Quantitative Microbial Risk Assessment: a catchment management tool to delineate buffer distances for on-site sewage treatment and disposal systems in Sydney’s drinking water catchments

by

Katrina Jane Charles

A thesis

presented to the University of New South Wales

in fulfilment of the

thesis requirement for the degree of

Doctor of Philosophy

in

Environmental Engineering

Sydney, New South Wales, Australia, 2009

©Katrina Charles, 2009

Borrower’s Page

The University of New South Wales requires the signatures of all persons using or photocopying this thesis. Please sign below, and give address and date.

iii

Acknowledgements

I would like to thank all those people who have been supported throughout this thesis, especially Nick Ashbolt, Jack Schijven and David Roser for their support and supervision; Daniel Deere for his unfailing enthusiasm and love of science; and Christobel Ferguson for kicking me and guiding me as appropriate.

I am grateful to the funding of the Cooperative Research Centre for Water Quality and Treatment for the provision of support enabling me to undertake a study tour of the US, attend the IWA Young Researchers Conference 2004 and participate in the Pathogen Roadshow 2004. Furthermore, thanks to Rachael Miller, Fiona Wellby and Dennis Mulchahy for their help in organising the above.

I would like to thank the following who helped, guided and supported this work: Martin Krogh, Robert McGuinness, Steve Manson, Bob Banens, Alan Shea, Sanjay Athavale, Malcolm Hughes, Tony Paull and Gary Bownds from the Sydney Catchment Authority; Danielle Baker and the team at Ecowise Environmental; Cheryl Davies, Christine Kaucner, Lyn Menzies, Robbie Smith and the team at the Centre for Water and Waste Technology; and Eric Evers & the modellers group, and Peter Teunis at the RijksInstituute voor Volksgezondheid und Milieu; the NSW Department of Health and other stakeholders who participated in the workshop and other events; and the friendly Americans who showed me around Jirka Simunek (UCLA), Charlie Olson (NYDEP) and Claire Welty (University of Maryland).

I would like to gratefully acknowledge the contribution from:

• Danielle Baker and the team at Ecowise Environmental who undertook the field work (sampling and analysis, excluding phage) for papers V, VI, VII and VIII;

• Freya Souter and AMS labs who undertook the phage analysis in paper VI, for the inactivation experiments in paper VII, and the field experiments in paper VII and VIII; and

• Cheryl Davies who undertook the inactivation analyses and assisted with analyses of column samples in paper VIII.

iv

Dedication

This dissertation is dedicated to Adam, my family and Seymour and Nibbler.

To the Australian/New Zealand Onsite Sewage Industry for their support and cooperation.

v

Abstract

On-site sewage systems, such as septic tank-absorption trenches, are used by approximately 20 000 people who live within the catchments that supply Sydney’s drinking water. These systems discharge sewage, treated to varying degrees depending on the system type and level of maintenance, to the environment. This can result in contamination of drinking water supplies if systems are not designed or managed appropriately. The aim of the project was to develop a methodology to define appropriate buffer distances between on-site sewage systems and waterways in Sydney’s drinking water catchments, to ensure the protection of drinking water quality. Specific objectives included: identifying the current status of on-site sewage management; assessing the effluent quality and treatment performance of septic tanks, aerated wastewater treatment systems (AWTS) with disinfection and an amended material sand mound; and development of an appropriate methodology for delineating buffer distances and assessing development applications.

Viruses were used as a focus for delineating the buffer distances due to their mobility and robustness in the environment, and the potential health consequences of their presence in drinking water. A Quantitative Microbial Risk Assessment (QMRA) model was developed to calculate the cumulative impact of the on-site sewage systems in the Warragamba catchment based on data from literature and experiments, with consideration of virus loads from sewage treatment plants within the catchments. The model enabled consideration of what was a tolerable impact in terms of the resulting infections within the community. The QMRA the tolerable loads of viruses from the Warragamba catchment were 108 viruses per year in raw water and 104 viruses per year in treated water. A log reduction method was developed to facilitate individual site development assessments. This method was compared to other management approaches to development assessment: fixed minimum buffer distances of 100m, reducing failure rates to zero, and the use of a preferred system. Each of these methods had a limit for how much they could reduce virus loads to the catchment due to either failure or short buffer distances at some sites. While the log reduction method is limited by the failure rates, the method provides a quantitative measure of risk by which maintenance inspections can be prioritised.

vi

vii

Table of Contents Borrower’s Page...... iii Acknowledgements...... iv Dedication...... v Abstract...... vi Table of Contents...... viii List of Tables ...... x List of Figures...... xi Abbreviations and special names...... xii List of Papers...... xiii 1 Introduction: Pathogens in our drinking water? ...... 1 1.1 Sydney’s drinking water catchments ...... 4 1.2 Quantitative Microbial Risk Assessment...... 6 1.3 Dissertation structure ...... 7 2 Aims, Rationale and Approaches ...... 8 2.1 Aims...... 8 2.2 Research rationale by paper ...... 10 2.3 Research approaches used...... 14 3 On-site sewage systems: Performance and Management...... 19 3.1 Management...... 20 3.2 Failure ...... 22 4 Virus fate and transport...... 24 4.1 Overview of virus inactivation and transport...... 24 4.2 Virus transport pathways from on-site sewage systems ...... 28 4.3 Modelling virus fate and transport...... 32 5 Quantitative microbial risk assessment ...... 45 5.1 Hazard identification...... 45 5.2 Exposure assessment...... 46 5.3 Dose-response...... 55 5.4 Application to catchment management...... 56 6 Short Summary of Results ...... 59 6.1 Paper I - Australasian standards for on-site sewage management ...... 59 6.2 Paper II - Buffer distances for on-site sewage systems...... 59 6.3 Paper III - Designing on-site sewage disposal systems ...... 60 6.4 Paper IV - Impacts of centralised versus decentralised systems...... 61 6.5 Paper V - Effluent quality from 200 on-site sewage systems ...... 62

viii

6.6 Paper VI - Disinfection in Aerated Wastewater Treatment Systems...... 62 6.7 Paper VII - Fate and transport in a mound system...... 63 6.8 Paper VIII - Virus fate and transport: laboratory and field studies...... 64 6.9 Paper IX QMRA: buffer distances for septic systems ...... 65 7 Discussion ...... 68 7.1 System performance...... 70 7.2 Buffer distance modelling...... 73 7.3 On-site sewage system management options...... 78 7.4 Recommendations & Further Research ...... 82 8 Conclusions...... 87 9 References...... 88 Paper I Australasian Standards for on-site sewage management: Application in the Sydney drinking water catchments ...... 102 Paper II Buffer Distances for On-Site Sewage Systems in Sydney’s Drinking Water Catchments ...... 118 Paper III Designing on-site sewage disposal systems to protect public health...... 132 Paper IV Centralised versus decentralised sewage systems: a comparison of pathogen and nutrient loads released into Sydney’s drinking water catchments...... 145 Paper V Effluent quality from 200 on-site sewage systems: Design values for guidelines 160 Paper VI Disinfection Performance in Aerated Wastewater Treatment Systems ...... 173 Paper VII Fate and transport of viruses during sewage treatment in a mound system... 186 Paper VIII Virus fate and transport from onsite sewage systems in Sydney drinking water catchments: laboratory and field studies...... 213 Paper IX Quantitative microbial risk assessment modelling to aid management of onsite sewage systems in Sydney’s drinking water catchment, Australia...... 248

ix

List of Tables Table 1 Types of on-site sewage treatment systems by local government area (Paper I)... 20 Table 2 Comparison of Design Criteria for on-site land application areas in Australia.... 21 Table 3 Summary of assumptions and data for household infection and excretion (Paper IX)...... 49 Table 4 Summary of assumptions and data for on-site sewage system performance and failure (Paper IX)...... 50 Table 5 Summary of assumptions and data for drinking water treatment (Paper IX)...... 55 Table 6 Summary of scenarios...... 57 Table 7 Conditions and actions for improved virus management for on-site systems ...... 61 Table 8 Comparison of actual septic tank effluent nutrient loadings with guideline predictions ...... 63

Table 9 Log10 virus load per year from Wingecarribee region, as modelled for the drinking water offtake at Warragamba Dam, under different management scenarios...... 67

x

List of Figures Figure 1 Sydney Catchment Authority area of operation (Sydney Catchment Authority, undated) ...... 5 Figure 2 Virus pathways for septic tank and absorption trench system...... 28 Figure 3 Virus pathways for AWTS with irrigation...... 29 Figure 4 Comparison of two temperature-dependent virus inactivation models, one devised by Yates and Yates (1987a) and the second employed by Beavers and Gardner (1993), with inactivation coefficients for a range of viruses published by Schijven and Hassanizadeh (2000) and Yates (2002)...... 36 Figure 5 Schematic of the risk assessment structure and development (Paper II) ...... 47 Figure 6 Sydney Drinking Water Catchments by subcatchment (Ferguson, 2005)...... 48

Figure 7 Log10 Pinf for from drinking water ingestion based on virus input from on-site sewage systems with (black) and without (grey) the inputs from STPs (percentiles indicated are for with STP)...... 66 .

xi

Abbreviations and special names AWTS Aerated Wastewater Treatment System

DLG NSW Department of Local Government

DoH NSW Department of Health

OSMS On-site sewage management system

QMRA Quantitative Microbial Risk Assessment

SCA Sydney Catchment Authority

USEPA US Environment Protection Agency

UV Ultraviolet

xii

List of Papers

Paper I Charles, K., N. Ashbolt, D. Roser, D. Deere and R. McGuinness (2001). "Australasian standards for on-site sewage management: Implications for nutrient and pathogen pollution in the Sydney drinking water catchments." Water Journal of the Australian Water Association 28(2): 58-64.

Paper II Charles, K., D. Roser, N. Ashbolt, D. Deere and R. McGuinness (2003). "Buffer distances for on-site sewage systems in Sydney's drinking water catchments." Water Science and Technology 47(7-8): 183-189.

Paper III Charles, K. J., J. F. Schijven, C. Ferguson, D. J. Roser, D. A. Deere and N. J. Ashbolt (2003). Designing on-site sewage disposal systems to protect public health. On-site '03 Future directions for on-site systems: Best management practice, Armidale, Lanfax Labs: 101-108.

Paper IV Charles, K. J., N. J. Ashbolt, C. Ferguson, D. J. Roser, R. McGuinness and D. A. Deere (2004). "Impacts of centralised versus decentralised sewage systems on water quality in Sydney's drinking water catchments." Water Science and Technology: 48(11-12):53-60.

Paper V Charles, K. J., N. J. Ashbolt, D. J. Roser, R. McGuinness and D. A. Deere (2005). "Effluent quality from 200 on-site sewage systems: design values for guidelines." Water Science and Technology 51(10): 163-169.

Paper VI Charles, K. J., N. J. Ashbolt, D. A. Deere and D. J. Roser (2003). Disinfection in Aerated Wastewater Treatment Systems. Ozwater Convention and Exhibition: Innovations in Water, Perth, AWA.

xiii

Paper VII Charles, K. J., F. C. Souter, D. L. Baker, C. M. Davies, J. F. Schijven, D. J. Roser, D. A. Deere, P. K. Priscott and N. J. Ashbolt (2008). "Fate and transport of viruses during sewage treatment in a mound system." Water Research 42(12): 3047-3056.

Paper VIII Charles, K. J., C. M. Davies, D. L. Baker, C. J. Charles, F. C. Souter, N. J. Ashbolt, J. F. Schijven, P. K. Priscott and D. A. Deere (Submitted). "Virus fate and transport from on- site sewage systems in Sydney drinking water catchments: laboratory and field studies." Submitted to Environmental Science and Technology.

Paper IX Charles, K. J. and N. J. Ashbolt (Submitted). "Quantitative Microbial Risk Assessment: a catchment management tool to delineate setback distances for septic systems." Submitted to Journal of Water and Health.

Page xiv

1 Introduction: Pathogens in our drinking water?

In 1997, an outbreak of Hepatitis A in NSW resulted in the reporting of 467 cases of hepatitis A and the death of one person (Conaty et al., 2000). The source of the outbreak was determined to be oysters from Wallis Lake, NSW. While the lake had several potential sources of faecal contamination, on-site sewage systems were suspected as the likely source. The following year, Cryptosporidium and Giardia were detected in Sydney’s drinking water supply resulting in three boil water alerts being issued, although no incident of associated disease was reported. The inquiry into Sydney’s incident again highlighted the lack of knowledge on pathogens from on-site sewage systems (McClellan, 1998). Such contamination incidents in Australia and internationally have highlighted the need for research into pathogen sources in catchments, including the role of on-site sewage systems.

On-site sewage systems provide collection, treatment and disposal of household sewage. In Australia and many other parts of the world, they are commonly used where connection to a centralised sewerage system is not available. There is a range of contaminants of concern in on-site sewage system effluent, including nutrients and pathogens. When there is an infection in a household, the effluent may contain human pathogens including bacteria (e.g. Salmonella, Campylobacter), enteric viruses (e.g. rotavirus, norovirus, hepatitis A virus), and parasitic protozoa (e.g. Cryptosporidium, Giardia). Through a variety of processes in on-site sewage treatment systems, and through treatment in the soil, these systems can effectively treat sewage on-site, reducing pathogens, nutrients and other contaminants to acceptable levels. However, when they are poorly designed or when there is a failure in the system there is a risk of contamination of source drinking water. Despite the decentralised nature of these systems, this risk of contamination of water resources needs to be managed on a centralised basis such as through planning, inspections and maintenance. Effective centralised management is particularly important where contamination of water resources can have significant consequences, for example in drinking water supply catchments.

In the Sydney drinking water supply catchments, viruses were considered the class of pathogens of highest concern in on-site sewage system effluent (Paper I). Human

Page 1

enteric viruses are the only contaminant of concern whose source is almost exclusively human sewage. Nutrient sources include fertilisers and animal faeces. Pathogenic bacteria and protozoa are commonly zoonotic, meaning that they infect animals that may spread the infection to humans, typically via animal faecal contamination of water and/or food. Human enteric viruses (viruses) are considered the most mobile and may also be persistent in the environment (Ferguson et al., 2003). Their small size relative to other pathogens increases the transport pathways available to them, both overland and through the soil or aquifer matrix (Davies et al., 2005).

Internationally, on-site sewage systems have been attributed to outbreaks from noroviruses (Anderson et al., 2003), Cryptosporidium spp. (Hrudey et al., 2004), and E. coli (Health Canada, 2000) from drinking water. Pathogens from septic tank effluent are known to be responsible for many instances of well contamination in the USA (Bechdol et al., 1994). In the United States septic systems were reported to be responsible for 11 % of a total 320 waterborne disease outbreaks between 1971 and 1980 (Craun, 1993 in Curry, 2000).

The use of sand and soil to treat effluent is widely used in applications such as sand filtration. On-site sewage systems commonly dispose of partially treated effluent underground to use soil as part of the treatment process. Movement of effluent across the surface can also provide additional sewage treatment, however, a higher degree of treatment is usually required before disposal in this case, due to the greater risk of humans coming into contact with the effluent. Research has shown soil treatment to be an effective method of removing nitrogen, bacteria, viruses and protozoa (Wilson et al., 1995). The distance between where the effluent is applied to the land and a water resource such as a river or groundwater well is called a setback or buffer distance. By defining a minimum distance to a groundwater bore or stream, an additional level of treatment is assumed in various international regulations.

A number of studies have reported the transport of pathogens from on-site sewage systems, via land-application of effluent, to groundwater (Yates, 1995b; Scandura et al., 1997; Curry, 2000). Such transport has been linked to drinking water outbreaks in the US (Scandura et al., 1997). Runoff from unsewered urban areas was reported to be the

Page 2

fourth most significant source of Cryptosporidium oocysts in the Sydney catchments (Swanson et al., 2000).

The human health impacts of living in areas serviced by septic tanks has been studied in the USA, where many unsewered areas also rely on groundwater for their drinking water. Epidemiological studies have shown increasing incidence of infectious diarrhoea and viruses detected in groundwaters with increased density of septic tanks (Lewis et al., 1993; Borchardt et al., 2003a; Lambertini, 2008). Lewis and Stark (1993) reported an increased risk of enteric virus infection with exposure to groundwater from wells located in subdivisions with on-site sewage disposal. Yates et al. (1985) reported outbreaks of hepatitis A, typhoid, Norwalk virus and echovirus that were traced back to contamination of drinking water wells with effluent from on-site systems.

While research has indicated that pathogens are removed by less than 0.6 m (2 ft) of soil transport (Nicosia et al., 2001), this research and numerous field studies has indicated that the removal is insufficient to protect groundwater (Curry, 2000). Studies have identified virus transport of up to 920 m at a sewage effluent irrigation area in New Zealand (Noonan et al., 1979), with vertical transport of up to 90 m (Powell et al., 2003).

Contamination of surface waters from on-site sewage systems has also been linked with rainfall events. A report for the Cooperative Research Centre for Water Quality and Treatment (Roser et al., 2007) highlighted that unsewered urbanised areas in Australia significantly impact surface water quality as a result of rainfall events, increasing microbial loads by a factor of between 18 for Cryptosporidium and 42 for Campylobacter, which was attributed to on-site sewage systems. Rainfall events reduce the performance of the effluent disposal system through increased hydraulic loading, as well as increasing the mobility of micro-organisms in overland runoff across grassed or bare soil surfaces (Davies et al., 2004; Ferguson et al., 2007). Rainfall can also increase the subsurface transport of viruses (Duboise et al., 1976; Nicosia et al., 2001).

There are generally no guidelines for virus concentrations in drinking and recreational waters, other than that they should be undetectable by culturing methods (WHO, 2004), which at the most sensitive is equivalent to a concentration of 1 per 100 L. The concentration of viruses in drinking water that is considered tolerable, where tolerable is

Page 3

defined as potentially resulting in less than one infection per 10 000 people per year, was calculated to be < 2 x 10-7 viruses per L (Regli et al., 1991). Hence, in drinking water catchments, the degree of virus reduction required (through inactivation, attachment and dilution) may be as high as 17 log10; being based on an infection in a household providing 1010 viruses per L (Paper III) and the tolerable level of 10-7 per L for an untreated groundwater. Other targets reported include a 7 log10 reduction required by the World Health Organization (WHO) (Yates et al., 1985) or a 4 log10 attenuation required by treatment systems under the US Groundwater Rule (USEPA, 2006). The low concentrations considered tolerable in drinking water are well below what can be routinely assayed, so other approaches are needed to estimate and manage site specific viral risks. The approach addressed in the research presented here was to apply quantitative microbial risk assessment (QMRA) that utilised a range of models to estimate virus fate and transport to aid in characterising virus exposure.

This dissertation deals with how to assess the contribution of viruses from on-site sewage systems to a drinking water supply catchment, and how to manage the impact of these systems through planning controls. It is implicit, although not directly addressed that controlling viral risks from on-site systems will likely lead to control of other pathogens from on-site systems.

1.1 Sydney’s drinking water catchments The Sydney Catchment Authority (SCA) area of operation (Figure 1) covers 16 000 km2, including land uses that range from protected forests to unsewered urban areas.

These catchments provide potable water for over four million people. The SCA aims to protect water quality in these catchments, including minimising risks to human health and the environment from pathogens and nutrients.

Page 4

Figure 1 Sydney Catchment Authority area of operation (Sydney Catchment Authority, undated)

Page 5

1.2 Quantitative Microbial Risk Assessment Quantitative microbial risk assessment is a formalised process based on the National Academies of Science’s chemical risk assessment paradigm (Haas et al., 1999). Most uncertainty in applying QMRA comes from the characterising pathogen exposure (Medema et al., 2006). Relevant to estimating virus exposure are the various mathematical models that have been used to predict virus fate and transport in groundwater. Yates et al. (1986) described virus transport in terms of groundwater velocity and temperature dependent inactivation. More recent models included consideration of virus sorption (Yates et al., 1995; Schijven et al., 2002c), filtration (Pang et al., 2005) and fate and transport in the unsaturated zone (Yates et al., 1992). However, as models have increased in complexity, the amount of data required has also increased. The data required includes sorption and inactivation coefficients which will vary with soil type, temperature, pH, organic content, etc. (Schijven, 2001). Modelling virus fate and transport over larger scales (> 100 metres) presents further issues due to heterogeneity within soils and aquifers, which has lead to the recommendation for a stochastic approach (Rehmann et al., 1999).

The complexity and specificity of the data required for virus modelling can restrict the application of the models. Hence, a broader risk assessment approach to address the issue of managing the development of septic systems was identified as a research priority at the National Research Needs Conference: Risk-based decision making for on-site wastewater treatment (Cliver, 2001).

Risk assessment is a methodology that enables variability and uncertainty to be built into the model. Schijven et al. (2006) employed a risk assessment methodology combined with virus transport modelling to calculate the required setback distance around groundwater wells in the Netherlands, to achieve an annual level of risk per person of < 10-4 infections per year. This included a worst case scenario of inactivation as the sole mechanism for removal. Previously Faulkner et al. (2003) had employed a risk assessment methodology combined with virus transport modelling to calculate the probability of transport of viruses through the unsaturated zone, based on the assumption of no effective removal in the saturated zone. However, there appears to be no QMRA-based approach to advising the management of on-site systems reported in the literature.

Page 6

1.3 Dissertation structure The structure of this dissertation is based on the European model of presenting a summary of the work, including general discussion and conclusions, with more complete details of the experiments and modelling included in published and submitted papers included at the end of this dissertation. This dissertation therefore includes: in Chapter 2, a summary of the research aims, as well as the research questions addressed and approaches used in each of the papers; Chapters 3 and 4 expand on the literature review below, focusing on on-site sewage systems and virus transport and fate; Chapter 5 outlines the risk assessment model which is the basis for the methodology for delineating buffer distances; Chapter 6 summarises the results from the papers which provide supporting data for the risk assessment, as well as the risk assessment results; and Chapter 7 provides the discussion of the papers and the dissertation results. For complete details of the experimental methodologies used and the detailed results, please refer to the relevant papers at the end of this dissertation. At the front of each of these papers, is a contribution table to clearly define the role of the candidate and the other contributors.

The candidate was enrolled full-time from 2001 until 2005, and then part-time until submission in 2007, and hence all the papers, including those from as early as 2001, are part of the candidate’s work during the period of enrolment.

Page 7

2 Aims, Rationale and Approaches

The basis of this dissertation was a call by the Sydney Catchment Authority to better understand pathogen risks from on-site sewage systems, given the local incidents of the hepatitis A outbreak at Wallis Lakes and the Cryptosporidium and Giardia crisis in Sydney’s drinking water discussed above. This provided a broad scope including issues of rates of infection and excretion, treatment system performance for common and alternative technologies, disposal system performance for common and alternative technologies, system management and system siting. In order to define the focus of the thesis, a workshop was undertaken in 2001 (Tierney, 2001) with industry stakeholders to identify and prioritise knowledge gaps within the New South Wales (NSW) water industry. Stakeholders included representatives from the Sydney Catchment Authority (SCA), academic researchers, and NSW Department of Health.

The workshop highlighted the problems with on-site system management in Australia, including many areas of low density systems making routine maintenance visits unfeasible and limitations on legal and financial recourse by councils to enforce system performance. In order to minimise the impact of these issues, the issues around setback distance and more generally siting of new systems were raised. The installation of new systems require a development application to be approved by the local council, and, in the SCA catchments, it will also be reviewed by the SCA. Appropriate siting of a system can take advantage of the capacity of the soil to provide reliable treatment. While there are several aspects to appropriate siting, it was decided that the research would focus on the capability of a buffer distance to protect water quality. The research reported here formed part of a program focused on providing data to aid catchment management and to prioritise faecal load rectification projects for the SCA.

2.1 Aims The primary aim of the research was to develop a methodology to delineate buffer distances between on-site sewage systems and waterways in Sydney’s drinking water catchments, to ensure the protection of drinking water quality. Whereas ‘set-back’ distance is assumed to refer to the absolute distance between the sewage disposal area and a waterway, the term ‘buffer’ was adopted to incorporate both this absolute distance and the capability of the land within the set-back distance to treat the effluent, such that

Page 8

the appropriate buffer distances will be a function of soil type, slope, depth to groundwater, effluent application rate and effluent quality. Viruses were selected on the basis of being the most mobile pathogen class in domestic sewage, and due to the ability to establish acceptable levels in water based on health outcomes. Nutrients were also considered a major concern due to their potential environmental impact, and hence were also included in experimental studies.

The workshop and initial research highlighted the knowledge gaps in the literature required for assessing the appropriate siting of on-site systems. Hence, after initial development of a methodology to delineate buffer distances, it was deemed necessary to address a number of supporting research questions in order to meet the thesis aims. For each research question, a paper was published (or submitted) that presented the results of the research, experiments and/or modelling that was undertaken to address that question. The culmination of these is Paper IX, the QMRA methodology which was developed based on the research and results of these previous papers. The specific research questions and associated papers were:

• What was the current status of on-site sewage management in Sydney’s drinking water catchments and more generally in Australia? (Paper I)

• What methodologies were available to define what are appropriate buffer distances? (Paper II)

• What design approaches will improve virus removal? (Paper III)

• What were the comparative loads of nutrients and pathogens to the catchments from decentralised (on-site) and centralised wastewater treatment systems? (Paper IV)

• What were the performances of septic tanks and Aerated Wastewater Treatment Systems (AWTS), and how do they compare with current guideline criteria? (Paper V)

• How effective were AWTS disinfection systems at removing pathogens? (Paper VI)

• What were the mechanisms and capability of an amended material sand mound for virus removal from sewage? (Paper VII)

Page 9

• What was the fate and transport of viruses in soils from Sydney’s drinking water catchments, and how did effluent and viruses behave in the field? (Paper VIII)

• Could a Quantitative Microbial Risk Assessment model be used to define appropriate buffer distances throughout a catchment, and assess management options for on-site sewage system installations on a catchment as well as a site by site basis? (Paper IX)

Based on the nine research aims described above, further context and research rationale (Section 2.2) and approaches (Section 2.3) are summarised, split between the nine papers that constitute this dissertation. The limitations of the research approaches for each individual paper are discussed in the appropriate paper, and are not repeated in any chapter except where relevant to the final QMRA.

2.2 Research rationale by paper

Paper I It was important to identify the current status of on-site sewage management, in the catchments and more broadly, to identify what management tools are available to aid in centralised management to mitigate contamination of water resources. Understanding the regulatory framework that exists provides the background for developing a new methodology for setting buffer distances.

Furthermore, as literature research identified limited information on the on-site sewage systems in the SCA area of operations, it was necessary to undertake a survey to identify the number and types of systems in use. While initiatives have been promoted by the NSW Health’s SepticSafe Program to develop databases of on-site sewage systems at a local government level, limited resources have restricted the collation of such information.

Paper II Concern about the potential water quality impacts of on-site sewage systems in Sydney’s drinking water catchments raised questions about how to manage on-site sewage systems to mitigate the effect on water quality. Powers were granted to the Sydney Catchment Authority to assess development applications within the catchments,

Page 10

which enabled consideration of planning tools in the management of on-site sewage systems. Buffer distances are one planning tool that can, depending on the site and soil conditions, provide high levels of effluent treatment with a high level of reliability. However, the variation between site and soil, as well as debate over the definition for what is a target level of treatment, can lead to difficulties in defining what distance is appropriate.

Paper III Design issues for on-site systems have traditionally focused on hydraulic and nutrient loading. Pathogens have been assumed to be adequately removed, as measured by faecal coliform removal in subsurface disposal and disinfection systems. However, viral and protozoan pathogens are generally more resistant to disinfection, and the small size of viral pathogens mean they are more mobile in subsurface environments. Given the lack of guidance for pathogen control, issues were highlighted that may lead to improved viral removal for on-site systems.

Paper IV A review of the sewerage infrastructure and performance of centralised sewage treatment plants for Sydney had been undertaken (CH2M HILL, 2001). However, little was known about the catchment wide contribution of nutrients and pathogens from the unsewered areas, and how that compared with the contribution from centralised sewage treatment plants. Paper IV reports research undertaken to assess potential pathogen and nutrient loads from centralised and decentralised systems within the Sydney drinking water catchments.

Paper V Values of effluent quality are typically taken from regulatory guidelines or standards, however as few effluent quality studies have been published, the information on which to base guidelines is limited. The definition of effluent quality is typically limited to the solids and organics content, with limited monitoring of nutrients or microbial composition. Effluent quality is a function of treatment system design, maintenance and the composition of the influent. Quality will vary over time due to diurnal fluctuations in household activity, time since last pump-out or maintenance visit, advances in on-site treatment technology and the changing nature of household chemical/detergent use.

Page 11

Effluent quality, including nutrients, is used as a basis for sizing effluent disposal areas. However, the NSW guidelines (DLG, 1998; NSW Health, 1998) differ from reported effluent quality from published research for septic tanks, secondary treatment systems and AWTS. In particular, several studies report poorer septic tank effluent quality than the NSW guidelines, the implication of which is an increase in the likelihood of disposal system failure. Similarly, studies of AWTS have reported high levels of non-compliance with the criteria for surface disposal.

For sustainable on-site sewage disposal, appropriate design values for effluent quality are required. This may require updating existing guidelines and approaches. For example, designs based on the average loads will potentially result in failure in approximately 50 % of systems due to underestimation of the solids and nutrient loading. Hence for design purposes, the average and/or median are not considered to be the most appropriate statistic. The design value for septic effluent should aid in reducing disposal system failure associated with contaminant overloading.

Paper VI The planning controls in the Sydney drinking water catchments, and more broadly in NSW, have resulted in a preference for AWTS due to the provision of secondary treatment and disinfection, and the use of spray irrigation reducing space requirements. However, the health implications of this policy had not been addressed. Disinfection system performance in AWTS were based on the principles of large centralised sewage treatment systems and the measurement of the removal of bacterial indicators, with no further experiments to verify pathogen removal or disinfection by-product formation in ultraviolet (UV) or chlorine systems.

Paper VII Amended sand mounds are based on a traditional sand mound design, with an amending material added to the sand to increase the removal of nutrients and pathogens. They require minimal maintenance, which makes them particularly attractive to planners and councils, including the Sydney Catchment Authority. In Western Australia, these systems are amended with pH-neutralised red mud (bauxite slag) and have reported excellent performance in terms of nutrients and bacterial indicators (Bowman, 1996) and viruses (Ho et al., 1991). When they were introduced to NSW, the amending

Page 12

material was changed to blast furnace slag, due to the lack of available red mud. And while high phosphorus removal capacities have been reported for blast furnace slag (Cheung et al., 2006), they are also quite variable. There was limited testing of the NSW design, with no data available on the life expectancy of amended sand mounds with regard to phosphorus removal, or the effect that the high pH of the system might have on pathogen fate and transport.

