Tropical Cyclone 'Roger' Storm Surge Assessment

THE UNIVERSITY OF QUEENSLAND

SCHOOL OF

CIVIL ENGINEERING

TROPICAL CYCLONE ‘ROGER’

STORM SURGE ASSESSMENT

J. STEWART, D. CALLAGHAN and P. NIELSEN

RESEARCH REPORT CE162

CIVIL ENGINEERING RESEARCH REPORTS

This report is one of a continuing series of Research Reports published by the School of Civil Engineering at the University of Queensland. Lists of recently-published titles of this series and of other publications are provided at the end of this report. Requests for copies of any of these documents should be addressed to the School Secretary.

The interpretation and opinions expressed herein are solely those of the author(s). Considerable care has been taken to ensure accuracy of the material presented. Nevertheless, responsibility for the use of this material rests with the user.

School of Civil Engineering The University of Queensland Brisbane QLD 4072 AUSTRALIA

Telephone: (61 7) 3365 3619 Fax: (61 7) 3365 4599

URL: http://www.eng.uq.edu.au/civil/

First published in 2010 by Jared STEWART, David CALLAGHAN and Peter NIELSEN School of Civil Engineering The University of Queensland, Brisbane QLD 4072, Australia

This book is copyright

ISBN No. 978-1-74272-002-9

The University of Queensland, St Lucia QLD Tropical Cyclone ‘Roger’ Storm Surge Assessment

by Jared STEWART Masters Student, School of Civil Engineering, The University of Queensland, Brisbane QLD 4072, Australia Mob.: (61 4) 0280 5321, Email: [email protected]

David CALLAGHAN Lecturer, School of Civil Engineering, The University of Queensland, Brisbane QLD 4072, Australia Ph.: (61 7) 3365 3517, Fax: (61 7) 3365 4599, Email: [email protected] and Peter NIELSEN Associate Professor, School of Civil Engineering, The University of Queensland, Brisbane QLD 4072, Australia Ph.: (61 7) 3365 3510, Fax: (61 7) 3365 4599, Email: [email protected]

RESEARCH REPORT No. CE162

ISBN 978-1-74272-002-9

School of Civil Engineering, The University of Queensland July 2010

ABSTRACT The consideration and accurate prediction of storm surge to ensure the protection of coastal communities and infrastructure is very important for coastal areas. While the basic processes which are responsible for storm surge are widely accepted and understood, there remain recorded storm surge events which have not been able to be fully explained using accepted methods and theories. A detailed investigation into one such storm surge event which resulted from Tropical Cyclone ‘Roger’ in 1993 has been completed with the use of a hydrodynamic model and hindcast meteorological data. A MIKE 21 FM model was developed and simulated using hindcast wind and pressure data. The effect of wave radiation stresses and tides on the modelled surge was also investigated. The modelled storm surge was generally found to under-predict the tidal anomalies recorded during the event. The inclusion of tides in the model was not found to affect the modelled surge, while the inclusion of wave radiation stresses was found to improve the fit between the modelled and recorded surge. However, there remained areas of the recorded tidal anomalies which were not well replicated by the model or easily rationalised. The two key recommendations of the study are for the simulation of additional historical storm surge events and the permanent installation of current recording instrumentation (i.e. bottom mounted ADCP) on offshore tide stations.

Keywords: Storm Surge Modelling, Tropical Cyclone Roger, Storm Surge, Wave Setup, Gold Coast Seaway.

TABLE OF CONTENTS Page

Abstract ii Keywords ii Table of contents iii List of symbols v

1. Introduction 1

2. Background 3 2.1. Tropical Cyclones 2.2. Storm Surge 2.3. Storm surge prediction in south east Queensland 2.4. Tropical Cyclone ‘Roger’ – March 1993

3. Storm surge modelling 33 3.1. Models 3.2. Model forcings 3.3. Bathymetry 3.4. Grid/mesh size 3.5. Parameters

4. Model Setup 42 4.1. Model code 4.2. Domain 4.3. Bathymetry 4.4. Meteorological input 4.5. Parameters 4.6. Tidal constituents 4.7. Wave simulation 4.8. Additional notes

5. Results 63 5.1. Initial simulation results 5.2. Sensitivity analysis results

6. Discussion 81 6.1. Pumicestone Passage to Mooloolaba surge

iii

6.2. Wind stress and wave radiation simulation results 6.3. Wave setup vs. wave radiation stresses 6.4. Other processes not considered 6.5. Coastline representation 6.6. Comparison with 1d analytical solution 6.7. Anomalies in anomalies 6.8. Summary of results

7. Conclusions 94

8. Acknowledgements 96 8.1. Assistance 8.2. Data

APPENDICES Appendix A - Tidal Anomaly Plots 97 Appendix B - NCEP DOE comparison with observed parameters 109 Appendix C - Rainfall data for March 1993 156

REFERENCES 164 Internet references 167 Open Access Repositories 168 Bibliographic reference of the Report CH77/10 169

LIST OF SYMBOLS The following symbols are used in this report:

CD free surface drag coefficient;

CB bottom drag coefficient; F force (N); f Coriolis parameter; g gravity constant (m/s2); h water depth (m); h0 water depth at the coastline (m);

Horms deepwater RMS wave height (m);

Hsig significant wave height (m); L continental shelf width (m); s numerical model grid spacing;

Sxx wave radiation stress in the shore normal direction (Pa); t time (s); U absolute wind velocity (m/s);

U10 absolute 10m wind velocity (m/s);

Ux x component of wind velocity (m/s);

Uy y component of wind velocity (m/s); u current velocity (m/s); x distance normal to the shoreline (m); z vertical distance (m) positive upwards, with z = 0 at the bed;

α continental shelf slope (radians); η water elevation (m);

ηc water elevation at the SWL shoreline (m);

ηs water elevation at the MWL shoreline (m); Ω angular velocity of the earth (7.29 x 10-5 radians/second); φ latitude (degrees); 3 ρa air density (kg/m ); ρ water density (kg/m3);

τb bed shear stress (Pa);

τw wind shear stress (Pa);

Abbreviations 2D two-dimensional 3D three-dimensional ADCP Acoustic Doppler Current Profiler AEST Australian Eastern Standard Time

AHD Australian Height Datum AWN Australian Weather News BOM Bureau of Meteorology BPA Beach Protection Authority DERM Department of Environment and Resource Management (previously EPA) ECL East Coast Low (low pressure weather system) GPS Global Positioning System HAT Highest Astronomical Tide LAT Lowest Astronomical Tide MHL Manly Hydraulics Laboratory MSL Mean Sea Level MSLP Mean Sea Level Pressure MSQ Maritime Safety Queensland MWL Mean Water Level NSW New South Wales QLD Queensland RMS Root Mean Square RMSE Root Mean Square Error SEQ South East Queensland SWL Still Water Level TC Tropical Cyclone UTC Universal Time Convention

vi 1. INTRODUCTION In today’s world there are numerous natural hazards which have the potential to cause large scale destruction, loss of lives and interruption to vital services and trade. These include floods, bushfires, earthquakes, tropical cyclones, tsunamis and landslides. Of these natural disasters tropical cyclones represent a significant hazard to coastal communities potentially affected by tropical cyclones. In Australia between the period of 1967 to 1999, approximately 154 deaths and 8.5 billion dollars in damages were attributable to tropical cyclones, accounting for approximately 24% of the estimated costs caused by natural disasters in Australia (BTE 2001). Globally, there have also been numerous natural disasters attributable to tropical cyclones, notably the devastation caused by a cyclone in the Bay of Bengal in November 1970 which resulted in approximately 300,000 deaths (Frank & Husain 1971) and Hurricane Katrina in 2005 which resulted in approximately 2000 deaths and estimated total damages of US 250billion (Glantz 2008). The elevation of the mean sea level during tropical cyclones is perhaps the most dangerous hazard associated with tropical cyclones (BOM 2009a, GA 2009). Referred to as storm surge, this process is caused by strong onshore winds and low atmospheric pressure which combine to elevate the sea level along coastlines, for periods of hours to days (Simpson & Riehl 1981). The consideration and accurate prediction of storm surge to ensure the protection of coastal communities and infrastructure is therefore very important for coastal areas (U.S. ACERC 1977). While the basic processes which are responsible for storm surge are widely accepted and understood, there remain recorded storm surge events which have not been able to be fully explained using accepted methods and theories. This report describes a project undertaken to assess such a storm surge which resulted from Tropical Cyclone ‘Roger’ in 1993. The aim of the project was to better understand storm surge components and their relative magnitude during Tropical Cyclone ‘Roger’. This event was selected for this project as it generated a notable storm surge in areas of south east Queensland despite being a relatively weak cyclone which did not cross the coastline. In order to assess the storm surge resulting from Tropical Cyclone ‘Roger’, a hydrodynamic model has been established to model the event and the resulting storm surge. The development of the model, including the data used in the model, adopted parameters, model sensitivity analysis, and the model results are discussed. The results from the model have also been compared with a simplified 1-d analytical solution. A review of storm surge assessments for SEQ, including a review of studies which have determined appropriate guidelines for storm surge levels, has also been undertaken to compare the storm surge recorded during Tropical Cyclone ‘Roger’ with such guidelines. Additionally, a review of previous

1 modelling of storm surges has been undertaken to assess the methods and values of key parameters adopted in previous studies. This report is structured as follows: Section 2 covers the background of the project including definition of the terminology used throughout the report, theory behind the basic processes of storm surge generation, a summary of the history of storm surge prediction in south-east Queensland, and a discussion of Tropical Cyclone ‘Roger’ including presentation of data recorded during the event. Section 3 discusses the role of storm surge modelling in storm surge prediction, focusing on the models, processes, and parameters used in previous studies. Section 4 describes the model developed and employed in the current study, including the model inputs and adopted parameters. Section 4 also outlines the simulations which were run using the developed model. Section 5 presents the model results and Section 6 provides a detailed discussion of the model results and further analysis of recorded data in an effort to rationalise differences between the modelled and recorded data. The recorded data discussed in Section 6 includes data recorded during Tropical Cyclone ‘Roger’ together with data collected at the Gold Coast Seaway during a storm surge event in 2009. The conclusion in Section 7 provides a concise summary of the outcomes of the current project and recommendations for further work.

2 2. BACKGROUND 2.1. TROPICAL CYCLONES Tropical cyclones have been defined as a violent type of moving, non-frontal revolving storm, with low central barometric pressure, which forms over warm tropical waters. The origin of the term ‘cyclone’ is based upon the Greek work ‘kyklos’, meaning circle or coil, which describes the typical cylindrical shape of tropical cyclones and the inward spiralling winds towards the centre, or ‘eye’ of a tropical cyclone. Tropical cyclones are known in other parts of the world as hurricanes and typhoons, depending upon the oceanic basin in which they form. The flow of air towards the ‘eye’ of a tropical cyclone spirals inwards in a clockwise direction in the southern hemisphere and in an anti-clockwise direction in the northern hemisphere (Terry 2007). The formation of tropical cyclones is attributable to unsteady atmospheric conditions in the low to mid latitudes (Terry 2007, Lighthill 1998). The basic prerequisites for tropical cyclone formation consist of the presence of an existing synoptic scale low pressure system or tropical disturbance (i.e. trough), a warm and moist atmosphere, a warm ocean with surface temperatures in the range 26- 27°C, all occurring in a location at least 5° of latitude from the equator so that the Coriolis force is sufficient to force the inward flowing air to be deflected into a spiralled motion (Sturman & Tapper 1996, Terry 2007). For these reasons, tropical cyclones form during the hotter months of the year in the low to mid latitudes (5-25°) in locations where elevated ocean temperatures are sustained. The typical structure of a tropical cyclone, as shown in Figure 2-1 and Figure 2-2, is formed due to the convective winds drawing warm saturated air into the intense low pressure at the centre of the cyclone. As the air is drawn into the ‘eyewall’, the air ascends rapidly up into the upper Troposphere, releasing moisture and rapidly cooling as it ascends, resulting in the formation of large Cumulonimbus clouds and heavy rainfall in the area surrounding the ‘eye’. It is also in the area of the ‘eyewall’ that the strongest winds and heaviest precipitation are found. Once air reaches the top of the Troposphere (Tropopause), it is shed outwards from the eye, which is important in allowing the continual inflow of warm moist air at the ocean’s surface. The cyclone’s ‘eye’ is typically 20-40km in radius and is typified by a calm zone with little to no wind, and a very slight downward draft of air (Terry 2007, Sturman & Tapper 1996). In addition to the intense cloud around the cyclone ‘eyewall’, ‘feeder bands’ or ‘rainbands’ consisting of organised bands of dense cloud are also present in tropical cyclones. These organised bands act as an integral part of the cyclone as they feed the heat and moisture required to sustain the cyclone.

Figure 2-1 - Schematic of a tropical cyclone (Source Terry 2007)

Figure 2-2 - Schematic of a hurricane (anti-clockwise convection) (Source USGS 2009)

The Australian Bureau of Meteorology define a tropical cyclone as a non-frontal low pressure system of synoptic scale developing over warm waters having organised convection and a maximum mean wind speed of 34 knots or greater extending more than half-way around near the centre and persisting for at least six hours (BOM 2009a). Tropical cyclones in Australia are classified on a scale of one to five according to the maximum mean wind speed as shown in Table 2-1 (BOM 2009b).

Table 2-1 Tropical cyclone categories in Australia (adapted from BOM 2009b)

Category Maximum mean wind Indicative central pressure speeds (km/h) (hPa) 1 63 - 88 > 985 2 89 - 117 985 - 970 3 118 - 159 970 - 955 4 160 - 199 955 - 930 5 > 200 < 930

A ‘mature’ or ‘classical’ tropical cyclone typically exhibits a symmetric form with a circular ‘eyewall’ at the centre of the cyclone, around which are the most destructive winds (Terry 2007). The extent of the gale force winds associated with a classical cyclone is typically less than 100km from the eye of the cyclone (BOM 2009c). Alternatively, tropical cyclones can exhibit an asymmetrical form. Such tropical cyclones or ‘hybrid cyclones’ are typically large, less intense cyclones that result in destructive winds at much greater distances from the ‘eye’ of the cyclone (BOM 2009c) as illustrated in Figure 2-3.

Figure 2-3 - Diagrams of classical and hybrid cyclones (Source BOM 2009c)

Tropical cyclone ‘David’ in was an example of a ‘hybrid’ cyclone which crossed the Queensland coastline on 19 January 1976. Tropical cyclone ‘David’ had a central pressure of approximately 970hPa and caused extensive damage in the Maryborough region and resulted in flooding as far south as Moreton Bay (BOM 2009c, Callaghan & Helman 2008). The interaction of tropical cyclones with other synoptic scale weather patterns can also occur, resulting in strong winds at significant distances from the centre of the cyclone. Although more likely to occur in the case of a hybrid cyclone, the strengthening of winds between a high pressure and low pressure system can occur, with the compression of barometric pressure isobars (i.e. the reduction in the pressure gradient between the two weather systems) resulting in strong winds. This type of interaction was evident during Tropical Cyclone ‘Roger’ and is discussed further in Section 2.4. The damage which is caused during cyclones is largely the result of three mechanisms; the direct force imposed on structures by high speed winds, flooding due to intense rainfall, and inundation of low lying areas due to storm surge (BOM 2009a, Harper 2001, Lighthill 1998). Often, the most destructive of these mechanisms is the storm surge (Simpson and Riehl 1981, BOM 2009a, GA

5 2009).

2.2. STORM SURGE The term ‘storm surge’ is one that is commonly confused due to various definitions on its meaning and the various components which combine to produce storm surges. Three different definitions are offered by Dean and Dalrymple (2002), Kamphuis (2000) and U.S. Army Coastal Engineering Research Centre (U.S. ACERC) (1977). The definition provided by U.S. ACERC (1977), Harper (1998) and Simpson and Riehl (1981) has been adopted for the purposes of this report which define storm surge as being a rise in normal water level caused by the combined effects of surface wind stress and atmospheric pressure fluctuations. Figure 2-4 illustrates these two components which combine to produce a storm surge, raising the Mean Water Level (MWL) to a level greater than the non-storm surge affected MWL.

Figure 2-4 - Illustration of storm surge components

Near to the coastline, wave breaking can result in further elevation of the MWL above the storm surge level. This additional elevation of the MWL at the coastline is attributable to wave setup. It is noted that wave setup is not included in the adopted definition of storm surge. The components of storm surge together with wave setup are individually described below.

2.2.1. Pressure surge The surge which results from the reduction in atmospheric pressure or pressure surge, can be visualised as the water being pushed up into a low-pressure system by the surrounding high pressure air as pictured in Figure 2-5.

Figure 2-5 - Schematic of pressure surge

The process depicted in Figure 2-5 is historically termed ‘the inverse barometer effect’ and the surge resulting from this process can be calculated for the case of stationary pressure systems (steady state) using hydrostatic principles. For a steady state case, the difference of -1hPa results in approximately 1cm water level rise (Harper 1998). In the case of non-stationary pressure systems, the effects of water depth, speed of forward motion of the low pressure system, coastal bathymetry and coastal forms can result in dynamic amplification of the pressure surge (Nielsen et. al. 2008, Harper 1998). Sobey et. al. (1977 in Harper 1998) report that such dynamic amplification can result in the increase of the offshore (deep water) surge by a factor of approximately two. In practical terms, these effects lead to Islands with narrow continental shelves and in deep water normally experiencing only the static effects of the pressure-induced surge, and areas with wide continental shelves being more susceptible to dynamic amplification of the pressure surge (Harper 1998). This is principally due to the necessary forward motion of the low pressure system to achieve dynamic amplification which is proportional to the square root of the water depth (Nielsen et. al. 2008) (i.e. faster forward motions are required in deeper waters).

2.2.2. Wind surge and Coriolis effect Strong winds impart their energy into the ocean and water bodies generating surface currents and large waves. The surface currents generated by strong onshore winds are impeded in shallow water areas by bottom friction and by the coastline, causing the water level to rise or pile up against the coast. The force exerted on the water surface due to wind, known as the wind stress, cannot be determined theoretically and has been determined empirically through experiments (Dean & Dalrymple 2002). The empirical formula for wind stress is (Dean & Dalrymple 2002):

2 = ρτ Daw UC (2.1) where ρa is the density of the air at the water’s surface, CD is a dimensionless friction coefficient (with values of the order 0.0025) and U is the wind speed (in m/s) measured at an elevation of 10m. The application of surface wind stress to a control volume as shown in Figure 2-6 yields: dη −ττ )( = bw (2.2) ρghdx where is the mean elevation of the water surface, τb is the bottom stress and h is the water depth. 7 From these two equations (2.1 and 2.2) it can be seen that for the same wind speed (and therefore the same wind stress) the resulting increase in water level is greater in shallower water depths.

Figure 2-6 - Schematic of wind stress force balance

An important consideration for wind induced surge is the Coriolis effect which is the natural forcing that results from the rotation of the earth. The force imparted by the Coriolis effect is described by the equation (Sturman & Tapper 1996):

Ω= VF sin2 φ (2.3) where Ω is the angular velocity of the earth (7.29 x 10-5 radians/second), V is the velocity of the object, and φ is the latitude. The Coriolis parameter is defined as (Sturman & Tapper 1996):

f Ω= sin2 φ (2.4) In the southern hemisphere the Coriolis effect acts to force a moving object to the left of its direction of travel. Therefore, along the east coast of Australia the Coriolis effect forces north flowing longshore currents towards the coastline. For this reason strong southerly or south-easterly winds may also generate a wind driven surge along a north-south oriented stretch of coastline as well as a directly onshore (easterly) wind. This is demonstrated by the use of a simple analytical model proposed by Tilburg and Garvine (2004) which incorporates wind stress and the Coriolis effect. The simple model calculates the wind setup at a vertical beach-face based upon the force balance of the wind stress, bottom friction (along the coastline only) and gravity, taking into account the cross shore and long shore components of wind (including Coriolis forcing) for a continental shelf with a constant slope. The equation is presented below (Tilburg & Garvine 2004):

fLU ρ UC ρ UUC ⎛ αL ⎞ Dax yDa ⎜ ⎟ ηc = − ⎜1ln + ⎟ (2.5) g ρ UC xB gαρ ⎝ ho ⎠ where ηc is the elevation of the MWL at the coastline, f is the Coriolis parameter, L is the width of the continental shelf, U is the wind speed (absolute as well as in the longshore (x) and cross-shore

(y) directions), ρa and ρ are the densities of air and water respectively, α is the slope of the

8 continental shelf, CD is the free surface drag coefficient, CB is the bottom drag coefficient and h0 is the height of the vertical beach face (SWL water depth at the assumed vertical wall). The significance of longshore currents generated by longshore winds which combine with the Coriolis effect can be illustrated by plotting separately the longshore and cross-shore components of the wind setup expressed in Tilburg and Garvine’s analytical model as presented in Figure 2-7. This figure illustrates that the potential wind setup resulting from a longshore wind is more significant than a directly cross-shore wind for the case of a 40km wide shelf with a slope of 0.286°. The figure also illustrates that the case of a wind at an angle of approximately 50° to the coastline results in the maximum surge from the combined longshore and cross-shore components of wind surge. It is noted that this simplification ignores the effect of coastal features such as headlands or bays which may amplify or dampen the effect of the wind surge locally, dependent upon the orientation of the feature and the wind direction.

0.40 Longshore 0.35 Cross-shore Combined

0.30

0.25 ) (m) c η 0.20

0.15 Wind Surge ( Surge Wind 0.10

0.05

0.00 0 102030405060708090 Wind direction relative to shoreline (°) Figure 2-7 - Relative contribution of longshore and cross-shore winds to wind setup at coastline based upon Tilburg & Garvine (2004)

The parameters used in the model for the comparison presented in Figure 2-7 are; U = 25m/s, L = 3 3 -3 -3 40km, h0 = 1m, ρa = 1.2kg/m , ρ = 1025kg/m , CB = 2.0x10 , CD = 2.3x10 , latitude = 28°S, α = -3 5.0x10 (radians) (max ηc = 0.34m). It is noted that the shelf width and slope adopted are indicative of the continental shelf in south-east Queensland. The wind surge typically accounts for a larger percentage of the storm surge than the pressure surge. This is demonstrated by a simple example using the model proposed by Tilburg and Garvine (2004) and an estimation of the rise in sea level due to a reduction in pressure. For the case of a tropical cyclone with a central pressure of 980hPa and maximum winds of 30m/s the maximum

9 contribution due to the pressure surge would be approximately 0.3m (assuming 1cm/hPa and a neutral pressure of 1013hPa), while the maximum wind setup would be approximately 0.5m based upon the parameters adopted earlier for Figure 2-7. Again, it is noted that this simple example ignores the effect of any amplification or dampening of the wind or pressure surges due to coastal features and dynamic effects.

2.2.3. Wave setup and wave run-up Wave setup As waves break at the shoreline momentum is transferred from the breaking waves to the water column generating an elevation of the MWL at the shoreline as illustrated in Figure 2-8. It is noted that the Still Water Level (SWL) line in Figure 2-8 corresponds to a flat water surface profile (i.e. MWL in the absence of waves).

Run-up limit

‘Shoreline’

η Dune

SWL MWL

Sand-bar WAVE WAVE SETUP RUN-UP

Figure 2-8 - Wave setup and wave run-up - generalised diagram

As illustrated in Figure 2-8, a lowering of the MWL occurs outside of the wave breaking point as a result of wave shoaling and is known as wave set-down (note however that this has never been measured). Shoreward of the breaking point, wave radiation stresses reduce, resulting in an elevation of the MWL. This is known as wave setup and is described by the equation (Dean & Dalrymple 2002): dη 1 dS −= xx (2.6) ρghdx dx where x is normal to the shoreline and increasing in the shoreward direction (i.e. left to right in

Figure 2-8), h is the mean water depth and Sxx is the wave radiation stress in the direction normal to the shoreline. An approximation for the maximum wave setup at the shoreline (ηs ) is (Hanslow & Nielsen 1992):

η s = 0.4Horms (2.7)

10 where Horms is the deepwater RMS wave height and the shoreline is defined as the position of the MWL on the beach face (i.e. η hx = )0(, ). It is therefore common during wave events that the shoreline setup exceeds 1.5m. While it has been long understood that wave setup occurs along beaches, studies by Nielsen (1989) and Hanslow et al. (1996) have shown that wave setup is negligible in trained river entrances. This has been in part explained by transverse momentum fluxes due to the curvature of the wave crests on the breakwaters (Dunn et al. 1999). Wave run-up Beyond the ‘shoreline’, waves in the form of small bores or swash propagate further up the beach face or further inland. The landward limit of such propagating waves is defined as the run-up limit. The height of the run-up limit above the shoreline wave setup is dependant upon the height of the propagating bores/waves and the slope of the beach face, and is typically of the same order as the shoreline setup (ηs ) on typical beaches. For this reason, the process of wave run-up can result in notable inundation and damage to coastal properties beyond that caused by storm surge and wave setup.

2.2.4. Storm tide vs. storm surge The cumulative effects of storm surge and wave setup added to the elevation of the MWL due to astronomical tides is commonly referred to as the storm tide and is illustrated in Figure 2-9. That is the storm tide level = storm surge + wave setup + expected/predicted tide level. It is noted that wave run-up is also included in Figure 2-9 which is the most landward limit of inundation but is not included in the storm tide level which represents the mean water level.

