.

Infrastructure Highly Neither satisfied Highly Satisfied Unsatisfied facilities satisfied nor unsatisfied unsatisfied Water supply Road network /

transport Fire fighting

resources Emergency warning Emergency

healthcare

Response and Recovery

In this section we want to find out what actions you would take during a bushfire and how your community would respond to fires. This information will help to enhance the response and recovery mechanisms.

49. To whom would you report a fire if you see one? (Select all that apply) 1. 000 2. Local council 3. NSW rural fire services 4. NSW fire brigade 5. Community fire units 6. NSW parks and wildlife 7. Family, friends or neighbours 8. No one

50. How strongly you agree or disagree with this statement.”My family would evacuate in an event of life-threatening bushfire”.

1. Strongly agree 2. Agree 3. Neither agree nor disagree 4. Disagree 5. Strongly disagree

51. When is the right time to evacuate? 1. As soon as fire approaches the edge of the suburbs 2. When I see the fire in the media 3. When I feel heat and see smoke 4. When neighbours start to evacuate 5. I wait until authorities tell me to leave 6. Other (specify: ______)

339

52. How would you leave?

1. Car 2. Bike 3. Foot 4. Truck 5. Other (Specify: ______)

53. To where would you go? 1. Identified evacuation point 2. Community centre 3. Friend or relative’s house 4. Hotel/motel 5. Uncertain 6. Other (specify: ______)

54. Have you identified a potential evacuation route?

1. Yes 2. No 55. If you would not evacuate, what is the reason? (Select all that apply) 1. To save the property 2. We have a fire bunker 3. It is safer to stay than to evacuate 4. I have faith in bushfire management activities 5. Other (Specify: ______)

56. Who is responsible for the following emergency response and recovery activities in your area? (Select all that apply)

Fire management Local NSW rural fire NSW fire NSW Residents CFU activity Council services brigade Police Early warning Fire fighting Assisting people to evacuate from the area Providing relief and

recovery needs Remove debris after

an event

This is the end of the Questionnaire, Thank you for participating.

340

III – Pairwise Comparison Tool

Vulnerability Components

What makes people more vulnerable to bushfires?

EI VSI SI SLI EI SLI SI VSI EI

Demographic and Socioeconomic 1 Exposure and Physical Susceptibility (Location and Structures) Conditions

2 Exposure and Physical Susceptibility (Location and Structures) Response and Coping Capacities Demographic and Socioeconomic 3 Conditions Response and Coping Capacities EI = Extremely Important, VSI = Very Strongly Important, SI = Strongly Important, SLI = Slightly Important, EI = Equally Important

341

Indicators – Exposure and Physical Susceptibility

What makes people and properties more exposed to bushfires? EI VSI SI SLI EI SLI SI VSI EI 1 Proximity to bushland Flammable vegetation in the neighbourhood 2 Proximity to bushland Slope 3 Proximity to bushland Aspect 4 Proximity to bushland Land use type 5 Proximity to bushland Building density Density of critical infrastructure (hospital, day care , 6 Proximity to bushland preschool, etc.) 7 Flammable vegetation in the neighbourhood Slope 8 Flammable vegetation in the neighbourhood Aspect 9 Flammable vegetation in the neighbourhood Land use type 10 Flammable vegetation in the neighbourhood Building density Density of critical infrastructure (hospital, day care , 11 Flammable vegetation in the neighbourhood preschool, etc.) 12 Slope Aspect 13 Slope Land use type 14 Slope Building density Density of critical infrastructure (hospital, day care , 15 Slope preschool, etc.) 16 Aspect Land use type 17 Aspect Building density Density of critical infrastructure (hospital, day care , 18 Aspect preschool, etc.) 19 Land use type Building density Density of critical infrastructure (hospital, day care , 20 Land use type preschool, etc.) Density of critical infrastructure (hospital, day care , 21 Building density preschool, etc.) EI = Extremely Important, VSI = Very Strongly Important, SI = Strongly Important, SLI = Slightly Important, EI = Equally Important

342

Indicators – Emergency Response and Coping Capacity What increases the level of response and coping capacity EI VSI SI SLI EI SLI SI VSI EI 1 Distance to main highways Good road network 2 Distance to main highways Number of dwellings per Community Fire Unit 3 Distance to main highways Proximity to fire station 4 Distance to main highways Proximity to evacuation point 5 Distance to main highways Proximity to hospital Density of water hydrants and static water supply 6 Distance to main highways points 7 Good road network Number of dwellings per Community Fire Unit 8 Good road network Proximity to fire station 9 Good road network Proximity to evacuation point 10 Good road network Proximity to hospital Density of water hydrants and static water supply 11 Good road network points 12 Number of dwellings per Community Fire Unit Proximity to fire station 13 Number of dwellings per Community Fire Unit Proximity to evacuation point 14 Number of dwellings per Community Fire Unit Proximity to hospital Density of water hydrants and static water supply 15 Number of dwellings per Community Fire Unit points 16 Proximity to fire station Proximity to evacuation point 17 Proximity to fire station Proximity to hospital Density of water hydrants and static water supply 18 Proximity to fire station points 19 Proximity to evacuation point Proximity to hospital Density of water hydrants and static water supply 20 Proximity to evacuation point points Density of water hydrants and static water supply 21 Proximity to hospital points EI = Extremely Important, VSI = Very Strongly Important, SI = Strongly Important, SLI = Slightly Important, EI = Equally Important

343

344