Paper VIII The capacity of soil to treat sewage effluent, removing nutrients and pathogens, is highly soil dependent. Hence, across the catchments there will be high variability in the removal of nutrients and pathogens within the buffer distance. Furthermore, no studies had been undertaken to assess the attachment and inactivation of viruses in soils from the Sydney drinking water catchment area.

Paper IX The SCA requires developments to have a neutral or beneficial impact on water quality (Sydney Catchment Authority, 2006). On-site sewage systems have the potential to have a detrimental impact on drinking water quality from the release of pathogens to the environment. However, how can we define what is a negligible impact of on-site sewage systems in terms of viruses, when there is limited information on virus concentrations in on-site sewage systems, on virus transport and on what are tolerable concentrations of viruses in waterways? Adequate siting of on-site sewage systems involves consideration of the cumulative water quality impacts of the existing systems across the catchment, as well as the impact of the potential increase in system density likely to occur in the future. Paper IX describes a method to quantify the cumulative impact of on-site systems on water quality, and applies this to the investigation of how to manage on-site sewage systems on a catchment wide basis through appropriate buffer distances.

Page 13

2.3 Research approaches used

Paper I A preliminary survey was undertaken by local government areas within the catchment boundaries to approximate the number and types of systems, and the management practices and experiences of the local environmental health council officers.

Paper II A combination of approaches was used to develop the methodological approach for delineating buffer distances. Stakeholder consultation was undertaken with SCA and other industry stakeholders to establish the management and research needs of the industry, including identifying resources and the needs of local regulatory authorities. A survey of effluent quality in 48 septic tanks was undertaken to provide probability density function data for all key constituents in effluent, including nutrients, potential effluent trackers and pathogens (adenoviruses, enteroviruses, reoviruses and coliphages by culture and noroviruses, rotaviruses and hepatitis A virus by PCR). A pilot study of groundwater contamination from on-site systems was undertaken to trial potential effluent trackers and to gather data on soil types, groundwater conditions and suitable analytes. And finally, a literature review was undertaken of previous approaches to delineating buffer distances and transport times for viruses.

Paper III A review of literature was undertaken to investigate virus fate and transport in the environment as it relates to on-site sewage disposal, specifically, the role of moisture content, organic matter, loading duration, rainfall and point of disposal on virus attenuation to inform the design and placement of sewage disposal areas. The paper assesses potential critical control points for minimising risk that may be applicable to larger developments and/or sensitive sites. And the potential implications of these management practices on the performance of the disposal areas with respect to hydraulics, phosphorus removal and nitrogen removal.

Paper IV Based on the results of effluent quality studies, a budget was undertaken to calculate the contribution of Cryptosporidium, viruses, nitrogen and phosphorus from centralised and

Page 14

decentralised sewage systems. Sewage treatment plant (STP) nutrient data, as collated from a variety of project reports, was sourced from CH2M HILL (2001). Septic tank supernatant samples were collected from the outlet of 40 septic tanks at residential and community or multi-user sites.

Two scenarios were examined for contaminant loads. The expected nutrient load from the STPs was calculated as the average dry weather flow (ADWF), converted to litres per year, multiplied by the average (or 50th percentile) concentration. The worst case load was calculated using the 90th percentile or expected maximum for each STP. The Cryptosporidium load was calculated as the ADWF, converted to litres per year, multiplied by the method recovery-adjusted concentration of Cryptosporidium per litre. The expected and worst case scenarios represent the ADWF multiplied by the median and maximum concentrations respectively for each STP. The expected STP virus load was calculated based on a 17 % occurrence of 105 viruses per litre, which were removed with 30 % efficiency within the STP, and a background concentration of 102 per litre for the remainder of the time. The worst case load was calculated on a 100 % incidence rate of 105 per litre.

Expected and worst case scenarios for decentralised systems were calculated from system flow, converted to litres per year, multiplied by the average or 90th percentile nutrient concentration for that system type to provide the expected and worst case values respectively. Pathogen loads were calculated similarly to nutrient loads, however incidence rates of 8 % and 50 % were assumed for Cryptosporidium and enteric viruses respectively, based on the on-site occurrence data.

Paper V A series of studies were undertaken to assess the performance of on-site systems, including septic tanks and AWTS with chlorine or ultraviolet disinfection. In total, 48 septic tanks and 143 AWTS were sampled over six separate projects. Analyses included suspended solids, biochemical oxygen demand, nitrogen, phosphorus and bacterial indicators. Grab sample and average time-series data were combined into a single database for analysis. The key issue was to provide sufficient data points to calculate probability density functions for the main guideline parameters.

Page 15

Paper VI Manufacturers of AWTS with chlorine disinfection were approached through the AWTS Manufacturers Association. The manufacturers of AWTS with UV disinfection were approached directly. Ten AWTS of each type were selected by Ecowise Environmental Pty Ltd (ECOWISE) and SCA from a list of fifteen sites supplied by the manufacturers. This ensured that the chosen systems sampled were at least one year of age and properly maintained. The time since last maintenance was up to the recommended maximum of 3 months.

Sampling and analysis was undertaken by ECOWISE. Pre-disinfection grab samples were taken immediately prior to disinfection, and post-disinfection grab samples were taken from the irrigation hose (UV) or pump well (Chlorine). The UV light was triggered 5 minutes prior to sampling to allow adequate warming up time for the lamp. Field sheets recording site-specific information, including the conditions of the chlorinator or UV lamp, were prepared for each site and photos taken. Samples were analysed in the field for pH, electrical conductivity (EC), temperature and free chlorine

(Free Cl2) for chlorine disinfection systems. Samples were analysed in the ECOWISE laboratory for suspended solids (SS), biochemical oxygen demand (BOD), total organic carbon (TOC), ammonia, thermotolerant coliforms, Escherichia coli, enterococci (ENT), Clostridium perfringens, and somatic coliphages. Samples were sent to Australian Water Quality Centre for laboratory analysis of trihalomethanes, chlorohydrate, and chloroacetic acid.

Additionally, one AWTS with UV disinfection and one AWTS with chlorine disinfection were selected for the experimental phage spiking with MS2 bacteriophage. Bacteriophage was added to the septic tank of each system. Sampling of effluent pre- and post-disinfection was undertaken to assess the reduction of faecal microorganisms in each disinfection system. At the time of MS2 application a small bottle of the same suspension was placed in the ground close to the AWTS tested, to act as an on-site control, providing information as to the longevity of the coliphage at that site. A sub- sample of this control and effluent samples from pre- and post- disinfection were taken at each sampling event, and were assayed on the day of sampling. All phage samples were tested on the day of sampling by AMS Labs, Sydney.

Page 16

Paper VII Experiments were undertaken on an amended material sand mound, installed in June 2001 near Sydney to treat sewage from four households and two public toilet blocks. Conservative tracers, bromide and Rhodamine WT, were spiked into the septic tank to measure actual hydraulic detention times. Timeseries sampling was undertaken fortnightly for 12 months, with samples of influent, effluent and within the mound collected, and analysed for a suit of chemical, physical and biological parameters. Statistical analyses for each sampling point were undertaken in SPSS (SPSS 11.5.2.1). Removal percentages were calculated in MS Excel (2002) as the difference between the influent and effluent concentrations at each sampling run.

Inactivation (timeseries) experiments were undertaken in the laboratory for MS2 and PRD1 bacteriophage. Water samples were collected from a sampling well in the sand mound located approximately 1.3 meters from the influent distribution trench. Fresh samples were collected, spiked with MS2 and PRD1 and stored at 4 ºC and 22 ºC. Samples were taken over 28 days. The inactivation experiments were designed to provide virus inactivation rates relevant to what may occur in the field and provide values required in the HYDRUS fate and transport model. Inactivation rates were calculated using Mathematica 4 for Students (v4.2.0.0, Wolfram Research, 2002).

The bacteriophages MS2 and PRD1 were spiked in the influent pump well, and their breakthrough into the influent distribution trench, sampling wells within the cell and effluent monitored in separate experiments. Due to limitations in the available loading data, removal was calculated as the reduction in peak concentration between the influent and the effluent. These results were compared with the results for somatic coliphages which were included in the timeseries sampling and analyses.

Investigations were undertaken into the persistence of viruses in the presence of the soil, and in raw and treated sewage. A column study was undertaken to simulate worst case transport, and field studies were undertaken to assess the performance of a system in- situ.

Paper VIII Virus behaviour in soils was determined through laboratory column and survival experiments in five catchment soils. Repacked column experiments are known to

Page 17

overestimate the removal of viruses in the field, and hence field experiments were undertaken at two of the sites that the soil was sampled from to establish removal at a field scale, and provide comparisons between column and field removal. Fields experiments were undertaken at two households within the drinking water catchments. Monitoring wells were installed up-gradient and down-gradient of the treatment and disposal systems. Septic tank effluent and groundwater were monitored routinely for nutrients and bacterial and viral indicator organisms. A spike of bacteriophage PRD1 was undertaken under simulated rainfall conditions to specifically monitor virus transport at these field sites. CXT-Fit and HYDRUS models were used to fit the breakthrough curves of the virus transport experiments and compare laboratory and field results.

Paper IX The approach chosen to investigate the cumulative virus loads from on-site sewage systems in Sydney’s drinking water catchments was QMRA, based on the data available in literature and from experiments reported in the previous papers. The QMRA model was developed using the MS Excel add-on @Risk (v4.1, Palisades Inc.) and incorporated modules on: household infection and excretion; system performance and failure; buffer distance transport through unsaturated and saturated soil, and in overland flow; sewage treatment plant inputs; catchment transport; removal in the water treatment plant; and pathogen dose-response. SCA data from their geographic information system (GIS) was used extensively to determine approximate system locations and a range of site specific data based on these locations.

Page 18

3 On-site sewage systems: Performance and Management

In Sydney’s catchments, 30 % of the 109 400 population live on unsewered properties (CH2M HILL, 2001). Septic tanks, which are the most common on-site sewage treatment system in Australia (Geary, 1992), consist of a single tank designed to provide a minimum of 24 hours retention of sewage (DLG, 1998) prior to subsurface disposal, commonly via absorption trenches. The retention of sewage provides preliminary treatment of sewage with sedimentation of solids and flotation of scum (e.g. oils and fats), with the retained solids and scum partially degraded by anaerobic digestion. Septic tanks can be enhanced with baffles, to maximise the duration of retention in the tank, and filters, to prevent overflow of solids.

Aerated Wastewater Treatment Systems (AWTS) consist of multiple chambers which provide secondary biological treatment with chlorine or ultraviolet (UV) disinfection. The treatment includes primary sedimentation of solids and flotation of scum, aerobic degradation of organic matter by bacteria and secondary sedimentation to clarify the effluent prior to disinfection. The higher level of treatment reduces the concentration of solids, organics and bacteria, allowing for a range of disposal options such as surface spray irrigation and drip irrigation.

The passage of sewage through soil also provides treatment. Soil treatment systems include sand filters and mound systems, and also natural soils where effluent is disposed. One of the key features of soil treatment is filtration of solids, as well as many of the nutrients and pathogens associated with solids. Usually, where sewage enters the soil, a “biomat” or clogging layer forms from the accumulation of organic matter, the carbon and nutrients in which provide the basis for a very active biomass. This has the additional benefits of reducing the rate of effluent infiltration into the soil, providing greater filtration than the soil alone and biological degradation of the effluent increasing nutrient and pathogen removal. Passage of effluent through unsaturated soil provides similar treatment to fixed film bioreactors (USEPA, 2002). Additionally, contaminants are removed during passage through saturated or unsaturated soil via attachment to soil and vegetation uptake.

Alternative on-site sewage systems include amended material mound treatment systems,

Page 19

composting toilets, vermiculture, hybrids, and greywater systems. Tanks designed to collect sewage that is subsequently removed via pump-out and transported to a remote location for treatment are also included due to the potential for impacts on catchment water quality where failure occurs.

The main on-site sewage treatment systems in the Sydney drinking water catchments (Table 1) are septic tanks mostly with absorption trenches and AWTS. Alternative systems include sand filters, amended material sand mounds and composting toilets. Disposal has traditionally been via an absorption trench for septic tank effluent or spray irrigation for AWTS effluent, however alternative forms of irrigation such as surface or subsurface drip irrigation are increasingly being adopted.

Human enteric viruses and parasitic protozoa (particularly Cryptosporidium oocysts), have been shown to survive common on-site treatment systems (Paper II).

Table 1 Types of on-site sewage treatment systems by local government area (Paper I)

Local Government Septic tanks AWTS Pump out Alternative Area (%) (%) (%) systems (%) Blue Mountains 58 9 33 0.2 Eastern Capital 90 10 - - Greater Argyle 92.4 7.6 - - Lithgow 70 30 - 0.2 Oberon 100 - - - Shoalhaven 71 20 9 0.2 Wingecarribee 79 20 1 0.4 Wollondilly 60 31 9 0.2 Wollongong 45 16 35 4.7

3.1 Management The regulations for on-site systems are generally enforced at the local government level. Local government provide planning permission, based on the available guidance such as the Australian/New Zealand Standard for On-site domestic wastewater management (Standards Australia, 2000) and, in NSW, the Department for Local Government’s Environment and Health Protection Guidelines (DLG, 1998). In accordance with the latter, local government are also responsible for getting landowners to register existing

Page 20

systems, including inspecting the system to ensure that adequate maintenance is being undertaken. As discussed above, the SCA have a role in the granting of planning permission for on-site systems in their area of operations.

AS/NZS 1547:2000 provides a basis for improvement in the sustainability of these systems through detailed site and soil assessment and management recommendations

(Paper I). However, regulation in Australia continues to be based on BOD5, SS and thermotolerant coliforms, with minimal consideration of nutrient or pathogen loads or removal. Furthermore, AS/NZS 1547:2000 does not provide guidance on set-back distances (Table 2).

Table 2 Comparison of Design Criteria for on-site land application areas in Australia

Design Criteria Australiaa NSWb Victoriac Queenslandd Western South Australiae Australiaf

Daily flow (unrestricted, 360+ 300 400 - 250† L per bedroom)

Buffer distances (m) (none)

Underground water tank 250 15 15/6*

Water supply bore 20 100 50 (minimum) up to 500 in Priority 1 areas Surface water 100 60 50/30/10 100 50

Permanent waters in 100- 100 100 supply catchment 150++

Intermittent waters in 100++ supply catchment

Reservoir or wetland 150++ 300 2000 50 (reservoirs), 50 (wetlands) a (Standards Australia, 2000); b (DLG, 1998); c (EPA, 2003); d ; e (Department of Water, 2006); + Flow per 2 people; † 125 L per person, minium design for 6 people ++ Sydney Catchment Authority requirements only; * primary / secondary / advanced secondary treated effluent.

Page 21

There is however, significant variation in advice on set-back distances between the states (Table 2). The set-back adopted by the SCA was 100 m from permanent or intermittent waterways, or 150 m from water storages and their main tributaries. Victoria (EPA, 2003) allows setback distances to be halved where there is advanced secondary treatment which is able to be enforced by council, and a slope of less than five percent. In Tasmania, buffer distances may be based on virus transport as calculated by TrenchTM (Australian Institute Environmental Health, Brisbane), with Queensland adopting a similar method in 2002 (NRM, 2002).

The Queensland Department of Natural Resources and Mines (NRM, 2002) also specified vertical separation distances to groundwater of 1.2, 0.6, and 0.3 m for primary, secondary and advanced secondary treated effluent. In contrast, most states regulate unsaturated soil depth as a potential hazard, for example Victoria (EPA, 2003) recommends a low risk rating for sites with a soil profile depth greater than 2 m, and high risk rating for less than 1 m.

3.2 Failure Failure of on-site systems includes the failure of the treatment system to comply with effluent quality criteria, mechanical failures in the treatment system, or failure of the disposal or land-application system to adequately treat and dispose of the effluent. The most common method of assessing this last failure mode is through visual inspection, and includes identifying surfacing effluent or areas of lush plant and weed growth.

A compilation of studies on the rate of failure for on-site systems was undertaken for Paper I. Absorption trench failure rates were between 30 % to 40 % for visible surface flow and 73 % for unsatisfactory performance (Geary, 1992; O’Neill et al., 1993). More recently, A survey in South Australia reported visible surcharge of effluent in 12 % of absorption trenches, with 50 % of septic systems reported to be under-performing (Arnold & Gallasch, 2001). In Paper V, the average loads of solids, organics and nutrients were shown to be considerably higher than the values that are recommended to be used as a basis for designing the disposal areas.

Failure rates of 95 % for AWTS have been reported (Coote, 1995), based on guideline effluent compliance. However, the performance of AWTS in the study reported in Paper V was significantly less (with 49 % compliance with the current requirements [NSW

Page 22

Health, 1998]), of which free chlorine concentrations alone accounted for 22 % of the overall non-compliance.

The primary causes of performance failure are poor design, including siting and sizing, and inadequate operation and maintenance, which are often caused by inadequate understanding of design limitations and failure modes, financial pressures and change of usage or owner.

The methods used for determining failure do not address groundwater contamination. Incidences of groundwater contamination from on-site systems, especially septic tanks, are often reported (Chen, 1988; Yates, 1995b; Scandura et al., 1997; Curry, 2000; Borchardt et al., 2003b), including several disease outbreaks (Craun, 1985; Anderson et al., 2003; Parshionikar et al., 2003; Fong et al., 2007). However, there are no guidelines for what is an acceptable level of contamination in groundwater by which failure can be defined. The effluent quality criteria were designed to reflect treatment performance, however there is inadequate data on what the implications are in terms of nutrients and pathogen loads. Furthermore, thermotolerant coliforms will not reliably indicate the presence of viruses and parasitic protozoa, as they are more susceptible to chlorination, filtering in soil matrices and adverse environmental conditions (Ferguson et al., 2003).

Prevailing standards and guidelines seek to address failure, but gaps in current knowledge limit the performance of these guidelines, particularly in relation to soil capacity to immobilise wastewater constituents and minimum setback distances from waterways.

Page 23

4 Virus fate and transport

The key aim of this dissertation was to develop a methodology to delineate buffer distances between on-site sewage systems and waterways in Sydney’s drinking water catchments, so ensuring the protection of drinking water quality based on virus transport. This chapter aims to summarise the key background research on viruses and their fate and transport as it relates to on-site sewage treatment and disposal.

4.1 Overview of virus inactivation and transport The fate and transport of viruses relevant to on-site sewage treatment systems, and the modelling thereof was reviewed and presented in Papers III, VII & VIII.

While generalisations about virus transport are drawn, it is important to remember that there is inevitably variability between the behaviour of different viruses (Goyal et al., 1979). For this reason a cocktail of coliphages is recommended for use in experimental work to cover a range of properties, and adoption of ‘worst case scenario’ virus properties suggested for use in modelling applications for planning purposes. Schijven and Hazzanizadeh (2000) recommends the use of at least two of the phages MS2, PRD1 and ϕX174. Additionally, bacteria such as E. coli and C. perfringens spores should be used, along with the more traditional thermotolerant (faecal) coliforms.

Inactivation (fate) Inactivation of viruses removes their ability to cause infections, and hence means that they no longer pose a risk to human health. There are a number of factors that influence inactivation rates: temperature, sunlight, desiccation/soil moisture content, aggregation/adsorption, microbial antagonism and pH. Reactivation of bacterial pathogens is also possible under certain conditions (Jagger, 1967; Alvarez et al., 2000), however, it is not considered relevant for viruses. Generally, inactivation rates can be assumed to be greatest in aerosols in the sunlight, then in overland flow, unsaturated soil and lowest in groundwater.

Temperature. Temperature is considered to be the most important factor in virus inactivation in the subsurface (Yates et al., 1985), with inactivation rates increasing with temperature. Schijven (2000) undertook a regression analysis of inactivation rates with temperature, which indicated that MS2 and other F-RNA coliphages and Echovirus

Page 24

were the most sensitive to temperature, whereas PRD1 phage, poliovirus and hepatitis A virus were the most resistant of the enteric viruses examined to temperature.

Sunlight. Kapuscinski and Mitchell (1983) observed significantly higher inactivation rates of E. coli and the bacteriophages MS2, ΦX174 and T7 in sunlight (sea water, 10 °C) than in dark conditions, with E. coli being the most susceptible, and MS2 the least. Sinton et al. (1999; 2002) reported somatic coliphages had the highest resistance to sunlight in seawater compared with other indicators, whereas F-RNA coliphages were the most resistant in freshwater.

Moisture Content. Desiccation of viruses is one of the more efficient inactivation methods. In the atmosphere, enteroviruses survive best at higher humidity (Haas et al., 1999). In soil, inactivation is increased by unsaturated conditions. Powelson et al. (1990) reported that despite poor attachment, removal of MS2 was greater in unsaturated than saturated columns, indicating increased inactivation under unsaturated conditions. The removal of MS2, PRD1 and poliovirus 1 were three times greater in unsaturated columns (Powelson et al., 1994). Air-water-solid interfaces exist under unsaturated conditions, which provide additional attachment sites with increases surface tension stresses. It is theorised that at these interfaces viruses are inactivated due to unfolding of coat proteins (Jin et al., 2000), in a similar way to the temperature effect, which is virus dependent. For example, Jin et al. (2000) reported MS2 to be more sensitive to decreases in moisture content than φX174.

Aggregation/Adsorption to particles. Viruses may be released from the host as aggregates within faeces. The aggregation of viruses can prolong their survival as most virion particles must be inactivated before the aggregate looses infectivity (Ferguson et al., 2003). Virus adsorption to soil or organic matter may increase survival, however, some surfaces may cause inactivation of the attached virus, such as oxides and hydroxides (Jin et al., 2002).

Microbial activity. Generally, inactivation rates are slower in sterile environments, although there are conflicting reports on the significance of this biotic effect. Sobsey (1980) reported inactivation of poliovirus and reovirus in settled sewage was generally slower under sterile conditions; as was the inactivation of poliovirus, echovirus and Hepatitis A virus in groundwater, primary and secondary effluents (Sobsey et al., 1986).

Page 25

However, Hurst et al. (1980) reported limited differences in sterile versus nonsterile soil. Yates’ et al. (1988) study of over 30 groundwater samples elicited no correlation of viral inactivation with the presence or absence of indigenous bacteria. pH. Most enteric viruses are stable between pH 3 and pH 9 (Yates et al., 1988), so pH should have little impact on most soil/groundwater systems. Nonetheless, MS2 is strongly affected by high pH (> pH 9.5), whereas PRD1 and indigenous somatic coliphages are more resistant to high pH (Logsdon, 1994; Governal et al., 1997).

In modelling the rate of inactivation, a first order or log-linear decay rate is commonly adopted: Ct = C0 exp (-µ t) where t is time, Ct is the concentration of viruses at time t and C0 is the concentration of viruses at time t = 0 and µ is the inactivation rate. The inactivation rate can be expressed as a function of temperature and other environmental parameters: Reddy et al. (1981) developed empirical models for inactivation of microorganisms that incorporate the effects of temperature, soil moisture, pH and adsorption. Hurst (1991) used linear and polynomial regression models to investigate variation of virus inactivation rate coefficients associated with various external controlling factors. Alternative models for environmental inactivation include biphasic (Cerf, 1977), time dependent (Anders et al., 2006) and the efficiency factor horn model (Thurston-Enriquez et al., 2003). The biphasic model describes non-linear behaviour, where there are two different rates of decay. This may represent aggregation (Gassilloud et al., 2005) or other differences (Cerf, 1977) within the virus population. The virus concentration, C, is described by C = C0 [α.Exp(-µ1.t) + (1-α).Exp(-µ2.t)], where C0 is the virus concentration at time = 0, t is time, α is the proportion of viruses that undergo rapid initial inactivation, and µ1 and µ2 are inactivation rates for the initial rapid inactivation and subsequent lower inactivation (Cerf, 1977; Petterson et al., 2000). Similarly, Anders and Chyrsikopolous (2006) developed a time dependent model, such that the rate of inactivation decreases with time.

Transport Virus transport is a function of the water flow rate and reversible and irreversible adsorption and is influenced by a range of factors including soil properties, heterogeneity and water quality (McKay et al., 1993; Rehmann et al., 1999; Schijven, 2001; Taylor et al., 2004; Foppen et al., 2006). Virus adsorption to surfaces can be

Page 26

either reversible or irreversible. While adsorption is the most important process for attenuation, inactivation and irreversible adsorption are required for removal. In subsurface flow, the flow rate will depend on the hydraulic conductivity of the soil, and similarly, adsorption rates will depend on the soil properties. In overland flow, the flow rate can be much greater than in the subsurface resulting in rapid transport over distance. Adsorption can still occur, such as on plants and soil, and can result in additional removal via sedimentation. The rate of adsorption of viruses, and hence their transport, at a particular site will be a function of effluent quality (e.g. pH, ionic strength, organic content, solids content, surfactants), and the soil (e.g. structure, pH, surface charge, organic matter), as well as hydrology and climate factors. Water flow, as well as transporting the viruses, can affect adsorption rates as increased flow velocity reducing contact time of viruses with the attachment sites and thereby reducing attachment (Schijven, 2001).

The effect of organic matter on virus behaviour depends on the form of the organic matter. Dissolved organic matter will decrease virus attachment to soil by competing with viruses for attachment sites (Jin et al., 2002). In contrast, bonded organic matter may increase attachment by providing additional attachment sites (Powelson et al., 1991). While reducing organic matter increases virus adsorption, the reduction achieved from primary to secondary treated effluent is not necessarily sufficient to affect virus transport (Gerba et al., 1978), however the reduction achieved from secondary to tertiary treatment has been reported to increase virus adsorption (Dizer et al., 1984). Additionally, dissolved organic carbon and anionic surfactants have been reported to decrease the inactivation rates of bacteriophages (Ryan et al., 2002). Where there are expected to be an abundance of attachment sites, such as in clay soils, the organic load may not be relevant (Gerba et al., 1978).

Adsorption of viruses to soil is known to be proportional to the ionic strength of the groundwater, with greater adsorption with increasing ionic strength. Conversely, an influx of water of low ionic strength can elute adsorbed viruses, e.g. rain water has been reported to desorb viruses (Gerba et al., 1975; Wellings et al., 1975; Schijven, 2001). There are a number of parameters that vary with soil type, and therefore affect virus transport. The pH of the soil and water will affect the surface charge of a virus and soil, and hence their interactions. In general, decreasing pH will result in greater attachment

Page 27

of viruses to soil (Yates et al., 1987b). Increases in isoelectric point, cation exchange capacity, exchangeable iron, iron oxide, specific surface area, ionic strength, multivalent cations or the clay/mineral content will favour attachment.

4.2 Virus transport pathways from on-site sewage systems The transport of viruses from an on-site sewage land-application or disposal systems will depend firstly on the system type and performance, and then on the environmental conditions. The pathways for septic tank-absorption trenches and for AWTS with irrigation are summarised in Figure 2 and Figure 3, respectively. The pathways and the factors that affect them are discussed in more detail below.

Household Normal conditions Failure Septic tank Wet weather

Absorption trench Surfacing effluent

Biomat

Overland flow Unsaturated soil

Saturated soil Catchment waterways

Figure 2 Virus pathways for septic tank and absorption trench system

Page 28

Household Normal conditions Failure AWTS Wet weather

Irrigation Overland flow

Unsaturated soil

Saturated soil Catchment waterways

Figure 3 Virus pathways for AWTS with irrigation

Household The household is the start of the virus transport pathway. The virus concentrations will be a function of the infection rates of people within the household and their rates of virus excretion.

Treatment system The fate of viruses within the treatment system is a function of the treatment type and performance of the system.

Limited studies have been undertaken on the occurrence of viruses in septic tanks effluent or the removal within the septic tank. Virus removal within septic tanks has been reported to be approximately 75 % (Payment et al., 1986; Rao et al., 1986; USEPA, 2002). The available studies on virus concentrations in septic tank effluent have reported only relatively low concentrations. Lewis et al. (1993) reported concentrations of enteric viruses in household septic tank effluent of between 0.07 viruses L-1 and greater than 59 L-1. Similarly low concentrations have been reported for communal facilities, with concentrations of 0.26 to 4.4 viruses L-1 in a school septic tank serving 350 people (1998a). These results are also significantly lower than in sewage from centralised systems where virus concentrations have been reported to

Page 29

range from 104 – 105 per L (Gerba, 2000a; van den Berg et al., 2005), although concentrations of up to 107 norovirus RNA-containing particles per litre have been reported (Lodder et al., 1999).

Virus removal in large-scale activated sludge treatment has been reported at 98 % (Payment et al., 1986), but has been assumed to be lower in smaller systems, such as

AWTS, with 1 log10 removal assumed (USEPA, 2002). Alternative mound systems have limited data on removal although Ho et al. (1991) have reported greater than 7 log10 removal of poliovirus in 65 cm columns of red-mud-amended sand.

Disposal systems As illustrated in Figure 2, an absorption trench performing well will release effluent to the unsaturated zone through the biomat or clogging zone. The effluent may then travel through the unsaturated zone to the saturated zone. An absorption trench which is not performing well due to poor design, or carry over of solids from the septic tank, may result in effluent travelling up to the surface where it may either be evaporated, be exported as runoff or re-infiltrate into the unsaturated soil.

The type of disposal system was based on the treatment system type, such that all septic tank and alternative system effluent was disposed via absorption trenches, and AWTS effluent via surface spray irrigation. Absorption trenches were assumed to result in discharge to the unsaturated zone at a depth of 0.5 m with vertical transport to groundwater or a confining layer, then horizontal transport to a stream. Surface spray irrigation was assumed to discharge on the surface and follow the same pathway as absorption trench effluent in dry weather. During dry weather, viruses were assumed to accumulate on the ground surface, with total numbers reduced by decay. During wet weather, these viruses that accumulated during the antecedent dry period were assumed to be available to be mobilised in surface runoff. Connectivity is defined as where sewage disposal is effectively direct to waterways. It may occur naturally, such as rock fissures or macropores that result in rapid transport to waterways, accidentally, such as spray irrigation into drains, and anecdotal evidence suggests this may arise from deliberate tampering with systems.

The design of the sewage disposal system will also affect virus transport with decreased attachment with increasing loading rates (Lance et al., 1984; Yates et al., 1988).

Page 30

Adsorption rates will also decrease with distance travelled through soil (Schijven et al., 2000), and over time as attachment sites are utilised by organics and microbes (Schaub et al.).