Figure 2-9 - Diagram showing components of storm surge and storm tide (source: (New Zealand Ministry for the Environment 2004))

The storm tide level is often considered of most significance when considering protection of coastal communities and infrastructure, as it defines the maximum mean water level which is reached during a storm event. It is typically referenced to HAT, as this provides a measure of the height of the surge above the maximum tidal range. The magnitude of the storm surge on the other hand is not referenced to a level or datum and is typically quantified by the deviation of the recorded water level from the predicted tide/water level at tide recording stations.

2.2.5. Measurements of storm surge The measurement of storm surges in Australia is most commonly achieved through the periodic measurement of water levels, often at tide gauge recording stations located in major ports and harbours and, in fewer cases, at offshore storm surge gauges. Due to the natural fluctuation of the MWL due to tidal fluctuations, the height of a given storm surge is typically calculated by the subtraction of the recorded water level from the predicted water level. Figure 2-10 illustrates the calculation of the storm surge recorded during Tropical Cyclone ‘Roger’.

1.0 2.0 Storm Surge Predicted water level 0.5 Recorded water level 1.5 a

0.0 1.0

-0.5 0.5 a Tide Level (m)

Storm surge / Tidal anomaly (m) anomaly / Tidal surge Storm -1.0 0.0

-1.5 -0.5 14/3/93 15/3/93 16/3/93 17/3/93 18/3/93 19/3/93 20/3/93 21/3/93 Date Figure 2-10 - Tidal anomaly recorded during Tropical Cyclone ‘Roger’ at Caloundra Public Jetty - height ‘a’ indicates the maximum difference between the predicted and recorded tides.

While this method of calculating the height of a storm surge is appropriate, consideration must be given to factors which may also influence the recorded water level and therefore the calculated storm surge level. Since many of the tide gauges in south-east Queensland are located in enclosed coastal areas, recorded water levels may be affected by river flows during flood events and localised wind or wave setup. Additionally, the accuracy of the predicted tides may also affect the calculated

12 storm surge. One further mechanism which can affect recorded tides is non-linear tidal effects as a result of notable changes in flow depths during a storm surge event. The propagation of the tidal ‘wave’ in limited water depths can vary notably with water depth. Therefore, if the mean water depth varies notably during a storm surge event, subsequent changes to the frequency and amplitude of the tidal ‘wave’ are likely to occur. The difference between the recorded and predicted water level is commonly referred to as the tidal anomaly rather than storm surge. This is due to the definition of storm surge and the fact that storm surge is not directly measured at tide gauge stations. For this reason ‘tidal anomaly’ is used in the following sections of this report to describe the difference between recorded and predicted water levels while ‘storm surge’ has generally been reserved for the description of modelling results.

2.2.6. Summary Tropical cyclones form due to atmospheric instability in the low to mid latitudes during the hotter months of the year. A ‘classical’ or ‘mature’ cyclone exhibits a symmetric form with the most intense winds located around the cyclone’s ‘eye’, while a ‘hybrid’ cyclone is typically weaker but larger in size with damaging winds often extending further from the cyclone’s ‘eye’. The total tropical cyclone hazard is comprised of three main mechanisms; storm surge, heavy rainfall and direct forces on structures due to strong winds. Of these, storm surge represents the most significant hazard and is comprised of wind and pressure surge components. Along open coastlines, wave setup is also a significant contributor to the elevation of the MWL in addition to storm surge. Therefore, the prediction of storm surge is important for both open coastlines, which are affected by storm surge and wave setup, and inland areas where flooding may be intensified by storm surge.

2.3. STORM SURGE PREDICTION IN SOUTH EAST QUEENSLAND On average Queensland is subjected to 4.7 tropical cyclones each year and there have been 207 known cyclone impacts along the coast since 1858 (BOM, 2009c). In addition to tropical cyclones, Queensland also experiences intense low pressure systems which have also caused large scale destruction in coastal areas due to strong winds and flooding. Given that 85% of Queensland’s population reside in the coastal zone (DERM 2009a), the prediction of storm surge is vitally important for the planning of new infrastructure, assessing risk to existing infrastructure and forecasting storm surge levels during tropical cyclone events. The vulnerability of Queensland’s coastal settlements to storm surge inundation has been assessed in a number of previous studies. These studies have aimed to quantify the magnitude and expected

13 frequency of storm surge events for various locations in the state. Three studies; Storm tide threat in Queensland - History prediction and relative risks (Harper, 1998), the Gold Coast Broadwater Study - Storm Tide Return Periods and 1974 Floodwater Modelling (McInnes et. al. 2000) and the Queensland Climate Change and Community Vulnerability to Tropical Cyclones – Ocean Hazards Assessment (Harper, 2004) are discussed in the following sections. A summary of the three studies is also presented in Section 2.3.4.

2.3.1. Storm tide threat in Queensland – History prediction and relative risks (Harper, 1998) This study was undertaken by the Coastal Management Branch, Department of Environment and Heritage (now the Department of Environment, Water, Heritage and the Arts) to assess and summarise the level of understanding of the risks of storm surge in Queensland and provide recommendations for further studies. The report provides an introduction into the mechanisms responsible for storm surge, a concise summary of historical storm surge events in Queensland since 1858, discussion of the variability of predicted storm surge levels along the Queensland coast, and a discussion of the numerous storm surge studies undertaken throughout Queensland including the basis of the studies. The report also provides predicted storm tide levels for return periods up to 10,000 years for 14 major population centres in Queensland including Brisbane and the Gold Coast. Many of the predicted storm tide levels presented in the report are based upon the results of numerical storm surge modelling using the generalised 2-dimensional model SURGE combined with statistical analysis of tidal levels carried out between 1975 and 1983. Due to computer limitations, the earlier studies undertaken using the SURGE model were based upon only 9 simulations for each location, comprising three fixed cyclone tracks and three fixed cyclone intensities. For all simulations the radius to maximum winds was fixed at 30km, the forward velocity of the cyclone was a constant velocity of 28km/h, the grid resolution was approximately five nautical miles, and water level fluctuations due to astronomical tides were not incorporated in the model. Additionally, validation of the SURGE model was limited to comparison against a single storm surge event resulting from tropical cyclone ‘Althea’ which struck Townsville in 1971. The simplifications which were made to reduce the number of model simulations were largely based upon a number of findings from previous studies. These included (Harper, 1998): • Storm surge magnitude can be regarded as directly proportional to the cyclone intensity for a given coastal site • the peak surge occurs approximately at the position of the radius to maximum winds and approximately at the time of landfall • In general, the highest surge will be generated for the case of a coast-crossing cyclone

14 • While some interaction between the astronomical tide and the generation of the surge does occur, usually its effect is quite small From the information provided in the report, the previously studies were concerned only with severe tropical cyclones, with the exception of studies undertaken for the Gold Coast. Studies for the Gold Coast also considered coastal low pressure systems together with tropical cyclones. On the basis of figure 16 presented in the report the predicted storm tide levels for the Gold Coast were estimated to be approximately 1.15mAHD and 1.7mAHD for the 50 and 10,000 year return period events respectively. Figure 17 also presented these levels referenced to HAT which were approximately 0.15mHAT and 0.7mHAT. The recommendations provided in the report included provisions for further studies to increase the spatial coverage of the predictions, account for changes in development patterns, investigate non- cyclonic storm tide generation and the development of more advanced numerical models and wind field parameterisation techniques.

2.3.2. Gold Coast Broadwater Study - Storm Tide Return Periods and 1974 Floodwater Modelling (McInnes et. al. 2000) The Gold Coast Broadwater Study - Storm Tide Return Periods and 1974 Floodwater Modelling (McInnes et. al. 2000) was undertaken specifically for the Gold Coast Broadwater and Seaway. The study derived predicted storm tide levels at the Gold Coast Seaway and the two branches of the Coomera River where they meet the Broadwater. The report describes the basis for the predicted storm tide levels which are compared against previous predictions, together with results of two specific case studies undertaken. The study utilised a deterministic 2-dimensional model, GCOM2D, which incorporated the effects of tides, wave setup, wind setup and barometric setup. Model simulations were conducted for both tropical cyclones and east coast low pressure systems. The model domain consisted of a course grid with 1000m grid spacing extending from approximately 25 °S to 28.6°S and from 153°E to 155°E and a nested grid with 100m grid spacing covering approximately 27.5°S to 28°S and extending approximately 10km offshore from the Gold Coast Seaway. It is noted that the source of the bathymetry data used in the GCOM2D model is not cited. The Holland model was used for the generation of wind fields for simulated tropical cyclones and NCEP reanalysed data between 1977 and 1996 was used for the simulation of east coast low pressure systems. Please note that a brief description of both the Holland model and reanalysed datasets is provided in Section 3.2.1. Wave setup was assessed by the inclusion of the horizontal gradients of the wave radiation stress terms in the GCOM2D model. A WAM wave model with a grid spacing of 5km was used to calculate wave heights, directions and periods which were then interpolated to the nested GCOM2D

15 model grid. The predicted storm tide levels for the Seaway presented in table 1 of the report are 1.9mMSL and 2.1mMSL for the 50 and 100 year return periods. These are significantly higher predicted levels than those presented in Harper (1998), being approximately equal to 0.9mHAT and 1.1mHAT respectively (compared with approximately 0.15mHAT and 0.2mHAT). Comparisons of these levels to those in a previous study by Blain et. al. (1985) revealed that the results of the McInnes et. al. (2000) study were much higher than the previous estimates. The 100 year return period storm tide was estimated as 1.3mMSL (approximately 0.3mHAT) in the Blain et. al. (1985) report compared with 1.9mMSL. The McInnes et. al. (2000) report concluded that this was a result of the inclusion of wave setup in the predictions, which were not incorporated in the earlier predictions by Blain et. al. (1985). The contribution of wave setup to the calculated storm tide levels is not explicitly provided in the report, but is estimated to be in the order 0.8 to 1.0m based upon the information provided in figure 11 and paragraph one, page 21 of the report. The two case studies presented in the report are for the storm surge events recorded on 25 April 1989 and during Tropical Cyclone Wanda in 1974. The storm surge recorded in the 1989 event was 0.43m above the predicted tidal levels in the Broadwater, and the modelled surge level was approximately 0.1m less that the recorded level (23% less). This is in spite of the inclusion of wave setup in the model which accounted for approximately 70% of the modelled surge. The model results from the simulation of the 1974 event which was predominantly a flood event (not storm surge), showed a discrepancy of up to approximately 0.20-0.25m in the modelled and observed water levels during the event (modelled less than observed) and the peak water level during the event could not be compared as there is a 48hr gap in the observed data covering the peak of the event.

2.3.3. Queensland Climate Change and Community Vulnerability to Tropical Cyclones – Ocean Hazards Assessment (Harper, 2004) The Queensland Climate Change and Community Vulnerability to Tropical Cyclones project represents the most comprehensive individual storm surge study undertaken for Queensland. This study, commissioned by the Queensland Government and undertaken in association with Systems Engineering Australia (SEA), the Australian Bureau of Meteorology (BOM) and James Cook University (JCU), was comprised of four main stages and undertaken over a period of approximately four years. The four stages of the study were broken down as follows: • Stage 1 – Review of Technical Requirements • Stage 2 – Tropical Cyclone-Induced Water Levels and Waves: Harvey Bay and Sunshine

16 Coast • Stage 3 – Surge Plus Tide Statistics for Selected Open Coast Locations along the Queensland East Coast • Stage 4 – Development of a Cyclone Wind Damage Model for use in Cairns, Townsville and Mackay The Stage 1 report provides a detailed discussion of storm surge history, modelling and statistics. The mechanisms of storm surge are discussed in a great level of detail, as are the proposed modelling techniques and assumptions. As a whole the report represents a comprehensive literature review on the subject of storm surge with a particular emphasis on storm surge modelling for Queensland. The Stage 3 report presents predicted storm surge levels for 50 open coast locations along Queensland’s coastline following extensive storm surge modelling using JCU’s storm surge model MMUSURGE. The large coverage of the predictions required that model domains be broken down into different levels of resolution using nested grids. Three levels of grids were adopted, with one A grid, eight B grids and twenty C grids. The resolution of the C grids was approximately 550m. Bathymetry data for the grids was primarily based upon navigational charts together with bathymetric survey data provided by the Environmental Protection Agency (EPA). Approximately 500 storms were simulated for each of the twenty C grids. The results of these simulations were used as the basis for the predicted storm surge levels presented in table 2 and appendix A of the report. The MMUSURGE simulations did not incorporate the effect of astronomical tides or wave setup. Additionally, the simulations considered only tropical cyclone-induced water levels. That is, non- cyclonic events such as east coast low pressure systems were not modelled or incorporated in the statistical analysis. This is specifically noted in the report and the effect of ignoring such events is quantified. The increase due to non-cyclone events is estimated in the report to be approximately 0.2m for a 10 year return period, reducing to 0m by between the 100 and 200 year return periods. The storm tide levels and storm surge heights from the Stage 3 report are presented in Figure 2-11. The storm tide levels presented in table 2 of the report are 1.22mAHD, 1.37mAHD, 1.42mAHD for the 100, 500 and 1000 year return periods respectively. These equate to approximately 0.10mHAT, 0.25mHAT and 0.30mHAT respectively (based upon HAT = 1.11mAHD as presented in Figure A51 of the report). Additionally the storm surge heights (taken from figure 2.6 in the report) are approximately 0.25m, 0.6m and 0.8m for the 10, 100 and 1000 year return periods respectively. Accounting for the 0.2m allowance for non-cyclonic events, the 10 year storm surge becomes approximately 0.45m and storm tide approximately 1.20mAHD.

Figure 2-11 - Predicted storm surge and storm tide levels for Surfers Paradise (Source: Stage 3 report Hardy, Mason & Astorquia 2004)

2.3.4 Summary The three reviewed studies each adopted varying techniques for the estimation of storm surge levels and return periods and consequently have derived varying estimates for storm surge levels and return periods. Some of the key parameters and methodologies adopted for the individual analyses are summarised on Table 2-2 and the predicted storm tide levels from each of the three studies are presented in Figure 2-12. It is noted that adjustments have been made to the storm tide levels reported in the Gold Coast Broadwater Study and Ocean Hazards Assessment studies to remove the effect of wave setup and add the effect of ECLs respectively.

Table 2-2 - Summary of modelling methodologies adopted in previous studies

Study Model Min. grid Wind field ECL’s Wave setup Tides resolution included? included? modelled? Harper 1998 SURGE ~9300m Parametric Y N N GCBS 2000 GCOM2D 100m Parametric & Y Y Y Hindcast (NCEP) OHA 2004 MMUSURGE 550m Parametric N N N

18 0.6 Storm tide threat in Queensland Gold Coast Broadwater Study* 0.5 Ocean Hazards Assessment

0.4

0.3

Storm tide level (mHAT) 0.2

0.1

0 10 100 1000 10000 Storm tide return interval (years) Figure 2-12 - Predicted storm tide levels for Surfers Paradise and Gold Coast Seaway referenced to meters above HAT) – *note 0.8m wave setup has been removed from the Gold Coast Broadwater Study levels and 0.2m has been added to the ocean hazards assessment levels at 10 year return period to account for non-cyclonic events.

2.4. TROPICAL CYCLONE ‘ROGER’ – MARCH 1993 Tropical Cyclone ‘Roger’ developed out of a tropical depression in the north Coral Sea near the Solomon Islands during the morning of 13 March 1993. The cyclone slowly intensified as it travelled in a south to south-westerly track. On 14 March ‘Roger’ was upgraded to a Category 2 tropical cyclone. Tropical Cyclone ‘Roger’ (TC ‘Roger’ hereafter) continued on its south to south-westerly track until it reached a stationary position on the 17 March. At this time the cyclone was positioned approximately 400km northeast of Brisbane and had an estimated central pressure of 987hPa. Over the ensuing days the cyclone moved slowly in a north-easterly direction away from the Queensland coast and reduced in intensity. The cyclone passed to the south of New Caledonia before reducing into a low pressure system. The track of cyclone ‘Roger’ is shown in Figure 2-13.

19 0 Papua New Guinea N

-5

Solomon Islands

11/03 -10 12/03

13/03 -15 Vanuatu 14/03 Coral Sea

15/03 Latitude -20 20/03 19/03 21/03 16/03 New Caledonia 18/03 17/03 22/03 -25 Australia South Pacific Ocean

-30

-35 140 145 150 155 160 165 170 175 180 Longitude Figure 2-13 - Tropical Cyclone ‘Roger’ track and approximate centre location at approx 10:00hrs AEST (track from JTWC best track data, JTWC 2009)

Despite not crossing the coast, TC ‘Roger’ combined with high pressure systems in the Tasman Sea over an extended period of days resulting in strong onshore winds extending over a large region of southern Queensland and northern New South Wales. These strong easterly and south-easterly winds produced large seas and swell and resulted in the closure of major ports in south-east Queensland including Brisbane, Mooloolaba, and Gladstone (BPA 1996). The maximum significant wave heights recorded at the Brisbane Waverider Buoy during cyclone ‘Roger’ of 7.36m remain the largest waves recorded at this site since its installation in 1976 (DERM 2009b). Storm surges recorded in south east Queensland were as high as 0.74m, recorded in the Gold Coast Seaway at 6am on the 17 March 1993 (BPA 1996). The combination of the high pressure systems and TC ‘Roger’ is illustrated in a mean sea level pressure chart taken from the United States National Weather Service’s (NWS) NCEP DOE reanalysed meteorological dataset in Figure 2-14.

20 extending asfarapproximately 5° (500-600km) from thecentreofcyclone. from the‘eye’.Thisis illustrated inFigure2-15 whichshowswindsspeedsgreater than20m/s cyclone asithadamoderate in (approximately 650-800km). Furthermore, thecycl outermost closedisobar(aspresentedin Tropical cyclone‘Roger’wasalarge cycloneon the 2.4.1. Wind andpressure during thepassageofTC‘Roger’arefurt Wind andatmosphericpressure,storm surge,wave ‘Roger’ whichresultedinstrongwi Figure 2-14 illustratestheclosely spaced isobars Figure 2-14-Meansealevelpressurechart 2 21

tensity butgaleforcewindswhic nds insouth-eastQueensland. her discussedinthefollowingsections. 17 March199310:00hrsAEST(Source:NWS, 2009) Terry 2007),whichwasapproximately 6-7° and rainfall andstream-flow observationsmade in theregionto thesouthoftropical cyclone scribed as a ‘hybrid’ one wouldbebestdescribedasa‘hybrid’ basis oftheradiusstormcentreto h extended asignificant distance

MSLP (hPa)

-10

-15

15 -20 eed (m/s)

10 Latitude -25 10m Wind sp

-30 5

-35 0 145 150 155 160 165 170 Longitude Figure 2-15 - Tropical cyclone ‘Roger’ 10m wind velocities at 17 March 1993 10:00hrs AEST (Data source NWS 2009 - NCEP DOE dataset)

Recorded wind speeds and atmospheric pressures at three BOM weather stations; Cato Island, Sandy Cape (Fraser Island) and Cape Moreton (Moreton Island) are presented in Figure 2-17 and Figure 2-18 respectively. The location of the three weather stations is also shown in Figure 2-16.

22 -16 14/03 N

-18 15/03 Townsville

-20 20/03

Mackay 21/03 19/03 -22 16/03

Latitude Cato Island (200601)

-24 17/03 Sandy Cape (39085) Hervey Bay QLD -26

Cape Moreton (40043)

-28 Gold Coast

NSW -30 146 148 150 152 154 156 158 160 162 164 Longitude Figure 2-16 - Location of weather stations relative to track of tropical cyclone ‘Roger’ (red line)

Cape Moreton

Sandy Cape

20 Cato Island

Wind speed (m/s) Wind 10

0 11/3/93 13/3/93 15/3/93 17/3/93 19/3/93 21/3/93 23/3/93 25/3/93 Date Figure 2-17 - Recorded wind speeds during tropical cyclone ‘Roger’ (data provided by BOM)

23 1025

1020

1015

1010

1005 Cape Moreton

1000 Sandy Cape Cato Island

Atmospheric pressure pressure Atmospheric (hPa) 995

990

985

980 11/3/93 13/3/93 15/3/93 17/3/93 19/3/93 21/3/93 23/3/93 25/3/93 Date Figure 2-18 - Recorded atmospheric pressures during tropical cyclone ‘Roger’ (data provided by BOM)

From these two figures it can be seen that wind speeds generally peaked between 15-20m/s at each of the stations, and the central pressure of the cyclone was approximately 985hPa when it passed over Cato Island in the early morning of 17 March 1993 (AEST).

2.4.2. Storm Surge The storm surge at the Gold Coast Seaway resulted in a peak storm tide level of 0.22mHAT, recorded at 4am 17 March 1993 (BPA 1996). Other tidal anomalies recorded during the event included a 0.57m anomaly recorded at Munna Point (Noosa) at 2pm 17 March 1993, with a peak storm tide level of 0.28mHAT at 5am 17 March 1993 (BPA 1996) together with a 0.7m anomaly recorded at the Caloundra Public Jetty at 2pm 17 March 1993 with a peak storm tide level of - 0.06mHAT. A summary of the recorded storm surge levels during the passage of tropical cyclone ‘Roger’ is presented in Table 2-3. Time series of the recorded tidal anomalies are also presented in Figure 2-20 and Figure 2-21. The location of tide stations and wave rider buoys in south-east Queensland and northern New South Wales is also presented in Figure 2-19. In addition to Figure 2-20 and Figure 2-21, predicted and recorded tidal levels together with tidal anomalies have been plotted for all stations for which data was available in Appendix A. It is noted that the quality of the tidal predictions varies between the different sites. This is due to the quality of the tidal constituents used to derive the tidal predictions, which is most often related 24 to the period of tidal records at each site. The standard ports have long periods of tidal records and subsequently have more accurate tidal predictions, while the opposite often applies to the non- standard ports. The tidal predictions at the standard ports of Mooloolaba, Brisbane Bar, Gold Coast Seaway and each of the NSW sites, generally show good agreement with the recorded tides before and after the passage of TC ‘Roger’. Poor tidal predictions are evident in Figure 2-20 and Figure 2-21 and Appendix A at the Noosa, Caloundra, Pumicestone Passage (The Farm) and Paradise Point sites, which exhibit obvious semidiurnal signals in the tidal anomalies.

Table 2-3 - Tidal anomalies recorded during tropical cyclone ‘Roger’ (data provided by MSQ & MHL)

Location Max tidal Max tidal anomaly Max Level Max level time anomaly (m) time (mHAT) Noosa River - Munna Point† 0.57 17/03/93 14:00 0.28 17/03/93 05:00 Mooloolaba Storm Surge 0.44 17/03/93 09:59 -0.24 18/03/93 05:59 Caloundra Public Jetty† 0.74 17/03/93 13:59 -0.06 17/03/93 04:59 Pumicestone Passage - Military Jetty† 0.59 17/03/93 12:59 17/03/93 05:59 Pumicestone Passage - The Farm† 0.62 17/03/93 09:59 17/03/93 06:59 Brisbane Bar 0.34 17/03/93 07:59 -0.48 18/03/93 06:59 Brisbane River - Gateway Bridge† 0.35 17/03/93 11:00 18/03/93 07:00 Brisbane River - Port Office† 0.34 17/03/93 07:59 -0.58 18/03/93 06:00 Paradise Point† 0.56 17/03/93 12:59 -0.11 17/03/93 05:59 Gold Coast Seaway 0.76 17/03/93 06:00 0.24 17/03/93 04:00 Tweed Regional* 0.64 17/03/93 13:00 Tweed Offshore 0.30 17/03/93 12:00 Brunswick Heads 0.52 17/03/93 16:45 Ballina 0.28 17/03/93 23:45 Yamba 0.19 17/03/93 20:45 Yamba Offshore 0.21 17/03/93 23:00 * Tweed Regional tidal anomaly taken from the NSW Ocean Tide Levels Annual Summary 1992/93 (MHL 1993) † Non-standard port

25 -26

Noosa N Mooloolaba Caloundra Pumicestone Passage -27

Brisbane Bar QLD Brisbane

Paradise Point -28 Gold Coast Seaway Gold Coast Tweed Offshore Latitude

Brunswick Heads

Ballina Byron Bay -29

Yamba Yamba Offshore NSW

-30 151 152 153 154 155 156 Longitude Figure 2-19 - Tide recording stations (blue) and Waverider buoy (red) locations in SEQ & northern NSW

26 1000 Noosa (Munna Pt) 500

1000 Mooloolaba 500

1000 Caloundra 500

1000 Military Jetty 500 Pumicestone Passage

1000 The Farm Pumicestone Passage 500 Tidal Anomaly (mm) 0

1000 Brisbane Bar 500

1000 Paradise Point 500

1000 Gold Coast Seaway 500

-500 11/3/93 13/3/93 15/3/93 17/3/93 19/3/93 21/3/93 23/3/93 25/3/93 Date Figure 2-20 - Tidal anomalies recorded at QLD tide stations (data provided by MSQ)

27 1000 Tweed Offshore 500

1000 Brunswick heads 500

1000 Ballina 500

Tidal Anomaly(mm) 1000 Yamba Offshore 500

1000

500 Yamba

-500 11/3/93 13/3/93 15/3/93 17/3/93 19/3/93 21/3/93 23/3/93 25/3/93 Date Figure 2-21 - Tidal anomalies at NSW tidal stations (data provided by MHL)

From these two figures and Table 2-3 it is evident that the peak of the tidal anomaly at the Gold Coast Seaway gauge occurred earlier than at many of the other stations. The figures also appear to suggest that there were no Kelvin waves propagating north along the coast during the passage of tropical cyclone ‘Roger’, which would be evident by a tidal anomaly in the charts moving from left to right going up the page (sites are arranged in order of north to south).

2.4.3. Waves A summary of the recorded significant wave heights recorded at three Waverider Buoys during TC ‘Roger’ is presented in Table 2-4 and time-series plots of the significant wave heights and wave periods are shown in Figure 2-22 and Figure 2-23 respectively. As noted earlier in Section 2.4, the maximum significant wave heights recorded at the Brisbane Waverider Buoy during tropical cyclone ‘Roger’ (7.36m) remain the largest waves recorded at this site since its installation in 1976 (DERM 2009b). 28 Table 2-4 - TC ‘Roger’ Waverider wave heights (data provided by DERM)

Waverider Hs (m) Date/Time (AEST) Stn. Brisbane 7.36 17/03/1993 10:30 Gold Coast 5.73 17/03/1993 12:30 Byron Bay 5.54 18/03/1993 00:00

Brisbane 7 Gold Coast Byron Bay 6

5 (m)

s 4 H

0 11/3/93 13/3/93 15/3/93 17/3/93 19/3/93 21/3/93 23/3/93 25/3/93 Date Figure 2-22 - Significant wave heights recorded during Tropical Cyclone ‘Roger’ (data provided by DERM and MHL) 10

Brisbane Gold Coast 9 Byron Bay

8 (s)

z 7 T

4 11/3/93 13/3/93 15/3/93 17/3/93 19/3/93 21/3/93 23/3/93 25/3/93 Date Figure 2-23 - Wave periods recorded during Tropical Cyclone ‘Roger’ (data provided by DERM & MHL)

29 2.4.4. Rainfall and stream-flows In addition to strong winds and large waves, TC ‘Roger’ also produced the highest rainfalls at many rainfall stations in south east Queensland and northern New South Wales during the month of March 1993. Rainfall generally fell between the 11th and 20th of March 1993 with the highest rainfalls recorded to 9am on 17 March 1993. The highest rainfall in SEQ to 9am 17 March was 95mm recorded at both the Springbrook Forestry (station 040192) and Maleny (Tamarind St - station 040121) BOM stations. Maps showing 3 day rainfall totals for 15 to 17 March 1993 and daily rainfall for 17 March 1993 are shown in Figure 2-24 and Figure 2-25 respectively.

160 Noosa 140

120 ) mm ( Caloundra 100

80 Rainfall

0 Gold Coast QLD Tweed Heads

Ballina

NSW

Figure 2-24 - 3-day rainfall totals for 15 to 17 March 1993 (dots indicate rainfall stations) (data from AWN 2009)

30 100 Noosa 90 80

70 ) mm Caloundra ( 60

50 Rainfall

0 Gold Coast QLD Tweed Heads

Ballina

NSW

Figure 2-25 - Daily rainfall totals for 17 March 1993 (dots indicate rainfall stations) (data from AWN 2009)

Stream-flow records were obtained from DERM for two stream gauges located upstream of the Gold Coast Seaway. The two stations; Nerang River at Glenhurst (station 146002B) station and Coomera River at Army Camp (station 146010A), had recorded peak discharges of 2.9m3/s and 0.8m3/s respectively between the period 17-21 March 1993. Although these records represent only two locations within the catchment of the Gold Coast Seaway, the records suggest that stream-flow during the passage of tropical cyclone ‘Roger’ was minimal in this area. It is noted that data was also requested for a number of BOM stations including the Evendale gauge (station 040684) which is located in the lower reaches of the Nerang River. However, data was not

31 available at the requested stations during this period. Stream-flow records for the Tweed, Brunswick, and Richmond rivers were also requested from MHL for the period of March 1993. However, stream-flow records were not available for this period for these three rivers.

2.4.5. Discussion Tropical cyclone ‘Roger’ resulted in a peak storm tide level at the Gold Coast Seaway of 0.22mHAT which according to the results of the previous studies reviewed in Section 2.3 represents an event with a return interval of between approximately 100 and 300 years as indicated in Figure 2-26. It is noted that this is despite tropical cyclone ‘Roger’ coinciding with a neap tidal cycle in south east Queensland.

0.6 Storm tide threat in Queensland Gold Coast Broadwater Study* 0.5 Ocean Hazards Assessment

0.4

0.3 0.22mHAT

Storm tide level (mHAT) level tide Storm 0.2

0.1

0 10 100 1000 10000 Storm tide return interval (years) Figure 2-26 - Illustration of peak storm tide recorded in Gold Coast Seaway during tropical cyclone ‘Roger’ in relation to previous studies.

32 3. STORM SURGE MODELLING The assessment of storm surges using numerical models has been undertaken in Queensland since the mid 1970’s when the first detailed storm surge studies were undertaken (Harper 1998). The earliest studies were undertaken following tropical cyclone ‘Althea’ which struck Townsville in 1971 triggering a period of intense research on storm surge risks in Queensland (Harper 1998). This ‘reactive’ theme has also been seen following the hurricane seasons of 2004 and 2005 which saw hurricanes including ‘Ivan’ and ‘Katrina’ strike the Gulf of Mexico with devastating effects, provoking renewed international interest in storm surge modelling. The complex nature of the processes which combine to produce storm surges makes numerical modelling techniques well suited to storm surge prediction. This is due to the large number of computations required to predict the spatial and temporal variations in storm surges and the efficiency of computers in performing these calculations. However, over the decades of the development of storm surge models various studies have demonstrated the importance of model inputs and parameters to ensure accurate predictions by storm surge models. Herbich (1990) and Harper (2001) discuss the importance of storm surge model inputs and parameters and identify a number of crucial factors including: • the paramount importance of accurate wind fields • adequacy of model resolution, particularly in near shore zone • adequacy of bottom stress parameterisation • selection of appropriate boundary conditions along open boundaries • appropriate means of incorporating the effect of wave radiation stresses • appropriate selection of winds stress drag coefficient • accurate bathymetric data These factors and others are further discussed in the following sections.

3.1. MODELS A number of numerical models, both commercial and non-commercial, are currently used to simulate of storm surges. These models most typically solve the Non-linear Shallow Water Equations (NLSWE) using either finite difference or finite volume solutions. Finite difference and finite volume methods are types of numerical approximation methods for solving mathematical equations, which in the case of most storm surge models are the NLSWE (conservation equations of hyperbolic form). Finite difference methods are based upon the discretisation of the spatial domain into discrete cells and numerically assess the derivatives of the governing equations in order to approximate the solution. Finite volume methods as the name

33 suggests, are based upon the discretisation of the spatial domain into ‘finite volumes’ and numerically assess the integral of q (flux) over each of these volumes (LeVeque 2002). Finite difference and finite volume methods can be implemented using either explicit or implicit solution schemes. These schemes relate to the treatment of time in the numerical methods. Explicit or forward difference schemes solve for the next timestep based upon the solution of the current timestep. Implicit or backward difference schemes solve for the next timestep by solving equations which involve both the current timestep and the next timestep. Explicit schemes are more computationally efficient but must respect the Courant-Friedrichs-Lewy (CFL) criterion to avoid numerical dispersion (numerical errors). On the other hand, implicit schemes are more computationally intensive but can use larger timesteps than that required by the CFL criterion without causing large numerical errors. Thus, for all else being equal implicit schemes are faster than explicit schemes. The Courant-Friedrichs-Lewy (CFL) criterion requires that the model timestep is proportional to the grid spacing and water depth as follows (Abbott 1979): Δs t <Δ (3.1) 2gh where t is the model time-step, s the grid spacing and h is the water depth. The use of either scheme is often dependent upon the solution method (i.e. finite difference or finite volume) and the resulting ease of implementation of the two schemes, with explicit schemes being simpler to implement than implicit schemes. Together with the temporal aspect, the descretisation of the spatial domain is an important part of numerical models. As outlined above, finite volume methods are founded upon finite volumes and are easily formulated to allow for unstructured grids. For this reason finite volume methods often use unstructured meshes which allow for greater flexibility and can be refined in areas of interest. Finite difference methods on the other hand commonly use rectangular, rectilinear or curvilinear grids which generally don’t offer the same level of flexibility of the unstructured meshes. While finite volume models have been available for well over a decade, it has only been in recent times that they have been applied to storm surge modelling, with finite difference models representing the majority of models used for storm surge modelling prior to this. Studies including Harper (1977), Vries et. al. (1995), Harper (2004), McInnes et. al.(2000), Yin et. al. (2009), Xie et. al. (2008) and Moon et.al. (2009) have utilised finite difference models, while recent studies by Weaver and Slinn (2009), Salisbury and Hagen (2007) and Lane et.al. (2009) have utilised finite volume models. Storm surge modes are largely confined to 2D models, despite the availability of barotropic and baroclinic 3D models. This is largely due to the increased computational cost of 3D models and the proven adequacy of 2D models for storm surge prediction (Harper 2001).

34 3.2. MODEL FORCINGS The model forcings which are incorporated in storm surge models are those metrological forcings which combine to cause storm tides. These include wind, barometric pressure, astronomical tides and waves. Each of these model forcings are discussed individually in the following sections.

3.2.1. Wind As identified in Section 2.2, the wind driven surge typically accounts for a large percentage of the overall storm surge. Since the wind stress (τw) is proportional to the wind speed squared, accurate definition of the wind speed and direction is often the most critical input in storm surge models (Harper 2001, Herbich 1990, Vickery et.al. 2009). 2D and 3D storm surge models require spatially and temporally varying wind speeds and directions as inputs. There currently exist two accepted methods for the calculation of temporal and spatial varying wind fields. The first is via the use of a parametric wind field model, such as the widely used ‘Holland’ model, and the second is via the use of hindcast meteorological datasets such as those offered by the European Centre for Medium-Range Weather Forecast (ECMWF), the National Centres for Environmental Protection (NCEP), the Japanese Meteorological Agency (JMA) and the National Aeronautics and Space Administration (NASA). Most parametric wind field models are based upon a steady axisymmetric vortex solution based upon the Eulerian equations of motion (Harper 2001). Equations for the atmospheric pressure field are used to derive gradient level winds based upon the force balance between centrifugal, Coriolis and pressure gradient forces. These gradient level winds are then reduced to the surface level through consideration of boundary layer effects. In doing this, consideration is also given to asymmetry of the wind field through empirical means. Harper (2001) shows that parametric models can be applied successfully to for storm surge modelling of classical cyclones. However, he also acknowledges that the parametric models have many limitations owing to the complexities of real tropical cyclone systems and the simplifications made in the parametric models. Where tropical cyclones interact with other synoptic scale systems, which often occurs in the case of ‘hybrid’ cyclones, parametric models can be manually altered to reflect such interactions. However, this type of adjustment neglects the complexities of the boundary layer and is likely to result in a poorer reproduction of the wind field. The main alternative to parametric wind models is offered by ‘hindcast’ or ‘reanalysed’ meteorological datasets. Reanalysed meteorological datasets such as the NCEP DOE dataset are gridded datasets produced using the combination of meteorological models together with assimilated meteorological observations (Kalnay et al. 1996). The meteorological models are simulated and compared with the assimilated data to produce a best-estimate result. The purpose of

35 producing such datasets is to provide meteorological/atmospheric data fields which cover the globe, and in doing so making data available in places where no direct measurements have been collected. Various ‘reanalysis’ datasets exist, each being based on varying meteorological models and data assimilation techniques. Most datasets cover long periods (i.e. 40+ years) and contain numerous atmospheric data fields which vary from wind and pressure fields, to downward solar radiation flux and total cloud cover. The atmospheric fields which are available from the datasets often vary in their dependency upon the meteorological model and the assimilated dataset, with some variables being almost exclusively based upon the model and vice-versa. Reanalysed meteorological datasets are principally available through ECMWF and NCEP, and include temporal and spatially varying wind and pressure fields available from 1957 to 2009. Each of the two organisations has more than one reanalysed dataset, with each dataset being based upon the simulation of a slightly different meteorological model. The temporal and spatial resolution and years of coverage for three datasets is provided in Table 3-1.

Table 3-1 Data resolution and coverage years of reanalysed meteorological datasets

Dataset Temporal MSL Pressures 10m Wind Years covered resolution Velocities NCEP-DOE 6hrs 2.5° 1.875° (Gaussian) 1979-2009 ECMWF-Interim 6hrs 1.5° 1.5° 1989-2009 ECMWF-ERA-40 6hrs 2.5° 2.5° 1957-2002

A key step when using either sources for wind field data (parametric or reanalysed) is the validation of wind fields with recorded measurements from meteorological weather stations and other observations where available (Vickery et.al. 2009). Regardless of the adopted approach, it is acknowledged by Harper (2001) and Vickery et.al. (2009) that our current lack of understanding of tropical cyclone processes, particularly in the boundary layer, and the current parameterisation of wind stress represents a large weakness in current storm surge modelling and subsequently an area where further research is necessary.

3.2.2. Pressure As outlined in Section 2.2, variations in atmospheric pressure result in notable fluctuations in water levels, and as such are a necessary input into storm surge models. Since the contribution to the overall storm surge from pressure related effects are commonly less significant than the contribution due to wind stress, the importance of accurate pressure fields is less than for wind fields. As for wind fields, two methods for the calculation of atmospheric pressure fields at the surface

36 (MSL) are commonly used, parametric models and hindcast meteorological datasets. These use the same sources as previously mentioned for wind fields, being parametric models such as the Holland model, and hindcast meteorological datasets such as those made available by ECMWF and NCEP. While the effect of pressure effects on storm surge levels may be less significant than other effects, the validation of pressure fields with measured data is considered important.

3.2.3. Waves (wave radiation stresses) The action of breaking waves results in momentum being transferred from the breaking waves into the water column. This momentum flux is typically described by a wave radiation stress, the gradients of which are responsible for the elevation of the MWL at the shoreline (wave setup). As outlined in Section 2.2, wave setup occurs along the coastline where the currents generated by the momentum flux are impeded by the coastline. The effect of waves and wave setup on storm tide levels is therefore necessary where currents generated by breaking waves may be of significance (i.e. in a shallow bay) or predictions of storm tide levels at the shoreline along open coastlines are necessary. The effect of waves is most commonly included in storm surge models through the inclusion of wave radiation stresses which are typically calculated by a separate spectral wave model such as SWAN or WAM and then input into the model. As previously highlighted in Section 2.2.3, studies by Nielsen (1989) and Hanslow et al. (1996) have shown that wave setup is negligible in trained river entrances. This has been in part explained by transverse momentum fluxes due to the curvature of the wave crests on the breakwaters (Dunn et al. 1999). This implies that water level records from tidal gauges inside trained river entrances are not affected by wave setup. A study by Xie et.al (2008) assessed the effect wave-current interaction in a storm surge model through the coupling of a wave model and 3D storm surge model. The findings of the study suggest that the effects of wave-induced stresses can result in significant changes in storm tide levels and subsequent inundation extents. This is in-part due to the spatial variability in waves, which results in large spatial variations in their contribution to the storm tide levels. Yin et.al. (2009) assessed coupling of wave and surge model with 2’ (2 min) resolution finite difference model and reported a significant difference between coupled and non-coupled models. However the coarse near-shore grid (2’ ~ 4km) may have resulted in the over estimation of the wind and wave stresses as argued by Moon et al. (2009) which is discussed further in Section 3.5. The effect of wave age on wind stress coefficients has also been investigated in a number of recent storm surge modelling studies including Brown & Wolf (2009). This topic is further discussed below in Section 3.5.

37 3.2.4. Tides The consideration of astronomical tides in storm surge studies is necessary when the prediction of storm tide levels (not just storm surge levels) is required. However, the inclusion of astronomical tides is often simply added to storm surge levels rather than the explicit inclusion of tides in storm surge models. A number of studies including a study by Lee et al. (2008, in Moon et al. 2009) have shown that the interaction between tides and storm surges is negligible, with the inclusion of tides in storm surge models resulting in very little changes to the predicted storm surge. However, a study by Kim et al. (2008) concluded that storm tide predictions are improved through the inclusion of tides in the storm surge model, and the simple addition of storm surge heights to predicted tide levels is conservative in the prediction of peak storm tide levels. Additionally, a study by Brown and Wolf (2009) found that the inclusion of tides for a simulated event resulted in large differences in the predicted storm surge. They demonstrated that the tide can affect the timing and height of a predicted storm surge in areas where a large tidal range exists, concluding that the tidal range is an important consideration in storm surge models. From these studies it is evident that the explicit inclusion of tides in storm surge models may be important for the accurate prediction of storm tide levels in areas where tidal propagation is significantly non-linear.

3.3. BATHYMETRY Bathymetry and coastline geometry are important inputs into storm surge models as both have the potential to affect storm surge heights (Herbich 1990). In the cross-shore direction, the width and slope of the continental shelf are important factors which can be demonstrated by the application of the simple model presented in Section 2.2. In general, shallower depths and flatter bed slopes result in higher surges. A recent study by Weaver and Slinn (2009) investigated the effect of bathymetric fluctuations and their effect upon storm surge estimates using both 1D and 2D models. In the study, fluctuations in bathymetry of ±20% were found to result in differences of less than 5% in storm surge heights. The study concluded that, provided the larger scale bathymetric characteristics are well represented, smaller scale deviations in bathymetry owing to seasonal changes and storm events are likely to effect storm surge results by less than 10% RMSE. The study also demonstrated that accurate bathymetry data is most important near to the shoreline to a depth of between 25-40m depending upon the average bed slope. The inclusion of large coastline features such as bays and headlands is also important for accurate storm surge predictions. This is demonstrated in a study by Salisbury and Hagen (2007) which 38 illustrated variations in the predicted surge in the longshore direction attributable to the presence of large scale coastal features. This study also demonstrated that the inclusion of tidal inlets has little effect on the predicted storm surge along the open coast. Therefore the adequate representation of coastline features such as bays and headlands is important, while tidal inlets may be ignored.

3.4. GRID/MESH SIZE The spacing or resolution of grid cells in storm surge models can affect storm surge predictions. This has been demonstrated in studies by Moon et al. (2009) and Kim et al. (2008) which have demonstrated that storm surge heights are increased in finer resolution grids for finite difference models. The reason for this is primarily due to the finer resolution grids more accurate representation of shallow near shore areas where the slope of the storm surge in the cross-shore direction is relatively large. No studies were identified which have investigated the effect of mesh size on storm surge predictions for finite volume models, however these are expected to be consistent with their finite difference counterparts.

3.5. PARAMETERS Various input parameters are required for storm surge models including eddy viscosity, density, wind stress, bottom stress, and solution technique parameters. Of these, wind stress parameters have been a focus of many storm surge studies, past and present, due to the large contribution which wind stress typically makes to storm surge. A summary of studies which have investigated the parameterisation of wind stress, and the relative importance of wind stress parameters are discussed below.

3.5.1. Wind stress The transfer of momentum from the wind into the sea, via wind stress, is a primary mechanism for storm surge. Assuming that the wind field is known (i.e. calculated wind speeds and directions are accurate), the most important parameter in a storm surge model is the wind stress drag/friction coefficient (Harper 2001). One of the earliest studies on wind stress was by Charnock (1955) which presented an implicit formula derived on dimensional grounds relating roughness to the wind following anemometer measurements taken in a reservoir. The resulting relationship between the drag coefficient and wind velocity was roughly linear above wind speeds of approximately 5m/s, the slope being dependant upon an implicit constant (α). A later study by Smith and Banke (1975) presented results from anemometer measurements collected on a low level sand spit. Measurements of wind speeds were less than 22m/s and the

39 proposed relationship (3.3) had a reasonable fit (R2 = 0.74) with a dataset which included their measurements and previous data.

= ρτ Daw UUC 1010 (3.2)

+ 066.063.0 U10 CD = (3.3) 1000 Subsequent studies by Garratt (1977) and Wu (1982) proposed variations of the formulation proposed Smith and Banke following review of different datasets. The variations between the three formulations are presented in Figure 3-1 below. It is noted that the maximum differences in the calculated CD using the three different formulas is approximately 7%.

3.0

2.5

2.0 ) 3 1.5 Cd (x10 1.0 Smith & Banke (1975) 0.5 Garratt (1977) Wu (1982)

0.0 0 5 10 15 20 25 30 U10 (m/s) Figure 3-1 - Comparison of wind stress drag coefficients from three studies

A lack of measurements at high wind speeds left considerable uncertainty in the application of these formulations at wind speeds greater than 25-30m/s. A study by Powell (2003) presented the results of wind speed measurements taken in cyclonic conditions via the use of GPS sondes dropped from aircraft. The study concluded that the CD derived from the measurements generally agreed with the previously developed formulations for wind speeds up to approximately 33m/s, at which point the

CD levelled off and then further reduced at higher wind speeds. This finding was supported in a later study by Donelan et al. (2004) which concluded that the wind stress drag coefficient reached a maximum of 2.5x10-3 at wind speeds greater than 33m/s on the basis of results from laboratory experiments. These conclusions were further supported by Zedler et al. which presented the results of numerical simulations of an ocean circulation model simulated during a hurricane event. This study concluded that model results agreed best with measured data when a drag coefficient which levelled off and reduced at high wind speeds was used.

40 The effect of the sea-state on the wind stress drag coefficient has been investigated using storm surge models and physical measurements. Studies including Mastenbroek et al. (1992), Brown & Wolf (2009) have proposed wave-dependant wind stress drag coefficients, which have typically used a modifying roughness factor based upon wave age. While it is accepted that sea-state affects the wind stress drag coefficient, the coupling of the wave and surge models in these studies generally resulted in drag coefficients greater than those originally proposed by Smith and Bank, Garratt and Wu. That is the use of the originally proposed formulations such as Smith and Banke (1975) resulted in underestimation of storm surges. A recent study by Moon et al. (2009) argued that the wind stress drag coefficients in many models, including those using coupled surge and wave models, had overestimated drag coefficients owing to a lack of grid resolution. The study presented the storm surge model results based upon different grid resolutions and wind stress drag coefficients including Wu (1982), and demonstrated that the underestimation of surge heights using course resolution grids can be compensated by the use of high drag coefficients. This offers possible explanation for the reason why some models have relied upon high drag coefficients. Another recent study on the effect of waves on wind-sea interaction by Garcia-nava et al. (2008) presents results from wind and wave measurements collected in the Gulf of Tehuantepec. The results suggest that the wind stress drag coefficient proposed by Smith in 1980 (similar to Smith & Banke (1975)) is in the range 20-50% lower than that indicated by the measured data for wind velocities in the range 10-20m/s. In summary there is still much debate on appropriate values for wind stress drag coefficients, owing to the complex nature of the interaction between the wind and the sea. The interaction between wind and waves is crucial in the estimation of wind stress, however the size and speed of waves are functions of time, as well as wind speed, direction, fetch and water depth. This results in a very complex interaction between the two which is still poorly understood (Harper 2001). While this may be the case, Herbich (1990) argues that all models should employ a consistent drag coefficient, and that it should not be tuned as a calibration factor.

41 4. MODEL SETUP In order to investigate the storm surge generated by tropical cyclone ‘Roger’, a hydrodynamic model was developed to model the event and assess the resulting storm surge. The development of the model, including the data used in the model and the adopted parameters are discussed in this section.

4.1. MODEL CODE The hydrodynamic model developed for this study was developed using DHI’s MIKE 21 FM. This model is based upon a flexible mesh (FM) approach and was developed for oceanographic, coastal and estuarine settings. The model solves the depth averaged mass and momentum conservation equations using a cell-centered finite volume method using an explicit solution scheme, and includes options for the inclusion of temperature, salinity, and density effects (DHI 2009a). One of the key benefits offered by the flexible mesh approach is the ability to easily define areas with higher resolution through the use of smaller mesh elements. Mesh elements are either triangular or quadrangular in shape. Importantly, the model also allows for the input of time variant wind and pressure distributions over the model domain, a feature which has been used for this project to simulate tropical cyclone ‘Roger’.

4.2. DOMAIN

4.2.1. Extent The initial model domain was selected to ensure that any dynamic effects resulting from the moving wind and pressure surges during tropical cyclone ‘Roger’ would be captured by the model. As such, the domain was selected to easily cover the southern section of the cyclone’s track and extend beyond the tide stations in Queensland and northern New South Wales which recorded surges. The domain was then further extended north and south of the area of interest to minimize any effects from the model boundaries. The resulting model domain extended from 20°S to 35°S and 148°E to 170°E as illustrated in Figure 4-1 (extent A).

42 -5 Papua New Guinea Solomon Islands

-10 N

a r t

' r

g o Vanuatu -15 R ' Coral Sea

New Caledonia -20 Latitude -25 Australia

South Pacific Ocean -30 extent C

extent B

-35 extent A

-40 140 145 150 155 160 165 170 175 180 Longitude Figure 4-1 - Model extents

This extensive domain was later reduced in a number of simulations to assess the effect of varying the model extent. In total three different model extents were simulated in the model as outlined in Table 4-1. The three model extents are also illustrated in Figure 4-1 together with the initial model extent.