An absorption trench performing well will release effluent to the unsaturated zone through an organic clogging zone or biomat. The effluent may then travel through the unsaturated zone to the saturated zone. An absorption trench which is not performing well due to poor design, or carry over of solids from the septic tank, may result in effluent travelling up to the surface where it may either be evaporated, be exported as runoff or re-infiltrate into the unsaturated soil. Alternatively, transport can bypass the unsaturated zone straight to groundwater. For surface disposal systems, such as spray irrigation systems, transport can also include the wind-borne transport of aerosols.

For surface disposal systems, such as spray irrigation systems, transport can also include the windborne transport of droplets and aerosols (Carducci et al., 2000). Many models are available for modelling the transport of aerosols from irrigation applications, however there are limited models available for quantifying the spread of contaminants due to windborne aerosols/dust from on-site sewage disposal application. In aerosols a combination of temperature, sunlight and humidity affects the inactivation rate, however, a conservative inactivation value of zero has been assumed for risk modelling (Camann, 1980).

Subsurface transport Subsurface transport through unsaturated soil, saturated soil and groundwater is usually the intended route for wastewater disposal. Transport in the unsaturated soil generally results in greater removal of viruses from the water column than in saturated soil (Jin et al., 2000), and inactivation is higher in unsaturated conditions, as discussed above. However, wastewater can bypass the unsaturated zone due to the poor design or through water logging of the ground.

Overland flow Surface transport can be separated into waterborne (overland flow) and windborne (aerosols/dust). While work has been undertaken to establish the transport of bacteria and protozoa in runoff from agricultural applications, assessment of runoff from sewage application areas has largely been limited to centralised sewage treatment methods

Page 31

(Schaub et al., 1980). One exception is the work of Roser and Ashbolt (2007) where faecal indicators, biomarkers and coliphages were used to identify the connection of septic seepage to nearby streams during rain events.

4.3 Modelling virus fate and transport Previous reviews of models for virus transport and fate in the subsurface have been undertaken by Rehmann (1998), Jin and Flury (2002) and Tim and Mostaghimi (1991). Tim and Mostaghimi (1991) reported that most models were limited to steady state flow in saturated conditions, such as presented by Grosser (1984) and Corapcioglu and Haridas (1984).

Virus transport is commonly quantified in column experiments using repacked soil (Lance et al., 1976; Powelson et al., 1994; Sobsey et al., 1995; Schijven et al., 2002a; Van Cuyk et al., 2004), with results generally considered to be representative of field- scale. However, there are several limitations of column studies which are in part discussed here with the models used to interpret the results.

Rehmann’s (1998) review of laboratory models concluded that simplified laboratory models are not directly applicable to field conditions, due to the increased complexity of aquifer heterogeneity. The review of field-scale models of virus and colloid transport in aquifers concluded that over small distances, modelling has had varying degrees of success, while larger scale models have been unsuccessful.

Pathogen transport has been modelled as a solute or a colloid. Viruses are typically modelled as a solute (Yates et al., 1990; Schijven et al., 1999), with their transport and fate in the subsurface affected by inactivation, advection, dispersion and adsorption. Larger microorganisms (which could include viruses sorbed to particles), such as bacteria and protozoa, are more commonly modelled as colloids (Bradford et al., 2003) as they are affected by filtration effects, in addition to inactivation and adsorption, although there is an increasing adoption of the colloid filtration model for virus applications (Bales et al., 1991; Schijven et al., 1999; Pang et al., 2005).

Viruses will readily attach or sorb to interfaces, such as the soil-water, air-water or soil- air-water interface. Adsorption is generally considered to be reversible, although may be irreversible under certain conditions. At the air-water interface (AWI), virus and colloid

Page 32

sorption is essentially irreversible due to capillary forces (Sim et al., 2000). In this situation, inactivation may also be enhanced due to the surface tension at the AWI. Modelling of the interactions at the AWI is generally limited (Jin et al., 2002)

Three main modelling theories relevant to viruses are discussed here: equilibrium adsorption, kinetic adsorption and colloid filtration. Equilibrium and kinetic adsorption theories assume that the virus essentially behaves as a solute. Hence the transport is modelled using solute transport models such as the advection-dispersion equation. Equilibrium adsorption is based on the theory, experienced in batch experiments, that viruses in solution will adsorb to a surface until a steady state concentration is obtained. These can then be used to construct Freundlich or Langmuir isotherms. Equilibrium adsorption assumes all adsorption is reversible, and therefore inactivation is the only removal mechanism. Results from field-scale studies have led to the conclusion that the classical form of advection-dispersion equation is inadequate for describing field-scale solute transport, partly due to the increase in dispersivity with travel distance (Yates et al., 1991).

Kinetic adsorption models have generally been applied to model transport scenarios, rather than batch experiments (Jin et al., 2002). They use first-order kinetics and have been applied with a single or dual adsorption sites (Schijven et al., 2002c) to represent different types of attachment sites, such as soil and air-water-interface, or different types of attachment, such as reversible and irreversible attachment.

Colloids are defined as particles with effective diameters of less than 10 micrometers (Bradford et al., 2002), and so colloid filtration theory has been applied to model the transport of bacteria, protozoa and viruses. Colloid filtration includes straining, sedimentation, inertial impingement and diffusion (Gerba et al., 1991). Straining occurs when a particle is too big to pass through a pore. Filtration efficiency is affected by flow velocity and time due to the build up of clogging with time. Similarly, time-dependent porosity can be used to account for this build up of clogging.

As colloid size increases, removal is increasingly attributed to straining over attachment (Bradford et al., 2003). For microbes, because of the size dependent relationship, straining is usually only considered for the larger microbes (bacteria and protozoa). Size is also important in the modelling of particle exclusion which can result in more rapid

Page 33

transport of particles if they only have access to limited, larger, pathways (Bradford et al., 2003). For example, in coarse textured soil, Cryptosporidium has been reported to be transported 1.19 to 1.37 times faster than the solute chloride (Harter et al., 2000). However, this behaviour is also mistakenly assumed at times in laboratory experiments due to the higher number of orders of magnitude of detection capabilities for microorganisms.

Colloid filtration is based on a single collector removal efficiency, η, which represents the ratio of particle deposition rate to the convective transport of upstream particles 2 toward the projected area: η = I / U C0 π ac where I is the deposition rate, U is the velocity and ac is the radius (Ryan et al., 1996).

Typically colloid filtration models have focussed on the pathogenic micro-organisms and have not considered the behaviour of other colloids that might affect virus transport. For example, effluent plumes result in increases in pH, concentrations of dissolved organic matter and surfactants which would contribute to the mobilisation of colloids (Ryan et al., 1996). Increased ionic strength, such as during rain events, would reduce mobilisation (Ryan et al., 1996).

Virus models, for both inactivation and transport, consider viruses as single particles. In reality, viruses released from sewage treatment systems are likely to represent a mixture of single particles, aggregated particles and attached particles, although this will depend on the treatment process. Virus aggregates and viruses attached to particulates will exhibit behaviour, both in inactivation and mobility, that differs from single viruses. Aggregated or attached viruses will be subject to straining, or may break up to release several single infectious units. This can particularly affect the modelling of short- distance experiments, such as laboratory-scale experiments, resulting in issues when up- scaling to field-scale.

Aggregated or attached viruses may also exhibit different inactivation rates. Two-phase inactivation during disinfection (Sharp et al., 1975) has been attributed to aggregation and assumed to be due to shielding from disinfectants provided to viruses in the centre of aggregates. The implications of aggregation are that the inactivation rate of single virus particles will be underestimated using the first-order inactivation model, due to the inactivation of single virus particles being represented by the initial rapid inactivation.

Page 34

The secondary, decreased inactivation rate represents the aggregated viruses, where multiple viruses are detected as one plaque forming unit (pfu), and hence aggregates are detected as one pfu until all viruses are inactivated. Conversely, attachment will result in the inactivation rate being overestimated. The relative contributions of attachment and aggregation to virus removal in inactivation studies have been studied together (Gassilloud et al., 2005), with both attachment and aggregation found to be significant removal mechanisms over three weeks in groundwater.

Reviewing modelling tools used to estimate setback distances A review was undertaken of existing tools for modelling virus fate and transport in the subsurface for the purpose of estimating setback distances, including: Yates’ geostatistics approach based on temperature-dependent inactivation only which has been applied in many subsequent modelling tools; USEPA tools VIRALT and CANVAS; Australian modelling tool TRENCH; models by Sim and Chrysikopoulos; USEPA model Virulo; CXT-FIT and HYDRUS.

Geostatistical tool Yates et al. (1986) proposed a model whereby geostatistics (kriging and regression) were used to predict virus (horizontal) transport and fate in two-dimensional steady- state groundwater flow. The model uses the Darcy groundwater flow equation. Virus transport is modelled as equivalent to water movement, and is solely based on the Darcy equation with no consideration of attachment kinetics. Inactivation is included in the model as linear, temperature dependant function: µ = 0.018 T - 0.144 (Yates et al., 1987a); where µ is the inactivation rate, and T is temperature (°C). The model is designed to be applied to assess the probability of groundwater well contamination for a certain setback, or the setback required to meet a certain probability of contamination (Yates et al., 1991). The data requirements are basic: hydraulic conductivity, hydraulic gradient, effective porosity and groundwater temperatures. As the model is steady state it does not allow for rainfall events or other variability.

The simplicity of the theory makes it relatively easy to apply, however, it does requires knowledge of geostatistics for proper application. The model is limited by its assumptions that viruses are transported with water flow thereby not allowing consideration of attachment kinetics or preferential flow paths. Data requirements are

Page 35

low, although the suitability of the inactivation equation is limited as it is only valid over 9 degrees, and would appear to over predict virus inactivation at high temperatures (Figure 4).

1 0 5 10 15 20 25 30

HAV MS2 Probability FRNAPH of an event PRD1 0.1 Polio 1 Echo 1 Trench Yates

0.01

Figure 4 Comparison of two temperature-dependent virus inactivation models, one devised by Yates and Yates (1987a) and the second employed by Beavers and Gardner (1993), with inactivation coefficients for a range of viruses published by Schijven and Hassanizadeh (2000) and Yates (2002).

While this approach will not provide accurate information on the transport and fate of viruses, it may be suitable for preliminary analysis or a screening level assessment. However, the equation used for virus decay rate variation with temperature may need to be revised for application to the Sydney catchments.

VIRALT VIRALT (HydroGeoLogic, 1994) is a semi-analytical and numerical code that simulates virus transport and fate in unsaturated and saturated conditions in the subsurface. It was developed for assessing the need to disinfect water from public groundwater wells in the USA as part of the groundwater disinfection rule and is designed to delineate groundwater path lines and well capture zones in confined and unconfined aquifers, and model groundwater flow between injection (point or line

Page 36

source) and well extraction, including calculating virus concentrations in extracted well water.

Virus transport is based on the advection and dispersion of viral particles in the vertical dimension in unsaturated zone (1D), and along groundwater path lines in the saturated zone (2D) with linear, equilibrium adsorption and first order inactivation (based on the temperature-dependent equation above). The convective-dispersive hypothesis is used which assumes that viruses behave like solutes. The model assumes porosity, flux and velocity are constant; dispersivity and dispersion are constant; and that sorption is linear, instantaneous and reversible.

VIRALT includes steady-state and transient transport, however the transient transport is limited to ten time steps of the source. Hence, rainfall events cannot be included.

Yates (1995a) reported that VIRALT under-predicted virus concentrations at wells (even when no adsorption assumed), over-predicted travel times of the viruses, and predicted that each site would meet the natural disinfection criteria, whereas the observed concentrations exceeded the criteria significantly. This may be related to the default values assumed for some parameters, a problem that may be overcome with calibration. Yates and Jury (1995) reported that the retardation was the most sensitive parameter, and that a change of ± 50% resulted in a maximum concentration change of over eight orders of magnitude. Similarly inactivation variation (based on temperature variation) resulted in a maximum concentration change of over seven orders of magnitude.

While the application of this model to the field has suggested it is unsuitable for a range of situations, it has highlighted the importance of sensitivity analysis.

CANVAS CANVAS (HydroGeoLogic, 1995) is a modification of VIRALT (Yates et al., 2000), which allows multiple contaminant sources (1 and 2D) to be modelled and includes transverse dispersion, detachment terms and a colloid filtration term. It allows delineation of groundwater path lines and well capture zones in confined and unconfined aquifers. CANVAS is a computer program developed for the USEPA to assess whether a groundwater system is a risk of viral contamination, and consists of a

Page 37

composite analytical-numerical code for simulation of virus transport and fate in the unsaturated zone and groundwater. It includes three sections: 1D vertical unsaturated steady state water flow and virus transport (USCOL); 2D steady state saturated horizontal water flow (SZFLOW); and 2D steady state virus transport in saturated zone. Virus transport and fate equations include advection, dispersion, adsorption, inactivation, attachment and detachment kinetics and colloidal filtration. Adsorption is assumed to be controlled by rapid instantaneous equilibrium and a kinetic expression for a slower adsorption rate.

Unsaturated flow is based on Darcy’s law combined with a non-linear constitutive relationship between the water-phase relative permeability and the pressure head. Effects of hysteresis are neglected. Virus transport governed by 1D advection- dispersion equation (diffusive/dispersive transport governed by Fick’s law). The hydrodynamic dispersion coefficient is the sum of the coefficients of mechanical dispersion and molecular diffusion. Sorption of viruses is assumed to be described by a linear freundlich equilibrium isotherm. Virus inactivation is first order. Virus transport in the saturated zone is solved with the Galerkin finite element method, and is governed by the 2D vertically averaged advective-dispersive transport equation.

Assessments of the performance of the CANVAS were undertaken by Yates et al. (2000), based on field experiment results at two 60 cm (2 ft) lysimeters. The colloid filtration factor was assumed to be zero (the default) as no data was available. No calibration was undertaken. An inactivation rate of 0.414.d-1 was assumed. The breakthrough curves for microspheres were used, showing poor correlation. The virus breakthrough curve provided a better fit, but not as good as HYDRUS-2D (discussed below) for the same input parameters. The study concluded that CANVAS was easy to use but did not provide good predictions. However, as with VIRALT this may be a limitation of the lack of data and calibration.

TRENCHTM TRENCH™ (Australian Institute of Environmental Health Cromer, 1999; Cromer et al., 2001) is an MS Excel based program for the design of effluent land application systems. The program has numerical 2D capabilities that consider saturated and unsaturated transport. The set-back distances are based on virus transport according to the

Page 38

geostatistical theory outlined above. The virus groundwater model includes situations for a two layer soil with impermeable subsoil without water table, a uniform soil with water table, a two layer soil with more permeable subsoil with water table in bottom layer, and water bores in uniform soil. Where a water table is present, travel time to reach the water table is calculated. In the first situation without a water table, the program seems to assume that transport is in groundwater. The temperature is taken as the minimum effluent temperature. There is no consideration of attachment, filtration or preferential pathways. The water flow model used for saturated and unsaturated flow is not stated.

TRENCH™ is designed to facilitate the sizing and siting of effluent land application areas, and as such is a generalised program primarily concerned with average or usual conditions. It includes a large amount of supporting information as defaults or as ‘accepted’ values which minimise the amount of site specific information required. The accuracy of this information is as for the relevant guidelines and standards, and is occasionally limited or misleading. The move toward less prescriptive standards in recent years has also reduced the amount of information available. Geometry is set up numerically by stating the available area, slope, direction of slope and rating the drainage. Climate data is based on monthly averages. The calculation of set backs requires the depth of soil, permeability of the soil, depth to groundwater and depth to bedrock.

The program allows for the sizing of a trench, and provides a graph of the expected depth of wastewater in the trench throughout the year. No consideration is included as to the long term performance, such as the effect of the formation of a biomat.

This modelling tool was included in the review as it is used in Australia, including by the SCA. TRENCH™ provides information on the transport and fate of viruses, and may be suitable for preliminary analysis or a screening level assessment, however, it has the same limitations as the temperature-dependent virus survival model.

Sim & Chrysikopoulos models Sim & Chrysikopoulos have published a series of papers (Sim et al., 1995; Sim et al., 1996; Sim et al., 1998; Sim et al., 1999; Sim et al., 2000) on analytical models of virus transport and fate in unsaturated and saturated soils.

Page 39

Sim and Chrysikopoulos (2000) provides a numerical model for 1D virus transport in homogenous, unsaturated porous media, including sorption to liquid-solid and liquid-air interfaces, as well as inactivation in solution and at interfaces. It is validated with the results of hydrophilic colloids transport experiments. The model uses constant inactivation rates, but recognises that they are temperature and time dependant. Adsorption is based on mass transfer rates and concentrations, following a linear equilibrium relationship. Variably saturated flow described by modified Richards equation. Unsaturated hydraulic conductivity is based on van Genuchten – Mualem equation.

Sim & Chrysikopoulos (1998) published a 3D model of virus transport in saturated, homogenous porous media based on hydrodynamic dispersion in a uniform flow field. As far as the author is aware this is the only 3D model for virus transport. It was trialled with field data from Bales et al. (1997) and provided a reasonable fit, with the assumption that assuming longitudinal dispersivity is ten times greater than the lateral and vertical dispersivities.

Sim & Chrysikopoulos have used colloid filtration theory (Sim et al., 1995), investigated temperature dependent inactivation rates (Sim et al., 1996) and adsorption (Sim et al., 1995), investigated the effect of moisture variations on sorption (Sim et al., 1999) and trialled stochastic models (Sim et al., 1996). However, there is limited calibration or validation of the model with laboratory and field experimental data.

Virulo Virulo (Faulkner et al., 2002) was developed by the USEPA National Risk Management Laboratory to aid in the siting of septic tanks. It is a probabilistic (Monte Carlo methods) computer model that enables the calculation of the reduction of viruses in a hyrdogeologic barrier (1D vertical transport in the unsaturated zone) based on virus attenuation due to biological, physical and chemical factors. The model recommends a virus reduction of 4 log10. It is designed for a single, short-term discharge, such as a ‘spill’ or septic tank overflow, where the duration of virus inputs is significantly shorter than the residence time in the hydrogeologic barrier. Hence, it assumes that there is no saturation of absorption sites. Preferential flow is not included.

Page 40

The model uses the Buckingham-Darcy flow equation for water flux in the unsaturated zone. Assumptions inherent in the model include steady-state flow, gravity drainage only, homogeneous soil in terms of hydraulic properties, virus properties and geochemistry, no preferential flow and that water does not contain significant detergents or other materials that may change soil properties. It includes soil properties from Rosetta soils for 12 USDA soil types.

The virus transport model including sorption and inactivation of viruses in the unsaturated zone is based on Sim & Chrysikopoulos (2000) (see above), a numerical model for 1D virus transport in homogenous, unsaturated porous media, including sorption to liquid-solid and liquid-air interfaces, as well as inactivation in solution and at interfaces. Adsorption is based on mass transfer rates and concentrations, following a linear equilibrium relationship. Virulo incorporates temperature-dependant inactivation rates for guidance.

Virulo has been designed to be flexible as new data becomes available, and for integration with GIS, and so that regulators can use the model based on limited information such as boring logs, climate data and soil survey reports. However, sensitivity analysis identified saturated hydraulic conductivity, porosity and the virus inactivation rate as the most important main effects, which are also among the least likely to have field measurements taken. Faulkner et al. (2002) recommend “because of the large uncertainty in parameters needed to predict virus transport in the unsaturated zone, probabilistic models that encapsulate and propagate the uncertainty in those parameters should be used.”

CXT-FIT CXT-FIT (Toride et al., 1995) is a steady state 1D saturated flow solute transport program designed to fit data from column experiments. It includes inverse parameter estimation. 1D transport models include: convection-dispersion equation, the chemical and physical non-equilibrium CDE and a stochastic stream tube model based on local scale CDE equilibrium or non-equilibrium adsorption. Toride et al. (1995) modified CXT-FIT to include a one site kinetic model with first order inactivation and one-site attachment. Schijven (2001) reported issues with trying to fit breakthrough curves when the tail is much smaller than the peak. This was due to the use of sum of the squared

Page 41

residuals for fitting curves and was shown by a sensitivity analysis of the attachment and detachment coefficients and the inactivation rate for attached viruses.

HYDRUS HYDRUS (Šimùnek et al., 1998; Šimùnek et al., 1999) was developed by the USSL. HYDRUS is a numerical model for horizontal or vertical 1D (Šimùnek et al., 1998) or 2D (Šimùnek et al., 1999) transport of water, heat and solutes in variably saturated and non-homogenous media designed for column and field experiments. It includes inverse parameter estimation. The geometry is set up graphically in MESHGEN (HYDRUS- 2D) or in a graphical editor (HYDRUS-1D).

HYDRUS uses the convection-dispersion solute transport equation with up to three solutes in a decay chain reaction. A variety of time and space weighting schemes are included such as the Galerkin Finite Element with artificial dispersion, and with/without non-linear adsorption. First order inactivation and adsorption are included in the solute transport considerations.

HYDRUS has been modified for virus transport by incorporating reversible adsorption to two types of kinetic sites (Schijven et al., 2002c). The program has also been modified for colloid filtration theory (Bradford et al., 2003), however the size and density of the colloids are fixed.

HYDRUS can use a large amount of data, including soil properties, climate, hydraulic inputs and solute inputs, especially when variable boundary conditions are included and in inverse solutions. However, soil properties, as well as being put in manually, can be selected from a database or calculated with Rosetta based on the soil texture. Similarly constant boundary conditions reduce the inputs required. HYDRUS also has the flexibility to include, through the inclusion of multiple soils, surface runoff and water behaviour in trenches, although such work has not been published.

Yates et al. (2000) applied HYDRUS-2D to model results of field experiments. While HYDRUS was calibrated and performed satisfactorily for bromide transport, the assessment of the potential for modelling virus transport was limited as the soil adsorption coefficient was assumed to be zero, hence ignoring attachment altogether.

Page 42

The conclusion was that HYDRUS performed satisfactorily; however, a lot of data was required, even without attachment characteristics.

Selecting a model These models and modelling tools were reviewed for their potential application to delineating buffer distances for on-site sewage treatment systems in Sydney’s drinking water catchments. All the models were capable of predicting virus movement from an on-site sewage treatment system. However, there were a number of limitations.

To model the impact of the thousands of onsite sewage treatment systems on water quality in the catchments, it is necessary for a model to be able to incorporate multiple contaminant sources. Of the models available for assessing setback distances, only HYDRUS has that capability. However, as with many modelling endeavours, there is a balance between the detail of the model and amount of data needed. A catchment scale model will always require a large amount of data, and modelling the catchments with HYDRUS would not be realistic.

Another limitation of the models for the application under consideration is focus on groundwater flow as the only pathway by many of the models. This does not allow for the greater removal of viruses in unsaturated soil, or for transport in overland flow for irrigation systems. These modelling tools are also designed to assess transport under design conditions, without consideration of potential failure mechanisms such as heavy rainfall events or clogging in the disposal field.

Yates & Jury (1995) found that virus transport models for assessing setbacks are considered inadvisable due to the poor knowledge of virus transport and fate processes and the error introduced by using point estimates for uncertain parameters. For example, retardation and inactivation are the least likely parameters for the water utility to be able to get site specific information on and there is high variation in retardation and inactivation between viruses as well as between sites.

Due to the limitations of these models, and the limitations of data availability on a catchment scale, a risk assessment methodology was employed for the catchment model. HYDRUS was used to evaluate laboratory column and field-scale experiments, and used as a basis for extrapolating the results so that they could be used in a

Page 43

catchment risk assessment model.

Page 44

5 Quantitative microbial risk assessment

In the New York drinking water supply catchments the setback for absorption trenches has been set at 30 metres (100 ft). A study undertaken to assess the efficacy of this buffer in protecting water quality reported phage transport up to the buffer distance under conditions that existed in the catchments (Curry, 2000). However, no conclusions were able to be drawn regarding the implications for water quality and septic system management because of the lack of information about what impact a single system or a multitude of systems

This chapter deals with the development of a quantitative microbial risk assessment (QMRA) model, adopted to delineate buffer distances in this thesis. A buffer distance delineation methodology was derived based on QMRA and a goal of < 10-4 annual infection probability in drinking water and is presented in full in Paper IX. The results are presented in Chapter 6 and Chapter 7.

5.1 Hazard identification The hazard investigated was human enteric viruses, as represented by rotavirus. Rotavirus represents a ‘worst-case’ model for human enteric viruses due to its reported high infectivity and prevalence in NSW, as well as the high rates of excretion from infected individuals. Rotavirus is the leading cause of gastroenteritis in children under five, and is attributed to half of all hospitalisations in Australia for gastroenteritis in children under five (Carlin et al., 1998).

The illness rate for rotavirus has been reported at between 0.71 % in the UK and 1.4 % in the USA (Mead et al., 1999; Wheeler et al., 1999), although the incidence of asymptomatic excretion in the UK has been reported to be 14 % (Amar et al., 2007). In NSW the rate of hospitalisation for children under five was 0.86 % (Carlin et al., 1998) which is within the ranges of the aforementioned studies. Haas (1999) reported that morbidity for rotavirus is approximately 50 %, thereby making the rate of infection double the reported rates of illness. Hence, a uniform distribution between 1.4 % and 2.8 % was employed for the rate of infection.

Page 45

5.2 Exposure assessment Humans can be exposed to pathogens from disposal of effluent via on-site systems through direct accidental ingestion, aerosols from irrigation systems, consumption of vegetables irrigated with effluent, or consumption of drinking water contaminated with inadequately treated effluent. This risk assessment focuses on the last of these.

There are many pathways and variables to consider in the transport of viruses from effluent via on-site sewage systems to the drinking water consumer in a surface catchment area such as the Sydney catchments, some of which are highlighted in Figure 5 which provides an overview of the risk assessment model structure. The pathways used in the model are those used in Figure 2: the model was developed to include virus loads in sewage as a function of household size and excretion; removal of viruses via treatment in the on-site sewage treatment system; removal in the environment with pathways based on the type of disposal system used; and removal during transport in the catchments and drinking water treatment processes. Due to a lack of availability of information not all of the routes and variables identified in Figure 2 could not be included in the final QMRA model. The variables included and the model inputs are presented below and are summarised in Paper IX.

The QMRA focused on the most populous catchment, the Warragamba catchment (Figure 1), which was subdivided into the following regions: Blue Mountains, Mulwarree, Wingecarribee and Wollondilly. The Blue Mountains subcatchments (Figure 6) that drain to Warragamba Dam include the Upper Cox’s, Mid Cox’s and Lower Cox’s River and the Kowmung River subcatchments. The Mulwarree region included the Upper Wollondilly River and Mulwarree River subcatchments. The Wingecarribee region included only the Wingecarribee river subcatchment. The Wollondilly region included Wollondilly River, Nattai River, Werri Berri Creek and Little River subcatchments. Each region was modelled individually, with the total output the calculated for the catchment.

Page 46

Scenario Household & Management building Excretion, Maintenance, Replacement Risk Management Model Feedback Mechanisms

Effluent quality Disinfection system Disposal system inputs Type, Performance Type, Performance

Surface Runoff Soil Groundwater Rainfall, Slope Saturation, Depth Heterogeneity, Flow

Contaminant transport and fate in buffer Buffer distance zones Slope, Vegetation, Heterogeneity

Contaminant transport and fate Stream Water Quality in the catchment Cumulative impacts of on-site systems and comparison with other land uses, Off- take and Recreational Water Quality, Water Quality Objectives

Figure 5 Schematic of the risk assessment structure and development (Paper II)

Page 47

Figure 6 Sydney Drinking Water Catchments by subcatchment (Ferguson, 2005)

Household The model component predicted virus loads based on community infection and excretion rates, and household population size. Virus loads were a function of the rate of infection of individuals and the excretion of the viruses during the infection as described in Table 3 (Paper IX).

Page 48

Table 3 Summary of assumptions and data for household infection and excretion (Paper IX)

Item (units) Distribution (values) Source & assumptions [limits] Infection rate Uniform (0.014, 0.028) Reported rates of illness of 0.71- 1.4 % (Mead et al., 1999; Wheeler et al., 1999) with morbidity 50 % (Haas et al., 1999) Virus excretion Triangular (4, 7, 9) High excretion rates of up to 1012 -1 rate (log10 .g ) (Gerba, 2000b) have been reported, however lower triangular distribution assumed by others (Höglund et al., 2002b) Faeces excretion Uniform (100, 200) (Feachem et al., 1983) rate (g.d-1) Duration of Lognormal (6, 3) [2, 23] Limits from Gerba (2000b) shedding (d) Time dependent Exponential (0.1), Based on assumption that virus shedding (-) [0.000000001, 1] shedding will commonly peak within the first few days of infection, but may continue at a lower rate for up to several months. Household Poisson (Average house Australian 2001 census data, (ABS, population population) 2002)

The rationale for choosing rotavirus was presented in Section 5.1. Further QMRA assumptions included in Table 3 were based on the rate of excretion of faeces is reported to be 100 – 200 grams per day (Feachem et al., 1983), hence a uniform distribution between those limits was applied. The duration of shedding of viruses in faeces is reported to be 8 - 23 days (Gerba, 2000a). A lognormal distribution was selected. The rate of excretion of rotavirus in faeces is reported to have a lognormal 10 12 distribution of 10 – 10 viruses per gram (Gerba, 2000b). The household virus load Lh (viruses.d-1) was calculated by:

Lv = LnL fih .10.. λ (Equation 1)

-1 -1 Lf was the load of faeces per person per household (g faeces.person .d ). Lv was the -1 virus load per gram of faeces (log10 viruses.g ); as no detailed distributions are available on excretion of viruses over the duration of an infection, this was based on the maximum reported loads. A virus excretion distribution, λ, was introduced based on the

Page 49

assumption that the peak of virus excretion during infection is for a limited period, and extended periods of shedding were at lower concentrations. The number of infected people per household ni was based on a Poisson distribution of the community infection rate times the duration of shedding (d.y-1).

On-site sewage treatment system All assumptions used in the QMRA model for on-site system performance are provided in Table 4.

Table 4 Summary of assumptions and data for on-site sewage system performance and failure (Paper IX)

Item (units) Distribution (values) Source & assumptions [limits] Septic removal Lognormal (0.6, 0.1) [0,1] (Payment et al., 1986; Rao et al., (Log10 ) 1986) AWTS removal Lognormal (1, 0.1) [0,] (USEPA, 2002) (Log10) Disinfection Uniform (1.1, 1.8) Paper VI removal (Log10) Alternative Normal (1.7, 0.4) Paper VII systems Pump out removal All Septic failure Discrete: 10% failure Assumed AWTS failure Discrete: 9.5% failure Paper VI Disinfection Discrete: 8.3% failure Paper VI failure Alternative system Zero No data available. Assumption failure based on the fact that the systems are relatively new. Pump out failure Discrete: 10% failure Includes accidental overflow and illegal discharge. Connectivity assumed. Disposal system Discrete: 1% connectivity, Total failure 40% (O'Neill et al., failure 39% performance failure 1993). Connectivity implies no removal in buffer zone (Ferguson, 2005).