Table 4-1 - Simulated model extents

Model extent Latitude extents Latitude Longitude extents Longitude coverage coverage A 20°S, 35°S 15° 148°E, 170°E 22° B 23°S, 32°S 9° 150°E, 162°E 12° C 25°S, 30°S 5° 152.5°E, 158°E 5.5°

4.2.2. Grid size The size of elements in the initial flexible mesh was sized such that the main coastline features were

43 adequately represented whilst also maintaining sufficient resolution in the nearshore region. The initial flexible mesh is shown in Figure 4-2. The minimum element edge length in this grid is approximately 10km, corresponding to the most landward elements. It is noted that quadrangular elements were used along the continental shelf south of approximately 24 °S and that coastal inlets and rivers were omitted from the model domain. Latitude

Longitude Figure 4-2 - Model mesh A-1 (extent A, resolution 1)

The effect of varying the size of the flexible mesh elements was also investigated in later simulations. In total three different meshes were generated, each with varying minimum element edge lengths. The three different mesh resolutions investigated are summarized in Table 4-2.

Table 4-2 - Mesh Resolutions

Mesh resolution Min element edge Max element edge length (m) length (m) 1 10,000 110,000 2 1,000 100,000 3 100 20,000

It is noted that for the finer meshes, coastline features such as headlands are more accurately

44 represented, and the Gold Coast Broadwater was included in meshes 2 and 3 but not 1.

4.3. BATHYMETRY Two sources of bathymetry data were used for the generation of the model domain. Digitised admiralty chat data was used over an area extending from the Gold Coast to just north of Mooloolaba and up to approximately 50km offshore. Outside of this area, the General Bathymetry Chart of the Oceans (GEBCO) one minute gridded dataset was used. The area covered by the digitised chart data is shown in Figure 4-3.

-25.0

Hervey Bay N

-26.0 QLD

Mooloolaba -27.0 Extent of digitised chart data Latitude -28.0 Gold Coast

-29.0

NSW

-30.0 150.0 151.0 152.0 153.0 154.0 155.0 156.0 157.0 158.0 Longitude Figure 4-3 - Extent of digitised admiralty chart data

The GEBCO one minute grid (version 2.0) dataset is based upon digitised deep water bathymetric charts with standard contour intervals of 500m and down to 100 or 200m in some areas and is therefore unlikely to be sufficiently accurate in shallow waters (Goodwillie 2008). It is also noted that the GEBCO dataset also utilises the NASSA Satellite Radar Topography Mission (SRTM) data for land elevations. While the GEBCO data is likely to be suitable to define the continental shelf and regions further offshore for this study, its accuracy landward of the continental shelf was assessed using the digitised admiralty chart data. The GEBCO data was compared with the digitised admiralty chart data along a set number of defined transects. Figure 4-4 shows the location of the transects and the

45 differences between the two datasets are shown in Figures 4-5 to 4-9. In addition to this comparison, the two datasets were compared over the full domain of the digitised admiralty chart data. A plot showing the difference in depths is provided in Figure 4-10.

N -26.6 Transect #1

Mooloolaba Transect #2 -26.8

-27.0

Transect #3 -27.2 Moreton Bay Latitude -27.4 Extent of digitised chart data chart digitised of Extent

-27.6

-27.8 Transect #4

Transect #5 -28.0 Surfers Paradise

152.8 153.0 153.2 153.4 153.6 153.8 154.0 Longitude Figure 4-4 - Locations of transects

46 Transect 1 50

Digitised chart data 0

GEBCO data

-50

-100 10 5 -150 0 Elevation (m) -5

-200 -10 Elevation (m) -15 -20 -250 0 250 500 750 1000 1250 1500

Distance (m) -300 0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 Distance (m) Figure 4-5 - Comparison between bathymetry datasets along Transect 1 (inset of nearshore area)

Transect 2 50

Digitised chart data 0

GEBCO data

-50

-100 10 5 -150 0 Elevation (m) -5 -10 -200

Elevation (m) Elevation -15 -20 -250 -25 0 250 500 750 1000 1250 1500 Distance (m) -300 0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 Distance (m) Figure 4-6 - Comparison between bathymetry datasets along Transect 2 (inset of nearshore area)

47 Transect 3 50

0 Digitised chart data

-50 GEBCO data

-100

-150 30

-200 20 Elevation (m) 10 -250 0 Elevation (m) -300 -10

-20 -350 0 250 500 750 1000 1250 1500 Distance (m) -400 0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000 Distance (m) Figure 4-7 - Comparison between bathymetry datasets along Transect 3 (inset of nearshore area)

Transect 4 25

0 Digitised chart data

-25 GEBCO data

-50

-75 10 -100 5 Elevation (m) 0 -125 -5 -10

-150 Elevation (m) -15 -20 -175 0 250 500 750 1000 1250 1500 Distance (m) -200 0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 Distance (m) Figure 4-8 - Comparison between bathymetry datasets along Transect 4 with inset (inset of nearshore area)

48 Transect 5 25

0 Digitised chart data

-25 GEBCO data

-50

-75 10 -100 5 Elevation (m) 0 -125 -5 -10

-150 Elevation (m) -15 -20 -175 0 250 500 750 1000 1250 1500 Distance (m) -200 0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 Distance (m) Figure 4-9 - Comparison between bathymetry datasets along Transect 5 (inset of nearshore area)

49 N -26.6

-26.8

100 -27.0 40

-27.2 15

5 Latitude -27.4 -5 ence in depth (m) r

-27.6 -15 diffe

-40 -27.8 -100

-28.0

152.8 153.0 153.2 153.4 153.6 153.8 154.0 Longitude Figure 4-10 - Comparison of GEBCO dataset and digitised admiralty chart data (positive/green values indicate areas where chart data is shallower than GEBCO data)

Figures 4-5 to 4-9 illustrate that there are many areas where the GEBCO dataset differs by approximately 20m to the digitised chart data and that the GEBCO data is much more ‘smoothed’ than the chart data. The figures also show that the GEBCO elevations are generally higher than the chart data, which is also illustrated in Figure 4-10 by a domination of red. The GEBCO dataset is also consistently shallower than the chart data in areas between the shoreline and approximately 5- 10km offshore, exhibiting a flatter bedslope from the shoreline to approximately 20m depth. Comparison along the full length of the transects generally show that the large scale bed slopes of the GEBCO dataset are generally consistent with the digitised chart data, with the exception that

50 they are generally flatter in the near shore region. Based upon equation 2.5 presented in Section 2.2 the effect of the slightly flatter bed slopes in the near shore region may result in a small increase in the predicted surge where the GEBCO dataset has been used.

4.4. METEOROLOGICAL INPUT In order to model the wind and pressure surges generated during tropical cyclone ‘Roger’ temporally and spatially varying wind and pressure fields were generated for input into the model. For this study reanalysed meteorological datasets were used as the basis for the varying wind and pressure fields.

4.4.1. Wind As outlined in Section 3.2, the wind field is often the most critical input into storm surge models. Therefore considerable effort was taken in assessing the accuracy of available reanalysed wind field data sources. Four sources of reanalysed wind field data were assessed; ECMWF’s Interim and ERA-40 datasets, and NCEP’s DOE and NCAR datasets. Given the relatively coarse spatial density of the reanalysed grids (1.5-2.5° - approximately 170-280km), each dataset was interpolated spatially using a cubic spline interpolation. The resolution of the interpolated grids ranged between approximately 20- 25km. The datasets were also interpolated temporally using a linear interpolation, increasing the temporal resolution from 6hrs to 15mins. In order to illustrate the differences between the different datasets, plots for three of the datasets were generated from the spatially interpolated datasets at 10:00hrs on 17 March 1993. These plots are provided in Figures 4-11 to 4-13. It is noted that the contours are at 2.5m/s intervals (max contour 22.5-25m/s) and the vectors represent the wind direction and their origins are located at the centre of the native grids (non-spatially interpolated grids).

51 -10

-15

15 -20

Latitude 10 -25 10m Wind speed (m/s) 10m Wind speed (m/s)

-30 5

-35 0 145 150 155 160 165 170 Longitude Figure 4-11 - NCEP DOE data wind plot – 17-03-93 10:00hrs AEST (spatially interpolated) (data source NWS 2009)

-10

-15

15 -20

Latitude 10 -25 10m Wind speed (m/s)

-30 5

-35 0 145 150 155 160 165 170 Longitude Figure 4-12 - ECMWF Interim data wind plot – 17-03-93 10:00hrs AEST (spatially interpolated) (data source ECMWF 2009a)

52 -10

-15

15 -20

Latitude 10 -25 10m Wind speed (m/s)

-30 5

-35 0 145 150 155 160 165 170 Longitude Figure 4-13 - ECMWF ERA-40 data wind plot – 17-03-93 10:00hrs AEST (spatially interpolated) (data source ECMWF 2009b)

From these figures it can be clearly seen that the different datasets predict notably different 10m wind speeds during tropical cylone Roger. The general patterns between the NCEP DOE and ECMWF Interim winds are consistent, however the two ECMWF datasets exhibit lower wind speeds compared with the NCEP DOE dataset. It is also noted that all three datasets do not predict high winds near to the cyclone eye, which is understood to be due to the coarse resolution of the reanalysed meteorological datasets. However, for tropical cyclone ‘Roger’, which represents a ‘hybrid’ cyclone with a large radius to maximum winds, this limitation is not of concern. In order to assess the applicability of the datasets for the current project, comparisons of wind speed and direction were made between the interpolated reanalysed datasets and recorded wind data at a number of BOM weather stations. Wind data was generally available at 3 hourly intervals and recorded as 10 minute average wind speeds and directions. The reanalysed datasets were interpolated (temporally and spatially) to the time and location of the recorded data for the comparison. A summary of the comparisons, including the Root Mean Squared Errors (RMSE), mean errors and number of observations (n), are provided in Tables 4-3 to 4-6 and a map of the stations is presented in Figure 4-14. It is noted that no factors to account for the elevation of the weather stations were applied to the recorded wind speeds used for the comparisons presented in Tables 4-3 to 4-6. Additionally, a factor was not applied to the reanalysed data points over land as

53 recommended by ECMWF.

-16 Willis Island (200283) N Lihou Reef (200727) Flinders Reef (200783) -18

Marion Reef (200704) Townsville -20 Creal Reef (200736) Mackay -22 Gannett Cay (200831)

Latitude Cato Island (200601) -24 Sandy Cape (39085) Hervey Bay QLD -26

Cape Moreton (40043)

-28 Gold Coast

NSW -30 146 148 150 152 154 156 158 160 162 164 Longitude Figure 4-14 - Locations of BOM weather stations used for data comparison

Table 4-3 - NCEP-DOE comparison with recorded data summary (between 10/3/93-26/3/93) Wind Speed (m/s) Wind Direction (deg) Station Station RMSE Mean n RMSE Mean n Height (m) Error Error Sandy Cape 99 3.3 2.6 64 35.7 26.9 64 Cape Moreton 100 2.6 -0.7 111 26.5 14.7 111 Willis Island 8 1.6 0.5 127 18.7 0.8 127 Cato Island 6.5 4.3 0.5 79 21.1 7.3 79 Marion Reef 2 2.8 1.6 117 19.6 9 117 Lihou Reef 3 2.9 1.3 118 19 6.9 118 Creal Reef 1.5 2.4 1.3 118 39.6 -1.7 118 Flinders Reef 2.5 2.1 0.2 113 26.2 1.9 113 Gannett Cay 2.5 2.7 1.9 118 21.8 12.8 118

54 Table 4-4 - NCEP NCAR comparison with recorded data summary (between 10/3/93-26/3/93) Wind Speed (m/s) Wind Direction (deg) Station Station RMSE Mean n RMSE Mean n Height (m) Error Error Sandy Cape 99 2.6 0.4 64 39.7 23.0 64 Cape Moreton 100 3.8 -3.0 111 37.5 17.3 111 Willis Island 8 1.9 -0.3 127 28.1 -2.9 127 Cato Island 6.5 3.4 -1.4 79 23.2 5.7 79 Marion Reef 2 1.9 0.1 117 23.2 4.1 117 Lihou Reef 3 1.9 0.0 118 25.7 1.7 118 Creal Reef 1.5 2.4 0.5 118 44.8 3.9 118 Flinders Reef 2.5 2.0 -0.2 113 46.9 16.8 113 Gannett Cay 2.5 2.9 0.3 118 24.2 12.8 118

Table 4-5 - ECMWF Interim comparison with observed data summary (between 10/3/93-26/3/93) Wind Speed (m/s) Wind Direction (deg) Station Station Height RMSE Mean n RMSE Mean n (m) Error Error Sandy Cape 99 1.7 0.0 64 37.7 24.7 64 Cape Moreton 100 5.6 -5.1 111 16.5 9.2 111 Willis Island 8 1.2 -0.7 127 15.9 -7 127 Cato Island 6.5 3.7 -1.9 79 17.8 4.4 79 Marion Reef 2 1.3 -0.5 117 14.8 1.9 117 Lihou Reef 3 1.4 -0.3 118 15.1 -1.4 118 Creal Reef 1.5 1.5 0.5 118 31 -6.4 118 Flinders Reef 2.5 1.4 -0.5 113 27.1 -2.7 113 Gannett Cay 2.5 1.2 0.0 118 15.6 8.4 118

55 Table 4-6 - ECMWF ERA-40 comparison with observed data summary (between 10/3/93-26/3/93) Wind Speed (m/s) Wind Direction (deg) Station Station RMSE Mean n RMSE Mean n Height (m) Error Error Sandy Cape 99 2.3 -0.6 64 30.8 21.0 64 Cape Moreton 100 5.1 -4.4 111 15.9 0.9 111 Willis Island 8 1.3 -0.8 127 17.2 -4.4 127 Cato Island 6.5 4.5 -2.0 79 25.8 7.4 79 Marion Reef 2 1.5 -0.6 117 17.4 3.2 117 Lihou Reef 3 1.6 -0.3 118 15.4 -0.9 118 Creal Reef 1.5 1.7 -0.7 118 40.5 -7.7 118 Flinders Reef 2.5 1.6 -0.9 113 27.8 2.8 113 Gannett Cay 2.5 1.4 -0.3 118 16.0 9.3 118

A review of the data presented in Tables 4-3 to 4-6 shows that the NCEP DOE dataset is the only dataset with wind speeds marginally higher than the recorded wind speeds. All three other datasets tend to predict lower wind speeds than those recorded. For the ERA-40 dataset, this agrees with a newsletter released by the ECMWF which noted that statistical analysis of the ERA-40 ocean winds with observations indicated underestimation of high wind speeds (ECMWF 2004) (this under prediction is also evident in Figure 4-13). For the other two datasets, this under prediction may be a result of the adopted interpolation process which did not apply an interpolation factor to data points located over land. It is recommended that future comparisons of the ECMWF Interim dataset be made using the interpolation procedure described on their website. Wind direction is generally predicted to a similar accuracy level for all four datasets, with the exception of the NCEP-NCAR dataset for which the poorest agreement was calculated. It is noted that the winds recorded at Sandy Cape and Cape Moreton stations are affected by topographical effects. The Sandy Cape station experiences some sheltering from southerly to south- easterly winds, which is supported by the high wind speed mean error and positive mean wind direction error in the comparisons with the reanalysed data. The effect of topography on wind measurements at the Cape Moreton station have been assessed in a wind tunnel study by Ginger and Harper (date unknown). This study found that the ration between 10 min average wind speeds recorded at the station and +5m winds on flat land varied between 0.8 and 1.2 for winds from a southerly to easterly direction (approximately 1.2. for south-east). Given that winds were predominantly south-easterly during TC ‘Roger’ (average recorded wind direction approximately 135°), the mean wind speed error for the NCEP-DOE reanalysed dataset of -0.7 agrees well with the findings of the wind tunnel study and therefore the recorded data. 56 On the basis of the comparative analysis, the NCEP-DOE reanalysed meteorological dataset was selected as the wind field to be used for the wind forcing in the model. It is noted that comparisons at additional weather stations and plots of the reanalysed data against the recorded data are provided for the NCEP-DOE dataset in Appendix B.

4.4.2. Pressure As per the reanalysed wind datasets, reanalysed Mean Sea Level Pressure (MSLP) data was obtained and interpolated in order to compare it with recorded MSLPs and apply it in the model. For MSLP, two datasets were compared with the recorded data, ECMWF’s Interim dataset and the NCEP-DOE dataset. Interpolations were carried out spatially and temporally as per the wind data, with the exception that the movement of the cyclone was also interpolated with respect to time. This was included via the movement of the point of lowest MSLP in the spatially interpolated grid between timesteps. In this way the movement and shape of the MSLP field varied with respect to time according to a linear temporal interpolation. The comparison of the wind speed and direction with recorded data was repeated for the interpolated Mean Sea Level Pressure (MSLP) data. The results of the comparison are presented in Table 4-7.

Table 4-7 - Interpolated MSLP data comparison with recorded data (in hPa) Station NCEP-DOE ECMWF Interim RMSE Mean n RMSE Mean n Error Error Sandy Cape 0.8 0.6 64 0.6 0.4 64 Cape Moreton 1.2 0.8 111 1.2 0.9 111 Willis Island 0.7 -0.4 127 0.6 -0.4 127 Cato Island 2.5 1.0 80 1.5 0.6 80 Marion Reef 0.7 -0.6 117 0.6 -0.4 117 Lihou Reef 0.5 0.0 118 0.6 0.2 118 Creal Reef 0.5 -0.2 118 0.6 -0.4 118 Flinders Reef 0.7 -0.4 113 0.7 -0.6 113 Gannett Cay 0.7 -0.2 118 0.6 -0.2 118

From the comparative analysis results in Table 4-7 it is evident that both datasets generally predict the variation in MSLP well. One notable difference in the two datasets is the comparison at Cato Island which TC ‘Roger’ passed very near to. The higher resolution ECMWF Interim dataset (1.5° vs. 2.5°) better predicts the central pressure of the cyclone than the coarser NCEP-DOE dataset.

57 Given the small difference between the accuracy of the two datasets and the use of the NCEP-DOE dataset for the wind field, the interpolated NCEP-DOE MSLP data was selected for use in the model.

4.5. PARAMETERS

4.5.1. Wind Stress The wind stress coefficient proposed by Wu (1982) was adopted in the model. The formula being: + 065.08.0 U C = 10 (4.1) D 1000 As illustrated earlier in Figure 3-1, Wu’s formulation of the wind stress coefficient results in higher values than the formulation proposed by Smith and Banke (1975), but is consistent with the formulation proposed by Garratt (1977) at higher wind speeds (>15m/s). Investigation into the sensitivity of the wind stress coefficient was also carried out in a separate model simulation using of a constant wind stress coefficient of 5x10-3 (approximately double typical values).

4.5.2. Bottom friction A constant Manning bottom friction coefficient of 32m1/3/s was adopted throughout the model. This is the default value recommended in the MIKE 21 FM user manual, which states typical values of between 20-40 m1/3/s (DHI 2009b). Investigation into the sensitivity of this parameter was also carried out in a separate model simulation using a constant value of 64m1/3/s.

4.5.3. Other parameters Various other options and parameters adopted in the model are outlined in Table 4-8. As per the adopted bottom friction value, these reflect the default parameters recommended by DHI (2009b) with the exception of the timestep which has been adopted following consideration of the minimum grid spacing and depths within the model and a sensitivity analysis.

58 Table 4-8 - Parameters adopted in MIKE 21 FM simulations Parameter/Option Adopted Solution technique Low order, fast algorithm Min timestep (s) 0.01 Max timestep (s) 5 Critical CFL no. 0.8 Density type Barotropic Smagorinsky eddy Constant - 0.28 viscosity formulation Min eddy viscosity (m2/s) 1.80E-06

Max eddy viscosity (m2/s) 1.00E+10

Coriolis forcing type Varying in domain Tidal potential not included

4.6. TIDAL CONSTITUENTS In order to assess the possible interaction of tides with the storm surge during TC ‘Roger’, a model simulation was run with astronomical tides included. In order to simulate the model with tides, time variant water levels were defined along each of the open boundaries. The time variant water levels were derived from tidal constituents taken from DHI’s global tidal model (incorporated in the MIKE software).

4.7. WAVE SIMULATION In addition to the tidal simulation, the model was also simulated with the inclusion of wave radiation stresses. The wave radiation stresses were derived via a spectral wave model which was constructed for this assessment. The DHI MIKE 21 Spectral Waves FM package was used to simulate the wave conditions during TC ‘Roger’. The Brisbane Waverider buoy data was used to force the model via the eastern model boundary condition. While it is acknowledged waves at the deep offshore model boundary are likely to vary from the waves recorded at the Brisbane Waverider station, this approximation was considered sufficient for the current study which is focused upon assessing the effect of wave radiation stresses on the storm surge and not the direct modelling of waves. The Brisbane Waverider buoy was non-directional up until the first directional buoy was installed in March 1997 and consequently no wave direction data was available during TC ‘Roger’. In the absence of wave direction data, a review of the offshore wind speed and direction using the NCEP

59 DOE dataset was undertaken. A plot showing the wind speed and wind direction during TC ‘Roger’ at an offshore location is shown in Figure 4-15.

270 30 Wind Direction Wind Speed 225 25

180 20

135 15

90 10 (m/s) Speed Wind Wind Direction (deg) Direction Wind

45 5

0 0 12/03 13/03 14/03 15/03 16/03 17/03 18/03 19/03 20/03 21/03 22/03 23/03 24/03 Date (1993) Figure 4-15 - Wind direction and speed at 28.7°S 154.7°E from NCEP DOE dataset (data source NWS 2009)

On the basis of Figure 4-15 a mean wave direction of 135° was adopted for the entirety of the simulation. A directional standard deviation of 5° was also adopted for the simulation. It is noted that a review of DERM’s annual Queensland wave climate annual summaries (2000-2004) (DERM 2009c) for the Brisbane Waverider station revealed that the mean wave direction during large wave events (Hsig > 2.5m) is most commonly between approximately 120-145°. Due to computational limitations, the spectral wave model covered a smaller extent than the hydrodynamic model, which is illustrated in Figure 4-16. It is highlighted that the aim of the wave simulations was to assess on the effect on the surge at the Gold Coast Seaway. Accordingly, the developed mesh was refined with a high resolution in this area as illustrated in Figure 4-17 which shows the wave model mesh. The minimum element length in this mesh was slightly less than the minimum element length adopted in mesh resolution 3 (~80m).

60 -24 N extent C -25

-26

QLD -27

Wave model extent Latitude -28

-29

-30 NSW

-31 151 152 153 154 155 156 157 158 159 Longitude Figure 4-16 - Wave model extent and hydrodynamic model extent C

The fully spectral formulation and instationary time formulation options were used together with default values for the spectral discretisation. The low-order fast algorithm solution technique was used with minimum and maximum timesteps of 0.01 and 30 seconds respectively. Both bottom friction and white capping were turned off.

61 Gold Coast Seaway Latitude

Longitude Figure 4-17 - Wave model extent and mesh

4.8. ADDITIONAL NOTES In order to avoid errors comparing data from different time zones, all data inputs were converted to AEST (+10:00hrs UTC). It is also noted that daylight savings in NSW finished on 7 March 1993 (BOM 2009d) and therefore no time correction has been applied to observed data from NSW after this date.

62 5. RESULTS 5.1. INITIAL SIMULATION RESULTS An initial simulation of the model was carried out using the adopted model parameters outlined in Section 5. The flexible mesh used for this simulation was configured with the largest grid extent and medium resolution (extent ‘A’ and mesh resolution ‘2’ as defined in Section 4). The results of this simulation are presented below in Figures 5-1 and 5-2. The peak modelled storm surge is also summarised in Table 5-1 together with the recorded peak tidal residuals. It is noted that for the trained river entrances which were not incorporated in the mesh the model reporting station was located approximately 500m offshore.

Table 5-1 - Initial simulation results Location Max Recorded Max modelled Difference anomaly (mm) surge (mm) Noosa River - Munna Point* 570 350 -39% Mooloolaba Storm Surge 440 290 -34% Caloundra Public Jetty 740 355 -52% Pumicestone Passage - Military Jetty 590 300 -49% Pumicestone Passage - The Farm 620 310 -50% Brisbane Bar 340 65 -81% Paradise Point 560 80 -86% Gold Coast Seaway 760 210 -72% Gold Coast Seaway 2† 760 280 -63% Tweed Regional* 640 195 -70% Tweed Offshore 300 190 -37% Brunswick Heads* 520 210 -60% Ballina* 280 225 -20% Yamba* 190 155 -18% Yamba Offshore 210 145 -31% * reporting stations located ~500m offshore of actual location as inlet not incorporated in model mesh † Second Gold Coast Seaway station located ~500m offshore immediately south of the southern breakwater

63 1000 Noosa (Munna Pt) Recorded 500 Modelled

1000 Mooloolaba 500

1000 Caloundra 500

1000 Military Jetty 500 Pumicestone Passage

1000 The Farm Pumicestone Passage 500 Tidal Anomaly (mm) 0

1000 Brisbane Bar 500

1000 Paradise Point 500

1000 Gold Coast Seaway 500

-500 11/3/93 13/3/93 15/3/93 17/3/93 19/3/93 21/3/93 23/3/93 25/3/93 Date Figure 5-1 - Initial simulation results at QLD sites

64 1000 Tweed Offshore Recorded 500 Modelled

1000 Brunswick heads 500

1000 Ballina 500

Tidal Anomaly (mm) 1000 Yamba Offshore 500

1000

500 Yamba

-500 11/3/93 13/3/93 15/3/93 17/3/93 19/3/93 21/3/93 23/3/93 25/3/93 Date Figure 5-2 - Initial simulation results at NSW sites

The results in Figures 5-1 and 5-2 and Table 5-1 illustrate that the initial results under predict the storm surge at most locations. The locations where the best agreement between the modelled and recorded levels are Yamba, Ballina, Tweed offshore and Mooloolaba, while poor agreement is evident at the Gold Coast Seaway, Brunswick Heads, Paradise Point and Caloundra. It is noted that the earlier arrival of the modelled surge for Ballina and Yamba is a result of the NCEP DOE wind speed peaking prior to the recorded wind speeds as illustrated in the comparison plots contained in Appendix B (see stations 59040, 59030 and 58198).