The type of disposal system was assumed based on the treatment system type, such that all septic tank and alternative system effluent was disposed via absorption trenches, and AWTS effluent via spray irrigation. Absorption trenches were assumed to result in discharge to the unsaturated zone at a depth of 0.5 m with vertical transport to

Page 50

groundwater or a confining layer, then horizontal transport to a stream. Surface spray irrigation was assumed to discharge on the surface and follow the same pathway as absorption trench effluent in dry weather. During dry weather, viruses were assumed to accumulate on the ground surface, with total numbers reduced by decay. During wet weather, these viruses that accumulated during the antecedent dry period were assumed to be available to be mobilised in surface runoff. Connectivity is defined as where sewage disposal is effectively direct to waterways. It may occur naturally, such as rock fissures that result in rapid transport to waterways, accidentally, such as spray irrigation into drains, and anecdotal evidence suggests this may arise from deliberate tampering with systems.

Buffer distance transport Transport and removal of viruses in the subsurface was based on column and field experiments undertaken using soils from the catchments (Paper VIII). Transport in groundwater was modelled using HYDRUS-1D (United States Salinity Laboratory, Riverside, CA) (Šimùnek et al., 1998) in a 150 m section of soil, with a constant flux input equal to the groundwater velocity (calculated using Darcy’s law). Dilution effects were ignored as the model was load based, i.e. the total number of the viruses transported to water was considered not the viruses per volume. Parameters for inactivation, attachment and detachment were derived from experiments (Paper VIII).

The results were converted to log10-removal with distance such that:

R = asat . x + bsat (Equation 2),

where R is the log10 removal of viruses, asat and bsat are soil and condition specific constants and x is distance (m) and the parameter asat can be described as function of slope, such that:

d asat = c . (slope) (Equation 3)

where c and d are each soil specific constants. As the variation in the constant bsat was small, an average value for each soil was assumed. The derivation and values for the soil constants are provided in Paper IX.

Page 51

Similarly, unsaturated transport was modelled with HYDRUS-1D (Šimùnek et al., 1998) over the maximum soil thickness for the range of design loading rates specified in AS/NZS1547:2000. The same virus parameters were used as for the saturated transport modelling, however two virus adsorption sites were assumed to account for additional attachment at the air-water-interface (Thompson et al., 1998). Similarly to saturated transport, virus removal in unsaturated soil was modelled as:

R = aunsat . x + bunsat (Equation 4) where bunsat was the average intercept for all scenarios, and

d aunsat = c . q (Equation 5) where q is flow which includes the effluent land application rate and rainfall.

The assumptions for the unsaturated soil transport component of the model included: the slopes were defined based on a soil database, not on location specific data; the slope of the groundwater was assumed to be equal to the slope of the surface; and the effluent application rate was not considered significant for the groundwater flow scenario.

The transport of pathogens in runoff in Sydney’s drinking water catchments was studied with controlled rainfall simulations at laboratory (2004) and field (Ferguson et al., 2007) scales. The derivation of the models of removal for overland flow is discussed in Paper IX, using the average for bare and vegetated soil results. It was assumed for the overland flow model that as the slope decreased, the removal of viruses increased, such that at zero slope removal was complete. Similarly, as the slope increased, removal was assumed to decrease. The removal in overland flow was calculated as the sum of removal in runoff and infiltration. The model of surface runoff was based on fitting linear and log-linear models to virus removal (Ferguson et al., 2007). In order to address site variability, the surface runoff model was included as a triangular distribution, where the most likely value was the removal predicted by the linear model, and the minimum and maximum were defined by the linear model ± the difference between the general linear model and the linear model for the ungrassed plot only.

In addition to the removal equations, a degree of efficacy was incorporated to address the decrease in removal experienced at field scale due to subsurface heterogeneities such as fractures and macropores (Taylor et al., 2004), as well as surface runoff areas

Page 52

that were concave or firmly packed. Hence the virus load removal in the buffer distance

(RB) was:

RB = Eoverland flow .Roverland flow + Eunsaturated flow . Runsaturated flow + Esaturated flow . Rsaturated flow (Equation 6) where Rx is the log10 virus reduction during transport through matrix x and Ex is the percentage efficacy of that matrix (%).

Wet and dry weather was defined by historical data from the study area (SCA supplied data for 1928 - 2002), where rainfall equalled the binomial probability of rain Prain. Similarly, the antecedent dry period was defined as the product of the binomial probability of an antecedent dry period Pantec, and the duration of the antecedent dry period tantec (days). The build-up of viruses during this period was defined by the virus load multiplied by the duration of shedding with the viable fraction defined by the biphasic decay model (Petterson et al., 2001c), with a = 99.88 %, h1 = 2.48, and h2 = 0.51, with time defined as the antecedence dry period.

Catchment transport and fate The location of on-site sewage systems within the Warragamba catchment were derived from a combination of digital mapping and local council surveys. Spatial data was accessed through the SCA GIS (Geographic Information Systems) database and used to define catchment characteristics. The locations of potential systems were defined as the centroid of cadastral parcels, but limited by land use type of the cadastral parcel. Based on this location, the following parameters were collated from SCA GIS layers: distance to the nearest waterway, soil type, catchment, subcatchment and land-use. Based on the soil type, ranges of soil depth, slope and permeability were extracted and used in uniform distributions. A survey of local government authorities (Paper I) was used to provide information for discrete distributions of on-site sewage treatment system type.

Transport in waterways within the catchment was assumed to remove viruses by inactivation only. Sedimentation and resuspension was not included in the model. The removal was calculated as the inactivation rate (k) times the travel time from the drainage unit outlet to the offtake, including both stream and reservoir transport.

Page 53

Reservoir travel time was assumed to be equivalent to the stream travel time, equal to travel distance divided by stream velocity.

Due to the low inactivation rate assumed for viruses in stream water, the transport times were not particularly sensitive parameters. The transport times for viruses in waterways in the catchments were derived from historical flow data. Adenovirus and PRD1 were used as model viruses for the inactivation/removal in waterways due to their persistence in the environment.

The loads from sewage treatment plants were calculated as QADWF x 365 x Cv, where

QADWF was the Average Dry Weather Flow (AWDF) and Cv was the virus concentration per litre. QADWF was available from SCA, and varied from 100 to 300 L per equivalent persons per day. Cv was calculated as for the on-site sewage systems, based on the population served (CH2M HILL, 2001).

The concentration of viruses at the catchment off take (Cc) was defined as Cc = τ

Lcatchment / QAnnual (Equation 7), where τ is the fraction of catchment flow that goes to drinking water and QAnnual is the annual flow from catchment that goes to drinking water [ML.y-1].

Treatment, Distribution and Ingestion Sydney Water’s water filtration plants include screening; four stage coagulation: ferric chloride, polydadmac cationic polymer, lime water pH correction and polyacrylamide filter aid addition; filtration on sand or dual media (anthracite and fine grained sand); and chlorination/chloramination followed by fluoridation (Mazounie et al., 2000). The annual load from a catchment was calculated as the sum of the site loads minus the removal attributable to drinking water treatment and distribution. Assumed probability density functions (PDFs) for treatment and distribution are given in Table 5. The concentration of viruses in drinking water (Cdw) was calculated as:

Log Cdw = Log Cc - T1 - T2 - T3 + TD (Equation 8)

Where Ti are the log10 removal of viruses afforded by the different treatment (i = 1 to 3) and distribution (i = D) stages identified in Table 5.

Page 54

Table 5 Summary of assumptions and data for drinking water treatment (Paper IX)

Item (units) Distribution (values) Source & assumptions Chemical precipitation Triangular (0.3, 1.3, 1.6) (Westrell, 2004) (log10 reduction) GAC filtration (log10 Uniform (0.7, 1.2) (Westrell, 2004) reduction) Postchlorination Triangular (1.5, 2.0, 3.0) (Westrell, 2004) (log10 reduction) Distribution 0.5 Log10 Assumed to contribute viruses to the water supply due to accumulation in biofilms within pipes (Petterson et al., 2000)

Reports on the consumption of drinking water vary significantly by country (Mons et al., 2007). A very conservative assumption is two litres per day per person (NH&MRC et al., 1996), but that is generally used for total consumption, not the fraction that is unheated/untreated in the home. In Australia, consumption has been reported to vary between cities (Mons et al., 2007). A lognormal distribution for drinking water consumption was assumed with an average of 1.1 ± 0.63 L.pers-1.d-1 (Robertson et al., 2000).

5.3 Dose-response The total dose (D) was therefore estimated based on the virus concentration and volume of drinking water ingested:

D = Cdrinkwater x Vdw (Equation 9)

-1 -1 Where Vdw is the volume consumed (L. person . year ). The probability of infection of the drinking water consumer (Pinf) was assumed to follow the Beta-Poisson model (Teunis et al., 2000), which is valid when β ≥ 1 and α ≤ β:

-α Pinf (D; α, β) = 1 – ( 1 + D/ β) (Equation 10)

Where D is dose, α and β are dose-response coefficients, where α = 0.253 and β = 0.422 as described for rotavirus (Ward et al., 1986).

Page 55

5.4 Application to catchment management This model forms the basis for assessing the tolerable virus loads from each region, where tolerable loads were considered to be those resulting in an annual risk of illness from drinking water consumption of < 10-4 infections per person per year. However, it is infeasible for this model to be used on a site by site basis for development applications. Hence, a log10 reduction approach was adopted for assessing individual houses, with QMRA used to assess the impact of this approach. The log reduction method proposed for development assessment followed these basic assumptions:

• Each stage of treatment and transport can be assumed to provide a reduction in the virus load;

• Log reductions can be added; and

• A uniform log reduction for a catchment provides adequate safeguard for catchment water supplies.

The log reduction for a system was the sum of the log reductions for the treatment system, disinfection system, disposal system, and buffer distance based on the models described above. Log reductions were based on the average values adopted in the QMRA, although consideration was given to the variability. Management conditions were stipulated on how robust the system was, for example if there was high variability in system performance, or for novel systems where performance has not been well established, regular maintenance inspections should be a requirement for development approval.

For example, a site with clay loam soil and a 65 m buffer distance on a 10 % slope would provide, based on the groundwater transport model, 7 log10 removal in groundwater. For this site, the linear overland flow model would predict 4.6 log10 removal in runoff. Hence options for on-site systems might include an AWTS with spray irrigation, with a strict maintenance requirement, or subsurface disposal (Paper VII).

A number of scenarios (Table 6) were investigated. The baseline scenario included all system types with their associated failure rates. An outbreak of viral disease in the community scenario was modelled as a doubling of the community infection rate.

Page 56

Management scenarios were modelled to predict and compare the impact of the 100 m buffer distance, the impact of failure and the impact from a single type of system. The log reduction method for site assessment was also investigated. This latter scenario considered the implementation of a 7 log10 reduction guideline, such that the sum of log virus removal in the treatment system, disposal system and buffer were a minimum of 7 log10. The implementation of the log reduction guideline was also modelled with the inclusion of disposal system failure.

Table 6 Summary of scenarios

Scenario Description 1 Normal Baseline: all system types with their associated failure rates 2 Double population Total population in each catchment doubles 3 Outbreak An outbreak scenario was modelled as a doubling of the community infection rate 4 Minimum 100 m setback All setback distances less than 100 m are increased to 100 m 5 Minimum 100 m buffer with All setback distances less than 100 m are double population increased to 100 m, and the total population in each catchment doubles 6 All AWTS All systems are assumed to be AWTS 7 All AWTS with double All systems are assumed to be AWTS, and the population total population in each catchment doubles 8 All septics All systems are assumed to be septic tanks with absorption trenches 9 All amended material soil All systems are assumed to be amended material mounds soil mounds 10 No failure All system types are included, but failure rates are assumed to be 0 for all systems 11 Minimum 7 log10 reduction All systems with buffers which provide less than 7 log10 reduction are increased to provide 7 log10 removal 12 Double population 7 log10 All systems with buffers which provide less than reduction 7 log10 reduction are increased to provide 7 log10 removal, and the total population in each catchment doubles 13 Minimum 7 log10 reduction All systems with buffers which provide less than with disposal failure 7 log10 reduction were increased to provide 7 log10 removal, 14 All 7 log reduction total All systems provide exactly 7 log10 reduction

Page 57

Sensitivity assessment was undertaken as part of the model development to assess the relative significant of the model parameters. It was undertaken on a single system basis by section, with all parameters in one section compared with each other, but only the most sensitive from each section compared across the model. The modelling and sensitivity analyses were undertaken in @Risk V4.5 (Palisade Corp. USA) within Microsoft ExcelTM 2003, using 10 000 iterations.

Page 58

6 Short Summary of Results

This chapter provides a brief summary of the results from all the papers at the end of this thesis.

6.1 Paper I - Australasian standards for on-site sewage management The Australian/New Zealand Standard for on-site systems (AS/NZS 1547:2000) provided an improved basis for the design of on-site disposal system compared to its forerunner. However it provides limited direction regarding pathogens or nutrients. AS/NZS 1547:2000 does not specify buffer distances, but places the responsibility on local authorities to develop buffer distances appropriate for local conditions. Unfortunately, the lack of resources in such local authorities means that they are likely to defer to prescriptive generalised set-back distances specified in regional guidelines which show limited consideration of soil type, slope and local conditions. Therefore, while AS/NZS 1547:2000 provides flexibility it does not provide appropriate tools to support the necessary decisions. Hence, more research on buffer distances is required.

This paper also identifies the scale of the issue in the Sydney drinking water catchments. A survey of local government authorities identified over 18 000 systems located in the catchments; the majority of which were septic tank-absorption trenches (73 %) followed by AWTS (20 %) and pump out systems (6 %).

6.2 Paper II - Buffer distances for on-site sewage systems The main concerns from on-site sewage disposal were identified by stakeholder consultation as phosphorus, nitrogen, human enteric viruses and parasitic protozoa, and the critical need as buffer distance research to assist management control development within the catchments.

A QMRA approach was selected as the preferred method of delineation of buffer distances. To set an appropriate buffer distance it was concluded that risk implications to the catchments must be quantified, and that buffer transport data must be integrated into the wider context of catchment management to allow cumulative impact assessments. The risk management model was developed to identify the steps required to ensure that a suitable, scientifically defensible, buffer zone can be consistently applied between on-site systems and waterways. Using this model it is planned to

Page 59

calculate site specific performance-based buffer distances which consider the sources of risk as they relate to on-site system management, treatment, disinfection, land application and site factors.

The results of a survey of effluent quality in 40 septic tanks (SCA unpublished data, Ferguson et al., 2009), servicing a range of residential, public and commercial premises, were reported that identified an incidence rate of noroviruses of 34 % with a further 13 % in which the analysis was inhibited (no result possible, contamination possible), however the population size cannot be estimated due to presence of public amenities in the systems sampled. Rotavirus was only detected in effluent from one septic tank, but inhibition counted for 48 % of the samples analysed.

The pilot groundwater study investigated two sites at Robertson. One site, a residence with a septic tank and absorption trench, had groundwater at 3-4 metres, with a hydraulic conductivity of 0.1 m.d-1. The nitrate plume on this site was measured over 12 metres from the trench with a maximum nitrate concentration of 7.0 mg.L-1. At the other site, a residence with an aerated wastewater treatment system with spray irrigation, groundwater occurred at a depth of 1-2 metres, with hydraulic conductivities between 2.3 and 7.4 m.d-1. The nitrate plume was measured over 20 metres from the spray field with a maximum nitrate concentration of 26.2 mg.L-1 and an average concentration at 20 m of 9.8 mg.L-1. Human enteric viruses were detected in 40 % of septic tank effluent samples.

6.3 Paper III - Designing on-site sewage disposal systems The concentrations of viruses in septic tanks were calculated to range from an average peak concentration of 103 up to 107 noroviruses L-1 and 108 up to 1010 rotaviruses L-1. These calculations were based on one infected individual at a household served by a 3 000 L septic tank receiving 1 000 L.d-1 of wastewater, which were considered representative of modern Australian household systems. These estimations did not include potential virus inactivation or removal within the septic tanks.

Paper III also presented a review of critical control points for the management of on-site sewage treatment and disposal systems, based on maximising virus inactivation and minimising virus transport. Critical control points are key elements within a multiple barrier system where management can impact on water quality, the environment and

Page 60

public health. The critical control points identified, generalised for a range of systems such as septic tank – absorption trench systems and AWTS, are summarised in Table 7.

Table 7 Conditions and actions for improved virus management for on-site systems

Condition Action Main Effect Moisture Increase trench surface area; Decreasing saturation increases content Reduce irrigation loading rate; virus inactivation Improve distribution system to ensure uniform distribution of effluent Salt Use run-off mitigation measures Run-off mitigation reduces the concentration/ around disposal areas wash-off of viruses and reduces rainfall remobilisation of adsorbed viruses Organic Increase treatment and inform Decreasing organic matter matter household on chemical/surfactant increases adsorption use to reduce organic matter prior to land application Hydraulic Use alternative disposal systems, Mitigates the reduction in virus conditions such as alternating between two or adsorption capacity of the soil more trenches, or moving the experienced with prolonged sprinkler system regularly sewage application

6.4 Paper IV - Impacts of centralised versus decentralised systems Based on the available data, the ‘expected’ scenario (i.e. under average conditions) in the Sydney drinking water catchments was calculated to be that decentralised systems dispose of similar total loads to land that centralised systems dispose of, usually to water, for phosphorus (36 040 kilograms per year for decentralised systems versus 37 090 for centralised systems), Cryptosporidium oocysts (3.1 x 1010 per year for decentralised systems versus 1011 for centralised systems) and enteric viruses (8.4 x 1013 per year for decentralised systems versus 9.1 x 1013 for centralised systems), but higher loads for nitrogen. However, the loads of all nutrients and pathogens were higher from decentralised systems (disposal to land) than from centralised systems (disposal to waterways) under the worst-case scenarios. The implications of these results for catchment water quality are affected by the disposal methods and failure issues. Decentralised system disposal to land may afford a degree of mitigation that can be enhanced, if the degree of failure is reduced. Reducing failure rates is reliant upon adequate management, maintenance and design, based on soil type and topography. The

Page 61

high loads associated with the worst-case scenario indicate the significance of decentralised systems failure, which is due to the nature of treatment systems and poor management.

6.5 Paper V - Effluent quality from 200 on-site sewage systems Results from sampling of 46 septic systems highlighted the poorer quality of septic tank effluent compared to guideline assumptions and literature values, with only 13 % of systems within the limits of the current NSW guidelines for SS and 33 % for BOD. Similarly nutrients levels were higher that the guidelines in 77 % and 60 % of systems for total nitrogen (TN) and total phosphorus (TP) respectively.

Results from 141 AWTS were similar to the NSW Health Accreditation requirements for SS, BOD, thermotolerant coliforms and chlorine. It is anticipated that as for septic tanks, high nutrient loads will also be present in AWTS. In total 51 % of AWTS failed to comply with, of which 43 % were due to failure to comply with free chlorine concentrations. Mechanical failures were also observed with non-operational disinfection system in 2 of 24 systems (8.3 %), and blocked treatment systems in 2 of 21 systems (9.5 %).

Nutrient loads in sewage are used as a basis for effluent disposal area calculations. Hence the implications of the higher nutrient loads were assessed. Flow data from one residential septic tank was used to calculate the equivalent nutrient loadings to the land application area, and despite flow being considerably lower than guideline flows, the nutrient loadings were higher than the loadings calculated from guideline concentrations and flows (Table 8), indicating that disposal systems may be being under-sized.

6.6 Paper VI - Disinfection in Aerated Wastewater Treatment Systems From 14 AWTS with chlorine disinfection, thermotolerant coliforms were within guideline limits on all samples, with chlorination achieving, on average, a 3.6 log10 reduction of thermotolerant coliforms and 3.7 log10 reduction of E. coli. By comparison, the 13 AWTS with UV disinfection studied achieved a lesser reduction of thermotolerant coliforms, with three systems failing to meet the NSW Health (1998) thermotolerant coliforms criteria of 100 cfu.100mL-1.

Page 62

Table 8 Comparison of actual septic tank effluent nutrient loadings with guideline predictions

System Average flow Concentrations Site loads TN TP TN TP (L.d-1) (mg N.L- (mg P.L-1) (g.d-1) (g.d-1) 1) Residential 526 (132 L per person) 123 22 65 12 septic results Guideline 720 (180 L per 50 - 60 10 - 15 36 - 43 7 - 11 predictions person)* 900 (300 L per 50 - 60 10 - 15 45 - 54 9 - 14 bedroom)** * (Standards Australia, 2000); ** (DLG, 1998)

However, UV disinfection performed as well or better with regard to other microorganisms studied, including enterococci (3.0 log10 reduction with UV versus 1.6 log10 with chlorine), Clostridium perfringens (0.6 log10 versus 0.5 log10) and MS2 bacteriophage (1.8 log10 versus 1.0 log10).

Disinfection by-products were only measured in the chlorinated effluents. Post- disinfection, the concentrations of disinfection by-products were some two orders of magnitude above recommended guidelines for drinking of 0.25 mg.L-1 trihalomethanes, (NH&MRC et al., 1994).

6.7 Paper VII - Fate and transport in a mound system The routine monitoring of an amended mound system resulted in 98 % total phosphorus removal and 14 % total nitrogen removal from effluent. Thermotolerant coliforms th th removal was high (2.9 to 4.5 log10 removal at the 20 to 80 percentiles respectively) which was attributed to a combination of inactivation, filtration and absorption. There was less removal of other indicator organisms such as somatic coliphages (0.5 – 2.5 log10) and Clostridium perfringens (1.1 – 3.7 log10). The pH averaged 10.3 within the mound system, and 8.2 in the effluent.

Survival experiments were undertaken with MS2 and PRD1 bacteriophage in effluent sampled from the mound. MS2 inactivation was rapid, with 5 log10 reduction over two days. PRD1 inactivation was significantly slower, with less than 1 log10 removal over

Page 63

28 days. Transport experiments were undertaken with PRD1 in a 50 cm column using ‘fresh’ amended material and artificial rain water. Rapid breakthrough occurred after

0.34 pore volumes with a 0.36 log10 reduction in PRD1 concentration and 59 % recovery in effluent of the inoculated phage.

Inoculation of the mound was undertaken with MS2 and PRD1 in separate experiments. The average residence time, from inoculation of the mound influent with Rhodamine, was approximately 18 days in the sand mound, or equivalent to 28 kL which is roughly equivalent to the volume of the void space of the cell and the influent distribution trench below the level of the liner. Large reductions in MS2 corroborated the findings of the inactivation study, although the inactivation results predicted > 5 log10 reduction in four days compared to the 10 days in the field. A reduction of 2.9 log10 in PRD1 over the

1.8 m of the mound was achieved, with 0.7 log10 attributable to inactivation. Extrapolation of the laboratory column experiments using HYDRUS-1D predicted a reduction PRD1 concentration of 0.53 log10 over 1.4 m transport (0.73 log10 over 1.9 m transport). This low reduction, compared with the 2.9 log10 experienced in the field, was attributed to the conditions within the column, including higher flow rate, vertical flow rather than horizontal, use of rainwater instead of sewage, and no clogging layer. The lower reduction experienced in the column is expected to be more representative of new systems and systems under high loads.

6.8 Paper VIII - Virus fate and transport: laboratory and field studies The fate and transport of PRD1 and MS2 bacteriophages were studied in five soils from the Sydney drinking water catchment area ranging from sandy loams to clay loams.

There was limited removal in the sandy loam 0.5 m soil columns: 0.0 log10 and 0.2 log10 reduction of PRD1; 0.1 log10 reduction of MS2. There was greater removal in clay loam and loam soil columns but with high variability between them: 1.1 log10 and 1.9 log10 reduction of PRD1 and 1.2 log10 and 1.3 log10 reduction of MS2 in clay loams; 3.0 log10 reduction of PRD1 in loam.

Fate and transport of septic tank effluent, including PRD1 bacteriophage, was also undertaken at two field sites (a clay loam and a loam from the column experiments). At the clay loam site, rapid PRD1 breakthrough in groundwater was observed, albeit at a 5 log10 reduction. Phage concentrations increased over 5 days to a plateau of

Page 64

approximately 500 pfu.mL-1, which was sustained for over two weeks. The reduction in concentration was 3.8 log10 over 3 m compared to 2.8 log10 for routinely monitored somatic phage. Removal over 7 m was greater than 6.6 log10. No breakthrough was detected at the loam site.

The column results for the clay loam soil predicted greater removal of viruses than were experienced in the field. The decreased removal in the field was assumed to be due to a combination of much greater heterogeneity in the repacked soil columns and the preconditioning of the site with effluent and viruses, and may also have been influenced by the differences in water types in the laboratory and field experiments. Conversely, the column results for the loam soil predicted less removal than experienced at field- scale.

6.9 Paper IX QMRA: buffer distances for septic systems The QMRA was the culmination of the research undertaken in the rest of the papers reported here. It was developed based on research into other regulatory tools (Paper I; section 3.1) and in consultation with a broad audience (Paper II). It builds on the issues of the potentially high virus concentrations in on-site sewage systems and the importance of adequate management of on-site systems to protect water resources raised in Paper III and Paper IV, respectively. It utilises the data gathered about the effluent quality in septic tank (Paper V), AWTS (Paper V, Paper VI), and amended material soil mounds (Paper VII). And it relies on the modelling of virus transport in the catchments from the column and field work undertaken in Paper VIII.

The sensitivity analysis of each model parameter identified the virus loads excreted by infected individuals as the most sensitive parameter. On-site system type was not significant. In wet or dry conditions, the greatest parameter sensitivities were in order: buffer distances, slope, saturated soil removal, unsaturated soil removal (for dry conditions), conservative and maximum flows, and failure of removal in the saturated zone. System type differences resulted in average loads from sites with septic tanks that th were 0.9 log10 higher than for sites with AWTS (>2 log10 at the 95 percentile), 1.9 log10 higher than for sites with a septic pump out, and 0.8 log10 higher than for alternative systems.

Page 65

A total of 18 149 on-site systems (28 595 cadastral lots) in the catchments were identified. The calculated buffer distances, i.e. the shortest straight line distance between a centroid and a waterway, were 150 ± 132 m with a range from zero up to approximately 850 m in most of the four regions, but up to 1 842 m in Mulwaree region. In total, 44 % of existing systems had calculated buffer distances of less than 100 m. The tolerable load of viruses was calculated to be 108 virus per year in raw water and 104 viruses per year in treated water. The 80th percentile of predicted infections in Sydney’s drinking water consumers from virus contamination from sewage contamination in the drinking water supply was less than 10-4 infections per year (Figure 7).

0.250 20th percentile 80th percentile 0.200

0.150

0.100 Frequency 0.050

0.000 -16 -12 -8 -4 0

Log10 Pinf

Figure 7 Log10 Pinf for from drinking water ingestion based on virus input from on- site sewage systems with (black) and without (grey) the inputs from STPs (percentiles indicated are for with STP)

A number of different management options were explored, and in particular, management options that were considered feasible tools to be used as part of assessing development applications. The management options included fixed buffer distances, adoption of one type of treatment system, reduction of system failure, and application of a fixed log reduction that included treatment and buffer distance removal based on the models of virus removal in subsurface and surface transport models. The impact of differing management scenarios on the log10 load of viruses from Wingecarribee region

Page 66

(Table 9) illustrated that a minimum buffer distance of 100 m or a zero failure rate had the greatest reduction of virus load at the 95th percentile, down to zero virus load. However, these management options did not achieve the greatest reductions at the 99th percentile and of maximum loads. By comparison, the 7 log10 reduction resulted in lower loads at the 95th and 99th percentile and maximum.

Table 9 Log10 virus load per year from Wingecarribee region, as modelled for the drinking water offtake at Warragamba Dam, under different management scenarios

Scenario Percentile performance 95% 99% Max 1 Normal 1.6 5.9 8.7 2 Double population 3.9 6.8 9.3 3 Outbreak: double rate of infection 4.2 6.7 8.8 4 Minimum 100 m buffer 0 4.3 8.7 5 Minimum 100 m buffer with double population 3.0 6.2 8.9 6 All AWTS 0.3 4.7 9.0 7 All AWTS with double population 3.0 5.5 8.9 8 All septics 1.8 6.1 8.6 9 All amended material soil mounds 0.9 5.1 8.1 10 No failure 0 4.4 7.8 11 Minimum 7 log10 reduction 0.9 2.0 3.4 12 Double population 7 log10 reduction 1.5 2.3 3.6 13 Minimum 7 log10 reduction with disposal 1.9 5.2 10.8 failure 14 All 7 log reduction total 1.7 2.5 3.6

Page 67

7 Discussion

Attempts to define appropriate buffer distances by length alone ignore the importance of soil in effluent treatment, and the variability of treatment between different soil types. Furthermore, attempting to define buffer distances based on virus transport need to include a practical endpoint. It is not feasible to use zero viruses as an endpoint. One virus may be able to cause one infection, however, on a catchment scale this is not significant. To set an appropriate buffer distance for viruses the risk implications to the catchments must be quantified, and buffer transport data must be integrated into the wider context of catchment management to allow cumulative impact assessment.

It is important to remember that the water quality of the catchments at present is considered acceptable, and that the information gained through this work is for use in future development control and planning decisions and for highlighting high priority areas for upgrades and improved management. The research presented is not an assessment of catchment health.

To estimate the 95th percentile with confidence considerably more data is required that for an 80th percentile, so in part, the acceptable criteria will rely on the precision available from the current and ongoing work.

The definition of what is a tolerable buffer distance adopted here was based on the assumption that a tolerable public health risk from a single source is one infection per 10 000 for a person in a year (Regli et al., 1991). QMRA predicts the virus loads from the catchment at levels below what is measurable. Similarly the infection levels that are targeted are below the level of sensitivity for epidemiological studies, e.g. about 10 % (Hellard et al., 2001). The tolerable concentrations of viruses in treated drinking water for Sydney (10-7 .L-1) calculated by the QMRA were comparable with that calculated in Regli et al. (1991).

The virus loads from decentralised and centralised systems calculated in Paper IV were significantly higher than those calculated in the QMRA model, as the QMRA included removal, whereas in Paper IV calculations were based on limited removal by treatment, and no buffer distance removal for decentralised systems.

Page 68

The QMRA results for virus infections were equivalent to people excreting into septic tanks on 7.3 % of days, or an infection rate of approximately 1.2 % based on 6 days of virus excretion.

Due to the high concentrations of viruses in effluent, this parameter had a large influence on the QMRA results. There was limited data available on the concentrations of viruses in septic tanks or other on-site sewage systems, and no information on the temporal distributions of viruses during an infection of an individual in a household. Quantification of the maximum concentrations and temporal distribution of virus concentrations in a septic tank during an infection would present a number of challenges: prior permission; notification of illness; response time between notification and sampling; access for sampling; as well as requiring the infectious agent to be viral and detectable. Hence, it is not surprising that the concentrations of viruses detected in septic tanks are lower than expected, for example one study reported detections ranging from 0.07 viruses L-1 to greater than 59 L-1 for household septic tanks (1993). Similarly low concentrations have been reported for communal facilities, with concentrations of 0.26 to 4.4 viruses L-1 in a school septic tank serving 350 people (1998b). These results are also significantly lower than those reported for centralised systems which range from 104 – 105 per L (Gerba, 2000a). However, high concentrations of Cryptosporidium (average 8.7 x 104 oocyts.L-1) have been reported in on-site sewage systems, several order of magnitude higher than those reported in literature for centralised sewage (Ferguson et al., 2009).

The concentrations of viruses in septic tank effluent will also depend on the type of virus. One enteric virus that is shed in high numbers is norovirus. Peak loads of norovirus during infection have recently been reported to be as high as 12.2 log10 genomic copies/g faeces, with a median from the 16 participants of 11.0 log10 genomic copies/g faeces (Atmar et al., 2008). This would be expected to result in high loads of viruses in septic tank effluent even with high solids removal, and with advances in detection methods in recent years it may be feasible to do further research in the future to assess norovirus loads in septic tanks during infections in the household. Due to the difficulty in quantify the loads in septic tank effluent described above, it is to be expected that predictions of virus loads will be higher than the levels detected. Hence,

Page 69

the disparity between the modelling and reported measurements is not a concern, but it is an area where further research is required.

A survey undertaken by SCA, to quantify the presence of a range of viruses and protozoa but not the concentrations, identified a surprisingly high proportion of positive results with 40 % of systems positive for at least one type of pathogen. The results for one communal facility, a child day care centre (SCA, unpublished data), indicated the presence of enterovirus, norovirus and Cryptosporidium. Of 39 systems studied, five of the 14 detections of norovirus, one of the two enterovirus detections, and the sole reovirus were at the eight community systems. The high detection rates may relate to the number and duration of infections as well as the length of retention of viruses in septic tanks, which has been reported to be up to 137 days in one study (Lewis et al., 1993).

The result from the QMRA predicted an average of 1.3 x 10-7 infections with the modelled virus due to Sydney drinking water contaminated with viruses discharged to the catchments from on-site systems.

7.1 System performance Limited data was available on the performance of septic tanks with regard to virus removal, and no data was available on how the treatment performance of septic tanks affects virus removal. The reported virus removal in septic tanks is comparable to the reported virus retention in sewage sludge for primary treatment (Ferguson et al., 2003). Hence, the primary mechanism for removal of viruses in septic tanks was assumed to be physical settling of solids-associated viruses, and therefore one of the few markers of septic tank failure is the build-up of solids. A properly maintained tank is expected to have a settled solids layer less than a third the depth of the tank, allowing adequate capacity for the tank to function. The build up of solids, and the associated reduction in capacity, will reduce the solids removal in the tank and eventually result in the overflow of settled solids. In the model, this performance failure is assumed to reduce virus removal to zero. The estimated failure of septic tanks (10 %) was considerably lower than the proportion of septic tanks that had suspended solids concentrations above the NSW guidelines (87 %) (Paper V). For the studied systems, no data was available on influent concentrations for determination of the proportion of solids removal, and this

Page 70

high proportion above the guidelines was considered to be greater than the proportion that had no solids removal.

Similarly to septic tanks, the assumed 1 log10 virus removal in AWTS corresponds with solids removal of 80 – 90 % in activated sludge treatment (Ferguson et al., 2003). Failure rates were based on field experiment results for complete treatment failure such as blockages and visible solids overflow (Paper V). The rates of failure for SS and BOD, which are assumed to represent poor performance of the treatment system, compared to the NSW Health guidelines (1998) were 15 % each. However, as for septic tanks, the lack of data on influent concentrations or correlations between solids and virus removal performance resulted in the above assumption about complete failure.

Connectivity is defined as where sewage disposal is effectively direct to waterways through either accidental or deliberate means. No data was available on connectivity in failure of disposal systems. During the process of system registration inspections, council environmental health officers have reported incidents of drains built directly from the tank to a neighbouring stream (Tina Dodson pers. comm.).

There are a number of concerns about the use of surface spray irrigation as an effluent land application method. Surface application has the capacity to increase virus inactivation due to exposure to sunlight, as well as increasing nutrient uptake by plants, however, run-off of pathogens and nutrients is a concern. Surface application can also increase public health risks through direct human contact with sewage and aerosols.

One alternative system, an amended mound, was studied as part of this thesis. The performance of the experimental amended material mound provided greater removal of viruses from sewage than reported for other on-site sewage treatment systems. However, virus removal was significantly lower in a repacked column experiment using the amended material used in the mound than in the experimental mound. In the column study, the amended soil was not preconditioned and therefore was expected to have little microbial activity, and hence low inactivation (Jin et al., 2002). Additionally, artificial rainwater was used in the experiment instead of sewage and at a high flow rate.

The use of rainwater, rather than sewage, and high flow rates in the columns, were assumed to provide a worse-case scenario. The artificial rainwater was sterile which

Page 71

reduced inactivation due to predation. This sterility as well as lack of organics and solids found in sewage, reduced any clogging that would occur in the field and lower ionic strength reduced attachment. Rainwater also represented the water quality influxes during times of heavy rainfall or high loading periods (e.g. heavy use of washing machine) which are known to promote virus transport (Paper III). However, there are equally components in sewage that can promote virus transport. Virus attachment to colloids can decrease inactivation and facilitate transport (Schijven, 2001; Jin et al., 2002). Organic matter, including surfactants in the water have been reported to decrease adsorption due to competition for attachment sites and are also reported to increase desorption of viruses (Schijven et al., 2000).

Inactivation was not a significant contributor to removal in the column. Inactivation of PRD1 in the soil-water matrix without preconditioning was lower than in effluent from the soil matrix. Virus removal within the sewage treatment mound system that utilises this soil was greater in the field experiments (1.1 log10 in 0.7 m) than experienced in the column experiments carried out in the present study (0.36 log10 in 0.5 m).

In field experiments, flow through the mound system was considerably below the design flow. Increases in the loading on the system would be expected to result in a decrease in system performance. For example, phosphorus was found to be significantly (p<0.01) negatively correlated with flow, and microorganism inactivation and attachment would be expected to decrease during increased flows.

Modelled extrapolation of the column study to the field-scale mound system provided a prediction of virus transport in less well established systems, and under higher flow rates that were used in the column studies. At the reduced detention time modelled, seven days to maximum breakthrough compared to over 20 in the field, the removal was only 0.73 log10 over the 1.9 m width of the system. This was assumed to be due to the key differences between the column and field experiment:

• Rainwater, as used in the columns, was assumed to result in decreased attachment compared to sewage;

Page 72

• Heterogeneity in the field may increase travel velocity compared to the repacked columns, thereby decreasing attachment, and may affect removal over the site; and

• Lack of microbial activity in the microcosms, if extrapolated, may underestimate inactivation in the field.

The column study was therefore expected to underestimated removal in general field conditions. However, under high flow, or prior to clogging layer development, the results from column experiments may represent transport and hence a range of values were used for the QMRA.

The high pH environment within the mound strongly affected many of the indicator organisms studied. In the sand mound studied, the pH ranged from an average of 10.4 to a maximum value of 11.0. Human enteric viruses have been reported to be stable between pH 3 and pH 9 (Jin et al., 2002), although bacteriophages may be more sensitive (Grabow et al., 2001). It is not known how long this viricidal high pH may persist, but as it is also required for phosphorus removal it is a key management parameter. Furthermore, high pH is possibly more important than the availability of sorption sites, and calcium leaching is expected to result in a decrease in phosphorus sorption.

The modelled scenario where all on-site systems were amended material sand mounds adopted a range of performance based on those experienced in the field and column studies. The model did not achieve better results than modelling of the present scenario. However, analysis of the results of the field experiments did identify that these systems can achieve a better level of sewage treatment than septic systems and AWTS. More research is needed to quantify the relationship between performance and flow rate, the effect of poor mixing at installation, and the performance of the system over time, including monitoring and management of pH levels over time.

7.2 Buffer distance modelling The use of GIS enabled assessment of the cumulative inputs across the catchments based on actual locations within the catchments. Furthermore, it provided site specific data on slope, soil type and distance to waterways. Limited data availability on system

Page 73

the locations resulted in the assumption that the system was located on the centroid of selected plots, based on land use as well as survey data on the number of systems in a local government area. In practice on-site sewage systems are typically placed close to the house, so the location will vary depending on the size of the block. As the registration programme for on-site systems in NSW progresses, there will be improvement in data on the location of systems, at least in terms of cadastral parcel, and for each system information will be available on type, age, size of household (number of bedrooms), and the age of the system.

Due to limitations with what was able to be undertaken with the GIS program, the length of the buffer between the on-site sewage system and the nearest waterway was based on the shortest straight line distance. The surface travel distance would be improved through delineation based on the downslope flow distance. Buffer distance removal was extrapolated from novel but limited experiments. While attempts were made in the QMRA to take account of site and soil variability across the catchments, the complexity of the systems meant that it was not feasible to thoroughly address the changes in the site and soil between the assumed point of disposal and the assumed point of discharge into a surface water body. Heterogeneity in site and soil will impact virus transport on a small scale that cannot be resolved through GIS databases, such as through the presence of preferential flow paths discussed below. This is one of the issues of variability that the risk assessment is designed to address.

The modelling of virus transport in overland flow and in subsurface transport was based on the available data. One of the limitations of this arose from the low removal of viruses in the sandy loam soil columns. The resulting estimate of virus removal in sandy soils used in the QMRA therefore assumed equally low removal of viruses in sandy soils, such that the subsurface virus transport in areas with sandy soil was further than for surface transport. This outcome is counter intuitive due to the filtration provided by sandy soils. However, similar results to the low removal experienced in the sand column experiments has previously been reported in several studies (Powelson et al., 1994; Jin et al., 1997; Schijven et al., 2002a). There are few studies on the transport of viruses in overland flow. The available data for overland flow was limited to a single type of rain event, and would be likely to vary depending on the intensity of the rain and the volumes of resulting runoff. However, this data was from runoff on bare soil, thereby

Page 74

overestimating the transport that would occur on grassed surfaces. Because of the evidence from literature that there is low removal of viruses in sands compared to other soils, the robust albeit limited data on virus transport in overland flow, as well as the fact that many areas of sandy soils in the area are underlain by fractured sandstone, this modelling result of greater transport in sandy soils than in overland flow has not been altered.

McLeod et al. (2001) studied the transport of the bacteriophage Salmonella typhimurium 28B in four undisturbed, unsaturated 700 mm soil lysimeters, including coarse sand, sand, silt and clay. They reported the greater phage removal in the coarse sand (pumice) and silt, than in the sand which peaked after 0.3 pore volumes at 4 % of the spike concentration, and the clay which peaked after 0.1 pore volumes at 80 % of the spike concentration. The rapid transport in the water-repellent sand was attributed to the formation of finger-flow pathways, and was therefore assumed to be similar to removal under saturated flow conditions. The rapid transport in the clayey soil was attributed to bypass flow through macropores and fractures in the soil; an effect of soil heterogeneity that is well documented.

Physical heterogeneities, such as roots, macropores and rock fractures, can provide preferential water flow paths (McKay et al., 1993), while chemical and biological heterogeneity can also affect virus adsorption and transport (Rehmann et al., 1999). Preferential transport of viruses due to size-based pore exclusion can result in viruses travelling at statistically extreme velocities (Taylor et al., 2004). The use of sieved, disturbed soil columns reduces the impact of heterogeneity, and therefore, introduces several limitations to the direct application of the results to the field-scale. Sieving the soil increased the homogeneity of the soil, limiting the effects of soil structure, which may vary considerably across the site, with results providing an overall view of the soil- virus interactions. Removal of soil pods, rocks and roots would have also decreased the hydraulic conductivity of the soil, thereby increasing the contact time between the viruses and soil attachment sites. Furthermore, the removal of macropores and fractures restricts preferential flow, which can have a greater influence than the adsorption capacity of the soil at field-scale (McLeod et al., 2001). Repacked column experiments where the soil has been sifted to achieve a more uniform size distribution, such as were undertaken here, will therefore tend to overestimate virus removal, and that care should

Page 75

be taken when attempting to extrapolate the results to field scale. Soil columns by their nature introduce paths for preferential transport along the soil-column interface, however as repacked column experiments aim to establish the adsorption capacity of the soil and its influence on virus fate and transport, every attempt is made to minimise their influence.

The use of rainwater was assumed to provide a worst-case scenario for transport due to low ionic strength resulting in low attachment. Rainwater has been shown to increase the transport of viruses from sewage in soil (Wellings et al., 1975). However, there are several components in sewage what can also increase virus transport that where not present. For example, in sewage there is the potential for viruses to attach to colloids which may decrease inactivation and facilitate transport (Schijven, 2001; Jin et al., 2002). Furthermore, organic matter including surfactants in the water/sewage has been reported to decrease adsorption due to competition for attachment sites and is also reported to increase desorption of viruses (Schijven et al., 2000; Chattopadhyay et al., 2002). The relationship between virus transport with rainwater and with sewage was not studied.

Virus adsorption has generally been found to be non-linear with distance (Schijven et al., 2000), with the rate of removal expected to decrease with increasing distance. One of the main causes suggested for this is the soil/matrix heterogeneity. Soil physicochemistry affects virus adsorption, with virus retention tending to be greatest in soils with higher cation exchange capacity, exchangeable iron, iron oxides and specific surface areas (Schijven et al., 2000). This is due to a greater number of sites available for attachment. Similarly soil organic matter can provide hydrophobic adsorption sites in sandy soils, but will reduce adsorption sites in general (Schijven et al., 2000). Hence, heterogeneity in the soils will cause variation in removal over distance. Furthermore, the injection or infiltration of sewage into soil will create an organic matter gradient, which will affect virus transport. Other potential causes of non-linear removal are variability between subpopulations of viruses (e.g. variability in surface charge/coat); or colloid assisted transport. These non-linear relationships are expected to affect the scaling up of results from the column to the field-scale, and hence, field-scale experiments were undertaken to compare with the results of laboratory-scale experiments.

Page 76

While the experiment aimed to address virus fate and transport during land application of effluent from an on-site sewage system, artificial rainwater was used in the experiment instead of sewage. As discussed above, this was assumed to provide a worse-case scenario for virus fate and transport based on the assumption that lower ionic strength increases transport more than organic constituents in sewage. However, presence of surfactants in sewage can also increase virus transport (Dizer et al., 1984).

There were several limitations with the field experiments, including difficulties obtaining sufficient sites, interference from domestic animals and variable hydro- geological conditions. Nonetheless, field experiments were considered an important missing component, due to previous findings that laboratory column studies overestimate virus removal during soil transport (Curry, 2000). This overestimation is assumed to be due to a combination of factors including less heterogeneity in the laboratory soil columns, preconditioning of soil with sewage in field sites, and the fact that removal is typically not linearly correlated with distance.

The results from the column experiments were compared with those of the field experiments, with the aim of extrapolating to virus transport in soils at a catchment scale for application to the delineation of setback distances for septic tanks. The removal of PRD1 in the homogenous repacked soil columns (1 log10 reduction over the

0.5 m column) overestimated the reduction experienced in the field (3.8 log10 reduction in 3 m and >6.6 log10 in 7 m). The decreased removal in the field was assumed to be due to a combination of heterogeneity, the effluent quality, and the preconditioning of the site with effluent and viruses. Factors such as dilution, dispersion, organic matter and a biomat in the absorption trench, which would increase virus reduction compared to the column experiment, were apparently less significant than the factors that increased transport listed above.

The buffer removal model was developed from extrapolation of laboratory and field results. Aerosols were not included within the buffer distance transport model, as they were considered to be of low relative significance to other transport pathways considered, and therefore unlikely to significantly impact the loads of viruses in water.

Page 77

7.3 On-site sewage system management options The QMRA presented here provides a tool to quantify the infection risk to consumer from Sydney’s drinking water. However, application of this model as for assessing development applications is unfeasible. Hence, a number of different management options were explored and, in particular, management options that are feasible tools to be used as part of assessing development applications. The management options included fixed buffer distances, adoption of one type of treatment system, reduction of system failure, and application of a fixed log reduction that includes treatment and buffer distance removal.

Buffer distances In the Wingecarribee subcatchment, 39 % of existing systems had a buffer distance less than that adopted by the SCA of 100 m from permanent or intermittent waterways.

Increasing buffer distances up to a minimum of 100 m achieved a reduction of 1.6 log10 at the 95th and 99th percentile for the Wingecarribee region. It is worth noting that the buffer distance removal model predicts 7.0 log10 removal over 100 m in clayey loam soils, and 4.6 log10 removal in sandy soils. Virus transport over this distance is possible (Curry, 2000; Pang et al., 2005), although it would be expected to be at low concentrations generally. The failure of disposal systems was assumed to remain unchanged, which resulted in cases with high levels of transport. The practicalities of implementing a uniform buffer distance will include the issue of how to deal with existing sites within the 100 m buffer zone, including sites which are planning developments as well as existing on-site sewage systems.

Buffer distances vary nationally (see Table 2) and internationally. As discussed above, in New York the required buffer distance is approximately 30 m (100 ft) although viruses have been shown to be mobile over that distance (Curry, 2000). In New Zealand, filter factors for viruses and bacteria were used to define appropriate buffer distances for groundwater transport to achieve 7 log10 reductions (Pang et al., 2005). The results ranged significantly based on the different types and conditions of aquifers: from 33 – 61 m in clean sandy fine gravel aquifers to 1.7 – 3.9 km in contaminated coarse gravel aquifers. In this case, the 7 log10 reduction was based on a single source of viruses with an expected concentration in sewage of 104 enteroviruses per L and a drinking water standard of zero enteric viruses per 100 L. Similarly for surface waters,

Page 78

setback distances (based on subsurface flow only) were assessed based on recreational water requirements for E. coli (< 126 per 100 mL) and the drinking water guidelines for viruses (Pang et al., 2004). Modelling results suggested that the minimal setback distances were 16 m for E. coli and 48 m for enteric virus, in this case based on 105 viruses per L, which reflects the greater mobility of viruses in groundwater.

There are a number of limitations to these studies in New Zealand that are considered to be addressed in the QMRA model presented here. The setback distances above are based on a single virus source, whereas there is the potential for many sources, particularly where there is a community outbreak (e.g. the results of the increased infection scenario in Table 9). The setback for surface waters does not include consideration of the potential for surfacing effluent and therefore transport in overland flow, which is a common failure mechanism in Australia. As discussed in Paper III, the concentrations of viruses in sewage, particularly in septic tanks, may be considerably higher than 104 - 105 viruses per L, although it is dependent on the virus type and studies have only reported relatively low concentrations to date due to the difficulty in timing sampling so that it coincides with an infection (e.g. Lewis et al., 1993; Deborde et al., 1998a). Also the tolerable concentrations of viruses have been predicted to be much lower than the limit the detection (Regli et al., 1991). While a 7 log10 reduction is considered appropriate to safeguard drinking water quality in the Warragamba catchment, which is over 9 000 km2, it may not be appropriate on a smaller scale where the additional virus removal from dilution, inactivation and drinking water treatment does not exist. In particular, the 7 log10 reduction specified here is not designed to, and unlikely to, protect bathing water quality on a local scale where there is an outbreak within that community.

Preferred systems One clear management option is for a regulatory authority to stipulate a preferred system, particularly for high risk sites. To assess this option the difference in virus loads to the catchments from septic tanks, AWTS and amended material sand mounds where compared. Of these, the greatest reduction at the 95th and 99th percentiles in virus loads compared to the current base scenario (73 % septic tanks, 20 % AWTS, 6 % pumpout, 1 % alternative systems) was from AWTS. AWTS were assumed to use spray irrigation for disposal. In wet weather, this resulted in transport in overland flow, which was

Page 79

modelled to provide less removal than in groundwater, and therefore potentially higher virus loads transported to waterways. The amended material soil mound was assumed to have no complete failure. The broad distribution of treatment performance based on field and laboratory experiments was assumed to account for performance failure. The column studies, with high flow rates, low organic matter, and little virus removal, were assumed to represent treatment in new systems, before establishment of clogging layers, and under high flows. While the field studies, which had lower flow rates, higher organic matter, and higher virus removal, were assumed to represent treatment in well managed, established systems. However, complete failure (i.e. no removal or much lower removal of viruses) could potentially occur where the impermeable liner is breached, allowing infiltration to groundwater before treatment is complete, or poor construction resulting in heterogeneity in the amending material distribution, providing transport pathways at a lower pH and with less attachment sites.

System failure Removal of all failure in system performance and buffer systems achieved a reduction th th of 1.6 log10 and 1.5 log10 at the 95 and 99 percentile, respectively, for the Wingecarribee region. This was comparable with the reduction for the uniform buffer distance above, however, in this case, there was considerable transport from sites with short buffer distances. In terms of implementation, the registration system for onsite sewage systems in NSW (DLG, 1998) will improve system performance in general by systematically identifying failing systems and providing education or enforcement to reduce failure. However, the assumption of no failure is not realistic.

The key types of failure that management should address were considered to be those that would result significant increases in the transport of viruses, namely no or reduced solids removal in treatment systems and transport in runoff rather than groundwater such as from surfacing effluent. Ensuring effective solids removal in treatment systems can be achieved through regular pump outs of the system which is undertaken or recommended to the householder as part of routine maintenance. For newer AWTS, this is currently achieved through quarterly service visits (NSW Health, 1998). However, for other systems and older AWTS this is achieved on an ad hoc basis by the householder, and can be prompted by council inspections based on complaints or the registration process. Centralised management of pump outs by councils would ensure that they are

Page 80

undertaken on a routine basis, and could be combined with a cursory inspection of the disposal field to identify the risk of surfacing effluent from subsurface disposal systems.

Log reduction The log reduction system developed here will provide assistance in implementing flexible buffer distances to deal with limitations on individual sites. The key benefit of the log reduction system for site assessment is the flexibility that it offers in the siting of new systems and adoption of new technology. This system provided a rapid method that is easy to apply on a site-by-site basis, with a simple algorithm for calculating virus removal in buffer distances based on soil type and slope. The model was tailored to the Sydney catchments, but could readily be modified for other locations. To improve the overall performance of on-site systems, an appropriate management system could be implemented on a site-by-site basis; for example, where there is uncertainty of the capability of the treatment system to routinely achieve the calculated log reduction, regular maintenance requirements should be stipulated. Furthermore, the log reduction systems enables novel systems to be installed based on estimates of removal from the literature, with monitoring criteria built into development application approval. Similarly, as a conservative estimation of removal in the buffer is used, where there are better soils, the removal can be increased based on proper testing and management/controls.

This log reduction system has been specifically designed and modelled for Sydney’s drinking water catchments. Application to other catchments may require further intensive modelling to calculate the appropriate level of removal. However, 7 log10 removal is in line with recommendations from WHO (2004). The QMRA methodology used can also be developed to refine log removals for buffer distances in different conditions and establish guidelines for nutrient removal. However, as for application to other catchments the further intensive modelling would be required.

While the methodology has been developed to be easy to use, it may pose more problems in its application, and the regulating of installations, than a traditional set-back distance approach. For this model, site conditions and soil types were estimated using the available GIS data, however the limitations of this data are recognised, and applying the methodology in the field should be based on field inspections. Hence, there is a level

Page 81

of training required to be able to undertake site inspections, evaluate the soil texture, evaluate the relevant slope and the likely flow of water, and identify how this will impact on the transport of not just the solids and nutrients in sewage, but also the pathogens. Similarly to other methods of assessing development applications for on-site sewage systems, consideration needs to be given to whether the system will be maintained and what will the impact be of failure of a component in the system.

The log reduction results highlight that, to achieve a reduced virus load from on-site sewage systems in the drinking water supplies examined, a combination of the following are required: appropriate buffer distances, appropriate system types and improved management to reduce failure. The log reduction method aims to address what are appropriate systems and buffer distances, based on the site characteristics, while providing information on which sites represent a high risk of impacting on catchment waters for prioritising management initiatives and stipulations.

7.4 Recommendations & Further Research The QMRA methodology presented here provided a useful tool for predicting the impact of on-site sewage systems on catchment water quality and public health that would not be possible via sampling or epidemiological methods. It has provided key criteria for which to base regulatory tools on, such as what virus concentrations are ‘tolerable’ in raw water, and how much virus removal is required before discharge to waterways.

There were a number of limitations identified during this research relating to the management of on-site sewage systems including the way failure is defined and managed. And it is these limitations that form the basis for the recommendations for improving on-site sewage management. A number of studies have reported on the failure of on-site sewage treatment and disposal systems to perform adequately. For AWTS in NSW, there are defined effluent quality criteria on which to base this assessment. However, for other treatment systems and for disposal systems there are limited, often subjective, criteria. Development of an effective, standardised method for failure detection during inspections would provide improved information on the potential public health and environmental impact. Installation of inspection ports in disposal fields, as are required in treatment systems, provide for monitoring of water

Page 82

levels and quality, reflecting the importance of the treatment capacity of the soil. Where systems are failing, guidance needs to be provided on how to rehabilitate the site, on subsequent monitoring of failing systems, or appropriate criteria for when to decommission, upgrade or replace a system.

The reduction in the failure of treatment systems can be achieved by undertaking regular inspections at appropriate intervals by adequately trained inspectors who are assessing the appropriate parameters. Regulation of this might include a register of when maintenance inspections were performance, such as for AWTS inspections, and accreditation of the inspectors.

Thermotolerant coliforms are widely used indicators of faecal contamination, which can be assayed using relatively inexpensive and rapid methods. However, for over twenty years, microbiologists have reported that thermotolerant coliforms will not reliably indicate the presence of viruses and parasitic protozoa. Coliforms do not survive as long, especially with systems using chlorination (Sobsey, 1989; Rennecker et al., 2000), and, being larger than the viruses, they are filtered out more effectively in subsurface environments (Lance et al., 1984; Scandura et al., 1997; Macler et al., 2000). Yet more suitable index organisms for these pathogens are not yet routinely used (Ashbolt, 2001).

AS/NZS 1547:2000, similarly to its predecessor, while recognising the inadequacy of thermotolerant coliforms as an indicator of public health risk still uses this as the sole microbial analyte, and places the responsibility to ensure health risks are tolerable on local authorities who tend to have limited microbiological expertise.

Testing for a range of faecal bacteria (including spore-forming Clostridium perfringens), as well as coliphages (viruses to selected coliforms), provides more reliable information on the possible presence of pathogens. Additionally, coliphages have been widely used as models of human enteric viruses in studies of fate and transport in groundwater (Deborde et al., 1999; Curry, 2000). The US EPA Ground Water Rule (2006) recommends E. coli, enterococci and coliphage be used as monitoring tools for faecal contamination. Adoption of a similar selection of microbes for assessing or accrediting the performance of new systems would provide a more robust estimation of performance, as highlighted by the considerably higher removal of

Page 83

faecal bacteria than viruses or bacterial spores in the amended material soil mound (Paper VII).

The results from Paper V highlighted that effluent quality from septic tanks was considerably more contaminated than what is currently described in design values used to size disposal areas. These higher values were assumed to be due to household changes, such as in the types of detergents used and the larger volumes of wastewater generated, and would be a contributing factor in disposal system failure. Increasing the design values for sizing of disposal areas such that they are comparable with the concentrations detected in septic tanks would help to reduce failure of these disposal systems. Furthermore, disposal system design should take account of the critical control points discussed in Paper III, including soil moisture content, salt concentrations, organic matter content and hydraulic conditions.

From Paper VI, the disinfection of effluent in AWTS provided increased removal of bacteria and viruses. Of the systems included in the QMRA model and based on the available data, the AWTS resulted in the lowest virus discharge loads. However, the model only considered the risk of infection from drinking water, and did not include risks from exposure to the effluent (i.e. direct contact and indirect via food). While regulations aim to minimise the potential for direct contact with spray irrigation, accidental exposure can occur; e.g. from aerosols, direct contact during irrigation or contact with irrigated surfaces. The treatment and disinfection processes from AWTS reduce the risk of infection, but does not eliminate it. Of particular concern is the allowed use of spray irrigation, where risk to the householder and/or the environment is considerably higher than from a well designed sub-surface effluent application area. An infection model was developed using the QMRA model to compare the annual risk of ingestion from garden irrigation (based on 1 mL per event, 90 events per year [Natural Resource Management Ministerial Council et al., 2006]). Where there was an infection within the house, the average load of viruses in spray irrigation was 1.9 log10 more than after 1 m subsurface transport. The resulting probability of infection from accidental ingestion of irrigation water was between 0.025 and 0.25, however it was not able to account for existing immunity.

Page 84

To better quantify the risks posed by these system, the presence of viruses in system effluent needs to be assessed. This includes improving understanding of virus concentrations in sewage from a single household during and after an infection and virus concentrations in sewage from a single household during an asymptomatic infection. More information is also need to improve understanding of virus removal in treatment systems and the impact of different flow regimes and sewage compositions on this performance, such as for the amended material soil mound, but also for other more common systems.

The amended material soil mound provided a 2.9 log10 reduction of PRD1 in field experiments, as well as high removal of nutrients. However, there is limited data available on the life expectancy of amended material soil mounds or management protocols for decommissioning and/or their replacement. In relation to the mound system investigated, from laboratory studies pathogen inactivation was identified as one of the most important removal mechanisms in the amended material. Therefore the results would not necessarily hold for a similar design with a different (e.g. lower pH) amendment, such as the neutralised red mud used by Bowman (1996). Further, new systems being installed should have a minimum of one observation well installed in each mound to enable monitoring of water levels and sampling for analyses of water quality. Soil sampling should initially be undertaken on after five years, and then annually on a small number of systems, to determine how rapidly the phosphorus sorption capacity is being utilised.