5.2. SENSITIVITY ANALYSIS RESULTS Following the initial model simulation, a number of additional simulations were run with varying parameters and mesh files. These simulations were focused on two key objectives: investigation of the model stability and general sensitivity, and sensitivity of adopted parameters linked to the

65 physical storm surge process. Simulations were initially run to assess the first of these using variable model extents and mesh resolutions. The second set of simulations were run with variable parameters and model forcings.

5.2.1. Model sensitivity results Model domain The results of the simulation of the three model extents (A,B,C) presented in Section 4.2 are summarised in Table 5-2. It is noted that model extent A corresponds with the initial simulation model extent.

Table 5-2 - Model extent sensitivity results

Model Yamba offshore Ballina Noosa Mooloolaba extent Peak surge Differenc Peak surge Differenc Peak surge Differenc Peak surge Differenc (mm) e (mm) e (mm) e (mm) e A 145 - 225 - 349 - 290 - B 146 0.7% 219 -2.7% 345 -1.1% 286 -1.3% C 125 -13.8% 209 -7.1% 344 -1.4% 280 -3.5%

Overall the results in Table 5-2 show that the modelled surge generally reduces with a reducing grid extent and that the distance downwind from the model boundary has a key influence. The Noosa and Ballina stations are both located approximately 1.5° of latitude from the model boundaries in the extent C model. Despite this, the difference for the extent C case is lower at the Ballina site, suggesting that the wind direction (which is largely south-east during TC ‘Roger’) has some effect on these results. By-in-large the modelled storm surge at stations well inside the model boundaries (greater than approximately 1.5°) is only affected marginally (1-4%) by the reduction in grid extent. Only at stations nearer to the model boundaries (such as the Yamba offshore station in extent C) do the modelled surges differ appreciably for the reduced grid extent. Mesh resolution In order to assess the effect of the mesh resolution, three simulations using the three different meshes outlined in Section 4.2.2 were run. The results of these simulations were compared at a location in the model which was least affected by the changes in the coastline as a result of the increasing mesh resolution (the definition of the coastline increases with increasing mesh resolution - resulting in appreciable local changes in the modelled storm surge). The selected location was a point approximately 500m offshore of Stradbroke Island’s eastern coastline. The modelled peak storm surge for each of the three meshes is presented in Table 5-3 together with the percentage

66 difference between each. It is noted that mesh resolution 2 corresponds with the initial simulation mesh resolution.

Table 5-3 - Model resolution results at Stradbroke Is.

Mesh resolution Peak surge (mm) Difference 1 256 - 2 316 19% 3 316 19%

These results demonstrate that the coarsest resolution mesh (1) under predicts the surge compared with the finer resolution meshes. The results also show that the increased resolution from meshes 2 to 3 (1km to 100m minimum element edge length) results in no change in the predicted surge at a distance of 500m offshore (approximately 10m depth). It is noted that this conclusion may be different in the case of shallower waters. This sensitivity analysis generally highlighted that storm surges vary considerably spatially as a result of coastal features and therefore the resolution of model meshes should be such that the locations of recording stations or calibration locations are represented at an appropriate level of accuracy.

5.2.2. Storm surge parameter sensitivity results Wind stress coefficient A simulation using an increased wind stress coefficient was completed to assess the increase in wind stress necessary to reproduce the recorded tidal anomalies during TC ‘Roger’. The adopted linear relation for the wind stress coefficient is presented in Figure 5-3 together with those presented earlier in Section 3.5.1.

67 7.0

Smith & Banke (1975) 6.0 Garratt (1977) Wu (1982) Model sensitivity Cd 5.0 )

3 4.0

Cd (x10 Cd 3.0

2.0

1.0

0.0 0 5 10 15 20 25 30

U10 (m/s)

Figure 5-3 - Sensitivity simulation wind stress coefficient (CD)

Figure 5-3 illustrates that the adopted wind stress coefficient was approximately double the value of that proposed by Wu (1982) (adopted for initial model simulation) at a wind speed of 20-25m/s. The results from the simulation are illustrated in Figures 5-4 and 5-5 and summarised in Table 5-4.

68 1000 Noosa (Munna Pt) Recorded 500 Modelled

1000 Mooloolaba 500

1000 Caloundra 500

1000 Military Jetty 500 Pumicestone Passage

1000 The Farm Pumicestone Passage 500 Tidal Anomaly (mm) Anomaly Tidal 0

1000 Brisbane Bar 500

1000 Paradise Point 500

1000 Gold Coast Seaway 500

-500 11/3/93 13/3/93 15/3/93 17/3/93 19/3/93 21/3/93 23/3/93 25/3/93 Date Figure 5-4 - Increased wind stress coefficient simulation results at QLD sites

69 1000 Noosa (Munna Pt) Recorded 500 Modelled

1000 Mooloolaba 500

1000 Caloundra 500

1000 Military Jetty 500 Pumicestone Passage

1000 The Farm Pumicestone Passage 500 Tidal Anomaly (mm) Anomaly Tidal 0

1000 Brisbane Bar 500

1000 Paradise Point 500

1000 Gold Coast Seaway 500

-500 11/3/93 13/3/93 15/3/93 17/3/93 19/3/93 21/3/93 23/3/93 25/3/93 Date Figure 5-5 - Increased wind stress coefficient simulation results at NSW sites

70 Table 5-4 - Increased wind stress coefficient simulation results

Location Max Recorded Max modelled Difference anomaly (mm) surge (mm) Noosa River - Munna Point* 570 493 -14% Mooloolaba Storm Surge 440 409 -7% Caloundra Public Jetty 740 532 -28% Pumicestone Passage - Military Jetty 590 449 -24% Pumicestone Passage - The Farm 620 495 -20% Brisbane Bar 340 99 -71% Paradise Point 560 83 -85% Gold Coast Seaway 760 292 -62% Gold Coast Seaway 2† 760 434 -43% Tweed Regional* 640 261 -59% Tweed Offshore 300 267 -11% Brunswick Heads* 520 298 -43% Ballina* 280 342 22% Yamba* 190 240 26% Yamba Offshore 210 223 6% * reporting stations located ~500m offshore of actual location as inlet not incorporated in model mesh † Second Gold Coast Seaway station located ~500m offshore immediately south of the southern breakwater

These results show that the use of a much higher wind stress coefficient provides a better fit between the recorded tidal anomalies and the modelled surge. However, Figures 5-4 and 5-5 show that at a number of stations the recorded tidal anomalies have an additional component which is evident prior to the peak in the modelled storm surge. The stations where this is most notable are Noosa, Caloundra, the two Pumicestone Passage stations, Gold Coast Seaway. It is noted that poor agreement between the recorded and modelled results is still evident at sites affected by an apparent negative storm surge in the model (Brisbane Bar and Paradise Point). Bottom friction A similar analysis to that completed for the wind stress coefficient was completed for the bottom friction coefficient. The adopted bottom friction for this assessment was double that adopted for the initial simulation (64 vs. 32m1/3/s). The results from this simulation are presented in Figures 5-6 and 5-7 and Table 5-5.

71 Table 5-5 - Increased bottom friction simulation results

Location Max Recorded Max modelled Difference anomaly (mm) surge (mm) Noosa River - Munna Point* 570 440 -23% Mooloolaba Storm Surge 440 380 -14% Caloundra Public Jetty 740 460 -38% Pumicestone Passage - Military Jetty 590 400 -32% Pumicestone Passage - The Farm 620 365 -41% Brisbane Bar 340 95 -72% Paradise Point 560 75 -86% Gold Coast Seaway 760 165 -78% Gold Coast Seaway 2† 760 356 -53% Tweed Regional* 640 200 -69% Tweed Offshore 300 215 -28% Brunswick Heads* 520 260 -50% Ballina* 280 250 -10% Yamba* 190 160 -16% Yamba Offshore 210 160 -24% * reporting stations located ~500m offshore of actual location as inlet not incorporated in model mesh † Second Gold Coast Seaway station located ~500m offshore immediately south of the southern breakwater

72 1000 Noosa (Munna Pt) Recorded 500 Modelled

1000 Mooloolaba 500

1000 Caloundra 500

1000 Military Jetty 500 Pumicestone Passage

1000 The Farm Pumicestone Passage 500 Tidal Anomaly (mm) 0

1000 Brisbane Bar 500

1000 Paradise Point 500

1000 Gold Coast Seaway 500

-500 11/3/93 13/3/93 15/3/93 17/3/93 19/3/93 21/3/93 23/3/93 25/3/93 Date Figure 5-6 - Increased bottom friction simulation results at QLD sites

73 1000 Tweed Offshore Recorded 500 Modelled

1000 Brunswick heads 500

1000 Ballina 500

Tidal Anomaly (mm) 1000 Yamba Offshore 500

1000

500 Yamba

-500 11/3/93 13/3/93 15/3/93 17/3/93 19/3/93 21/3/93 23/3/93 25/3/93 Date Figure 5-7 - Increased bottom friction simulation results at NSW sites

These results illustrate that the increase in bottom friction results in appreciable increases in the modelled surge at a number of the sites, particularly the QLD sites. Tides A simulation was run with astronomical tides incorporated to assess if tides had any affect upon the storm surge during TC ‘Roger’. A simulation was fist run with the tidal forcings (time varying water level boundary conditions) but without the wind or pressure forcing, and a second simulation run with both the wind and pressure forcing and the tidal forcings. The difference in the modelled water levels between the second and the first simulations was then taken as the storm surge height. The modelled levels at Mooloolaba and the Gold Coast Seaway were compared with the predicted tides as shown in Figures 5-8 and 5-9.

74 1.5 Predicted Modelled 1.0

0.5

0.0 Water Level Water (m) -0.5

-1.0

-1.5 13/03/93 14/03/93 15/03/93 16/03/93 17/03/93 18/03/93 19/03/93 20/03/93 Date

Figure 5-8 - Comparison of predicted and modelled tides at Mooloolaba

1.5 Predicted Modelled 1.0

0.5

0.0

-0.5

-1.0

-1.5 13/03/93 14/03/93 15/03/93 16/03/93 17/03/93 18/03/93 19/03/93 20/03/93

Figure 5-9 - Comparison of predicted and modelled tides at the Gold Coast Seaway

The comparisons shown in Figures 5-8 and 5-9 show that the modelled tides marginally over- predict the high tide and under-predict the low tide and there is a slight phase error (modelled ahead of predicted). However, the simulated tides were considered adequate for the purposes of the comparative analysis being undertaken. The results of the comparison between the with and without 75 tide cases are summarised in Table 5-6.

Table 5-6 - Results for tide simulation

Location Modelled surge Modelled surge Difference without tides (mm) with tides (mm) Noosa River - Munna Point* 349 344 -1% Mooloolaba Storm Surge 290 286 -1% Caloundra Public Jetty 353 358 1% Pumicestone Passage - Military Jetty 301 311 3% Pumicestone Passage - The Farm 309 409 32% Brisbane Bar 64 75 17% Paradise Point 81 84 4% Gold Coast Seaway 211 225 7% Gold Coast Seaway 2† 282 276 -2% Tweed Regional* 194 194 0% Tweed Offshore 192 188 -2% Brunswick Heads* 206 204 -1% Ballina* 225 213 -5% Yamba* 156 151 -3% Yamba Offshore 145 140 -3% * reporting stations located ~500m offshore of actual location as inlet not incorporated in model mesh † Second Gold Coast Seaway station located ~500m offshore immediately south of the southern breakwater

These results illustrate that the effect of the tide upon the modelled storm surge is minimal at the majority of tide stations. The only exceptions to this are those stations which have been located in shallow areas of the model domain (The Farm, Brisbane Bar). Wave radiation stresses The effect of wave radiation stresses upon the modelled surge during TC ‘Roger’ was assessed via a wave simulation, with the results of this simulation incorporated into the hydrodynamic model together with the wind and pressure forcings. The results of the wave simulation are illustrated in Figures 5-10 to 5-13 which show the recorded vs. modelled significant wave heights and mean wave periods at the Gold Coast and Brisbane Waverider stations. It is noted that the close agreement at the Brisbane Waverider station is a result of the use of this station’s wave climate data to force the model.

76 7

Recorded 6 Modelled

Hsig (m) Hsig 3

0 12/03/93 14/03/93 16/03/93 18/03/93 20/03/93 22/03/93 24/03/93 26/03/93 Date

Figure 5-10 - Modelled vs. recorded sig. wave heights (Hsig) at Gold Coast Waverider station

12 Recorded Modelled

Tz (sec) 6

0 12/03/93 14/03/93 16/03/93 18/03/93 20/03/93 22/03/93 24/03/93 26/03/93 Date Figure 5-11 - Modelled vs. recorded mean wave period (Tz) at Gold Coast Waverider station

77 8

Recorded 7 Modelled

4 Hsig (m) Hsig 3

0 12/03/93 14/03/93 16/03/93 18/03/93 20/03/93 22/03/93 24/03/93 26/03/93 Date

Figure 5-12 - Modelled vs. recorded sig. wave heights (Hsig) at Brisbane Waverider station

Recorded 12 Modelled

Tz (sec) 6

0 12/03/93 14/03/93 16/03/93 18/03/93 20/03/93 22/03/93 24/03/93 26/03/93 Date

Figure 5-13 - Modelled vs. recorded mean wave period (Tz) at Brisbane Waverider station

From Figure 5-10 it is evident that the model slightly over-predicts the significant wave height at the Gold Coast Waverider station, but is considered to be within acceptable tolerances for the purpose of this analysis. The effect of the mean wave direction on the modelled wave heights and periods was also briefly assessed as part of the wave simulations. This revealed that as the wave direction shifted further to the south, the wave heights reduced at the Gold Coast Waverider station, 78 most likely as a result of sheltering from the Tweed headland. The results of the hydrodynamic simulation incorporating the wave radiation stresses together with the wind and pressure forcing are summarised in Table 5-7 and in Figure 5-14. The results have only been provided for the Gold Coast tide recording stations due to the limited extent of the wave model. It is also noted that the results are included for the two simulations, one using mesh resolution 2 and one using mesh resolution 3 (fine mesh).

Table 5-7 - Wave radiation simulation summary results for Gold Coast Seaway

Mesh resolution 2 Mesh resolution 3 Model reporting Case location Modelled surge Differenc Modelled surge Differenc (mm) e (mm) e Without wave 210 - 175 - radiation Gold Coast Seaway Including wave 325 +55% 265 +52% radiation Without wave 280 - 285 - radiation Gold Coast Seaway 2 † Including wave 385 +38% 405 +42% radiation † Second Gold Coast Seaway station located ~500m offshore immediately south of the southern breakwater

1000 Recorded Incl wave rad No wave rad 500

0 Tidal Residual (mm)

-500 11/3/93 13/3/93 15/3/93 17/3/93 19/3/93 21/3/93 23/3/93 25/3/93 Date Figure 5-14 - Comparison of model results with and without wave radiation stresses included at Gold Coast Seaway

From these results it is clear that the inclusion of wave radiation stresses increases the modelled surge in the Gold Coast Seaway. However, the magnitude of the increase is very small, being approximately 0.1m at the two Gold Coast Seaway locations. A review of the surge at a distance of 500m offshore from the beach just south of the seaway

79 revealed that the surge increase was 38% compared with 55% inside the Seaway (for mesh resolution 2). The peak current velocity at the same offshore location for the with and without wave radiation cases were 0.93m/s and 0.58m/s (60% difference) respectively. This illustrates that the inclusion of the wave radiation stresses results in significantly higher longshore currents during this event. It also highlights the magnitude of the currents which may have been experienced during this event, with longshore currents predicted by the model to exceed 1.0m/s in the nearshore zone as shown in Figure 5-15.

Figure 5-15 - Modelled currents at 18:00hrs 17/03/93 including wave radiation stresses (extent covers Tweed Headland to South Stradbroke Island) (mesh resolution 2)

It is noted that the results of the wave simulation were found to be particularly sensitive to the bathymetry, due to the influence of the bathymetry on the location at which waves break and the flow-on effect to the wave radiation stresses. It is also noted that the results in Table 5-7 illustrate that the difference in modelled wave setup attributable to the changes in mesh resolution is less than 5%, but that the increased resolution results in a lower predicted surge within the broadwater.

80 6. DISCUSSION The results presented in Section 5 are discussed further in this section with regard to the objectives of this study and implications of the findings. An initial discussion of the results at four particular stations is provided, which is followed by the discussion of particular processes or factors. These include: • Wind stress and wave radiation simulation results • Wave setup vs. wave radiation stresses • Other processes not considered in model A comparison of the model results with the simplified 1-d analytical solution presented in Section 2.2 is also presented.

6.1. PUMICESTONE PASSAGE TO MOOLOOLABA SURGE The results from the initial model simulation showed that the surge at Mooloolaba was slightly under-predicted, but with similar timing and shape. Contrary to this, the modelled surge at the Caloundra public jetty peaked much later than the recorded surge and there was a significant difference between the peak surge and peak tidal anomaly. These two stations are located only 30km from one another, but are each located in embayments which are oriented in opposite directions (north-east vs. south-east). In order to further explore these differences a figure was generated to show the recorded tidal anomalies and modelled surges at these stations together with the wind speed at Cape Moreton. This figure is shown as Figure 6-1 and also includes a hypothetical surge equal to the modelled surge plus 35% to illustrate the possible increase due to wave radiation stresses (following from the findings presented in Section 5.2).

81 1000 Recorded Modelled 500 Mooloolaba Modelled +35%

1000 Caloundra ? Rainfall? 500

1000 Military Jetty Pumicestone Passage

nomaly (mm) 500 A

Tidal 0

1000 The Farm Pumicestone Passage 500

100 Moreton Is. Recorded 75 Wind Speed NCEP DOE

25 Wind Speed (km/h) Speed Wind

0 13/3/93 14/3/93 15/3/93 16/3/93 17/3/93 18/3/93 19/3/93 20/3/93 21/3/93 22/3/93 Date Figure 6-1 - Recorded vs. modelled (initial sim. results) and modelled plus 35% comparison

This figure shows that the modelled surge plus 35% (black line) at Mooloolaba agrees well with the recorded tidal anomaly (blue line). It also highlights that the same adjustment at the Caloundra and Pumicestone Passage sites results in an improved fit, with the exception of the recorded anomalies between the period 06:00 17 March to 00:00hrs 18 March. While this period corresponds with the peak winds (and also waves) recorded in SEQ, the increase in the surge level appears disproportionate to the increase in wind speed (and also waves). One possibility, as suggested in

82 Figure 6-1, is the presence of notable freshwater flows at this time, resulting in an additional component to the tidal anomaly. To explore this possibility a plot of the daily rainfall for 17 March 1993 was generated, focusing on this area (Figure 6-2).

100

Mooloolaba 60 Rainfall (mm)

Caloundra 40

Figure 6-2 - Daily rainfall map for 17/03/93 (dots indicate rainfall station locations)

While it is evident in Figure 6-2 that heavier rainfall was experienced in Caloundra itself, similar rainfall also fell in the Mooloolah River catchment (generally to the south-west of Mooloolaba). The one difference between these to river systems is that Mooloolaba has a trained, north-facing outlet, while Pumicestone Passage’s outlet to the north is un-trained and faces east-ward as shown in Figures 6-3 and 6-4. This raises the possibility of sedimentation (i.e. deposition of sand) within the river outlet/entrance adjacent to the Caloundra station. Sedimentation would reduce the flow area and hydraulic capacity of the entrance and subsequently exacerbate the effect of rainfall runoff (i.e. would further raise water levels). It is noted that the presence of the same irregularity in the tidal anomalies at the two Pumicestone Passage sites would intuitively dismiss the possibility that the Caloundra station was subject to any notable wave related increases in the mean water level. The process of sedimentation may also apply to other easterly facing river entrances which experienced large waves from a south-easterly direction during this event. This would include the Gold Coast Seaway, Tweed River and Brunswick Heads.

Figure 6-3 - Location of Mooloolaba tide station and outlet of Mooloolah River (source Google Earth)

Figure 6-4 - Location of Caloundra tide station (source Google Earth)

84 6.2. WIND STRESS AND WAVE RADIATION SIMULATION RESULTS The results from the wind stress sensitivity simulation showed a better agreement between the modelled surges and recorded anomalies at most of the sites. However, at a number of the NSW stations (Yamba and Ballina) the resulting modelled surge was greater than the recorded anomalies. This over-prediction may have resulted from the wind stress or the wind speeds being too high. The results of the wave radiation simulation results showed an increase of 55% at the gauge location in the Gold Coast Seaway. While the resulting modelled surge was still grossly less than the recorded anomaly in the Seaway, this additional component would likely result in an over- prediction when combined with the increased wind-stress at the other stations. However, without the simulation of waves over the full model domain, and incorporation of the resultant wave radiation stresses in the hydrodynamic model, this remains unconfirmed.

6.3. WAVE SETUP VS. WAVE RADIATION STRESSES As outlined in Section 3.2 studies by Nielsen (1989) and Hanslow et al. (1996) have shown that wave setup is negligible in trained river entrances, generally implying that water level records from tidal gauges inside trained river entrances should not be affected by wave setup. While these studies were both conducted for a specific trained entrance (Brunswick Heads, NSW), recent data supporting these findings has been collected for the Gold Coast Seaway at the university’s Gold Coast Spit research station (location shown in Figure 6-7). Illustrated in Figures 6-5 and 6-6, the collected data has shown that the mean sea level at 500m offshore varies only marginally to the mean sea level inside the seaway for a range of wave conditions. The small differences observed have been attributed to small phase differences in tidal variations.

85 2.0 Broadwater Shoreline Setup 500m Offshore 1.5

1.0

0.5

0.0 Mean Water Level RL (mAHD)

-0.5

-1.0 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 Time (11/03/09) Figure 6-5 - Data collected during passage of TC ‘Hamish’ 11 March 2009 at the Gold Coast Spit research facility (Hsig ~ 4m)

2.0 Broadwater Shoreline Setup 500m Offshore 1.5

1.0

0.5 Mean Water Level Level (mAHD) Water Mean

0.0

-0.5 9:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 Time (21/5/09)

Figure 6-6 - Data collected on May 21 2009 at the Gold Coast Spit research facility (Hsig ~ 5m)

Figure 6-7 - Location of the manometer tubes at the Gold Coast Spit research facility

Two figures showing the results of the wave simulation are shown in Figures 6-8 and 6-9. These figures show the modelled significant wave heights and water levels at 18:00hrs 17 March 1993 respectively. While the mesh for the wave simulation (Figure 6-8) is a little coarser than desirable in the nearshore area, it shows the large waves breaking well outside of the breakwaters, and only much smaller waves propagating into the breakwater.

Figure 6-8 - Plot of significant wave heights at 06:00hrs 17/3/93 near Gold Coast Seaway (mesh resolution 3)

Figure 6-9 - Plot of water levels at 15:00hrs 17/03/93 near Gold Coast Seaway (includes wave radiation stresses) (mesh resolution 3)

Figure 6-9 illustrates that the predicted surge at the mouth of the Broadwater is generally consistent with the wider surge predicted along the coastline, and is not subject to localised elevation of the water level as a result of wave setup or any other process. In fact, the results show a reduction in the water surface elevation around the tip of the southern breakwater as a result of the increase longshore velocities in this area. Figure 6-9 also shows a localised setup on the southern side of the Seaway due to the obstruction presented by the breakwater. Therefore, in significant wind and wave events the 500m manometer tube could potentially be affected by a localised surge. In such cases the comparison of the water levels from the 500m manometer tube with the Broadwater should be made with caution. Regardless of this, the previous studies together with the model results provide strong evidence that wave setup does not occur within trained river entrances and that as a result tide gauges located inside trained river entrances and harbours are unlikely to be affected by wave setup, even during extreme wave conditions.

6.4. OTHER PROCESSES NOT CONSIDERED The storm surge processes included in this study and the adopted modelling approach are considered relatively complete, but by no means exhaustive. The effect of the East Australian Current (EAC) on the storm surge, resulting from localised ‘back-eddies’ in this strong offshore oceanic current, have not been considered. This is due to the difficulty of implementing the EAC in the MIKE 21 FM model. In addition to this, vertical density and current variations have been ignored through the use of a 2d model. While a fully 3d model could be implemented, the computational demand of such a model would be excessive.

6.5. COASTLINE REPRESENTATION The way in which coastline features were represented in the model was found to have a significant effect upon the predicted surge at certain locations. The more coarse resolution grids result in a coarser and typically ‘smoothed’ representation of the coastline, neglecting subtle features and the shapes of headlands and embayments. Depending on their location, these omissions were found to result in appreciable changes in the modelled surges. In consideration of this, it is recommended that the tidal stations which were not explicitly incorporated in the TC ‘Roger’ model meshes, be included to a sufficient level of detail to assess if any localised effects are evident.