The QMRA model was developed to assess the cumulative risk from the on-site sewage systems in Sydney’s drinking water catchments, and it can be further interrogated to identify the drainage units and individual lots that present the greatest risk to water quality. This can enable prioritisation of resources for inspections. The key research needs for improving the model are considered to be:

• improved information on system location and type, which is being addressed in part by the on-site sewage system registration process and would be further aided by development of a centralised database and the provision of hand-held geographic positioning systems for the officers inspecting systems;

Page 85

• continuing research into system performance and novel technologies to provide information on the occurrence, concentrations, fate and transport of viruses, as well as other parameters; and

• establishment of monitoring bores in new disposal systems and an appropriate monitoring program, which will inform failure assessments and provide information on the log reductions for assessing development applications.

The log reduction method was developed for use as a development assessment tool. The method aimed to provide a simple method for site assessment based on: sewage treatment and disposal system types, basic soil type, slope and distance to surface water. It is recommended that, in particular, this method should be applied to aid in assessments where the distance to water is less than 100 m.

Page 86

8 Conclusions

This project aimed to develop a methodology to delineate buffer distances between on- site sewage systems and waterways in Sydney’s drinking water catchments, to support the protection of drinking water quality. A methodology was sought to enable calculation of the cumulative impact of on-site sewage systems in the catchments, with the goal to definition what is an appropriate buffer distance (between on-site systems and watercourses). A quantitative microbial risk assessment (QMRA) approach was used. The QMRA model was based on data from literature and experiments, with consideration of the inputs of viruses from other human sources, in this case assumed to be only treated centralised sewage system discharges. This approach enabled the definition of what was an appropriate buffer distance to be based on what was a tolerable impact in terms of the resulting infections within the drinking water community.

The QMRA model focused on the most populous catchment, known as the Warragamba catchment. The model predicted that loads from on-site sewage systems were within the limit of what was considered tolerable (less than one infection per 10 000 people per year via drinking water). The tolerable virus load from the Warragamba catchment was calculated to be about 108 viruses per year in raw water and 104 viruses per year in treated drinking water. A log reduction method was developed to facilitate individual site development assessments. This method was compared to other management approaches to development assessment: i.e. fixed minimum buffer distances of 100 m, reducing failure rates to zero, and the use of a preferred system. Each of these methods had a limit for how much they could reduce virus loads to the catchment due to either failure or short buffer distances at some sites. While the log reduction method is limited by on-site system failure rates, the method provides a quantitative measure of risk by which maintenance inspections can be prioritised, and indeed system performance validated by controlled spike tests; both of which fit well with the risk management approach promoted in the Australian (NHMRC, 2004) and WHO (2004) drinking water guidelines.

Page 87

9 References

ABS. (2002). "2001 Census of Population and Housing Basic Community Profile." Retrieved 20/11/03, 2003, from http://www.abs.gov.au/ausstats/. Adams, M. H. (1959). Bacteriophages. New York, Interscience Publishers Inc. Alvarez, M. E., M. Aguilar, A. Fountain, Gonzalez.N, O. Rascon and D. Saenz (2000). "Inactivation of MS-2 phage and poliovirus in groundwater." Can J Microbiol 46(2): 159-165. Amar, C. F., C. L. East, J. Gray, M. Iturriza-Gomara, E. A. Maclure and J. McLauchlin (2007). "Detection by PCR of eight groups of enteric pathogens in 4,627 faecal samples: re-examination of the English case-control Infectious Intestinal Disease Study (1993-1996)." Eur J Clin Microbiol Infect Dis 26(5): 311-23. Anders, R. and C. V. Chrysikopoulos (2006). "Evaluation of the factors controlling the time-dependent inactivation rate coefficients of bacteriophage MS2 and PRD1." Environ Sci Technol 40(10): 3237-42. Anderson, A. D., A. G. Heryford, J. P. Sarisky, C. Higgins, S. S. Monroe, R. S. Beard, C. M. Newport, J. L. Cashdollar, G. S. Fout, D. E. Robbins, S. A. Seys, K. J. Musgrave, C. Medus, J. Vinje, J. S. Bresee, H. M. Mainzer and R. I. Glass (2003). "A waterborne outbreak of norwalk-like virus among snowmobilers - Wyoming, 2001." Journal of Infectious Diseases 187: 303-6. APHA (1998). Standard Methods for the Examination of Water and Wastewater. Washington D.C., American Public Health Association. Arnold K, Gallasch T (2001) 'Mount Lofty Ranges Waste Control System Project. The nature, function and potential for catchment impact of domestic on-site waste control and disposal in the Mount Lofty Ranges.' Adelaide Hills Council, Adelaide Hills. Ashbolt, N. J., Grabow, W.O.K. and Snozzi, M. (2001). Water Quality: Guidelines, Standards and Health. Risk assessment and management for water-related infectious disease, Chapter 13. . L. Fewtrell and J. Bartram. London, IWA Publishing: 289-315. Atmar, R. L., A. R. Opekun, M. A. Gilger, M. K. Estes, S. E. Crawford, F. H. Neill and D. Y. Graham (2008). "Norwalk virus shedding after experimental human infection." Emerg Infect Dis 14(10): 1553-7. Ausland, G., T. K. Stevik, J. F. Hanssen, J. C. Kohler and P. D. Jenssen (2002). "Intermittent filtration of wastewater - removal of fecal coliforms and fecal streptococci." Water Research 36(14): 3507-3516. Bales, R. C., S. R. Hinkle, T. W. Kroeger and S. K. (1991). "Bacteriophage adsorption during transport through porous media: chemical perturbations and reversibility." Environmental Science and Technology 25: 2088-2095. Bales, R. C., S. Li, T.-C. J. Yeh, M. E. Lenczewski and C. P. Gerba (1997). "Bacteriophage and microsphere transport in saturated porous media: Forced- gradient experiment at Borden, Ontario." Water Resources Research 33(4): 639- 648. Beavers, P. D. and E. A. Gardner (1993). "Prediction of Virus Transport Through Soils."

Page 88

Bechdol, M. L., A. J. Gold and J. H. Gorres (1994). Modeling viral contamination from onsite wastewater disposal in coastal watersheds. Corapcioglu, M.Y., Haridas, A., 1984. Transport and fate of. Proceedings of Seventh National Symposium on Individual and Small Community Sewage Systems, 11– 13 December 1994, Atlanta, GA, ASAE Publishing. Blanc, R. and A. Nasser (1996). "Effect of effluent quality and temperature on the persistence of viruses in soil." Water Science and Technology 33(10-11): 237- 242. Borchardt, M. A., P. D. Bertz, S. K. Spencer and D. A. Battigelli (2003a). "Incidence of Enteric Viruses in Groundwater from Household Wells in Wisconsin." Appl. Environ. Microbiol. 69(2): 1172-1180. Borchardt, M. A., P. H. Chyou, E. O. DeVries and E. A. Belongia (2003b). "Septic system density and infectious diarrhea in a defined population of children." Environmental Health Perspectives 111: 742-748. Bowman, M. (1996). "On-site tertiary treatment using Ecomax systems." Desalination 106: 305-310. Bradford, S. A., J. Simunek, M. Bettahar, M. T. Van Genuchten and S. R. Yates (2003). "Modeling colloid attachment, straining, and exclusion in saturated porous media." Environmental Science & Technology 37(10): 2242-2250. Bradford, S. A., S. R. Yates, M. Bettehar and J. Šimùnek (2002). "Physical factors affecting the transport and fate of colloids in saturated porous media." Water Resources Research 38(12): 1327-. Camann, D. E. (1980). A model for predicting dispersion of microorganisms in waste water aerosols. In Waste Water Aerosol and Disease. USEPA: EPA 600/9-80- 028. H. Patison and W. J. Jerbowski. Washington D. C., United States Evironmental Protection Agency. Canter, L. W. and R. C. Knox (1985). Septic tank system effects on ground water quality. Chelsea, Michigan, Lewis Publishers. Carducci, A., E. Tozzi, E. Rubulotta, B. Casini, L. Cantiani, E. Rovini, M. Muscillo and R. Pacini (2000). "Assessing airborne biological hazard from urban wastewater treatment." Water Research 34(4): 1173-1178. Carlin, J. B., P. Chondros, P. Masendycz, H. Bugg, R. F. Bishop and G. L. Barnes (1998). "Rotavirus infection and rates of hospitalisation for acute gastroenteritis in young children in Australia, 1993-1996." Medical Journal of Australia 169(5): 252-256. Cerf, O. (1977). "Tailing of survival curves of bacterial spores." J Appl Bacteriol 42(1): 1-19. CH2M HILL (2001). Review of Catchment Sewerage Needs. Prepared for Sydney Catchment Authority. Australia, CH2M HILL. Charles, K., N. Ashbolt, D. Roser, D. Deere and R. McGuinness (2001). "Australasian standards for on-site sewage management: Implications for nutrient and pathogen pollution in the Sydney drinking water catchments." Water Journal of the Australian Water & Wastewater Association 28(2): 58-64. Charles, K., D. Roser, N. Ashbolt, D. Deere and R. McGuinness (2003a). "Buffer distances for on-site sewage systems in Sydney's drinking water catchments." Water Science and Technology 47(7-8): 183-189. Charles, K. J. (2007). Quantitative Microbial Risk Assessment: a catchment management tool to delineate buffer distances for on-site sewage treatment and

Page 89

disposal systems in Sydney’s drinking water catchments. Civil and Environmental Engineering. Sydney, University of New South Wales. PhD. Charles, K. J. and N. J. Ashbolt (2004a). Quantitative Microbial Risk Assessment: a catchment management tool to delineate setback distances for septic systems. Young Researchers Conference 2004. P. Lens and R. Stuetz. Wageningen, the Netherlands, IWA: 139-146. Charles, K. J., N. J. Ashbolt, D. A. Deere and D. J. Roser (2003b). Disinfection in Aerated Wastewater Treatment Systems. Ozwater, Perth. Charles, K. J., N. J. Ashbolt, C. Ferguson, D. J. Roser, R. McGuinness and D. A. Deere (2002). Impacts of centralised versus decentralised sewage systems on water quality in Sydney's drinking water catchments. Small Water & Wastewater Treatment Plants, Istanbul, IWA. Charles, K. J., N. J. Ashbolt, D. J. Roser, R. McGuinness and D. A. Deere (2005). "Effluent quality from 200 on-site sewage systems: design values for guidelines." Water Science & Technology 51(10): 163-169. Charles, K. J., C. M. Davies, D. L. Baker, C. J. Charles, F. C. Souter, N. J. Ashbolt, J. F. Schijven, P. K. Priscott and D. A. Deere (Submitted). "Virus fate and transport from on-site sewage systems in Sydney drinking water catchments: laboratory and field studies." Environmental Science and Technology. Charles, K. J., J. F. Schijven, D. Baker, D. J. Roser, D. A. Deere and N. J. Ashbolt (2004b). Transport and fate of nutrients and pathogens during sewage treatment in a mound system. On-site Wastewater Treatment X: Proceedings of the Tenth National Symposium on Decentralised Sewage Treatment and Disposal Systems. Sacramento, USA, American Society for Agricultural Engineers: 460 - 469. Charles, K. J., F. C. Souter, D. L. Baker, C. M. Davies, J. F. Schijven, D. J. Roser, D. A. Deere, P. K. Priscot and N. J. Ashbolt (2008). "Fate and transport of viruses during sewage treatment in a mound system." Water Research 42(12): 3047. Chattopadhyay, D., S. Chattopadhyay, W. G. Lyon and J. T. Wilson (2002). "Effect of surfactants on the survival and sorption of viruses." Environmental Science and Technology 36(19): 4017-4024. Chen, M. (1988). "Pollution of groundwater by nutrients and faecal coliforms from lakeshore-septic systems." Water, Air and Soil Pollution 37: 407-417. Cheung, K. C. and T. H. Venkitachalam (2006). "Kinetic studies on phosphorus sorption by selected soil amendments for septic tank effluent renovation." Environ Geochem Health 28(1-2): 121-31. Cliver, D. O. (2001). Research Needs in Decentralised Wastewater Treatment and Management: Fate and Transport of Pathogens. National Research Needs Conference Proceedings: Risk-Based Decision Making for Onsite Wastewater Treatment. K. Carns. St Louis, Missouri, Electric Power Research Institute. Conaty, S., P. Bird, G. Bell, E. Kraa, G. Grohmann and J. M. McAnulty (2000). "Hepatitis A in New South Wales, Australia, from consumption of oysters: the first reported outbreak." Epidemiology and Infection 124(1): 121-130. Coote, B. (1995). Aerated Septic Systems for Camden Council. Sydney, AWT Report 95/194. Sydney Water Corp. Corapcioglu, M. Y. and A. Haridas (1984). "Transport and fate of microorganisms in porous media: A theoretical investigation." J. Hydrology 72: 149-162. Craun, G. F. (1985). "A summary of waterborne illness transmitted through contaminated groundwater." J Environ Health 48(3): 122-127.

Page 90

Cromer, B. (1999). TrenchTM 3.0 - A computer application for site assessment and system sizing. On-site ’99 Making On-site Wastewater Systems Work, Armidale, Australia, Lanfax labs. Cromer, W. C., E. A. Gardner and P. D. Beavers (2001). An improved viral die-off method for estimating setback distances On-site '01 Advancing On-site Wastewater Systems, Armidale, Australia, Lanfax labs. Curry, D. S. (2000). Final Report for the Septic Siting Study. New York, New York City Department of Environmental Protection: 210. Dahling, D. R., R. S. Safferman and B. A. Wright (1989). "Isolation of enterovirus and reovirus from sewage and treated effluents in selected Puerto Rican communities." Applied and Environmental Microbiology 55(2): 503-506. Davies, C., C. Kaucner, N. Altavilla, N. Ashbolt, W. Hijnen, G. Medema, D. Deere, M. Krogh and C. Ferguson (2005). Fate and transport of surface water pathogens in watersheds. Report 91078F. Denver, American Water Works Research Foundation. Davies, C. M., C. M. Ferguson, C. Kaucner, N. Altavilla, D. A. Deere and N. J. Ashbolt (2004). "Dispersion and transport of Cryptosporidium oocysts from fecal pats under simulated rainfall events." Applied and Environmental Microbiology 70(2): 1151-1159. Davies, C. M., C. M. Ferguson, C. Kaucner, M. Krogh, N. Altavilla, D. A. Deere and N. J. Ashbolt (2003). Dispersion and transport of Cryptosporidium oocysts from faecal pats under simulated rainfall events. IWA Health Related Microbiology, Johannesburg, SA, IWA. Davies, C. M., M. R. Logan, V. J. Rothwell, M. Krogh, C. M. Ferguson, K. Charles, D. A. Deere and N. J. Ashbolt (2006). "Soil inactivation of DNA viruses in septic seepage." J Appl Microbiol 100(2): 365-74. Deborde, D. C., W. W. Woessner, Q. T. Kiley and P. Ball (1999). "Rapid transport of viruses in a floodplain aquifer." Water Research 33: 2229-2238. Deborde, D. C., W. W. Woessner, B. Lauerman and P. Ball (1998a). "Coliphage prevalence in high school septic effluent and associated ground water." Water Research 32(3781-3785). Deborde, D. C., W. W. Woessner, B. Lauerman and P. N. Ball (1998b). "Virus occurrence and transport in a school septic system and unconfined aquifer." Ground Water 36(5): 825-834. Department of Water (2006). Wastewater treatment – onsite domestic systems. Water Quality Protection Note 70, September 2006, Department of Water, Western Australia. Dizer, H., A. Nasser and J. M. Lopez (1984). "Penetration of different human pathogenic viruses into sand columns percolated with distilled water, groundwater, or wastewater." Applied and Environmental Microbiology 47(2): 409-415. DLG (1998). Environment and Health Protection Guidelines: On-site Sewage Management for Single Households. Sydney, NSW, Department of Local Government. Duboise, S. M., B. E. Moore and B. P. Sagik (1976). "Poliovirus survival and movement in a sandy forest soil." Applied and Environmental Microbiology 31(4): 536-543.

Page 91

Enriquez, C. E., C. J. Hurst and C. P. Gerba (1995). "Survival of the enteric adenoviruses 40 and 41 in tap, sea, and waste water." Water Research 29(11): 2548-2553. EPA (2003). Land Capability Assessment for Onsite Domestic Wastewater Management. Melbourne, Environmental Protection Authority, Victoria: 16. Estes, M. K., D. Y. Graham, E. M. Smith and C. P. Gerba (1979). "Rotavirus stability and inactivation." Journal of General Virology 43(2): 403-409. Faulkner, B. R., W. G. Lyon, A. F. Khan and S. Chattopadhyay (2003). "Modeling leaching of viruses by the Monte Carlo method." Water Research 37: 4719- 1729. Faulkner, B. R., W. G. Lyon, F. A. Khan and S. Chattopadhyay (2002). Predicting Attenuation of Viruses During Percolation in Soils: 1. Probabilistic Model, EPA/600/R-02/051a. Contract Number 68-C-98-138. Cincinnati, OH 45268, National Risk Management Research Laboratory, US-EPA. Feachem, R. G., D. J. Bradley, H. Garelick and D. D. Mara (1983). Sanitation and disease health aspects of excreta and wastewater management. Chichester, England, John Wiley & Sons (for The World Bank). Ferguson, C. (2005). Deterministic model of microbial sources, fate and transport: a quantitative tool for pathogen catchment budgeting. Sydney, Australia, University of New South Wales - Centre for Water and Waste Technology. Ferguson, C., K. Charles and D. Deere (2009). "Quantification of Microbial Sources in Drinking-Water Catchments." Critical Reviews in Environmental Science and Technology 39(1): 1-40. Ferguson, C., C. M. Davies, C. Kaucner, M. Krogh, J. Rodehutskors, D. Deere and N. J. Ashbolt (2007). "Field scale quantification of microbial transport from bovine faeces under simulated rainfall events." Journal of Water and Health 5(1): 83-95. Ferguson, C. M., A. M. de Roda Husman, N. Altavilla, D. Deere and N. J. Ashbolt (2003). "Fate and transport of surface water pathogens in watersheds." Critical Reviews in Environmental Science and Technology 33(3): 299-361. Fong, T. T., L. S. Mansfield, D. L. Wilson, D. J. Schwab, S. L. Molloy and J. B. Rose (2007). "Massive microbiological groundwater contamination associated with a waterborne outbreak in Lake Erie, South Bass Island, Ohio." Environ Health Perspect 115(6): 856-64. Foppen, J. W., M. van Herwerden and J. Schijven (2007). "Measuring and modelling straining of Escherichia coli in saturated porous media." J Contam Hydrol 93(1- 4): 236-54. Foppen, J. W. A., S. Okletey and J. F. Schijven (2006). "Effect of goethite coating and humic acid on the transport of bacteriophage PRD1 in columns of saturated sand." Journal Of Contaminant Hydrology 85(3-4): 287-301. Gantzer, C. and L. Schwartzbrod (2002). Enteroviruses: Occurence and persistence in the environment. Encyclopedia of Environmental Microbiology. G. Bitton. Gassilloud, B. and C. Gantzer (2005). "Adhesion-aggregation and inactivation of poliovirus 1 in groundwater stored in a hydrophobic container." Appl Environ Microbiol 71(2): 912-20. Geary, P. (1992). "Diffuse pollution from wastewater disposal in small unsewered communities." Aust. J. Soil & Wat. Conserv. 5(1): 28-33. Geeraerd, A. H., V. P. Valdramidis and J. F. Van Impe (2005). "GInaFiT, a freeware tool to assess non-log-linear microbial survivor curves." International Journal for Food Microbiology 102(1): 95-105.

Page 92

Gerba, C. (2000a). Domestic wastes and waste treatment. Environmental Microbiology. R. M. Maier, I. L. Pepper and C. P. Gerba. San Diego, Academic Press: 505- 534. Gerba, C. P. (2000b). "Assessment of enteric pathogen shedding by bathers during recreational activity and its impact on water quality." Quantitative Microbiology 2(1): 55-68. Gerba, C. P. and J. C. Lance (1978). "Poliovirus removal from primary and secondary sewage effluent by soil filtration." Applied and Environmental Microbiology 36(2): 247-251. Gerba, C. P., J. B. Rose, C. N. Haas and K. D. Crabtree (1996). "Waterborne rotavirus: A risk assessment." Water Research 30(12): 2929-2940. Gerba, C. P. and G. E. Schaiberger (1975). "Effect of particulates on virus survival in seawater." Journal of the Water Pollution Control Federation 47(1): 93-103. Gerba, C. P., M. V. Yates and S. R. Yates (1991). Quantitation of Factors Controlling Viral and Bacterial Transport in the Subsurface. Modelling the Environmental Fate of Microorganisms. C. J. Hurst. Washington DC, USA, American Society for Microbiology: 77-88. Governal, R. A. and C. P. Gerba (1997). "Persistence of MS-2 and PRD-1 bacteriophages in an ultrapure water system." Journal of Industrial Microbiology & Biotechnology 18(5): 297-301. Goyal, S. M. and C. P. Gerba (1979). "Comparative adsorption of human enteroviruses, simian rotavirus, and selected bacteriophages to soils." Applied and Environmental Microbiology 38(2): 241-247. Grabow, W. O. K., M. B. Taylor and J. C. de Villiers (2001). "New methods for the detection of viruses: call for review of drinking water quality guidelines." Water Science and Technology 46(12): 1-8. Grosser, P. W. (1984). "A one-dimensional mathematical model of virus transport, predicting virus movement through soils." Ground Water 29: 251-259. Haas, C. N. (1996). "How to average microbial densities to characterize risk." Water Research 30(4): 1036-1038. Haas, C. N., J. B. Rose and C. P. Gerba (1999). Quantitative Microbial Risk Assessment. New York, John Wiley & Sons, Inc. Harter, T., S. Wagner and E. R. Atwill (2000). "Colloid transport and filtration of Cryptosporidium parvum in sandy soils and aquifer sediments." Environmental Science and Technology 34(1): 62-70. Harvey, R. W. and J. N. Ryan (2004). "Use of PRD1 bacteriophage in groundwater viral transport, inactivation, and attachment studies." FEMS Microbiology Ecology 49: 3-16. Havelaar, A. H. and W. M. Hogeboom (1984). "A method for the enumeration of male- specific bacteriophages in sewage." Journal of Applied Bacteriology 56: 439- 447. Health Canada (2000). Waterborne outbreak of gastroenteritis associated with a contaminated municipal water supply, Walkerton, Ontario, Canadian Communicable Disease Report: 170-173. Hellard, M., M. I. Sinclair, A. B. Forbes and C. K. Fairley (2001). "A randomized blinded controlled trial investigating the gastrointestinal health effects of drinking water quality." Environmental Health Perspectives 109(8): 773-778.

Page 93

Ho, G. E., R. A. Gibbs and K. Mathew (1991). "Bacteria and virus removal from secondary effluent in sand and red mud columns." Water Science and Technology 23(1-3): 261-270. Höglund, C., N. J. Ashbolt, T. A. Stenström and L. Svensson (2002a). "Viral persistence in source-separated human urine." Advances in Environmental Research 6(3): 265-275. Höglund, C., T. A. Stenström and N. Ashbolt (2002b). "Microbial risk assessment of source-separated urine used in agriculture." Waste Management and Research 20(3): 150-161. Hrudey, S. and E. Hrudey (2004). Safe Drinking Water - Lessons from Recent Outbreaks in Affluent Nations. London, UK, IWA Publishing. Hurst, C. J. (1991). Using linear and polynomial models to examine the environmental stability of viruses. Modelling the Environmental Fate of Microorganisms. C. J. Hurst. Washington DC, USA, American Society for Microbiology: 137-159. Hurst, C. J., C. P. Gerba and I. Cech (1980). "Effects of environmental variables and soil characteristics on virus survival in soil." Applied and Environmental Microbiology 40(6): 1067-1079. HydroGeoLogic (1994). VIRALT Version 3.0: Documentation and User's Guide. Virginia, USA, Prepared for the USEPA. HydroGeoLogic (1995). CANVAS Version 2.1 - Documentation and User's guide. Herdon, VA, USA, Prepared for the USEPA. ISO 10705-2 (2000). Water quality - Detection and enumeration of bacteriophages. Part 2: Enumeration of somatic coliphages. Geneva, International Organization for Standardization. Jagger, J. H. (1967). Introduction to Research in UV Photobiology. Englewood Cliffs, NJ, Prentice-Hall, Inc. Jelliffe, P. A., G. Sabburg and J. Wolff (1994). Key factors in minimising water pollution from unsewered areas. AWWA 16th Federal Convention. Jin, Y., Y. J. Chu and Y. S. Li (2000). "Virus removal and transport in saturated and unsaturated sand columns." Journal Of Contaminant Hydrology 43(2): 111-128. Jin, Y. and M. Flury (2002). "Fate and transport of viruses in porous media." Advances in Agronomy 77: 39-102. Jin, Y., M. V. Yates, S. S. Thompson and W. A. Jury (1997). "Sorption of viruses during flow through saturated sand columns." Environmental Science & Technology 31(2): 548-555. Johansson, L. (1999). "Blast furnace slag as phosphorus sorbents - column studies." The Science Of The Total Environment 229: 89-97. Kapuscinski, R. B. and R. Mitchell (1983). "Sunlight-induced mortality of viruses and Escherichia coli in coastal seawater." Environmental Science and Technology 17: 1-6. Lambertini, E., Spencer, S.K., Bertz, P.D., Loge, F.J., Kieke, B.A. and Borchardt, M.A. (2008). "Concentration of enteroviruses, adenoviruses, and noroviruses from drinking water by use of glass wool filters." Applied and Environmental Microbiology(74): 2990-2996. Lance, J. C. and C. P. Gerba (1984). "Virus movement in soil during saturated and unsaturated flow." Applied and Environmental Microbiology 47(2): 335-337. Lance, J. C., C. P. Gerba and J. L. Melnick (1976). "Virus movement in soil columns flooded with secondary sewage effluent." Applied and Environmental Microbiology 32(4): 520-526.

Page 94

Lewis, A. L. and L. M. Stark (1993). Florida on-site septic disposal system research project viral study summary. Tampa, Florida Department of Health and Rehabilitative Services: 97. Lodder, W. J. and A. M. de Roda Husman (2005). "Presence of noroviruses and other enteric viruses in sewage and surface waters in The Netherlands." Applied and Environmental Microbiology 71(3): 1453-1461. Lodder, W. J., J. Vinj, R. van de Heide, A. M. de Roda Husman, E. J. T. M. Leenen and M. P. G. Koopmans (1999). "Molecular detection of Norwalk-like caliciviruses in sewage." Applied and Environmental Microbiology 65: 5624-5627. Logsdon, G. S. (1994). The removal and disinfection efficiency of lime softening processes for Giardia and Viruses. Denver, Co., American Water Works Association Research Foundation and American Water Works Association. Macler, B. A. and J. C. Merkle (2000). "Current knowledge on groundwater microbial pathogens and their control." Hydrogeology Journal 8(1): 29-40. Marquardt, D. W. (1963). "An algorithm for least-squares estimation of nonlinear parameters." Journal of the Society for Industrial and Applied Mathematics 11(431-441). Mazounie, P., F. Bernazeau and P. Alla (2000). "Removal of Cryptosporidium by high rate contact filtration: the performance of the Prospect water filtration plant during the Sydney water crisis." Water Science and Technology 41(7): 93-101. McClellan, P. (1998). Sydney Water Inquiry. Introduction Recommendations and actions. Sydney, Premier's Department, NSW Government. 1. McKay, L. D., J. A. Cherry, R. C. Bales, M. T. Yahya and C. P. Gerba (1993). "A field example of bacteriophage as tracers of fracture flow." Environmental Science and Technology 27(6): 1075-1079. McLeod, M., J. Aislabie, J. Smith, R. Fraser, A. Roberts and M. Taylor (2001). "Viral and chemical tracer movement through contrasting soils." Journal of Environmental Quality 30: 2134-2140. Mead, P. S., L. Slutsker, V. Dietz, L. F. McCaig, J. S. Bresee, C. Shapiro, P. M. Griffin and R. V. Tauxe (1999). "Food-related illness and death in the United States." Emerging Infectious Diseases 5(5): 607-625. Medema, G., J.-C. Loret, T. A. Stenström and N. Ashbolt (2006). Quantitative Microbial Risk Assessment in the Water Safety Plan. Final Report on the EU MicroRisk Project. Brussels, European Commission. Mons, M. N., J. M. L. Van der Wielen, E. J. M. Blokker, M. L. Sinclair, K. F. A. M. Hulshof, F. Dangendorf, P. R. Hunter and G. J. Medema (2007). "Estimation of the consumption of cold tap water for microbiological risk assessment: an overview of studies and statistical analysis of data. ." Journal of Water and Health(in press). Moore, R. S., D. H. Taylor, L. S. Sturman, M. M. Reddy and G. W. Fuhs (1981). "Poliovirus adsorption by 34 minerals and soils." Applied and Environmental Microbiology 42(6): 963-975. Natural Resource Management Ministerial Council, Environment Protection and Heritage Council and Australian Health Ministers’ Conference (2006). National guidelines for water recycling: managing health and environmental risks (Phase 1). National Water Quality Management Strategy. NH&MRC and ARMCANZ (1994). Austalian drinking water guidelines. Canberra, National Health and Medical Research Council, Agriculural and Resource Management Council of Australia and New Zealand: 248.

Page 95

NH&MRC and ARMCANZ (1996). Australian drinking water guidelines - Summary. National Water Quality Management Strategy. Canberra, National Health and Medical Research Council; Agricultural and Resource Management Council of Australia and New Zealand: 40. NHMRC (2004). Australian Drinking Water Guidelines (ADWG). Canberra: National Health and Medical Research Council. Nicosia, L. A., J. B. Rose and L. Stark (2001). "A field study of virus removal in septic tank drainfields." Journal of Environmental Quality 30(6): 1933-1939. Noonan, M. J. and J. F. McNabb (1979). "Movement of bacteriophages in groundwater in New Zealand." Ann. Mtg. Am. Soc. Microbial: 221. NRM. (2002, June 2002). "On-site Sewerage Facilities. Guidelines for vertical and horizontal separation distance." Retrieved 8 March, 2004, from http://www.nrm.qld.gov.au/compliance/wic/pdf/separation_distance_gl.pdf. NSW Health (1998). Aerated Wastewater Treatment Systems (AWTS) Accreditation Guideline. Sydney, NSW Health Department. Nwachuku, N. and C. P. Gerba (2006). "Health risks of enteric viral infections in children." Reviews in Environmental Contamination and Toxicology 186: 1-56. O'Neill, R. A., G. K. Roads and R. N. Wiese (1993). On-site waste water treatment and disposal in New South Wales. School of Civil and Environmental Engineering. Sydney, University of Technology, Sydney. Pang, L., M. Close, M. Goltz, M. Noonan and L. Sinton (2005). "Filtration and transport of Bacillus subtilis spores and the F-RNA phage MS2 in a coarse alluvial gravel aquifer: Implications in the estimation of setback distances." Jounral of Contaminant Hydrology 77: 165-194. Pang, L., M. Close, M. Goltz, L. Sinton, H. Davies, C. Hall and G. Stanton (2004). "Estimation of septic tank setback distances based on transport of E. coli and F- RNA phages." Environ Int 29(7): 907-21. Parshionikar, S. U., S. Willian-True, G. Shay Fout, D. E. Robbins, S. A. Seys, J. D. Cassady and R. Harris (2003). "Waterborne outbreak of gastroenteritis associated with a Norovirus." Applied and Environmental Microbiology 69(8): 5263-5268. Payment, P., S. Fortin and M. Trudec (1986). "Elimination of human enteric viruses during conventional wastewater treatment by activated sludge." Can J Microbiol 32: 922-925. Petterson, S. R. and N. J. Ashbolt (2001a). "Viral risks associated with wastewater reuse: Modelling virus persistence on wastewater irrigated salad crops." Water Science and Technology 43(12): 23-226. Petterson, S. R., N. J. Ashbolt and A. Sharma (2001b). "Microbial risks from wastewater irrigation of salad crops: A screening-level risk assessment." Water Environment Research 73(6): 667-672 Errata: Wat. Environ. Res. 74(4):411. Petterson, S. R., N. J. Ashbolt and P. F. M. Teunis (2000). Viral risks associated with wastewater reuse: Modelling virus persistence on wastewater irrigated salad crops. 1st World Water Congress of the International Water Association - Health-Related Water Microbiology Symposium, Paris, Comit‚ Françaus de la Recherche sur la Pollution de l'eau. Petterson, S. R., P. F. M. Teunis and N. Ashbolt (2001c). "Modeling virus inactivation on salad crops using microbial count data." Risk Analysis 21(6): 1097-1107.

Page 96

Powell, K. L., R. G. Taylor, A. A. Cronin, M. H. Barrett, S. Pedley, J. Sellwood, S. A. Trowsdale and D. N. Lerner (2003). "Microbial contamination of two urban sandstone aquifers in the UK." Water Res 37(2): 339-52. Powelson, D. K. and C. P. Gerba (1994). "Virus removal from sewage effluents during saturated and unsaturated flow through soil columns." Water Research 28(10): 2175-2181. Powelson, D. K., J. R. Simpson and C. P. Gerba (1990). "Virus transport and survival in saturated and unsaturated flow through soil columns." Journal of Environmental Quality 19: 396-401. Powelson, D. K., J. R. Simpson and C. P. Gerba (1991). "Effects of organic matter on virus transport in unsaturated flow." Applied and Environmental Microbiology 57(8): 2192-2196. Rao, V. C., S. B. Lakhe, S. V. Waghmare and V. Raman (1986). "Virus removal in primary setlling of raw sewage." Journal of Environmental Engineering 107: 57- 59. Rayment, G. E. and F. R. Higginson (1992). Australian Laboratory Handbook of Soil and Water Chemical Methods. Port Melbourne, Reed International Books Australia P/L, trading as Inkata Press. Reddy, K. R., R. Khaleel and M. R. Overcash (1981). "Behavior and transport of microbial pathogens and indicator organisms in soils treated with organic wastes." Journal of Environmental Quality 10(3): 255-266. Regli, S., J. B. Rose, C. N. Haas and C. P. Gerba (1991). "Modeling the risk from Giardia and viruses in drinking water." Journal of American Water Works Association 83(11): 76-84. Rehmann, L. L. C. (1998). A Spectral Stochastic Model of Virus Transport in Aquifers. Department of Civil and Architectural Engineering. Philadelphia, Drexel University: 225. Rehmann, L. L. C., C. Welty and R. W. Harvey (1999). "Stochastic analysis of virus transport in aquifers." Water Resources Research 35(7): 1987-2006. Rennecker, J. L., B. Corona-Vasquez, A. M. Driedger and B. J. Marinas (2000). "Synergism in sequential disinfection of Cryptosporidium parvum." Water Science and Technology 41(7): 47-52. Robertson, B., A. Forbes, M. Sinclair and J. Black (2000). "How well does a telephone questionnaire measure drinking water intake?" Aust. N. Z. J. Public Health Care Manag Sci 24(6): 619 - 622. Roser, D. J. and N. J. Ashbolt (2007). Source Water Quality Assessment and the Management of Pathogens in Surface Catchments and Aquifers. Research Report 29. Bolivar, CRC for Water Quality and Treatment. Ryan, J. N. and M. Elimelech (1996). "Colloid mobilization and transport in groundwater." Colloids And Surfaces A-Physicochemical And Engineering Aspects 107: 1-56. Ryan, J. N., R. W. Harvey, Metge, M. Elimelech, Novigato and A. Pieper (2002). "Field and laboratory investigations of inactivation of viruses (PRD1 and MS2) attached to iron-oxide coated quartz sand." Environmental Science & Technology 36: 2403-2412. Sakadevan, K. and H. J. Bavor (1998). "Phosphate adsorption characteristics of soils, slags and zeolite to be used as substrates in constructed wetland systems." Water Research 32(2): 393-399.

Page 97

Scandura, J. E. and M. D. Sobsey (1997). "Viral and bacterial contamination of groundwater from on-site sewage treatment systems." Water Science and Technology 35(11-12): 141-146. Schaper, M., A. E. Duran and J. Jofre (2002). "Comparative resistance of phage isolates of four genotypes of F-specific RNA bacteriophages to various inactivation processes." Applied and Environmental Microbiology 68(8): 3702-3707. Schaub, S. A., H. T. Bausum and G. W. Taylor (1982). "Fate of virus in wastewater applied to slow-infiltration land treatment systems." Applied and Environmental Microbiology 44(2): 383-94. Schaub, S. A., K. F. Kenyon, B. Bledsoe and R. E. Thomas (1980). "Evaluation of the overland runoff mode of land wastewater treatment for virus removal." Applied and Environmental Microbiology 39(1): 127-134. Schijven, J. F. (2001). Virus removal from groundwater by soil passage: Modeling, field and laboratory experiments. Delft, University of Delft, The Netherlands. Schijven, J. F. and S. M. Hassanizadeh (2000). "Removal of viruses by soil passage: Overview of modeling, processes, and parameters." Critical Reviews in Environmental Science and Technology 30(1): 49-127. Schijven, J. F., S. M. Hassanizadeh and H. A. M. de Bruin (2002a). "Column experiments to study nonlinear removal of bacteriophages by passage thorugh saturated dune sand." Journal of Contaminant Hydrology 58: 243-259. Schijven, J. F., S. M. Hassanizadeh and R. H. A. M. de Bruin (2002b). "Two-site kinetic modeling of bacteriophages transport through columns of saturated dune sand." Journal Of Contaminant Hydrology 57(3-4): 259. Schijven, J. F., W. Hoogenboezem, S. M. Hassanizadeh and J. H. Peters (1999). "Modeling removal of bacteriophages MS2 and PRD1 by dune recharge at Castricum, Netherlands." Water Resources Research 35(4): 1101-1111. Schijven, J. F., J. H. Mulschlegel, S. M. Hassanizadeh, P. F. Teunis and A. M. de Roda Husman (2006). "Determination of protection zones for Dutch groundwater wells against virus contamination--uncertainty and sensitivity analysis." Journal of Water and Health 4(3): 297-312. Schijven, J. F. and J. Simunek (2002c). "Kinetic modeling of virus transport at the field scale." Journal Of Contaminant Hydrology 55(1-2): 113. Sharp, D. G., R. Floyd and J. D. Johnson (1975). "Nature of the surviving plaque- forming unit of reovirus in water containing bromine." Applied Microbiology 24(1): 94-101. Sim, Y. and C. V. Chrysikopoulos (1995). "Analytical models for one-dimensional virus tranpsort in saturated porous media." Water Resources Research 31(5): 1429-1437. Sim, Y. and C. V. Chrysikopoulos (1996). "One-dimensional virus tranpsort in porous media with time-dependant inactivaiton rate coefficients." Water Resources Research 32(8): 2607-2611. Sim, Y. and C. V. Chrysikopoulos (1998). "Three-dimensional analytical models for virus transport in saturated porous media." Transport in Porous Media 30: 87- 112. Sim, Y. and C. V. Chrysikopoulos (1999). "Analytical models for virus adsorption and inactivation in unsaturated porous media." Colloids & Surfaces A: Physicochemical & Engineering Aspects 155: 189-197. Sim, Y. and C. V. Chrysikopoulos (2000). "Virus transport in unsaturated porous media." Water Resources Research 36(1): 173-179.

Page 98

Šimùnek, J., M. Šejna and M. T. van Genuchten (1998). The HYDRUS-1D software package for simulating the one-dimensional movement of water, heat and multiple solutes in variably saturated media. Golden, Colorado, International Ground Water Modelling Centre, Colorado School of Mines. Šimùnek, J., M. Šejna and M. T. van Genuchten (1999). The HYDRUS-2D software package for simulating two-dimensional movement of water, heat and multiple solutes in variably saturated media. Golden, Colorado, International Ground Water Modelling Centre, Colorado School of Mines. Sinton, L. W., R. K. Finlay and P. A. Lynch (1999). "Sunlight Inactivation of fecal bacteriophages and bacteria in sewage-polluted seawater." Applied and Environmental Microbiology 65(8): 3605-3486. Sinton, L. W., C. H. Hall, P. A. Lynch and R. J. Davies-Colley (2002). "Sunlight inactivation of fecal indicator bacteria and bacteriophages from waste stabilization pond effluent in fresh and saline waters." Applied and Environmental Microbiology 68(3): 1122-1131. Sobsey, M. D. (1989). "Inactivation of health-related microorganisms in water by disinfection processes." Water Science and Technology 21(3): 179-195. Sobsey, M. D., C. H. Dean, M. E. Knuckles and R. A. Wagner (1980). "Interactions and survival of enteric viruses in soil materials." Applied and Environmental Microbiology 40(1): 92-101. Sobsey, M. D., R. M. Hall and R. L. Hazard (1995). "Comparative reductions of hepatitis A virus, enteroviruses and coliphage MS2 in miniature soil columns." Water Science and Technology 31(5-6): 203-209. Sobsey, M. D., P. A. Shields, F. H. Hauchman, R. L. Hazard and L. W. Caton (1986). "Survival and transport of hepatitis A virus in soils, groundwater, and wastewater." Wat. Sci. Technol. 18: 97-106. Standards Australia (2000). On-site domestic wastewater management. Sydney, Standards Australia. Straub, T. M., I. L. Pepper and C. P. Gerba (1992). "Persistence of viruses in desert soils amended with anaerobically digested sewage sludge." Applied and Environmental Microbiology 58: 636-641. Swanson, P., A. Davidson, P. Hawkins and M. Cunningham (2000). Sources of Cryptosporidium and Giardia in the Warragamba water supply catchment. Sydney, AWT. Sydney Catchment Authority (2006). Neutral or Beneficial Effect on Water Quality Assessment Guidelines. Sustaining the Catchments. The Regional Plan for the drinking water catchments of Sydney and adjacent regional centres. Sydney, SCA. Sydney Catchment Authority. (undated). "The SCA's drinking water catchments." Retrieved 29 October 2007, from http://www.sca.nsw.gov.au/publications/files/CatchmentsMapA4.pdf. Taylor, R., A. Cronin, S. Pedley, J. Barker and T. Atkinson (2004). "The implications of groundwater velocity variations on microbial tranpsort and wellhead prtection - review of field evidence." FEMS Microbiology Ecology 49: 17-26. Teunis, P. F. M. and A. Havelaar (2000). "The beta Poisson dose-response model is not a single hit model." Risk Analysis 20(4): 511-518. Thompson, S. S., M. Flury, M. V. Yates and W. A. Jury (1998). "Role of the air-water- solid interface in bacteriophage sorption experiments." Applied and Environmental Microbiology 64(1): 304-309.

Page 99

Thurston-Enriquez, J. A., C. N. Haas, J. Jacangelo and C. P. Gerba (2003). "Chlorine inactivation of adenovirus type 40 and feline calicivirus." Appl Environ Microbiol 69(7): 3979-85. Tierney, D. (2001). Performance of Popular OSMS in SEPP 58 Area - Workshop Minutes. Sydney, Tierney Page Kirkland Pty Ltd: pp 11. Tim, U. S. and S. Mostaghimi (1991). "Model for predicting virus movement through soils." Ground Water 29(2): 251-259. Toride, N., F. J. Leij and M. T. van Genuchten (1995). The CXTFIT code for estimating transport parameters from laboratory or field tracer experiments. Research Report No. 137. Riverside, California, USA, U.S. Salinity Lab, Agric. Res. Service, US Dep. of Agric., . USEPA (2002). On-site Wastewater Treatment Systems Manual, US EPA, Washington. USEPA (2006). "National Primary Drinking Water Regulations: Ground Water Rule; Final Rule." Federal Register 71(216): 65574-65660. Van Cuyk, S. and R. Siegrist (2004). Fate of viruses in the infiltrative surface zone of systems that rely on soil treatment for wastewater renovation. On-Site Wastewater Treatment X: Proceedings of the tenth national symposium on individual and community sewage systems. St Joseph, Michigan, American Society for Agricultural Engineers. van den Berg, H., W. Lodder, W. van der Poel, H. Vennema and A. M. de Roda Husman (2005). "Genetic diversity of noroviruses in raw and treated sewage water." Res Microbiol 156(4): 532-40. Vaughn, J. M., E. F. Landry and M. Z. Thomas (1983). "Entrainment of viruses from septic tank leach fields through a shallow, sandy soil aquifer." Applied and Environmental Microbiology 45(5): 1474-1480. Ward, R. L. and C. S. Ashley (1977). "Identification of the virucidal agent in wastewater sludge." Applied and Environmental Microbiology 33(4): 860-864. Ward, R. L., D. I. Bernstein and E. C. Young (1986). "Human rotavirus studies in volunteers: Determination of infectious dose and serological response to infection." Journal of Infectious Diseases 154: 871. Wellings, F. M., A. L. Lewis, C. W. Mountain and L. V. Pierce (1975). "Demonstration of virus in groundwater after effluent discharge onto soil." Applied Microbiology 29(6): 751-757. Westrell, T. 2004. Microbial risk assessment and its implications for risk management in urban water systems. Ph.D. dissertation. University of Linköping, Faculty of Arts and Science, Linköping, Sweden. Westrell, T., O. Bergstedt, T. A. Stenström and N. J. Ashbolt (2003). "A theoretical approach to assess microbial risks due to failures in drinking water treatment." International Journal Of Environmental Health Research 13(2): 181-197. Wheeler, J. G., D. Sethi, J. M. Cowden, P. G. Wall, L. C. Rodrigues, D. S. Tompkins, M. J. Hudson and P. J. Roderick (1999). "Study of infectious intestinal disease in England: rates in the community, presenting to general practice, and reported to national surveillance. The Infectious Intestinal Disease Study Executive." Bmj 318(7190): 1046-50. WHO (2004). Guidelines for Drinking-Water Quality. Geneva, World Health Organization. 3 Surveillance and control of community supplies: 238. Wilson, L. G., G. L. Amy, C. P. Gerba, H. Gordon, B. Johnson and J. Miller (1995). "Water quality changes during soil aquifer treatment of tertiary effluent." Water Environment Research 67(3): 371-376.

Page 100

Yates, M. V. (1995a). "Field Evaluation fo the GWDR's natural disinfection criteria." Journal of the American Water Works Association 97(2): 76-85. Yates, M. V. (1995b). "Septic tank density and groundwater contamination." Ground Water 23: 586-591. Yates, M. V. (2002). Bacteriophage of enteric bacteria: occurence and persistance in the environment. Encyclopedia of Environmental Microbiology. G. Bitton: 384-391. Yates, M. V., C. P. Gerba and L. M. Kelley (1985). "Virus persistence in groundwater." Applied and Environmental Microbiology 49(4): 778-781. Yates, M. V. and W. A. Jury (1995). "On the Use of Virus Tranpsort Modeling for Determining Regulatory Compliance." Journal of Environmental Quality 24: 1051-1055. Yates, M. V., W. A. Jury, S. R. Yates, D. L. Anderson, L. M. Stark and P. Sherblom (2000). Measurement of virus and indicator survival and transport in the subsurface. Denver, AWWA Research Foundation: 1-141. Yates, M. V. and Y. Ouyang (1992). "VIRTUS, a model of virus transport in unsaturated soils." Applied and Environmental Microbiology 58(5): 1609-1616. Yates, M. V. and S. R. Yates (1987a). "A comparison of geostatistical methods for estimating virus inactivation rates in ground water." Water Research 21: 1119- 1125. Yates, M. V. and S. R. Yates (1988). "Modelling microbial fate in the subsurface environment." CRC Critical Reviews in Environmental Control 17(4): 307-344. Yates, M. V. and S. R. Yates (1990). "Modeling microbial transport in soil and groundwater." American Society for Microbiology News 56(6): 324-327. Yates, M. V. and S. R. Yates (1991). Modelling Microbial Transport in the Subsurface: A Mathematical Discussion. Modelling the Environmental Fate of Microorganisms. C. J. Hurst. Washington DC, USA, American Society for Microbiology: 48-76. Yates, M. V., S. R. Yates, J. Wagner and C. P. Gerba (1987b). "Modelling Virus Survival and Transport in the Subsurface." Journal Of Contaminant Hydrology 1: 329-345. Yates, M. V., S. R. Yates, A. W. Warrick and C. P. Gerba (1986). "Use of geostatistics to predict virus decay rates for determination of septic tank setback distances." Applied and Environmental Microbiology 52(3): 479-483.

Page 101

Paper I Australasian Standards for on-site sewage management: Application in the Sydney drinking water catchments

Published as:

Charles, K., Ashbolt.N, D. Roser, D. Deere and R. McGuinness (2001). "Australasian standards for on-site sewage management: Implications for nutrient and pathogen pollution in the Sydney drinking water catchments." Water (Australia) 28(8): 58-64.

Reprinted here with minor modifications.

Author Contribution Role (all authors helped with reviewing and (%) editing) Charles, K 65 Research and writing Ashbolt, N 5 Supervisor Roser, D 20 Co-supervisor, helped with rewrite Deere, D 5 Industry advisor McGuinness, R. 5 Industry advisor

Page 102

Abstract Australasian Standard AS/NZS 1547:2000 provides the framework for improved design of new on-site sewage management systems (OSMS) through greater consideration of local site and soil conditions, and improved performance of existing systems through the promotion of proper operation and maintenance. However, its ability to protect the environment and public health is less satisfactory due to limited criteria for management of pathogens (viruses and protozoa) and nutrients (nitrogen and phosphorous).

In the Sydney Catchment Authority area of operation pathogen and nutrient pollution from OSMS is of particular concern. These drinking water catchments are the context used for assessing the applicability of AS/NZS 1547:2000 to sensitive environments, an assessment which has highlighted the importance of the management sections ability to improve the performance of OSMS in the short term through better operation and maintenance, and the education of stakeholders to ensure adequate consideration of qualitative criteria for pathogens and nutrients in OSMS design. It also identifies possible improvements for the design of new and the management of existing OSMS, including decision making tools for setting buffer distances to protect water quality.

Introduction Sustained functioning of on-site sewage management systems (OSMS) is required to protect the environment and public health. And proper design and operation and maintenance (O&M) is vital to ensure sustainability and continued performance. In sensitive catchments the risks and potential impact from poor system performance are greater still.

The Sydney Catchment Authority (SCA) area of operation (Figure 1) comprises several such sensitive catchments as they provide potable water for over 4 million people. The Sydney Catchment Authority (SCA) aims to protect water quality in these catchments, including minimising risks to human health and the environment from pathogens and nutrients. Hence, the high rates of OSMS failure reported in NSW are a concern for the SCA.

O'Neill et al. (1993) reported visible surfacing effluent at over 40% of absorption trenches. Coote (1995) reported 95% of aerated wastewater treatment systems (AWTS)

Page 103

failing to comply with at least one parameter in the NSW effluent guidelines (DLG, 1998). Australian Water Technologies (1999) reported densities of thermotolerant (faecal) coliforms in effluent pooled on the soil surface from absorption trenches and spray irrigation areas of up to 107 colony forming units (cfu) per 100 mL, representing a potential risk to human health. Similar data has been reported from other states. Thirty percent of absorption trenches had visible surface flow and 73% performed unsatisfactorily in South Australia (Geary, 1992). In Queensland, 39% of trenches had poor performance or surface seepage (Jelliffe et al., 1994) and 70% of AWTS failed to achieve the required effluent quality (Beavers et al., 1999).

The primary causes of performance failure are poor design, including siting and sizing, and inadequate O&M, which are often caused by inadequate understanding of design limitations and failure modes, financial pressures and change of usage or owner. The above studies did not specifically identify adverse environmental impacts, or directly measure effects on human health or ecology. However, potential adverse effects are expected if guideline values are not met. Therefore, the SCA is vitally interested in whether the new Australasian Standard for domestic wastewater management, AS/NZS 1547:2000 will reduce the number of failures, promote better human health and ecological protection in sensitive environments such as the Sydney drinking water catchments, and adequately address the pathogen and nutrient concerns?

AS/NZS 1547:2000 The new Australasian Standard for on-site sewage management, AS/NZS 1547:2000, integrates design and O&M with other management issues and is intended to promote a great improvement in OSMS performance. AS/NZS 1547:2000 requires that OSMS designs are sustainable and protect the environment and public health. This paper evaluates whether AS/NZS 1547:2000 will fully achieve these aims by assessing whether its pathogen and nutrient management recommendations and procedures reflect best possible management practice.

Overall AS/NZS 1547:2000 'On-site Domestic-Wastewater Management' is a significant advance on the previous standard, AS 1547:1994, especially with respect to improved effluent land application system design criteria and the introduction of management

Page 104

recommendations. AS 1547:1994 included design requirements for absorption, evapotranspiration and irrigation disposal systems, collection wells and grey water treatment tanks, with disposal area design based on the long-term acceptance rate (LTAR) of the site, a permeability-based design hydraulic loading. AS/NZS 1547:2000 introduces the term 'land application' to replace 'disposal' to acknowledge the treatment provided by the soil matrix and ecosystems, and improves OSMS sustainability through:

• increased consideration of local site and environmental conditions in site assessments and sizing of land application systems;

• replacement of the LTAR with a Design Loading Rate (DLR) based on a detailed soil assessment, including soil texture and structure; and

• inclusion of management recommendations and qualitative performance criteria.

In these and other respects AS/NZS 1547:2000 compares well with international on-site effluent disposal standards (Table 1). Further notable features of AS/NZS 1547:2000 are as follows.

Site Evaluation Aspects The AS/NZS 1547:2000 site evaluation addresses public health, environmental, legal and economic considerations, as well as site and soil characteristics, by requiring a desktop study, preliminary site and soil check (SSC) and detailed soil assessment. This provides information for selection of design parameters, treatment system selection and land application system location. The SSC ensures a broad range of site and environmental factors are considered in system siting and design. The soil assessment attempts to ensure land application area is sustainable by prediction of soil permeability through classification, hence, reducing failure due to surfacing effluent.

Design Aspects AS/NZS 1547:2000 recommends varying effluent land application areas based on design wastewater production rates, depending on water supply and usage factors, which are generally more conservative than previous values. Design wastewater

Page 105

production rates can be based on the number of bedrooms or occupants in a dwelling. The former, being controlled by council development processes, better guarantees usage will not exceed the design.

The DLR varies according to local soil properties, the land-application system type and the effluent quality. DLRs are generally lower than the LTARs and, with the change in design wastewater production rate, will generally result in the need for larger land application areas.

Page 106

Table 1 Comparison of Australasian and international OSMS standards

Standard/s Systems Site evaluation Land application Performance Criteria & Comments system design criteria Management

AS 1547:1994 Surface and Depth of soil, permeability, 300 L of wastewater Quantitative criteria for Prescriptive “Disposal systems for subsurface disposal climate, seasonal changes in produced per surface irrigation. standard. effluent from domestic systems. soil and groundwater, bedroom⋅day (town Inadequate premises” (Standards Collection wells seepage, runoff, impact on water). Permeability coverage of Australia, 1994) and grey water neighbours, life expectancy based loading rate. nutrients and treatment tanks. of system and area available pathogens 1. for primary and alternative systems. Australia Australia AS/NZS 1547:2000 Surface and Desktop study, site and soil Wastewater production Qualitative criteria for Performance based. “On-site domestic- subsurface land check and site and soil 180 L per person⋅day system performance, Inadequate wastewater application systems, assessment, including soil (town water). Loading management, construction coverage of management” including mounds. properties, vegetation, fill, rate from soil texture and installation and design nutrients and (Standards slope, exposure and salinity. and structure. 2. Criteria for irrigation of pathogens 1. Australia/Standards Consideration of effluent. New Zealand, 2000) effluent treatment. Management information for administration, education, monitoring, and O&M (for guidance only). Australasia Australasia

Page 107

Standard/s Systems Site evaluation Land application Performance Criteria & Comments system design criteria Management

D5879 “Surface Site Subsurface land Desktop study and site Wastewater production Nil Inadequate Characterisation for application systems, assessments. Subsurface is 568L per coverage of On-Site Septic including filter assessment includes limiting bedroom⋅day. Loading nutrients and Systems” (ASTM, beds. depth based on changes in rates, based on soil pathogens. 1995) permeability, rupture texture and structure, D5921 “Subsurface resistance, cementation, may be higher than Site Characterisation of penetration resistance, roots AS/NZS 1547:2000. Test Pits for On-Site and pores. Septic Systems” (ASTM, 1996a) D5925 “Preliminary Sizing and Delineation of Soil Absorption Field Areas for On-Site Septic Systems” (ASTM, 1996b) USA USA BS 6297:1983 “Code Disposal to Desktop study and site Wastewater production Includes installation No coverage of of practice for design waterways, surface assessment of potential is 120L per person⋅day. approval, system integrity nutrients and and installation of and subsurface land noise, potential for pollution Percolation based and alarms to indicate pathogens. small sewage treatment application systems. of waterways, slope, loading rates, lower failure or malfunction. works and cesspools” prevailing wind, flooding than AS/NZS (British Standards, potential, water table depth 1547:2000 for 1983) and percolation. equivalent 3 permeability. Consideration of effluent treatment. Europe Notes 1 Acknowledges that a low thermotolerant coliform count does not imply an absence of pathogens 2 Includes qualitative performance criteria for nutrients and pathogens 3 The available European standard, EN 12566-1:2000 (European Committee for Standardisation, 2000), only addresses septic tanks, however, additional standards are being developed for 'Soil Infiltration Systems' and 'Filtration Systems'.

Page 108

Performance Criteria & Management AS/NZS 1547:2000 incorporates qualitative performance criteria for wastewater systems, evaluation of site and soil characteristics, construction and installation, and management, and quantitative performance criteria for systems using irrigation (Table 2). The performance criteria require the life expectancy of an OSMS is above 15 years (unless otherwise nominated).

Table 2 Effluent Quality Requirements for OSMS

Parameter Median Maximum Applies to Biochemical oxygen ≤ 20 30 surface spray, covered demand (BOD) (g·m-3) surface drip, subsurface drip Suspended Solids (g·m-3) ≤ 30 45 surface spray, covered surface drip, subsurface drip Thermotolerant coliforms ≤ 10 20 surface spray (cfu·100 mL-1) (in 4 out of 5 samples) Total chlorine (g·m-3) ≥ 0.5 surface spray

Management in AS/NZS 1547:2000 aims to ensure sustainable long-term performance and protect public health, the environment and public amenity. Potentially, management can reduce the risk from OSMS failure and poor performance through improved O&M and through:

• education of stakeholders, including accreditation of designers and site assessors;

• recording OSMS information on the property title, and hence, changes of owner and information management;

• regular monitoring of system performance; and

• the preparation of O&M guidelines.

OSMS in the SCA area of operation A survey of councils in the SCA area of operation, undertaken by the Centre for Water and Waste Technology (CWWT) (unpublished data), has identified over 18 000 OSMS. Seventy-three percent were septic tank/absorption trench systems. AWTS were the

Page 109

second most popular (Table 3).

Table 3 OSMS Survey Results

System Type Number in SCA area of Percent of operation Total Septic tank/Absorption trench 13 500 73 AWTS 3 744 20 Pump out 1 168 6 Other* 53 <1 Total 18 465 100 *Includes sand filters, composting toilets, grey-water treatment systems, mounds, evapotranspiration beds and constructed wetlands

In the SCA area of operation, pathogen contamination of the drinking water catchments is the key public health risk. Nutrients, particularly phosphorus, are also of concern as they impact native vegetation and can cause eutrophication, which has public health and environmental implications. OSMS are potentially significant sources of both.

The primary pathogens of concern are human enteric viruses and Cryptosporidium. Pathogen transport from land-application of OSMS effluent to groundwater has been reported (Curry, 2000; Scandura & Sobsey, 1997) and has been linked to drinking water outbreaks in the US (Scandura & Sobsey, 1997). Runoff from unsewered urban areas was found to be the fourth most significant source of Cryptosporidium in the Sydney catchments (Swanson et al., 2000).

Phosphorous removal from land-applied effluent primarily occurs in unsaturated soil, with up to 98% being immobilised through sorption and precipitation (Bicki et al., 1984). In groundwater, phosphorous transport is retarded by reversible sorption reactions, however, phosphorous plume transport has been detected 180m from its source (Curry, 2000). Phosphorous transport impacts on surface water have been identified from absorption trenches within 50 m of water or where insufficient unsaturated soil depth occurs (Bicki et al., 1984). Additionally, nitrate contamination of groundwater is a ‘virtual certainty’ where groundwater recharge occurs (Brouwer, 1983). Hence, appropriate selection, sizing and management of land application areas is required to protect water quality.

Page 110

Weaknesses of AS/NZS 1547:2000 in the SCA area of operation The design criteria in AS/NZS 1547:2000 should reduce the risks to water quality in the Sydney catchments from OSMS in the long-term. However, concerns exist regarding the secondary effluent DLRs for absorption trenches, which result in small areas in high permeability soils, due to the potential for groundwater contamination or failure from clogging where effluent does not comply with criteria.