89 6.6. COMPARISON WITH 1D ANALYTICAL SOLUTION A simple comparison of the 2d model results with the 1d analytical model presented in Section 2.2 is presented in Figures 6-10 and 6-11. Figure 6-11 shows a timeseries plot of the 2d model results at a location south of the Gold Coast Seaway, approximately 500m offshore. It is noted that a pressure compensation (1cm per hPa) has been applied to the Tilburg and Garvine analytical model.

0.40 Approx mean wind Longshore direction during TC 0.35 'Roger' Cross-shore

Combined 0.30

0.25 c) (m) η 0.20

0.15 Wind Surge ( Wind Surge

0.10

0.05

0.00 0 102030405060708090 Wind direction relative to shoreline (°) Figure 6-10 - Wind setup at coastline based upon Tilburg & Garvine (2004) for wind speed of 23m/s

0.30

0.25 Tilburg & Garvine (2004) Modelled (initial sim) 0.20

0.15

0.10

0.05 Storm surge (m) surge Storm

0.00

-0.05 10/03/93 12/03/93 14/03/93 16/03/93 18/03/93 20/03/93 22/03/93 24/03/93 Date Figure 6-11 - Timeseries of wind setup at coastline predicted by Tilburg & Garvine (2004) (incl. correction for pressure) vs. initial model result at location 500m offshore just south of Gold Coast Seaway It is noted that the same parameters used in the model for Figure 6-11 are the same as those presented in Section 2.2.

Figure 6-11 shows a good agreement between the 2d model results and those predicted by the 1d analytical model during the main period of the event (16/03/93 to 19/03/93).

6.7. ANOMALIES IN ANOMALIES Following the model simulations and sensitivity there were a number of the irregularities identified in a number of the recorded tidal anomalies. These included the tidal anomalies recorded at the Gold Coast Seaway and the Tweed Offshore stations. The anomaly at the Gold Coast Seaway station began to rise far in advance of any of the other stations, including the Tweed Offshore station. The anomaly at the Tweed Offshore station is significantly lower than that recorded in the Seaway and is also followed by a similar surge a number of days following the TC ‘Roger’ surge. Figure 6-12 shows the recorded tidal anomalies at these two stations together with the NCEP DOE wind speed at the location of the Ballina BOM weather station. Two additional figures (Figures 6- 13 and 6-14) show the predicted, recorded and tidal anomalies at these two stations as per the figures in Appendix A.

1000 Gold Coast Seaway

500

1000 Tweed Offshore Tidal anomaly Tidal(mm) anomaly 500

100 200

75 150

50 100 ection (deg) ection r

25 Wind speed 50 Wind Speed (km/h) Speed Wind Wind dir Wind di 0 0 13/3/93 15/3/93 17/3/93 19/3/93 21/3/93 23/3/93 25/3/93 Date Figure 6-12 - Tidal anomalies recorded at Gold Coast Seaway and Tweed Offshore stations and NCEP DOE wind speed at Ballina This figure shows a second increase in the wind speed at Ballina following the passage of TC

91 ‘Roger’, and a southerly shift of the wind direction. However, the tidal anomaly recorded at the Tweed Offshore station is disproportionate to the wind speed when compared with the surge during the passage of TC ‘Roger’. This could either be explained by an error in the tidal anomaly during TC ‘Roger’ or during this second period (21/03/93 to 23/03/93), or alternatively the presence of an ‘isolated’ wind ‘event’ (considered very unlikely).

Figure 6-13 - Predicted and recorded tide and anomaly at Gold Coast Seaway

Figure 6-14 - Predicted and recorded tide and anomaly at Tweed Offshore

92 6.8. SUMMARY OF RESULTS This study has generally shown that the hindcast modelling of a storm surge event is complex. The numerous possible interactions between the physical processes responsible for the generation of storm surges alone necessitates that the various processes be incorporated or at least considered in the numerical model. Add to this the spatial and temporal variability in storm surge events, and the possible influences, apart from storm surge, which tidal stations are potentially subject to. Despite these complexities, this study has revealed a number of findings, both specific to the TC ‘Roger’ event as well as general storm surge modelling findings. The modelled storm surge levels from the hindcast simulation of TC ‘Roger’ were found to largely under-predict the recorded tidal anomalies. Better fits with the recorded data were found with the increase of wind stress and inclusion of wave radiation stresses. While a simulation incorporating wave radiation stresses throughout the full model domain has not yet been completed, it is expected that the inclusion of these stresses will greatly improve the modelled surge levels, without increases in the wind stress coefficient. The sensitivity analyses completed revealed that changes in the model extent (within the range assessed) resulted in minimal changes to the predicted surges, while refinement of the mesh resolution principally resulted in changes where representation of coastal features was enhanced by the increased resolution. Therefore the refinement of the mesh was found to improve predictions of local surge variations which can be significant. After consideration of these findings and others presented in this report, the key recommendations are for the simulation of additional storm surge events and the permanent installation of current recording instrumentation on offshore tide stations. Ideally, one of the additional storm surge events will be similar in magnitude to TC ‘Roger’. This will enable a comparison of the predictive capability of the model between the two events, and help to further explore some of the unexplained irregularities identified for TC ‘Roger’. The installation of current recording instrumentation on offshore tide stations would greatly assist in the calibration of storm surge models, which are typically only calibrated on the basis of water levels and not current velocities and directions. Measurement via Acoustic Doppler Current Profiler (ADCP) would enable measurement of current velocity and direction over a specified depth range.

93 7. CONCLUSIONS This report has presented the development of a storm surge model which has been used to investigate the storm surge recorded during Tropical Cyclone Roger in 1993. The overall aim of the project was to better understand the storm surge processes and their relative contribution to the recorded surge during this event. A review of previous storm surge predictions for the SEQ region was also undertaken in order to assess the relative magnitude of the Tropical Cyclone ‘Roger’ event. A MIKE 21 FM model was developed and simulated using NCEP DOE hindcast wind and pressure data. The effect of wave radiation stresses and tides on the modelled surge was also investigated through the inclusion of these processes in the model. Numerous sensitivity analyses were undertaken as part of the modelling exercise, investigating the model domain, mesh size, and parameters such as the wind stress coefficient and bottom friction. A detailed comparison of four hindcast meteorological datasets revealed that the NCEP DOE dataset compared best with recorded observations at BOM weather stations during Tropical Cyclone ‘Roger’. The other datasets investigated (ECMWF Interim and ERA-40 and NCEP NCAR datasets) were found to generally under-predict the higher wind speeds recorded during the peak of the event. In the case of the ECMWF Interim dataset, this was in spite of being based upon a higher resolution dataset. The modelled storm surge levels from the hindcast simulation of Tropical Cyclone ‘Roger’ were found to largely under-predict the recorded tidal anomalies. Better fits with the recorded data were found with the increase of wind stress and inclusion of wave radiation stresses. While a simulation incorporating wave radiation stresses throughout the full model domain has not yet been completed, it is expected that the inclusion of these stresses will greatly improve the modelled surge levels, without increases in the wind stress coefficient. Additionally, the refinement of the coastline and bathymetry in the locations of the tide gauges is also considered likely to result in better prediction of the recorded tidal anomalies. The sensitivity analyses completed revealed that changes in the model extent (within the range assessed) resulted in minimal changes to the predicted surges. They also illustrated that the refinement of the model’s mesh resolution beyond a minimum element length of approximately 1km only resulted in changes in the surge where the representation of the coastline was notably improved or altered as a result of the increased resolution. Some of the general findings with respect to storm surge modelling include: • Wind speed input into models should be of primary importance – with the adopted wind

stress coefficient (CD) secondary.

94 • Hindcast wind datasets suitable for storm surge modelling for large systems – however rigorous checks with measured data are still important. While the findings of this study have been somewhat inconclusive with respect to the surge generated during Tropical Cyclone ‘Roger’, the study has lead to a number of other relevant findings. The results demonstrated that the inclusion of wave radiation stresses in the model resulted in a small increase to the predicted surge within the Gold Coast Broadwater (~0.1m increase). This indicates that wave setup is unlikely to have contributed significantly to the recorded water levels within the Broadwater during TC ‘Roger’. This is consistent with measurements taken in the Broadwater and previous detailed measurements at other trained river entrances which have found that wave setup does not occur within such entrances. It is therefore considered that the addition of the wave setup along the open coastline to tailwater levels in flood models is not appropriate. A review of previous storm surge studies revealed that storm surge models are commonly based upon limited calibration data and often exclude wave radiation stresses. The findings of this study highlight that for events such as East Coat Lows (ECLs) or large ‘Hybrid’ Cyclones which generate notable surges and waves, the inclusion of wave radiation stresses is important. Occurrences of ‘Hybrid’ Cyclones and ECLs in South East Queensland are far more frequent than mature coastal crossing cyclones. Based upon historical records, these events are also more likely to coincide with significant rainfall events. Therefore it is considered important that research into storm surge generation during such events, as well as for more significant tropical cyclone events, be continued to develop a greater understanding of the processes and subsequently greater confidence in our predictions for such events. After consideration of the findings of this report, two key recommendations are for the simulation of additional storm surge events and the permanent installation of current recording instrumentation on offshore tide stations. The simulation of additional events should incorporate wave radiation, wind and pressure forcings in the model together with accurate bathymetric data in specific areas of interest. The installation of current recording instrumentation on offshore tide stations would greatly assist in the calibration of storm surge models, which are typically only calibrated on the basis of water levels. Additionally, further research into tropical cyclone processes, particularly in the boundary layer, and the parameterisation of wind stress is considered necessary.

95 8. ACKNOWLEDGEMENTS The authors thank Dr Ian Teakle and Dr Nick Cartwright for their reviews of the report and valuable comments.

8.1. ASSISTANCE The assistance of a number of people helped to make this project possible. These included: • Jeff Callaghan – meteorological and general guidance • Ivor Blockley – assistance with hindcast meteorological datasets • Stefan Paul Szylkarski – for the use of the DHI software

8.2. DATA Data used in this project has been obtained from many different sources. The time taken for those who have compiled and provided this data is greatly appreciated. The sources of data and those who provided the data are acknowledged below. • NSW Tide and wave data – MHL – Matthew Smith • QLD Tide data – MSQ – Daryl Metters & Ray Pedderson • Weather station data - BOM • QLD wave data - DERM - Jim Waldron • NCEP datasets - http://nomad2.ncep.noaa.gov/ncep_data/ • ECMWF datasets - http://data.ecmwf.int/data/ • Bathymetry Data - GEBCO dataset - http://www.bodc.ac.uk/data/online_delivery/gebco/select/

96 APPENDIX A - TIDAL ANOMALY PLOTS

-18

15/03 -19

-20 Tropical Cyclone 'Roger' Track

-21 Mackay Hay Point 19/03

-22 16/03

-23 Rosslyn Bay 18/03

Gladstone -24 17/03

-25

Latitude Urangan

-26 Noosa Mooloolaba Caloundra -27 QLD Brisbane Bar

-28 Gold Coast Seaway Tweed Offshore Brunswick Heads Ballina -29

Yamba

-30

-31 NSW

-32 148 149 150 151 152 153 154 155 156 157 158 Longitude Figure A-1 Tide gauge station locations and TC ‘Roger’ track (JTWC 2009) (dates indicate approximate cyclone centre at 10:00hrs AEST)

100

101

102

103

104

105

106

107

108 APPENDIX B - NCEP DOE COMPARISON WITH OBSERVED PARAMETERS

Comparison with recorded data for period 10/3/93-26/3/93

Wind Speed (m/s) Wind Direction (deg) MSLP (hPa) Station Stn height (m) RMSE Mean Error n RMSE Mean Error n RMSE Mean Error n St Lawrence Post Office 17 5.4 4.9 79 64.5 10.9 79 0.9 -0.1 79 Lady Elliot Island 3.5 4.4 4.0 73 19.5 4.3 73 1.6 -0.8 73 Sandy Cape 99 3.3 2.6 64 35.7 26.9 64 0.8 0.6 64 Gladstone 75 4.9 3.7 127 40.6 20.8 127 0.6 -0.1 127 Bundaberg 27 4.1 2.6 110 36.0 -7.9 110 1.1 -0.9 110 Seventeen Seventy 34 3.4 1.6 91 30.3 3.3 91 0.8 0.0 90 Cape Moreton 100 2.6 -0.7 111 26.5 14.7 111 1.2 0.8 111 Coolangatta 6 5.1 4.1 114 43.6 -27.0 114 0.5 -0.1 114 Gold Coast Seaway 3 5.2 4.0 116 31.8 -12.6 116 0.6 0.3 116 Rainbow Beach 14.5 7.0 6.3 80 35.7 -8.1 80 0.6 -0.2 80 Yamba 29 3.4 2.1 80 48.2 -21.3 80 0.7 0.1 80 Ballina 1 5.2 3.6 64 60.7 -20.8 64 N/A N/A N/A Smoky Cape 117 2.9 0.3 111 50.5 -25.3 111 0.5 0.1 111 Coffs Harbour 5 3.5 2.5 106 64.8 -40.9 106 0.6 -0.2 104 Willis Island 8 1.6 0.5 127 18.7 0.8 127 0.7 -0.4 127 Norfolk Island 112 5.8 5.4 127 20.8 9.6 127 0.7 -0.4 98 Cato Island 6.5 4.3 0.5 79 21.1 7.3 79 2.5 1.0 80 Marion Reef 2 2.8 1.6 117 19.6 9.0 117 0.7 -0.6 117 Lihou Reef 3 2.9 1.3 118 19.0 6.9 118 0.5 0.0 118 Creal Reef 1.5 2.4 1.3 118 39.6 -1.7 118 0.5 -0.2 118 Flinders Reef 2.5 2.1 0.2 113 26.2 1.9 113 0.7 -0.4 113 Gannett Cay 2.5 2.7 1.9 118 21.8 12.8 118 0.7 -0.2 118

109

110

111

112

113

114

115

116

117

118

119

120

121

122

123

124

125

126

127

128

129

130

131

132

133

134

135

136

137

138

139

140

141

142

143

144

145

146

147

148

149

150

151

152

153

154

155 APPENDIX C - RAINFALL DATA FOR MARCH 12-20 1993

Station Station # Latitude Longitude 12 13 14 15 16 17 18 19 20 ACHIL 40499 -26.2239 151.2986 0 1.6 0 0 0 0 0 0 0 ALDERLEY 40224 -27.4181 153.0025 22 0.1 0 10.9 2 11.2 5.8 1.2 1.4 AMAMA 40491 -26.3667 152.5661 0 0 0 11 0 7.5 0 5 0 AMAMOOR FORESTRY 40003 -26.3664 152.6236 26 -- -- 10.03 5.6 13.2 0 2.6 0 AMBERLEY AMO 40004 -27.6294 152.7114 2.8 0 0 0 0 0.4 1.2 0 0.4 ARATULA ELIZABETH ST 40266 -27.9825 152.5475 4.2 0 0 0 0 1.2 1.2 0 0 ARCHERFIELD AIRPORT 40211 -27.5717 153.0078 ------ASHGROVE BOWLS CLUB 40326 -27.4433 152.975 18.8 0.4 0 5.6 1.4 14 6.4 3.4 1.4 BALAAM HILL 40300 -27.4533 152.3733 ------BALD HILLS POST OFFICE 40213 -27.3236 153.0097 13.8 0 -- 29.22 7.4 22.6 8.2 7 0.4 BANYO SEMINARY 40204 -27.3778 153.0883 12 1.4 0 32 8.4 21.8 12.6 10.4 0 BAROON POCKET DAM 40850 -26.715 152.8719 19 5 0 47 17 59 47.5 24.5 1.8 BEAUDESERT 40014 -28.0178 153.0131 0 0.3 0 0 0 2.3 -- 1.22 0.4 BEECHMONT BINNA BURRA ROAD 40015 -28.1467 153.1903 1 0.4 0.4 4 0 36 9.6 10 4.4 BEENHAM VALLEY RD 40686 -26.1842 152.8206 16.2 0 0 17.2 26.6 -- 37.42 0 2 BEENLEIGH BOWLS CLUB 40406 -27.7094 153.2014 3.8 0 0 8.4 1.8 25.2 13.4 3.4 2 BEERBURRUM FOREST STN 40284 -26.9567 152.96 26 0.4 0 7.2 16.2 25 28 4.2 0.8 BEERWAH FOREST 40017 -26.8564 152.9764 35 -- -- 23.23 11 59.2 32.2 18.8 0 BELLI PARK 40561 -26.5142 152.8314 13.8 0 0 27 10.2 36.2 18.2 6.8 3.8 BIGGENDEN POST OFFICE 40021 -25.5097 152.0456 19 0 0 4.6 3.2 0 0.2 0 0 BLACKBUTT FORESTRY STN 40769 -26.8919 152.105 5.6 -- -- 1.43 0 0 2.4 0 -- BLACKBUTT POST OFFICE 40020 -26.8844 152.1006 3.4 0 0 1.4 0.6 1 2 0 2.8 BOONAH BORDER GATE 40523 -28.2525 152.5192 3.2 0.4 4.4 5.6 0.6 23 20 12.2 3.4 BOONAH STARK AVE 40024 -27.9925 152.6919 13 0 0 0 0 0 1.2 0 0 BOONARA ROMLEY 40721 -26.0933 152.0522 15 0 0 12 2 0 0 0 0 BOONDALL BCC 40531 -27.3306 153.0711 11.4 1.6 0 19.4 6.8 22.6 11.2 5.8 1.2 BORUMBA DAM 40481 -26.5039 152.5861 9 0 0 5 7.2 5.4 3 4 0 BRAMBLE BAY BOWLING CLUB 40807 -27.2583 153.1017 0 0 0 23.3 9.6 28.5 19 0 0 BRIAN PASTURES 40428 -25.655 151.745 46 0 0 0.4 0.2 0 0 0 0 BRIBIE ISLAND QLD UNI 40685 -27.0467 153.1667 ------35.44 29.4 33 26 10 0 BRIGOODA 40377 -26.2575 151.4108 0.4 0 0 0 0 0 0 0 0 BRISBANE AERO 40223 -27.4178 153.1142 11.2 1.4 0 44.6 6.8 22.4 15.2 9.8 0 BRISBANE REGIONAL OFFICE 40214 -27.4778 153.0306 16.4 0 0 6.6 1.8 12.4 5.8 2.4 0.6 BRISBANE RPA HOSPITAL 40767 -27.4975 153.035 12.6 -- -- 4.03 1.6 18.2 7 2.4 -- BRISBANE SHOW GROUNDS 40216 -27.4511 153.0319 17 0 0 9.8 1.8 10.6 4.6 1.8 0 BROOWEENA LAHEY ST 40028 -25.6006 152.2614 64.21 0 0 3.6 5.4 0 2.4 0 0 BUARABA 40403 -27.3878 152.3492 0 0 0 0 0 0 0 0 0 BUDERIM POST OFFICE 40031 -26.6867 153.05 12.5 -- -- 58.43 19 53.2 23.3 18.1 -- BURBANK LEACROFT ROAD 40862 -27.5864 153.1583 13.6 0.2 0 12.6 6.8 28.2 17 5.4 1 BURPENGARY ULMANN RD 40035 -27.1414 153.0089 ------CABOOLTURE POST OFFICE 40038 -27.085 152.9517 16.8 -- -- 26.83 6.8 17.4 13.2 9.4 0 CALOUNDRA WATER TREAT 40496 -26.7928 153.1286 41 0 24 18 25 68 36 10 0 CALTEX REFINERIES 40320 -27.4164 153.1556 12.8 0 0 32.8 6.8 27.2 24 10 1 CAMP MOUNTAIN 40041 -27.3961 152.8664 28.6 0 0 5 0.8 15.2 6.4 0.8 0 CANNON HILLS BOWLS CLUB 40468 -27.4692 153.0803 ------16.44 -- 22.02 0 0 0 CANUNGRA FINCH ROAD 40042 -28.0142 153.1664 0 0 0 2 1.2 15 0.4 5.8 0

CANUNGRA LAND WARFARE CENTRE 40852 -28.0272 153.1794 0 0 0 0 1 18.4 1 6 6.4 CAPALABA WATER TREAT 40458 -27.5314 153.1825 13 1.2 0 29.2 6 38.2 28 9.8 0.4 CAPE MORETON LIGHTHOUSE 40043 -27.0328 153.465 29.8 14.8 1 33.6 6.6 20.8 12 2.8 0

156 Station Station # Latitude Longitude 12 13 14 15 16 17 18 19 20 CARBROOK LONGLAND ROAD 40408 -27.6847 153.28 ------CARNEYS CREEK 40490 -28.2167 152.5333 0 0 0 0 0 11.2 17 2.6 0 CARRARA BRADSTONE RD 40864 -28.0167 153.3517 0.4 1 0 13 8.6 24 19 3 1.2 CENTRAL KERRY 40413 -28.1561 153.0397 2.5 0 1 0 0 4 14 5 2 CHERMSIDE BOWLS CLUB 40268 -27.3931 153.0267 13.5 0.5 0 -- 32 19.2 5.2 3.2 0 CHRISTMAS CREEK RD 40725 -28.1683 152.9883 0 0 0 0 0 0 7.2 0 2.6 CLEARVIEW TM 40416 -28.0017 153.3083 ------CLEVELAND BOWLS CLUB 40047 -27.5253 153.2756 11.2 3.4 0 18.2 13.2 28.6 41.8 12 1.8 COMO 40057 -26.1842 152.9136 15.2 0 0 55.2 30 32 11.4 12 2.2 CONNEMARRA 40698 -26.7881 152.7308 14.6 -- -- 14.23 31.2 31.8 23.2 14.8 0.4 CONONDALE TOWNSHIP 40051 -26.7292 152.7172 0 20 0 6.6 13 26 17.2 8.6 0.4 COOLANGATTA AIRPORT 40717 -28.1661 153.5036 ------COOLUM BOWLS CLUB 40420 -26.53 153.0897 ------COOMBABAH WATER TREATMENT PLANT 40849 -27.9136 153.3656 4.6 0 0 13.8 9.6 37.5 24 21 0 COOMINYA POST OFFICE 40056 -27.3914 152.5014 16 0 0 2.8 0 0 1.4 0 0 COORAN 40755 -26.5575 152.7714 9.6 0 0.4 35.2 16 21 9.8 7 1.8 COOROY COMPOSITE 40059 -26.4181 152.9128 12 0.4 0 32.4 23.2 44 16 11.2 0 COORPAROO BCC 40798 -27.5153 153.0722 5.5 6 0 0 0 12 8.4 5 2.4 COORPAROO BOWLS CLUB 40220 -27.4922 153.0564 ------COOYAR POST OFFICE 40060 -26.9814 151.8311 26 0 0 0 0 0 0 0 0 CORBOULD PK RACECOURSE 40759 -26.785 153.0686 28 3 -- 36.02 23 38.2 28.4 17 3 COURAN COVE 40471 -27.8333 153.4083 ------COWAL 40013 -25.8167 152.6222 24.4 0 0 8.8 11 16 0 5.8 0 CRAGLANDS 40134 -26.6992 152.9039 11 0 0 22.6 27 45.4 36.6 19.2 0.2 CRESSBROOK DAM 40808 -27.2644 152.1961 0 0 0 2 0 4.2 0 0 3 CROHAMHURST 40062 -26.8094 152.87 21.5 -- -- 22.63 16.4 38.2 31.2 14.8 0 CROWS NEST 40382 -27.2711 152.0642 3.6 0 0 1.4 0 0 0 0 0 CURRIGEE 40484 -27.8881 153.4231 4.8 0 0 6 12.2 18.4 20.6 5.8 0 DANEWOOD VALE 40635 -26.835 152.49 26.6 0 0 2.4 18 10.2 3.2 8.2 1.8 DARLINGTON 40610 -28.2522 153.0872 2.8 0 3.2 11.4 6.4 53 45.6 22.2 7.2 DAYBORO 40578 -27.2133 152.8067 ------DAYBORO POST OFFICE 40063 -27.1967 152.8247 16 0 0 25 4 15 6 2 1 DEVERTON SAWPIT GULLY RD 40883 -27.6847 152.0575 0 0 0 0 0 0 0 1 0 DOOLANDELLA 40951 -27.6286 152.9831 7.1 0.8 0 3.1 0.8 12.5 2.4 0.2 1.1

DOUBLE ISLAND POINT LIGHTHOUSE 40068 -25.9319 153.1906 ------DUCKINWILLA CREEK 40069 -25.4 152.4333 64.6 0 9.2 5.2 4.2 0 0 0 0 DUNOLLIE 40455 -25.85 151.3667 0.2 0 0 0 0 0 0 0 0 DUNWICH 40537 -27.4961 153.4081 11.2 ------73.7 19.6 44.2 22.2 -- EAGLE FARM RACECOURSE 40212 -27.4303 153.0669 16.8 0 0 19 2.8 21.2 9.4 0 3 ELANORA WATER TREAT 40609 -28.1181 153.4456 0 3 0 19 13 25 43.6 8 4 EMBREYS BRIDGE 40500 -26.6431 151.9433 22.8 0 0 2.6 0 1.2 0 0.6 0 ENGLESBERG VILLAGE 40104 -27.9486 152.6236 ------ENOGGERA RESERVOIR 40225 -27.4447 152.9286 20 0 0 4.2 1.2 16 6 1.4 1.6 ESK POST OFFICE 40075 -27.2397 152.4225 8 0 0 0.8 0 0 0 0 0 ESK STATE FOREST R531 40431 -27.2881 152.3333 ------14.013 ------EUMUNDI - CRESCENT RD 40078 -26.4764 152.9456 ------FERNY GROVE CEDAR CK RD 40703 -27.415 152.915 24.4 0 0 4.5 0 15.8 6.4 0 0 FERNY HILLS AUST WOOLSHED 40461 -27.3947 152.9308 0 0 0 9 0.8 13 4.9 5 0 FLAGSTONE CREEK 40302 -27.6242 152.1158 0 0 0 0 0 0 0 0 0 FORDSDALE 40395 -27.7181 152.1211 1.6 0 0 0 0 0 0 0 0 FOREST HILL POST OFFICE 40079 -27.59 152.3567 17.4 0 0 0 0 0 0 0 0

157 Station Station # Latitude Longitude 12 13 14 15 16 17 18 19 20 FORESTDALE STAPYLTON RD 40512 -27.6542 152.9856 14 0 0 0 0 12.8 2 0 0.6 FORTLAND 40402 -28.1181 152.8522 0 0 0 0 0 0.8 3.6 0 0 FRANKLYN VALE 40374 -27.7594 152.4564 0.5 0 0 0 0 0 0 0 1 FRASER ISLAND EURONG 40478 -25.505 153.1292 14.2 35 0 0 0 27 19.2 3 4 GALLANGOWAN FORESTRY 40122 -26.4289 152.2928 30 -- -- 4.03 5.8 4.8 0.4 1.4 0 GATTON ALLAN STREET 40083 -27.5425 152.2981 0 0 0 0 0 0 0 0 0 GATTON QDPI RESEARCH STN 40436 -27.5456 152.3286 7.6 0 0 0 0 0 0 0 0.2 GLEN CAIRN 40080 -28.2619 153.0178 0 0 0 3.8 0 16.8 8.8 4.8 2.8 GLENAPP 40404 -28.2633 152.88 0.6 0 0 0 0.6 8.2 16.4 9.4 3.4 GLENAVEN 40301 -27.1881 151.9633 0 0 0 0 0 0 0 0 0 GOLD CREEK RESERVOIR 40230 -27.4606 152.8811 12.4 0 0 9.6 1 28.4 3.8 1.4 1.8 GOODGER STORE 40086 -26.6675 151.8169 ------GOODNA AMPOL 40226 -27.6081 152.8975 7.6 0 0 2.6 0 5 0 0 0 GOOMBOORIAN WEBSTER RD 40089 -26.0636 152.7739 37.4 0 0 22 11.6 20 9.8 2.6 4.4 GOOMERI POST OFFICE 40090 -26.1828 152.0678 17 0 0 0 0 0 0 0 0 GRANDCHESTER SYMES ST 40091 -27.6597 152.4675 ------GRANGE BOWLING CLUB 40488 -27.4236 153.0181 20 0 0 0 11.3 23.2 0 0 0 GREEN MOUNTAINS 40182 -28.2322 153.1361 0 4 1 12.5 6 49 34 24 11 GREENBANK THOMPSON ROAD 40659 -27.6958 152.9408 19.6 0 0 2.2 0.2 3.6 0 0.6 0 GREENSLOPES PRIVATE HOSPITAL 40383 -27.5119 153.045 14 -- -- 4.53 5.5 18 8 0 -- GREYSTONLEA 40670 -26.5639 151.4383 0 0 0 0 0 0 0 0 0 GUNALDA POST OFFICE 40682 -25.9992 152.5639 0 0 4.9 7.2 5.1 1 1.7 0 0 GYMPIE 40093 -26.1831 152.6414 17 0 0 7.4 8 10.2 3.2 3 1.6 HARRISVILLE POST OFFICE 40094 -27.8117 152.6675 0.2 0 0 0 0 0.8 0.2 0 1.6 HATTONVALE STORE 40095 -27.5675 152.4628 10 0 0 0 0.6 1.6 0 0 3.4 HELIDON POST OFFICE 40096 -27.5503 152.1236 0 0 0 0 0 0 0 0 0 HERVEY BAY WILDLIFE PARK 40765 -25.2903 152.8153 6.4 10.2 0 10.4 18.2 34 10.2 2 0.8 HIGHVALE 40693 -27.3789 152.8161 ------HINZE DAM 40584 -28.0481 153.2875 0 0 0 5 4 14 4 4.5 1 HOME PARK TM 40833 -25.7683 152.525 ------HOMELEIGH 40493 -27.7797 152.535 0 0 0 0 0 0 0 0 0 HOWARD POST OFFICE 40098 -25.3181 152.5625 48 2.2 0 7.2 6.8 12 4.2 1 2.6 IMBIL FORESTRY 40100 -26.4622 152.665 19.2 ------15.8 19 4 6.2 0 IMBIL POST OFFICE 40099 -26.4594 152.6761 19 0 0 9.8 6.6 22 4.2 6 0 INALA BCC 40530 -27.5853 152.9883 8 1 0 3.5 1 13 1 1.5 2 INDOOROOPILLY BOWLS CLUB 40229 -27.4992 152.9769 12.4 0 0 3.4 0.8 14.2 4.2 1.2 0 INNISPLAIN 40156 -28.1808 152.9411 0 0 0 4.8 0 0 0 0 0 INSKIP POINT LIGHTHOUSE 40210 -25.8167 153.05 ------IPSWICH 40101 -27.6117 152.7608 4.1 0 0 0.7 0 0.2 0.8 0.8 1.2 JIMBOOMBA MILLSTREAM ROAD 40768 -27.8719 153.0147 2.2 0 0 0 0 2.4 0 1 2.2 JIMEAL MONOGORILBY 40708 -26.0336 151.0603 0 0 0 0 0 0 0 0 0 JIMNA COMPOSITE 40102 -26.6656 152.4594 16.4 0 0 -- 17 29 4 11.2 0.5 KALINGA BOWLS CLUB 40222 -27.4117 153.0456 8 16 0 18.6 1.6 19.4 7.2 0 0 KANDANGA POST OFFICE 40105 -26.3869 152.6767 27.8 0 0 11.2 5.4 12 2 1.6 0.8 KANDANGA UPPER 40389 -26.3967 152.6147 ------KARRAGARRA ISLAND 40269 -27.6333 153.3667 6.4 5.2 0.4 22 31.2 39 41.8 15.6 0 KENILWORTH TOWNSHIP 40106 -26.5947 152.7228 8 0.6 0 17 6 15.8 13.2 4 0 KENMORE WAR VETS HOME 40295 -27.5292 152.9142 0 0 0 0 3 10.2 2.4 0.8 0 KEPERRA COUNTRY GOLF CLUB 40476 -27.4094 152.9525 0 0 0 4.4 5.6 9.4 8.4 4.8 -- KHOLO 40108 -27.5519 152.7478 14.6 0 0 2 0 0 0 8 3.2 KIA ORA SANDY RIDGES 40109 -26.5181 152.0139 ------KIAMBA 40525 -26.5936 152.9036 0 0 0 38 19.4 38.8 16.2 0 0 KILCOY POST OFFICE 40110 -26.9425 152.5647 26.2 -- -- 11.03 0 6 0 0 --

158 Station Station # Latitude Longitude 12 13 14 15 16 17 18 19 20 KILKIVAN POST OFFICE 40111 -26.0861 152.2381 5.2 0 0 2.6 2.2 0 0.4 0.2 0 KILLARA 40671 -26.2839 151.1897 0 0 0 0 0 2.8 0 0 0 KIN KIN POST OFFICE 40438 -26.26 152.8683 8.4 0 0 21 30.2 31.3 22 0 0 KINGAROY PRINCE STREET 40112 -26.5544 151.8456 0 0 0 1 0.1 0 0 0 0.2 KOORALGIN 40304 -26.9475 151.9567 3 0 0 1.6 0 0 0.2 0 1.2 KUMBIA POST OFFICE 40113 -26.6894 151.655 0 0 0 0 0 0 0 0 0 KURABY BEENLEIGH ROAD 40283 -27.6125 153.0989 ------LAIDLEY POST OFFICE 40114 -27.6322 152.3919 7.4 0 0 0 0 0 0 0 0 LAKE MANCHESTER 40115 -27.4914 152.7519 14 0 0 2 0 3 3.4 0 1.2 LAMB ISLAND PINE AVE 40851 -27.6294 153.3772 6.2 3.2 0 9.8 62.6 29 26 21 4 LANARK 40071 -26.3828 151.1922 0 0 0 0 0 3.6 0 0 0 LANDSBOROUGH POST OFFICE 40117 -26.8081 152.9642 31.2 -- -- 40.63 14.4 52.2 38 22.4 0 LINDFIELD 40247 -26.8422 152.5803 34.8 0 0 7.4 12.6 9.8 1.6 4.4 0.4 LINVILLE 40387 -26.8242 152.2742 ------LITTLE NERANG DAM 40524 -28.1467 153.285 2 0 0 10 5 44 17 12 1 LITTLE YABBA SFR 274 40118 -26.6236 152.6839 10.2 0 0 0 21.8 17 10 7.4 -- LOGAN CITY WATER TREATMENT 40854 -27.6836 153.1958 5.6 0 0 6 2.2 29.2 12.4 8.4 1.6 LOGAN VILLAGE TAMBORINE RD 40766 -27.7922 153.1069 3.6 0 0 3.6 0 9 1.8 0.8 0 LONG POCKET CSIRO LAB 40450 -27.5108 152.9975 14.8 0 0 3.6 1 14 4.6 1.4 0 LOWOOD DON ST 40120 -27.4622 152.575 15 -- -- 3.23 0 0.4 2.2 0 -- LUMEAH 40407 -28.0558 153.0328 0 0 0 0 0 0 3 0 7 LYNNDON PARK BOWLS CLUB 40368 -27.5247 153.0633 0 0 0 4.6 2 21.2 0 5.6 0 MACLEANS BRIDGE 40542 -27.7881 153.0158 12 0.4 0.2 0.8 0.2 5.2 0.2 0 0.2 MAIDENWELL 40287 -26.8469 151.7986 0 0 0 0.2 0 0.2 0 0 0 MALENY DENNING RD 40396 -26.7811 152.8211 36 0 0 15 18 19.4 25 12 0.8 MALENY TAMARIND ST 40121 -26.7528 152.8519 38 14 0 47 36 95 56 29 1 MANLY RAILWAY STATION 40231 -27.4567 153.1797 11 1.6 0 31.8 5.2 26.6 22.8 11 0 MAPLETON POST OFFICE 40123 -26.6225 152.8656 5 0.4 0 53.8 20 33 37 13.8 1 MARBURG - WARREGO HIGHWAY 40124 -27.5675 152.6053 9 0 0 0 2 2 0 0 0 MARODIAN HOMESTEAD 40469 -25.8694 152.3172 7.5 0 0 4 0 2.5 0 0 0 MAROON 40290 -28.1667 152.7167 0 0 0 0 5 0 8.8 3 1 MAROON DAM 40677 -28.1753 152.6553 0 0 0 0 0 1.6 2.5 1 0 MARSDEN 40704 -27.6833 153.1 ------MARYBOROUGH 40126 -25.5181 152.7111 7.8 0 0.2 23 9 15.6 1.2 1.2 1.4 MCKENZIE CREEK 40517 -27.1953 152.7539 20 0 0 11.8 9 3 5 1 0.6 MERMAID WATERS TIMANA AV 40724 -28.0497 153.425 0 8.8 0 25.8 10.8 25.2 42.6 4.8 1.4 MIAMI BARDON AVE 40417 -28.0717 153.4433 0 6.8 0.2 17 12 26.8 44.6 4.2 0.4 MINMORE 40385 -26.5539 151.6678 0 0 0 0 0 0 0 0 0 MOOGERAH DAM 40135 -28.0317 152.5517 6.8 0 0 0 0 1.2 0.8 0.4 0.3 MOOLOOLAH POST OFFICE 40136 -26.7661 152.9622 ------MOORANG 40400 -27.9064 152.4739 2.6 0 0 0 0 0 0 0 0 MORAYFIELD MARK ST 40774 -27.1036 152.9461 7.8 0 0 33 6.8 24 8.8 6.8 0.8 MORETON SUGAR MILL 40547 -26.6267 152.9567 4 -- -- 42.03 14 45 24 16 -- MOUNEFONTEIN 40138 -26.5094 151.5156 7.8 0 0 0 0 0 0 0 0 MOUNT BARNEY 40394 -28.2317 152.7833 0 0 0 0 0 0 26.4 7.6 0 MOUNT COTTON UNI FARM 40460 -27.6081 153.2381 8.8 0 0 19.4 12 37.5 39 8 0 MOUNT COTTON WEST 40141 -27.6158 153.2042 8.6 0 0 13.2 8.2 31.6 26.2 7 0 MOUNT GRAVATT BOWLS CLUB 40274 -27.5553 153.0797 16 0 0 5.2 2.2 21.6 8.6 3.6 0.8 MOUNT JOSEPH 40144 -25.7411 152.2364 26.6 0 0 5 2 1.8 4.2 1 0 MOUNT MC EUEN 40809 -26.2417 151.7453 30.4 0 0 0 0 0 0 0 0 MOUNT MEE FOREST STATION 40637 -27.0958 152.7011 25.1 0 0 37.1 10.7 18 10.3 9.5 2.8 MOUNT SYLVIA 40384 -27.7219 152.2239 ------MT ALFORD 40139 -28.0753 152.6169 ------

159 Station Station # Latitude Longitude 12 13 14 15 16 17 18 19 20 MT BAUPLE MAC FARMS 40470 -25.9064 152.5928 11.6 0 0 6.5 13.8 11.6 0 0 0 MT BERRYMAN 40310 -27.7239 152.3108 1 0 0 0 0 0 0 0 0 MT BRISBANE 40140 -27.1492 152.5781 0 0 0 3 1 5.6 0 0 0 MT COOT-THA ABQ 2 BCC 40533 -27.4664 152.945 14.1 0 0 4.1 1.5 13.6 4.9 1 0 MT CROSBY 40142 -27.5386 152.7997 12 0 0 2.4 0 2.6 2.4 0.2 2 MT GLORIOUS FAHEY RD 40308 -27.335 152.7714 13.8 0 0 12 2.8 19.4 9.2 2.4 2.2 MT MEE 40145 -27.0675 152.7808 19 0 0 33.2 15 18.6 16 7.2 0 MT MOWBULLAN 40435 -26.8833 151.6167 2 0 0 2.4 0.4 0.2 0 0 0.6 MT NEBO POST OFFICE 40147 -27.4 152.7883 0 0 0 8 0.5 13.6 3.4 0.4 0 MT STANLEY STATION 40674 -26.6614 152.2003 4 0 0 5 9.8 0 0 1.6 0 MT TAMBORINE FERN ST 40197 -27.9697 153.195 1.2 0 0 5.8 4.6 28.6 4.6 5.6 3.2 MT WHITESTONE 40397 -27.6692 152.1592 0 0 0 0 0 0 0 0 0 MUNDOOLIN 40150 -27.905 153.0933 0 0 0 0 4.6 0.4 0 0 0 MUNGAR JUNCTION 40151 -25.605 152.5892 13.4 0 0 13.4 14 12.2 1 2.4 0 MURARRIE ROAD CSIRO 40235 -27.4681 153.0975 12.2 0 -- 19.02 2 30 12.2 4 0 MURGON POST OFFICE 40152 -26.2425 151.9425 29 -- -- 0.63 2.8 0 0 0 0 NAMBOUR BOWLING CLUB 40157 -26.6208 152.9664 3.2 0.8 0 43.4 15 49.6 23.4 16 0 NAMBOUR DPI 40282 -26.6431 152.9392 4.8 -- -- 52.43 12.8 54 26 14.2 0 NANANGO EAST FOREST 618 40277 -26.6603 152.0711 0 0 0 9.4 1 0.4 2 0 0 NANANGO WILLS ST 40158 -26.6756 151.9939 7.6 0 0 3 0 1.8 0.6 0 1.4 NERANG GILSTON RD 40160 -28.0092 153.3175 0.2 0.2 0 15.8 6.6 28.4 10.4 3.8 1.8 NEW BEITH 40312 -27.7356 152.9442 -- -- 6.47 0.6 0 4.6 0 0.4 1.6 NUMINBAH 40550 -28.2467 153.2383 7 14 3 27 19 53 3 19 22 NUMINBAH STATE FARM 40162 -28.1633 153.2114 1.4 0 0 10.8 4.1 37.2 22.3 0 22.6 OAKWOOD 40870 -26.4472 152.4875 ------OCEAN VIEW 40536 -27.1375 152.8078 21.4 0 0 24 7.2 23.4 14.2 4.2 1 ORMEAU 40863 -27.7914 153.2739 3 0 0 8.4 5 35.4 11.4 3.4 1.6 ORMISTON COLLEGE 40770 -27.5156 153.2492 11.2 0 0 31 27.8 59.8 -- 127.4 0 OXENFORD 40166 -27.8936 153.3106 4.2 0 0 13.8 5.6 34.4 22.6 6.6 1.4 OXLEY GOLF RANGE 40463 -27.5544 152.9794 11.8 1 0 3.8 -- 18.42 3 1.2 3.8 PALEN CREEK CORRECTIONAL 40167 -28.3258 152.7694 ------PALMWOODS HOBSON STREET 40695 -26.6842 152.9589 5 0 49 12 61 26 14 0 0 PEACHESTER WOODFORD RD 40169 -26.8447 152.8806 70 0 1 16 15 0 60 12 0 PECHEY FORESTRY 40170 -27.3228 152.0533 1 0 0 -- 1.22 0 0 0 -- PERSEVERANCE DAM 40480 -27.2883 152.1242 3 0 0 2.8 0 2 0 0 1.2 PETRIE AUST PAPER MILLS 40171 -27.2692 152.9839 10.4 1 0 16 10.5 22 6.6 4.4 1 PLACID HILLS 40449 -27.5572 152.2308 2.8 0 0 0 0 0 0 0 0 POINT ARKWRIGHT 40783 -26.5483 153.0986 18.6 2.4 1.8 21.8 35.3 57 36 8.6 0.6 POINT LOOKOUT BOWLS CLUB 40175 -27.4275 153.5219 16.2 -- -- 43.23 15.2 17.4 12.6 6.8 0 POINTRO 40483 -28.1833 152.6667 ------POMONA POST OFFICE 40176 -26.365 152.8533 20 0 0 30 20.8 29.4 10.8 10 1.2 PROSTON POST OFFICE 40177 -26.1636 151.6039 0 0 0 0 0 0 0 0 0 RAINBOW BEACH 40856 -25.9 153.0886 12.8 5.8 0 11 17.6 46.8 44.4 9.8 1.8 RANGE VIEW 40317 -27.7508 152.6667 6.8 0 0 0 0 1.2 1.4 0 0 RATHDOWNEY POST OFFICE 40178 -28.215 152.8639 0 0 0 0 0 4 8.2 6.4 2.4 RAVENSBOURNE 40270 -27.3628 152.1594 2 0 0 1.8 0 0 0 0 6 REDCLIFFE COUNCIL 40697 -27.245 153.1006 7 1 0 15.8 12.4 16 15.4 12 0.4 REDGATE 40843 -26.2742 152.0256 16 0 0 3.4 0 0 0 0 0 REDLAND BAY GOLF CLUB 40853 -27.6022 153.2953 ------REDLAND BAY QLD UNI FARM 40291 -27.6192 153.3056 7.6 5.8 0 28 17.4 34.4 51.6 17.8 1.2 REDLANDS HRS 40265 -27.5278 153.25 11.6 -- -- 21.03 15.6 35.2 37.6 16 3.6 RHONDA 40447 -27.9906 152.4614 5.4 0 0 0 0 0 0 0 0 RIPLEY VALLEY 40314 -27.7189 152.8172 0 0 0 1 2 0 0.6 0.4 0

160 Station Station # Latitude Longitude 12 13 14 15 16 17 18 19 20 RIVERMEAD 40437 -27.4464 152.6447 92.4 0.4 0 2.8 0 1 4.2 0 3.2 ROCHEDALE SOUTH 40429 -27.5942 153.1181 8.4 0 0 7.8 4.2 21.4 10.4 6.8 0 ROCKY POINT SUGAR MILL 40319 -27.7347 153.3275 5 -- -- 5.03 10 27 54 13 0 ROMANI 40411 -27.8481 152.9061 2.8 0.6 0 1.4 0 2.4 0.6 1.4 1.6 ROSEVALE 40183 -27.8519 152.4797 1.4 0 0 0 0 0.6 0 0 0 ROSEWOOD MATTHEW ST 40184 -27.6361 152.5922 4.6 0 0 0.6 0 0.8 0.2 0 0 RUSSELL ISLAND 40185 -27.645 153.4 7.2 0 6.8 14 42.8 37.6 31.4 10 3.2 SALISBURY BOWLS CLUB 40240 -27.5519 153.0358 ------SAMFORD CSIRO 40241 -27.3617 152.8861 24 0 0 6.4 0.6 11.4 6 1.6 0 SAMSONVALE 40186 -27.2906 152.8214 12.2 0 0 8.2 7 2.4 6.2 0.8 0 SANDGATE POST OFFICE 40242 -27.3233 153.07 10 -- -- 23.03 6.4 27 0 6 0 SHAILER PARK OREGON DRVE 40715 -27.6511 153.1933 5.8 0 0 5.2 2.4 28 12.2 8 3 SHAMROCK MINE 40146 -26.22 152.2683 3.8 0 0 4.8 1.6 2.2 0.2 1.6 0 SIM JUE CREEK 40188 -27.2675 152.6275 9 0 0 5.2 0 14.8 4.8 4.2 0 SOMERSET DAM 40189 -27.1169 152.555 22 0 0 7 1.2 5.4 1 0 1.2 SOUTHPORT DRURY AVE 40190 -27.9869 153.4092 3.2 3 0 12 12.4 25 17 6.2 1.4 SPRING BLUFF RAILWAY STN 40421 -27.4636 151.99 0 0 0 0 0 0 0 0 0 SPRINGBROOK FORESTRY 40192 -28.2264 153.2786 9 -- 31 40 29 95 -- 51.02 -- SPRINGBROOK ROAD 40607 -28.2031 153.2717 4 14 5 23 13 86 21 31 12 SPRINGLEA 40130 -26.1692 151.8928 0 0 18 0 0 0 0 0 0 STRATHPINE COLONIAL DRIVE 40633 -27.2831 152.9583 12.6 0 0 21 9 22 6.6 5.6 0 SUNNYBANK BOWLS CLUB 40244 -27.5756 153.0583 10.2 0 0 3 2 17.6 5.6 2.6 0 TAABINGA 40280 -26.6661 151.7328 14 0 0 0.4 0 0 0 0 0 TALLEBUDGERA GUINEAS CK RD 40196 -28.1331 153.4167 0.4 2.6 0.8 28.2 19 28.6 28.2 12.2 1 TAROME 40198 -27.9764 152.5117 2.2 0 0 0 0 1 0 0 1.6 TARONG 40199 -26.7439 151.8433 8.4 0 0 0 0 0 0 0 0 TEDDINGTON WATERWORKS 40390 -25.6531 152.6633 9 0 0 15 17 21.4 1 3.2 2.2 TERRENE 40672 -27.5486 152.0239 0 0 0 0 0 0 0 0 0 TEWANTIN POST OFFICE 40264 -26.3919 153.0408 18 3.2 3 28.4 46 56 27.2 9.2 2.6 THE OVERFLOW 40497 -27.9319 152.8575 0 0 0 0 0 0 0 0 3.4 THEEBINE 40200 -25.9481 152.5419 23.2 0 0 15 7.8 5.2 2 1 0 TIARO 40203 -25.7289 152.5814 11 0 0 7.6 -- 18.42 0 -- 3.02 TIN CAN BAY COUNTRY CLUB 40272 -25.9283 152.9933 ------TOOGOOLAWAH POST OFFICE 40205 -27.0878 152.3756 25 -- -- 0.83 0 0 0.4 0 0 TOOGOOM 40412 -25.2469 152.6625 34 3.4 0 4.6 15 25.6 3.6 3 0.8 TOOLARA FORESTRY 40451 -25.9961 152.8339 9.6 -- -- 20.83 27 31 9.6 14 -- TOOMBUL BOWLS CLUB 40237 -27.3911 153.0628 14.8 0.4 0 41.4 4.2 23 8.6 5.6 0.8 TOOWONG BOWLS CLUB 40245 -27.4919 152.9933 12.4 0 -- 3.02 1.2 13 5 1 0 TOWNSON 40675 -27.9097 152.3886 3.6 0 0 0 0 0 0 0 0 TRAVESTON 40206 -26.325 152.7867 11.2 0 0 32 17.8 10.4 6.6 5 0 TUAN CREEK FOREST STN 40207 -25.6778 152.7928 6.6 0 0 15.4 16.8 25.6 8 2.6 1 UMARELLA 40289 -27.1161 151.8664 ------UNIVERSITY OF QUEENSLAND GATTON 40082 -27.5508 152.3358 6.4 0 0 0 0 0 0 0 0.2 UPPER MUDGEERABA WATER 40606 -28.1056 153.3289 0 0 0 8.6 6 38 15 6.4 -- UPPER TENT HILL 40388 -27.6342 152.2203 2 0 0 0 0 0 0 0 0 URANGAN HIBISCUS ST 40430 -25.2825 152.8994 5.6 11.6 0 18.4 26.4 28.4 20.4 2.4 0 VIEWMOUNT 40208 -27.5508 152.7258 14 2 0 1 0 0.4 4 0 2 VINCENT VALE 40307 -26.9656 151.7206 16.6 0 0 0 0 0 0 0 0 WALLEN WALLEN 40732 -27.5308 153.4186 7.8 -- -- 25.23 45 25.4 45.6 15.8 -- WAMURAN POST OFFICE 40343 -27.0403 152.8661 8 0 0 23.4 9 17.2 8.6 5.2 2 WANERVA 40246 -26.5539 151.4156 0 0 0 0 0 0 0 0 0 WEST HALDON 40424 -27.755 152.0814 4.2 0 0 0 0 0 0 0 0

161 Station Station # Latitude Longitude 12 13 14 15 16 17 18 19 20 WIDGEE 40583 -28.2714 153.0742 2.2 0 0 11 6 38 18 10.4 6.4 WIDGEE STATION HILL 40029 -26.1831 152.4828 0 0 6 2.6 6.8 0 0 0.8 0 WILSONS PEAK 40485 -28.25 152.5167 1.2 0 0 7 0 21.4 19.4 9 5.4 WONDAI POST OFFICE 40251 -26.3172 151.8756 9.4 0 0 0.4 0 0 1.8 0 0 WOODFORD BCC 40628 -26.9425 152.7608 15 0 0 4 10 11 8.5 3 0.5 WOODFORD ETONS LANE 40887 -26.9267 152.7031 14 0 0 6 8 10 3 0 0 WOOLOOGA 40365 -26.0544 152.3931 1.8 0 0 4 3.8 2.8 0 0 1.2 WOOROOLIN POST OFFICE 40255 -26.4103 151.8153 0 0 0 0 0 0 0 0 0 WUNBURRA 40534 -28.1558 153.2683 8 0 0 16.4 0 0 76 18 0 WYNNUM BCC 40532 -27.4247 153.1614 11.4 0.4 0 25.2 13.2 27.6 20.4 10 2 YABBA CREEK 40611 -28.3333 152.35 ------YABBA STATION 40486 -26.6167 152.5328 13 0 0 0 5 20 9 4.4 0 YANDINA POST OFFICE 40257 -26.5603 152.9556 7 -- -- 49.03 15 48 23 11 -- YARRAMAN POST OFFICE 40258 -26.8403 151.9803 4.4 0 0 1.4 0 0 0.5 0 0.8 YARRAMAN UPPER 40259 -26.8911 151.8953 0 0 0 4 0 0 0 0 0.4 YIELO 40492 -26.67 152.5131 18.2 0 0 0 12 20 16 0 0 ZILLMERE POST OFFICE 40263 -27.3589 153.0375 10.8 -- -- 29.43 15 29.2 0 7.4 0 ALSTONVILLE TROPICAL FRUIT RESEARCH STAT 58131 -28.8521 153.4556 6.2 13.8 3.2 20.8 20.8 30 6.2 10.6 13 BALLINA 58001 -28.8528 153.5691 0 24 12 45 30 22 10 10 3 BALLINA AIRPORT AWS 58198 -28.8353 153.5585 0 31 7.4 43.8 35 21.6 9.2 7.4 3.6 BENTLEY 58078 -28.7789 153.1122 9 0 0 16 3 39 18 16 2 BRAYS CREEK 58005 -28.3958 153.1929 1 12 8 14 7 20 18 16 11

BRUNSWICK HEADS BOWLING CLUB 58103 -28.5512 153.547 0 -- -- 49.03 44 27 7.6 0 -- BRUSHGROVE POST OFFICE 58006 -29.5665 153.0811 0.6 0.4 0 19.6 0 10 0.6 8.2 0.2 BYRON BAY 58009 -28.6388 153.6361 9.2 41.6 4.8 3.2 65.9 5.4 3.6 4.2 7 BYRON BAY 58007 -28.6369 153.588 0 0 0 61.8 73.6 13.4 2.4 7.6 0 CASINO AIRPORT 58063 -28.8755 153.0493 8.2 0.6 0.4 1.6 2.2 23.4 2.8 8 8.8 CHILLINGHAM 58011 -28.3125 153.275 0 13 1 10 5 25 8 10.5 4.5 CLUNES 58127 -28.7322 153.405 0 26 7 33 30 32 8 15 20 COPMANHURST 58073 -29.4945 152.7945 0 0 8 0 0 18 7.4 0 0 COPMANHURST POST OFFICE 58014 -29.5861 152.7759 9 0 1 1 0 11 10 5 0 CORAKI POST OFFICE 58015 -28.9875 153.2874 0 0 -- 18.22 6.2 41.8 7.8 9.8 0 CORNDALE 58133 -28.7178 153.3619 2 11.6 0.2 19.2 15.2 28.6 19.6 5.8 9.6 DAIRY FLAT 58194 -28.3811 152.7173 0 0 0 8 10 20 20 15.8 0 DOON DOON 58183 -28.5013 153.3059 0.8 11.6 5.6 15.4 11.4 36.2 19.2 30 13 DOON DOON 58019 -28.5311 153.3153 3.5 21 -- 51.02 49 74 29.1 44 25 DYRAABA CREEK 58022 -28.7567 152.8163 ------EDEN CREEK 58139 -28.5917 152.8958 0 0 0 13 0 49 23 16 10 ETTRICK 58088 -28.6724 152.9081 9.4 1.4 4 12 0 34.2 5.4 8.8 8.4 EUNGELLA 58193 -28.3538 153.293 ------FEDERAL POST OFFICE 58072 -28.6533 153.4542 ------39.011 48 26 15 23 -- GRAFTON LIGHTNING 58161 -29.7588 153.0278 ------GRAFTON OLYMPIC POOL 58130 -29.6833 152.9283 3.4 0 2.2 1.2 0 9.2 8.8 0.6 0.2 GRAFTON RESEARCH STN 58077 -29.6224 152.9605 ------GRAFTON SOUTH 58102 -29.7383 152.7842 2.2 0 0 0 0 7 6 4 0 GREEN PIGEON 58113 -28.4785 153.0862 2.4 8 1 20 37.6 54.4 26 28 16 GREVILLIA 58115 -28.3867 152.8778 1.8 1.4 2 9 8.6 37.6 46.8 25 8.2 GREVILLIA 58026 -28.4414 152.8296 0 0 0 13.8 2.4 35.4 20 12 0 HARWOOD ISLAND 58027 -29.4227 153.2533 0 1.6 1.2 5.4 1 45 5.6 18.2 2.6 ILUKA 58149 -29.4053 153.3526 0 4 3.4 -- -- 25.63 0 20.6 1 KANGAROO CREEK 58138 -29.8567 152.9025 0 0 3.6 0 0 6.4 15 1 0

162 Station Station # Latitude Longitude 12 13 14 15 16 17 18 19 20 KANGAROO CREEK 58074 -29.9333 152.8683 0 0 0 0 0 1.5 17.1 0 0 KINGSCLIFF 58137 -28.2583 153.58 2 11.1 5 15.2 124 7.6 2.6 3 4.6 KUNGHUR 58129 -28.4658 153.2632 0 6.8 10.2 10.4 8.4 24.2 18.2 17 10 KYOGLE 58146 -28.6211 152.9977 8 0 0.6 16.8 3.4 50.6 20 5 9.4 KYOGLE POST OFFICE 58032 -28.6225 153.0036 6.4 -- -- 21.03 3.4 47 20 12 -- LAWRENCE POST OFFICE 58033 -29.4967 153.1041 0 0 0 15.2 0 13 0.8 10.6 2.8 LILLIAN ROCK 58148 -28.5276 153.1519 2.6 8.6 2.4 18.8 19.2 44 10.4 21 26 LISMORE 58037 -28.807 153.2628 5.4 5.2 3.2 22.2 5.8 40.2 30.4 21.6 11 LOADSTONE 58141 -28.4119 152.9827 0.8 6 -- 14.02 13 38 30 27.5 16.5 MACLEAN 58038 -29.4521 153.2007 0.2 0.2 0 0.8 0 14 1.2 24.6 5.4 MEERSCHAUM VALE 58171 -28.9311 153.4139 2 11.2 2 12.2 11.6 26.8 6.4 11.2 25.4 MEERSCHAUMVALE 58135 -28.8926 153.4479 8 12 7 20 20 23 15 14.5 22 MOUNT NUMINBAH 58197 -28.2694 153.2386 5.5 11.1 1.9 14.3 4.5 25.4 11.8 14.3 10.2 MOUNT PIKAPENE FORESTRY 58039 -29.0375 152.6897 13 0 0 5 0 32 2 5 2 MULLUMBIMBY 58040 -28.5451 153.4948 1 44 1 12 32 27 4 18 7 MUMMULGUM 58004 -28.7856 152.7683 30 0 0 0 0 38 4 0 10 MURWILLUMBAH 58158 -28.3408 153.3784 4.8 10.8 1.6 20.4 14.6 35.8 8.4 11.4 6.8 MURWILLUMBAH 58036 -28.3106 153.2217 3 8 0 11 4 19 19 18 14 MURWILLUMBAH 58020 -28.2903 153.3642 9 12 3 15 0 60 10 7 3 NASHUA 58162 -28.7278 153.4622 ------NEW ITALY 58097 -29.1514 153.2792 ------NIMBIN 58125 -28.5467 153.2872 ------65.04 6 0 ------NIMBIN POST OFFICE 58044 -28.5966 153.2233 5 12.8 0 19.4 18 42 13 15 19 NYMBOIDA 58045 -29.9691 152.7265 13 0 1 0 0 5 0 0 0 PEARCES CREEK 58087 -28.7 153.5 ------PUMPENBIL 58054 -28.3644 153.1422 1 12.4 5 7.8 4.8 28.6 17.2 25.4 11.4 ROSEBANK 58070 -28.64 153.4122 1 25.6 9 39 50 16 28 13.4 20.4 ROSEBANK 58165 -28.6217 153.409 1.8 29 8 42.2 43 32.8 11.8 15.4 16.6 SOUTH GRAFTON POST OFFICE 58025 -29.7067 152.94 ------THE CHANNON 58147 -28.6696 153.2792 ------TOMEWIN 58067 -28.2406 153.3778 11 5 0 23 26 33 5 10 3 TWEED HEADS GOLF CLUB 58056 -28.2044 153.5489 0 13.8 3 21.2 33.4 15.2 23 4.2 1.8 TYALGUM 58109 -28.3652 153.1727 0 8 7 11 4 20 18 18 -- TYALGUM 58156 -28.3433 153.165 4 4 4 12 3 26 17 15 10 ULMARRA 58059 -29.6309 153.0286 0 0 0 ------UPPER COMMISSIONERS CRK 58182 -28.5067 153.3389 0 24 6 27 18 45 47 28 29 UPPER CRYSTAL CREEK 58150 -28.2583 153.32 11 9 3 25 12 53 11 12 7 UPPER MONGOGARIE 58192 -28.9883 152.8806 12 -- -- 4.43 -- 18.42 3.2 -- 10.42 WHIAN WHIAN 58060 -28.5988 153.3783 0 -- -- 89.03 40 -- 74.02 21 -- WHIPORIE POST OFFICE 58099 -29.2822 152.9888 5.6 0 2 7.6 0 21.6 2.4 16.6 0 WIANGAREE POST OFFICE 58195 -28.5063 152.9668 8.012 0 -- 16.02 -- 65.02 ------WOLLONGBAR 58093 -28.8194 153.4161 ------WOODBURN POST OFFICE 58061 -29.0713 153.3449 3.6 2.6 1 7.4 9.8 39.6 13.6 3.8 2.2 WOOLI BEACH 58080 -29.8694 153.2661 0 5 3 19 3 12 7 5 12 WOOLNERS ARM 58220 -28.705 152.8414 13 3.2 2.5 0.5 31 2.7 5 5.7 0 YAMBA PILOT STATION 58012 -29.4333 153.3633 1.2 2.6 5.6 2.4 7.4 34 2 17.6 7.4 (Data Source AWN (2009))

163 REFERENCES Abbott, MB 1979, Computational hydraulics: Elements of the theory of free surface flows, Pitman, London.

Beach Protection Authority 1996, Tropical Cyclone 'Roger' March 1993 Technical Report, Report No. BPA 30.

Brown, JM & Wolf, J 2009, 'Coupled wave and surge modelling for the eastern Irish Sea and implications for model wind-stress', Continental Shelf Research, vol. 29, no. 10, pp. 1329-42.

BTE (Bureau of Transport Economics) 2001, Economic Costs of Natural Disasters in Australia, 103, Canberra.

Callaghan, J & Helman, P 2008, Severe Storms on the East Coast of Australia 1770-2008, Griffith University, Gold Coast.

Charnock, H 1955, 'Wind stress on a water surface', Quarterly Journal of the Royal Meteorological Society, vol. 81, pp. 639-40.

Dean, RG & Dalrymple, RA 2002, Coastal Processes with Engineering Applications, Cambridge University Press, Cambridge.

DHI 2009a, MIKE 21 & MIKE 3 Flow model FM - Hydrodynamic and Transport Module - Scientific Documentation, 2009 edn, DHI, Horsholm.

DHI 2009b, MIKE 21 FLOW MODEL FM - Hydrodynamic Module - User Guide, 2009 edn, DHI, Horsholm.

Donelan, MA, Haus, BK, Reul, N, Plant, WJ, Stiassnie, M, Graber, HC, Brown, OB & Saltzman, ES 2004, 'On the limiting aerodynamic roughness of the ocean in very strong winds', Journal of Geophysical Research Letters, vol. Letter 31, no. L18306.

Dunn, SL, Nielsen, P & Madsen, PA 1999, 'Wave setup in jettied river entrances', in Proc. 14th Australasian Coastal and Ocean Engineering Conference, Perth, Western Australia, pp. 193-8.

Frank, NL & Husain, SA 1971, 'The deadliest tropical cyclone in history?', Bulletin American Meteorological Society, vol. 52, no. 6, pp. 438-44.

Garcia-Nava, H, Ocampo-Torres, FJ, Osuna, P & Donelan, MA 2009, 'Wind stress in the presence of swell under moderate to strong wind conditions', Journal of Geophysical Research, vol. 114, no. C12008, pp. 1-12.

Garratt, JR 1977, 'Review of Drag Coefficients over Oceans and Continents', Monthly Weather Review, vol. 105, pp. 915-29.

Glantz, MH 2008, 'Hurricane Katrina as a "teachable moment"', Advances in Geosciences, vol. 14, pp. 287-94.

Goodwillie, A 2008, Centenary Edition of the GEBCO Digital Atlas - User guide to the GEBCO one minute grid.

164 Hanslow, D, Nielsen, P 1992, 'Wave setup on beaches and in river entrances' Proceedings of the 23rd International Conference on Coastal Engineering, Venice, vol. 1, pp. 240-252.

Hanslow, DJ, Nielsen, P & Hibbert, K 1996, 'Wave setup at river entrances', Proceedings of the 25th International Conference on Coastal Engineering, Orlando, vol. 2, pp. 1335-1348.

Hardy, T, Mason, L & Astorquia, A 2004, Ocean Hazards Assessment - Stage 3 Report - Surge Plus Tide Statistics for Selected Open Coast Locations along the Queensland East Coast, Queensland Department of Natural Resources and Mines.

Harper, B 1977, Numerical Simulation of Tropical Cyclone Storm Surge Along the Quensland Coast - Parts I to X. Department of Civil and Systems Engineering, James Cook Univerity, November 1977.

Harper, B 1998, Storm Tide Threat in Queensland - History, predicition and relative risks, Department of Environment and Heritage.

Harper 2001, Ocean Hazards Assessment - Stage 1 Report - Review of Technical Requirements, Queensland Department of Natural Resources and Mines.

Harper 2004, Ocean Hazards Assessment - Project Synthesis Report, Queensland Department of Natural Resources and Mines.

Herbich, JB (ed.) 1990, Handbook of Coastal and Ocean Engineering, vol. 1, 3 vols., Gulf Publishing Company, Houston.

Kalnay, E., M. Kanamitsu, R. Kistler, W. Collins, D. Deaven, L. Gandin, M. Iredell, S. Saha, G. White, J. Woollen, Y. Zhu, M. Chelliah, W. Ebisuzaki, W. Higgins, J. Janowiak, K. C. Mo, C. Ropelewski, J. Wang, A. Leetmaa, R. Reynolds, R. Jenne, and D. Joseph, 1996, ‘The NMC/NCAR 40-Year Reanalysis Project’. Bull. Amer. Meteor. Soc., 77, 437-471.

Kamphuis, JW 2000, Introduction to Coastal Engineering and Management, vol. 16, Advanced Series on Ocean Engineering, World Scientific, Singapore.

Kim, SY, Yasuda, T & Mase, H 2008, 'Numerical analysis of effects of tidal variations on storm surges and waves', Applied Ocean Research, vol. 30, pp. 311-22.

Lane, EM, Walters, RA, Gillibrand, PA & Uddstrom, M 2009, 'Operational forecasting of sea level height using an unstructured grid ocean model', Ocean Modelling, vol. 28, no. 1-3, pp. 88-96.

LeVeque, RJ 2002, Finite-Volume Methods for Hyperbolic Problems, Cambridge University Press, Cambridge.

Lighthill, J 1998, 'Fluid Mechanics of Tropical Cyclones', Theoretical and Computational Fluid Dynamics, vol. 10, pp. 3-21.

Mastenbroek, C, Burgers, G & Janssen, P 1993, 'The Dynamical Coupling of a Wave Model and a Storm Surge Model through the Atmospheric Boundary Layer', Journal of Physical Oceanography, vol. 23, no. 8, pp. 1856-66.

McInnes, K, Hubbert, G, Oliver, S & Abbs, D 2000, Gold Coast Broadwater Study - Storm Tide Return Periods and 1974 Floodwater Modelling.

165 Moon, I-J, Kwon, J-I, Lee, J-C, Shim, J-S, Kang, SK, Oh, IS & Kwon, SJ 2009, 'Effect of the surface wind stress parameterization on the storm surge modeling', Ocean Modelling, vol. 29, no. 2, pp. 115-27.

MHL (Manly Hydraulics Laboratory) 1993, NSW Ocean Tide Levels Annual Summary 1992/93, Report MHL658, October 1993.

Nielsen, P 1989, 'Measurements of wave setup and the watertable in beaches', in Proc. 9th Australasian Conference on Coastal and Ocean Engineering, Adelaide, South Australia, pp. 275-9.

Nielsen, P, de Brye, S, Callaghan, DP & Guard, PA 2008, 'Transient dynamics of storm surges and other forced long waves', Coastal Engineering, vol. 55, no. 6, pp. 499-505.

Powell, MD, Vickery, PJ & Reinhold, TA 2003, 'Reduced drag coefficient for high wind speeds in tropical cyclones', Nature, vol. 422, pp. 279-83.

Salisbury, MB & Hagen, SC 2007, 'The effect of tidal inlets on open coast storm surge hydrographs', Coastal Engineering, vol. 54, pp. 377-91.

Simpson, RH & Riehl, H 1981, The Hurricane and its Impact, Blackwell, Oxford.

Smith, SD & Banke, EG 1975, 'Variation of the sea surface drag coefficient with wind speed', Quarterly Journal of the Royal Meteorological Society, vol. 101, pp. 665-73.

Sturman, AP & Tapper, NJ 1996, The weather and climate of Australia and New Zealand, Oxford University Press, Melbourne.

Terry, JP 2007, Tropical cyclones : climatology and impacts in the South Pacific, Springer, New York; London.

Tilburg, CE & Garvine, RW 2004, 'A Simple Model for Coastal Sea Level Prediction', Weather and Forecasting, vol. 19, pp. 511-9.

U.S. ACERC (Army Coastal Engineering Research Centre) 1977, Shore Protection Manual, vol. 1, U.S. Government Printing Office, Washington.

Vickery, PJ, Masters, FJ, Powell, MD & Wadhera, D 2009, 'Hurricane hazard modeling: The past, present, and future', Journal of Wind Engineering and Industrial Aerodynamics, vol. 97, no. 7-8, pp. 392-405.

Vries, H, Breton, M, Mulder, T, Krestenitis, Y, Ozer, J, Proctor, R, Ruddick, K, Salomon JC & Voorrips, A 1995, 'A comparison of 2D storm surge models applied to three shallow European seas', Environmental Software, vol 10, no. 1, pp. 23-42.

Weaver, RJ & Slinn, DN 2009, 'Influence of bathymetric fluctuations on coastal storm surge', Coastal Engineering, vol. 57, no. 1, pp. 62-70.

Wu, J 1982, 'Wind-stress coefficients over sea surface from breeze to hurricane', Journal of Geophysical Research, vol. 87, pp. 9704-6.

166 Xie, L, Liu, H & Peng, M 2008, 'The effect of wave-current interactions on the storm surge and inundation in Charleston Harbor during Hurricane Hugo 1989', Ocean Modelling, vol. 20, no. 3, pp. 252-69.

Yin, B, Xu, Z, Huang, Y & Lin, X 2009, 'Simulating a typhoon storm surge in the East Sea of China using a coupled model', Progress in Natural Science, vol. 19, no. 1, pp. 65-71.

INTERNET REFERENCES

AWN (Australian Weather News Daily climatic data archive BOM (Bureau of Meteorology) BOM 2009b, Tropical Cyclone BOM 2009c, Tropical Cyclones BOM 2009d, Daylight savings DERM (Department of Queensland's Coastal Policy DERM 2009b, Wave Monitoring DERM 2009c, Wave Monitoring ECMWF 2004, Newsletter 101 ECMWF 2009a, ERA-40 dataset ECMWF 2009b, Interim dataset GA (Geoscience Australia) 2009, 10/11/2009 Natural Hazards - Cyclone JTWC (Joint Typhoon Warning Hemisphere Best Track Data New Zealand Ministry for the Guidance Manual for local government in New Zealand

167 NWS (National Weather 22/03/2009 Service) 2009, NOMADS NCEP Data Server USGS (United States Geological Hazards: Hurricanes and Extreme Storms

OPEN ACCESS REPOSITORIES

EprintsUQ {http://eprint.uq.edu.au/} OAIster {http://www.oaister.org/} UQeSpace {http://espace.library.uq.edu.au/}

168 BIBLIOGRAPHIC REFERENCE OF THE REPORT CE162 The Civil Engineering Research Report series CE is a refereed publication published by the School of Civil Engineering at the University of Queensland, Brisbane, Australia.

The bibliographic reference of the present report is: STEWART, J, CALLAGHAN, D, and NIELSEN, P. (2010). "Tropical Cyclone ‘Roger’ Storm Surge Assessment" Research Report No. CE162, School of Civil Engineering, The University of Queensland, Brisbane, Australia, 171 pages (ISBN 978-1-74272-002-9).

The Research Report CE162 is available, in the present form, as a PDF file on the Internet at UQeSpace:

http://espace.library.uq.edu.au/

169 CIVIL ENGINEERING RESEARCH REPORT CE

The Civil Engineering Research Report CE series is published by the School of Civil Engineering at the University of Queensland. Orders of any of the Civil Engineering Research Report CE should be addressed to the School Secretary.

School Secretary, School of Civil Engineering, The University of Queensland Brisbane 4072, Australia Tel.: (61 7) 3365 3619 Fax : (61 7) 3365 4599 Url: http://www.eng.uq.edu.au/civil/ Email: [email protected]

Recent Research Report CE Unit price Quantity Total price STEWART, J, CALLAGHAN, D, and NIELSEN, P. (2010). AUD$10.00 "Tropical Cyclone ‘Roger’ Storm Surge Assessment" Research Report No. CE162, School of Civil Engineering, The University of Queensland, Brisbane, Australia, 81 pages (ISBN 9781742720029). CALLAGHAN, D.P., NIELSEN, P., and CARTWRIGHT, N. AUD$10.00 (2006). "Data and Analysis Report: Manihiki and Rakahanga, Northern Cook Islands - For February and October/November 2004 Research Trips." Research Report CE161, Division of Civil Engineering, The University of Queensland (ISBN No. 1864998318). GONZALEZ, C.A., TAKAHASHI, M., and CHANSON, H. (2005). AUD$10.00 "Effects of Step Roughness in Skimming Flows: an Experimental Study." Research Report No. CE160, Dept. of Civil Engineering, The University of Queensland, Brisbane, Australia, July (ISBN 1864998105). CHANSON, H., and TOOMBES, L. (2001). "Experimental AUD$10.00 Investigations of Air Entrainment in Transition and Skimming Flows down a Stepped Chute. Application to Embankment Overflow Stepped Spillways." Research Report No. CE158, Dept. of Civil Engineering, The University of Queensland, Brisbane, Australia, July, 74 pages (ISBN 1 864995297). HANDLING (per order) AUD$10.00 GRAND TOTAL

Note: Prices include postages and processing.

PAYMENT INFORMATION

1- VISA Card

170 Name on the card :

Visa card number :

Expiry date :

Amount : AUD$ ......

2- Cheque/remittance payable to: THE UNIVERSITY OF QUEENSLAND and crossed "Not Negotiable".

N.B. For overseas buyers, cheque payable in Australian Dollars drawn on an office in Australia of a bank operating in Australia, payable to: THE UNIVERSITY OF QUEENSLAND and crossed "Not Negotiable".

Orders of any Research Report should be addressed to the School Secretary.

School Secretary, School of Civil Engineering, The University of Queensland Brisbane 4072, Australia - Tel.: (61 7) 3365 3619 - Fax : (61 7) 3365 4599 Url: http://www.eng.uq.edu.au/civil/ Email: [email protected]

171