Stakeholder education, as recommended in the management section, is essential for the success of AS/NZS 1547:2000. To address this, and hence the above design concern, AS/NZS 1547:2000 recommends accreditation of certain stakeholders but no guidance on accreditation course content is included. Additionally, while the management section is informative, implementation is not required by AS/NZS 1547:2000 for compliance.

The O&M recommendations in AS/NZS 1547:2000 address many of the concerns regarding existing systems in the SCA area of operation. The success of these recommendations, however, may depend on the availability of resources for inspections. The current failure detection methods do not relate directly to environmental and public health impacts, and hence, may offer inadequate protection of water quality in the Sydney drinking water catchments. The inspection outcomes are also limited by a lack of guidance on remedial measures. And while AS/NZS 1547:2000 does consider life expectancy, good record keeping and a management plan to assess continuing life expectancy or remedial action are only assumed.

The performance criteria require that OSMS protect public health, the environment and public and community amenity. However, the standard provides limited guidance on how to achieve these aims, particularly on pathogen and nutrient issues that are implicit in these aims, and hence, does not appear to adequately address the primary water quality concerns in the Sydney drinking water catchments.

Pathogens For over twenty years, microbiologists have reported that thermotolerant coliforms will not reliably indicate the presence of viruses and parasitic protozoa. Coliforms do not survive as long, especially with systems using chlorination (Sobsey et al., 1989; Rennecker et al., 2000), and, being larger than the viruses, they are filtered out more

Page 111

effectively in subsurface environments (Macler and Merkle, 2000; Lance & Gerba, 1984; Scandura & Sobsey, 1997). Yet more suitable index organisms for these pathogens are not yet routinely used (Ashbolt et al., 2001).

AS/NZS 1547:2000, similarly to its predecessor, while recognising the inadequacy of thermotolerant coliforms as an indicator of public health risk still, uses this as the sole microbial analyte, and places the responsibility to ensure health risks are tolerable on local authorities who tend to have limited microbiological expertise.

Testing for a range of faecal bacteria (including spore-forming Clostridium perfringens), as well as coliphages (viruses to selected coliforms), provides more reliable information on the possible presence of pathogenic organisms. Additionally, coliphages have been widely used as surrogates of viruses in studies of fate and transport in groundwater (Curry, 2000; Deborde et al., 1999). The US EPA Proposed Ground Water Rule (2000) recommends E. coli, Enterococci and coliphage be used as monitoring tools for faecal contamination. Similar regulation in Australia would improve identification of risk from faecal contamination, and improve quantification of risk associated with OSMS on a local scale, such as from irrigation, as well as on a catchment scale.

The adequate implementation of the qualitative performance criteria, with regards to pathogens, during OSMS design relies on the education and local knowledge of the designer. For example, impacts of virus survival and transport in groundwater needs to be considered before recommending high DLRs. Without appropriate knowledge qualitative performance criteria are inadequate management tools.

Buffer zones are an important tool for attenuating pathogens and nutrients (Barling & Moore, 1992). AS/NZS 1547:2000 does not recommend buffer zones or provide tools for developing buffers appropriate for local conditions, placing responsibility on often poorly resourced local authorities.

Nutrients AS/NZS 1547:2000 does not include quantitative performance criteria for nutrients, other than BOD, but addresses them through identification of the impacts on water quality and vegetation in land application area design, and consideration of system life

Page 112

expectancy. These qualitative criteria, and associated nutrient balances, rely on designer expertise to adequately consider site and environmental conditions, and regulatory authorities.

AS/NZS 1547:2000 does not provide sufficient links to receiving water quality. For example, it is hard to see how design criteria for a single system can be related to the Healthy Rivers Commission (1998) water quality objectives of 50 µg/L total phosphorus and 700 µg/L total nitrogen, particularly when several landuses occur in the catchment. Nor is AS/NZS 1547:2000 yet linked to the new ANZECC & ARMCANZ (2000) guidelines (draft first circulated in 1999) which provide various guides for setting water quality objectives and include a shared national objective, with guidance on issues that may arise in individual circumstances and the flexibility to address these issues.

Conclusions The reported high rates of OSMS performance failure indicate a need for improved design and management to protect the environment, public health and community amenity. AS/NZS 1547:2000 addresses this need through consideration of a broader range of site and soil design factors and promotion of O&M and education.

However, AS/NZS 1547:2000 appears to be deficient in addressing pathogen and nutrient issues in sensitive catchments, such as the Sydney drinking water catchments. Following the ANZECC & ARMCANZ (2000) model, a number of possible improvements should be explored:

• Development of an effective, standardised method for failure detection during inspections, better methods for rehabilitation and subsequent monitoring of failing systems and appropriate criteria for decommissioning, upgrading or replacement, tracking of desludging and AWTS maintenance, and auditing of the service providers to improve O&M.

• Development of tools for undertaking nutrient balances and identifying buffer distances required for single lots and subdivisions, in wet and dry weather, to protect water quality, based on effluent quality, site conditions and receiving environment.

Page 113

• Enhanced but workable graded effluent quality criteria that include pathogens and nutrients, with monitoring of long-term compliance and performance of innovative systems.

• Means of linking AS/NZS 1547:2000 to water quality targets and other relevant environmental management tools.

The education of stakeholders, and hence the management section, is fundamental in ensuring the proper implementation of AS/NZS 1547:2000. Ultimately, the success of this standard will depend on the resources available to local government and other regulatory authorities to allow them to address the issues in their area, and enforce restrictions where necessary. For the SCA, this will include setting criteria for system performance and buffer distances, and promoting better methods for assessing pathogen and nutrient impacts.

Recognising the need for improvements complementary to AS/NZS 1547:2000, the SCA and the CWWT are developing a catchment OSMS risk assessment, encompassing septic tanks, AWTS (chlorine and ultraviolet disinfection) and Ecomax systems. The risk assessment will provide information on cumulative catchment-scale impact of OSMS and most effective methods of managing risk, and a mass balance of nutrient inputs from OSMS in the Sydney catchments where algal blooms are a problem for water quality managers. Ultraviolet disinfection and Ecomax systems are being investigated as potential alternatives for high-risk areas, especially their effectiveness for removing a range of pathogens, nutrients and indicators.

Additionally, experiments of fate and transport of pathogens and nutrients will be undertaken to provide a quantitative basis for buffer design and size. The intended outcome of this work is a comprehensive catchment management strategy for OSMS in sensitive catchments to which other local government and regulatory authorities will be encouraged to contribute.

Acknowledgements The assistance of those individuals who have aided in the development of this paper and in the associated research is greatly appreciated. In particular thanks to Stephen Manson, SCA, and the staff of the councils in the SCA area of operation.

Page 114

References ANZECC and ARMCANZ (2000). Australian and New Zealand Guidelines for Fresh and Marine Water Quality. (National Water Quality Management Strategy No 4.) Australian & New Zealand Environment & Conservation Council.

Ashbolt, N. J., Grabow, W. O. K. and Snozzi, M. (2001). Indicators of microbial water quality. In Fewtrell, L. and Bartram, J. (ed.), Water Quality: Guidelines, Standards and Health. Risk assessment and management for water-related infectious disease. WHO, (in press).

ASTM (1995). D 5879 - 95 Standard Practice for Surface Site Characterization for On- Site Septic Systems. American Society for Testing and Materials, Washington D.C..

ASTM (1996a). D 5921 - 96 Standard Practice for Subsurface Site Characterization of Test Pits for On-Site Septic Systems. American Society for Testing and Materials, Washington D.C..

ASTM (1996b). D 5925 - 96 Standard Practice for Preliminary Sizing and Delineation of Soil Absorption Field Areas for On-Site Septic Systems. American Society for Testing and Materials, Washington D.C..

Australian Water Technologies (1999). Assessments of the Environmental Impacts of Unsewered Areas in the Werriberri Creek Catchment. AWT Environment, Science and Technology, Sydney.

Barling, R.D. and I.D. Moore (1992). The Role of Buffer Strips in the Management of Waterway Pollution. In: The Role of Buffer Strips in the Management of Waterway Pollution from Diffuse Urban and Rural Sources. Proceedings of a Workshop, October 1992. (Eds.: Woodfull, J., B. Finlayson and T. McMahon) Land and Water Resources Research and Development Corporation, Canberra, p. 44.

Beavers, P., I. Tully and A. Woolley (1999). Performance evaluation of On-site Aerated Wastewater Treatment Systems. In: Proceedings of On-site '99 Conference: Making on- site wastewater systems work; 13th-15th July 1999, University of New England. (Ed.: Patterson, R.A.) Lanfax Laboratories, Armidale, pp. 45-52.

Page 115

Bicki, T.J., R.B. Brown, M.E. Collins, R.S. Mansell and D.F. Rothwell (1984). Impact of On-site Sewage Disposal Systems on Surface and Groundwater Quality. Florida Department of Health and Rehabilitative Services, USA.

British Standards (1983). BS 6297:1983 Code of practice for Design and installation of small sewage treatment works and cesspools. British Standards Institution, London.

Brouwer, J. (1983). Land Capability for Septic Tank Effluent Absorption Fields. Part B: Review of Research and Regulations in Australia and Overseas for on-site Waste Water Disposal. (Australian Water Resources Council Technical Paper No. 80.) Australian Government Publishing Service, Canberra.

Coote, B. (1995). Aerated Septic Systems for Camden Council. AWT Report 95/194. Sydney Water Corp., Sydney.

Curry, D.S. (2000). Final Report for the Septic Siting Study. New York City Department of Environmental Protection, New York.

DeBorde, D.C., W.W. Woessner, Q.T. Kiley and P. Ball (1999). Rapid transport of viruses in a floodplain aquifer. Wat. Res. 33:2229-2238.

DLG (1998). Environment and Health Protection Guidelines: On-site Sewage Management for Single Households. Department of Local Government, NSW.

European Committee for Standardization (2000). EN 12566-1:2000 Small wastewater treatment systems for up to 50 PT - Part 1: Prefabricated septic tanks. CEN, London.

Geary, P. (1992). Diffuse pollution from wastewater disposal in small unsewered communities. Aust. J. Soil & Wat. Conserv. 5(1):28-33.

Healthy Rivers Commission (1998). Independent Inquiry into the Hawkesbury Nepean River System, Final Report. Healthy Rivers Commission, Sydney.

Jelliffe, P.A., G. Sabburg and J. Wolff (1994). Key factors in minimising water pollution from unsewered areas. AWWA 16th Federal Convention :85-90.

Lance, J. C. and Gerba, C. P. (1984). Virus movement in soil during saturated and unsaturated flow. Appl. Environ. Microbiol. 47, 335-337.

Page 116

Macler, B.A. and J.C. Merkle (2000). Current knowledge on groundwater microbial pathogens and their control. Hydrogeology J. 8(1):29-40.

O'Neill, R.A., G.K. Roads and R.N. Wiese (1993). On-site waste water treatment and disposal in NSW. School of Civil and Environmental Engineering, UTS,.

Rennecker, J.L., A.M. Driedger, S.A. Rubin and B.J. Mariñas (2000). Synergy in sequential inactivation of Cryptosporidium parvum with ozone/free chlorine and ozone/monochloramine. Wat. Res. 34(17):4121-4130.

Scandura, J.E. and M.D. Sobsey (1997). Viral and bacterial contamination of groundwater from on-site sewage treatment systems. Wat. Sci. Tech. 35(11-12):141- 146.

Sobsey, M.D. (1989). Inactivation of health-related microorganisms in water by disinfection processes. Wat. Sci. Tech. 21(3):179-195.

Standards Australia (1994). Disposal systems for effluent from domestic premises. Standards Australia, Homebush.

Standards Australia/Standards New Zealand (2000). AS/NZS 1547:2000 On-site domestic-wastewater management. Standards Australia, Homebush.

Swanson, P., A. Davidson, P. Hawkins and M. Cunningham (2000). Sources of Cryptosporidium and Giardia in the Warragamba water supply catchment. AWT, Sydney.

U.S. Environmental Protection Agency (2000). National Primary Drinking Water Regulations: Ground Water Rule; Proposed Rules. Federal Register, Vol. 65, No. 91 ed. US-EPA.

Page 117

Paper II Buffer Distances for On-Site Sewage Systems in Sydney’s Drinking Water Catchments

Published as:

Charles, K. J., D. J. Roser, N. J. Ashbolt, D. A. Deere and R. McGuinness (2003). "Buffer distances for on-site sewage systems in Sydney's drinking water catchments." Water Science & Technology 47(7-8): 183-189.

Reprinted here with minor modifications.

Author Contribution Role (all authors helped with reviewing and (%) editing) Charles, K 80 Research and writing, designed experiments and devised risk assessment approach Roser, D 5 Co-supervisor Ashbolt, N 5 Supervisor, assisted with experimental design Deere, D 5 Industry advisor, assisted with experimental design McGuinness, R. 5 Industry advisor Other Experiments designed by undertaken by AWT

Page 118

Abstract Pathogens and nutrients released from on-site sewage systems represent a risk to surface and ground water quality, particularly where there are sensitive receiving waters such as in drinking water catchments. Buffer zones between on-site systems and waterways are one barrier used to protect water quality. The increased time and distance they provide increases the opportunities for the effluent purification functions of the soil to occur. A risk management model is proposed to assess the efficacy of the buffer zones in Sydney’s drinking water catchments. The model is the basis for the development of performance based setback distances for on-site systems from waterways, and incorporates stochastic analysis of pathogen and nutrient transport in the environment and consideration of the effluent quality variability from on-site systems. Catchment- scale integration of contaminant transport is employed to facilitate a risk assessment of on-site systems. The risk management model also allows for the impact of on-site system management and maintenance on catchment water quality to be assessed through scenario building and feedback mechanisms.

Introduction The impact of on-site sewage treatment systems on water resource quality is of particular concern in areas of surface drinking water supply, groundwater drinking water supply and surface waters used for recreation. An assessment of failing on-site systems (Charles et al., 2001) has revealed that poor design, operation and maintenance are the major causes of water pollution from these systems. Prevailing standards and guidelines seek to address these problems, but gaps in current knowledge limit the performance of these guidelines, particularly in relation to soil capacity to immobilise wastewater constituents and minimum setback distances from waterways. The function of buffers/setbacks is to provide adequate travel time and distance to ensure the sustainable removal of contaminants from the effluent plume. This is achieved through natural transformations such as oxidation and nitrification, decay, sorption to soil, microbial antagonism, plant uptake, chemical reactions and dilution.

Studies of groundwater quality provide direct evidence of contamination of adjacent water resources with sewage where inadequate buffers are in place. Lewis & Stark (1993) reported an increased risk of enteric virus infection with exposure to groundwater from wells located in subdivisions with on-site sewage disposal. Yates

Page 119

(1985) reported outbreaks of hepatitis A, typhoid, Norwalk and echovirus that were traced back to contamination of drinking water wells with effluent from on-site systems. The proposed USEPA (2000) Groundwater Rule has been designed so that viruses in well water result in less than one infection per 10 000 people a year or a concentration of <2 x 10-7 viruses per litre (Regli et al., 1991) for non-disinfected water. Such low concentrations are well below what can be routinely assayed, so other approaches are needed to estimate and manage site specific viral risks. To address this, the USEPA rule permits the use of pathogen groundwater transport models to assess the horizontal groundwater travel time or setback distance criteria required to protect drinking water sources. One model developed to calculate site specific criteria bases the setback on the time taken to achieve the required reduction in viruses (based on a first order decay rate for viruses) and uses Darcy’s law for groundwater transport to assess the distance required to provide this length of time (Yates and Yates, 1989). This approach is widely applied due to its simplicity and can be extended to surface water runoff and transport. However, it does not account for the difference between water and virus transport. Other detailed models consider the influence of virus, soil and environmental properties (Schijven, 2001). While the application of these models to small sites is difficult due to the amount of data required (Yates et al., 2000), they are well suited to risk assessment and decision management approaches.

Nonetheless, for current setback distances in Australia, as well as in the United States, these models have yet to be used to develop regulations. In the New York drinking water supply catchments the setback for absorption trenches has been set at 30 metres (100ft). A study undertaken to assess the efficacy of this buffer (Curry, 2000) in protecting water quality reported phage transport up to the buffer distance under conditions that exist in the catchments. No conclusion was drawn regarding the implications for water quality and hence septic system management.

The 18 000 on-site systems in Sydney’s drinking water catchments are a concern primarily due to the release of phosphorus, nitrogen, human enteric viruses and parasitic protozoa (Charles et al., 2001). As in the New York catchments, the buffer distances for on-site systems in Sydney’s catchments of 100 metres from waterways, or 150m from water storages and their main tributaries, are under review. This paper details the review process undertaken to develop performance based buffer distances for Sydney’s

Page 120

drinking water catchments, including the use of risk assessment techniques to the application of pathogen and nutrient fate and transport modelling knowledge.

Methods Model development has been undertaken in three phases (Figure 1). The aim of Phase 1 was to provide an overview and rationale for research, including developing mechanisms to meet the Sydney Catchment Authority’s (SCA) management needs effectively. This involved consultation between SCA and other industry stakeholders to establish the management and research needs of the industry, including identifying resources and the needs of local regulatory authorities. A literature review, undertaken concurrently to address the needs, identified gaps in knowledge where research is required. A conceptual model of the fate and transport of pathogens and nutrients from surface and subsurface disposal on-site systems was developed in Phase 2 to address the needs identified. It was developed as a basis for designing the quantitative contaminant transport models required to delineate buffer distances, and to identify experiments and data needs. Phase 3 involves the development of quantitative models and undertaking experiments to gather calibration, validation and input data.

Literature Assess Sydney Catchment Stakeholder review Authority Needs consultation

Phase 1 Develop risk management model

Develop conceptual contaminant Identify key indicators transport and fate model and data needs Phase 2

Develop stochastic transport and fate model Phase 3 Calibrate and validate model

Figure 1 The buffer distance review process

Sample collection and analyses are currently being undertaken by ECOWISE and Sydney Water analytical laboratories in consultation with SCA and UNSW. Initial

Page 121

experimental works include the following:

A survey of effluent quality in 48 septic tanks (SCA, unpublished data Ferguson et al., 2009) has been undertaken to provide probability density function data for all key of constituents in effluent, including nutrients, potential effluent trackers and pathogens (adenoviruses, enteroviruses, reoviruses and coliphages by culture and Norwalk-like viruses, rotaviruses and hepatitis A virus by PCR).

A pilot study of groundwater contamination from on-site systems has been undertaken to trial potential effluent trackers and to gather data on soil types, groundwater conditions and suitable analytes.

A long term (12 month) study is being developed for measuring groundwater contamination and nutrient and pathogen transport, as well as further studies of effluent quality variability. The Wingecarribee reservoir catchment and surrounds (Figure 2) has been chosen for the experiments due to the relatively high population density and rainfall. The catchment covers an area of 40 square kilometres (SCA, 2000) and supports a population of 1,426 or approximately 555 on-site systems (CH2MHILL, 2001). The SCA has recorded an average annual rainfall of 1,000 mm in the past 13 years in Robertson, while the NSW Department of Public Works and Services (1998) reports an average of 1,665 mm. The extension of the model to the other areas of the catchment will be facilitated by expanding the groundwater study outlined above in a location in the outer parts of the catchments, and will also rely on available GIS data and coordination with other SCA projects.

Page 122

Wingecarribee Dam

Wingecarribee Swamp Caalang Creek

Burrawang

Robertson

1km N

Figure 2 Wingecarribee Catchment

Results & Discussion

Risk Management Model Framework Stakeholder consultation confirmed phosphorus, nitrogen, human enteric viruses and parasitic protozoa as the main concerns from on-site sewage disposal, and the critical need as buffer distance research to assist management control development within the catchments (Tierney, 2000). As evidenced by the lack of useful management related conclusions from the New York study (Curry, 2000), understanding pathogen and nutrient transport principles alone is insufficient to meet management’s needs. To set an appropriate buffer distance the risk implications to the catchments must be quantified, and buffer transport data must be integrated into the wider context of catchment management to allow cumulative impact assessment. The risk management model (Figure 3) was developed to identify the steps required to ensure that a suitable buffer zone, with a scientific basis, can be consistently applied between on-site systems and waterways. Using this model it is planned to calculate site specific performance-based buffer distances which consider the sources of risk as they relate to on-site system

Page 123

management, treatment, disinfection, land application and site factors.

Contaminant Transport and Fate Model Development The conceptual transport and fate model (Figure 4) outlines the processes and the data needs for phase 3. To address the sources of risk and variability, the conceptual model includes consideration of household and management activities that impact on the loads of pathogens and nutrients entering the environment, as well as consideration of the factors that impact on the movement and survival of these constituents in the environment. The effluent disposal system is considered the first point of contact of the effluent with the environment in a properly functioning system and affects the fate of the effluent constituents. Absorption trenches and spray irrigation are the main forms of disposal used in Sydney’s catchments (Charles et al., 2001) and hence, the initial focus of the model. To encompass both these designs, and variability in performance, effluent transport through both soil and groundwater and in surface runoff is being included.

The movement of effluent in the environment is a function of the effluent loads, soil characteristics, slope, rainfall, evapotranspiration, hydraulic conductivity and groundwater depth. However, contaminant transport and fate differs from that of effluent.

Page 124

Scenario Management building Operation, Maintenance, Replacement Risk Mana

Effluent Treatment system Disinfection system Disposal system g quality ement Model Feedback inputs Type, Performance Type, Performance Type, Performance

Surface Runoff Soil Groundwater Contaminant Rainfall, Slope Saturation, Depth Heterogeneity, Flow transport and fate in buffer zones Buffer distance Slope, Vegetation, Heterogeneity

Catchment transport Stream Water Quality and fate Cumulative impacts of on-site systems and comparison with other land uses, Off-take and Recreational Water Quality, Water Quality Objectives

Figure 3 Schematic of the risk management model

Page 125

Population, health, age, water and chemical use

Treatment system Sedimentation, inactivation, Maintenance, hydraulic load, disinfection solids load Disinfection system

Design, age, size, condition, clogging Disposal system Adsorption, sedimentation, inactivation, plant uptake

Slope, soil type, evaporation, rainfall, vegetation, Surface runoff Adsorption, inactivation, temperature sedimentation, plant uptake

Soil size and chemistry, organic matter, temperature, Soil Sorption, precipitation, hydraulic conductivity, rainfall, inactivation, plant uptake evapotranspiration

Groundwater Reversible P sorption, pathogen Soil size and chemistry, adsorption inactivation organic matter, temperature, flow velocity, heterogeneity

Surface water Sedimentation, inactivation, uptake

Figure 4 Conceptual transport and fate model

Nitrogen, particularly nitrate, is highly soluble and travels rapidly in groundwater with effluent; however, its concentration is affected by biological transformations and plant uptake (White & Kookana, 1998). Phosphorus travels at a retarded rate in groundwater due to sorption reactions, and is primarily removed where there is adequate unsaturated soil (Harman et al., 1996). Both P and N are also subject to transport in surface runoff, especially when the soil is saturated or infiltration rates are exceeded. The main nutrient model requirements are the loads of nutrients in their various forms. Information on the P-sorption properties of the soil, other reactions and plant uptake of nutrients is also being gathered from the sites and literature.

Human enteric viruses and parasitic protozoa (particularly Cryptosporidium oocysts), have been shown to survive common on-site treatment systems (Lewis and Stark, 1993) and be transported in groundwater and surface runoff. Indicator bacteria, commonly

Page 126

used to assess water quality, do not survive as long the pathogens they are supposed to represent, nor necessarily travel in the same manner, and hence the focus here is on virus and protozoan oo/cyst transport. Pathogen movement depends on the microbe structure, physical soil matrix and the chemistry that affects sorption. Data is required on infection prevalence, pathogen loads in effluent, and on factors that affect inactivation in the environment, including temperature, moisture, sorption to particulate matter and microbial activity (Schijven, 2001). Information on pathogens taken from literature will include inactivation rates and sub-population inactivation rates, size, surface chemistry and infectivity.

Stochastic model calibration, validation and inputs Stochastic modelling is to be employed for two reasons. Firstly, as the model is to assess risk a stochastic basis is preferred to give probabilistic outcomes for management use. Secondly, for modelling virus at a scale of 100 metres or more, a stochastic approach that allows consideration of subsurface heterogeneity has been shown to provide better estimates of virus transport rates (Rehmann et al., 1999), which should similarly apply to nutrient and protozoan oo/cyst transport. The stochastic model should allow scenario construction to evaluate the influence of surface conditions, subsurface variability and on-site treatment system performance. The latter may allow a better scientific basis for enforcement of on-site sewage system management and auditing regimes in sensitive environments through the risk management model. The stochastic model also allows for the inherent variability in the catchments, however, the acceptable percentile has yet to be decided. To estimate the 95th percentile with confidence considerably more data is required that for an 80th percentile, so in part, the acceptable criteria will rely on the precision available from this project and ongoing work.

Variability between systems has been assessed through the study of septic tank effluent quality. The data has been used to develop probability density functions for effluent constituents, which will be used as inputs to the model. Cumulative lognormal distributions provided an excellent fit for each analyte in the septic tank effluents (Table 1). Electrical conductivity, and hence dissolved solids, and ammonium and ortho-phosphorus were highly correlated, as were suspended solids with biochemical oxygen demand and total phosphorus. Surprisingly human enteric viruses were detected in 40% of septic tank effluent samples. Overall, the variability of constituent

Page 127

concentrations between tanks was high, for example suspended solids concentrations ranging between 12 and 29 000 mg.L-1.

The pilot groundwater study investigated two sites at Robertson. One site, a septic tank with absorption trench, had groundwater at 3-4 metres, with a hydraulic conductivity of

0.1 m.d-1. The nitrate plume on this site was measured over 12 metres from the trench with a maximum nitrate concentration of 7.0 mg.L-1. At the other site, an aerated wastewater treatment system with spray irrigation, groundwater occurred at a depth of

1-2 metres, with hydraulic conductivities between 2.3 and 7.4 m.d-1. The nitrate plume was measured over 20 metres from the spray field with a maximum nitrate concentration of 26.2 mg.L-1 and an average concentration at 20 m of 9.8 mg.L-1.

Table 1 Cumulative lognormal probability density function parameters (SCA, unpublished data)

Analyte Units Geometric Standard R2 (Log mean Deviation (log10) distribution) Biochemical oxygen demand mg.L-1 165 0.45 0.990 Suspended solids mg.L-1 177 0.69 0.995 Specific electrical µS.cm-1 conductivity 1336 0.28 0.996 Ammonium Nitrogen mg.L-1 84 0.43 0.998 Total Kjeldahl Nitrogen mg.L-1 116 0.41 0.998 Total Nitrogen mg.L-1 117 0.40 0.999 Total Phosphorus mg.L-1 18 0.38 0.996 Ortho Phosphorus mg.L-1 10 0.50 0.988

Based on these findings, further calibration experiments will measure the transport of effluent in the soils and groundwater in the catchment area, and the variability encountered over a year. Four experimental sites are to be characterised through hydraulic conductivity testing, effluent tracing and soil analysis, and have groundwater monitoring wells installed along the effluent plume centreline. Groundwater and effluent sampling will be undertaken for twelve months to assess effluent transport and seasonal variability, including continuous monitoring of electrical conductivity, pH, temperature and depth fluctuations throughout the period. Pathogen transport and breakthrough will be assessed through coliphage and fluorescent microsphere spiking experiments. The surface water runoff model will be based on the site characteristics,

Page 128

including soil type and infiltration rate, and literature on unsewered area and agricultural runoff, however, no runoff sampling will be undertaken.

Conclusions The proposed risk management model aims to provide a scientific basis for the design of site specific performance based buffer zones in Sydney’s drinking water catchments. The research for this model has been structured with a view to developing an effective management tool for development assessment of both new and existing on-site systems, and to assess the impact of improved management and maintenance of these systems on the catchments.

The variability experienced in septic tank effluent highlights the need for the stochastic basis of the contaminant transport and fate models included in the risk management model framework, and for such variability to be better understood. The definition of the worst-case scenario, the severity and the likelihood of occurrence are required to fully appreciate the risks to catchment water quality, as well as understanding of the lesser risks. Similarly the variability experienced between the sites used in the pilot study illustrates the need to assess the variability over greater distances than commonly studied. The ability of the project to cover the variability experienced throughout the 1.6 million hectares of catchments will be limited to information available from literature and SCA.

Acknowledgements The authors would like to thank Martin Krogh, Christobel Ferguson and Sanjay Athavale of the Sydney Catchment Authority, Mark Angles of Sydney Water and Declan Page of ACTEW for their assistance with this project.

References CH2MHill, 2001, Review of Catchment sewerage needs. Prepared for Sydney Catchment Authority. CH2MHill, Sydney.

Charles, K., N. Ashbolt, D. Roser, D. Deere and R. McGuinness (2001). Australasian Standards for on-site sewage management: Implications for nutrient and pathogen pollution in the Sydney drinking water catchments. Water (Aust.): December 2001.

Page 129

Curry, D.S. (2000). Final Report for the Septic Siting Study. New York City Department of Environmental Protection, New York, 210 pages.

Harman, J., W.D. Robertson, J.A. Cherry and L. Zanini (1996). Impacts on a Sand Aquifer from an Old Septic System: Nitrate and Phosphate. Groundwater 34(6):1105- 1114.

Lewis, A.L. and L.M. Stark (1993). Florida OSDS research Project. Viral Study Summary. Tampa Branch Laboratory/Virology, Florida.

NSW Department of Public Works and Services (1998) Robertson Sewerage Feasibility Study Report No. WE9864R, Report prepared for Wingecarribee Shire Council and Department of Land and Water Conservation.

Regli, S., J.B. Rose, C.N. Haas and C.P. Gerba (1991). Modeling the risk from Giardia and viruses in drinking water. J. AWWA 83:76-84.

Rehmann, L.L.C. and C. Welty (1999). Stochastic analysis of virus transport in aquifers. Wat. Resources Res. 35:1987-2006.

Schijven, J.F. (2001). Virus Removal from Groundwater by Soil Passage: Modeling, Field and Laboratory Experiments. Delft Technology University, Amsterdam.

Sydney Catchment Authority (SCA), 2000, Wingecarribee Swamp and Special Area Plan of Management 2000. SCA, Sydney.

Tierney, D. (2000) Performance of popular OSMS in SEPP 58 Area. Workshop Minutes. TPK, Sydney.

U.S. Environmental Protection Agency (2000). National Primary Drinking Water Regulations: Ground Water Rule; Proposed Rules. Federal Register, Vol. 65, No. 91 ed. US-EPA.

White, R.E. and R.S. Kookana (1998). Measuring nutrient and pesticide movement in soils: benefits for catchment management. Aust. J. Expt. Agric. 38:725-743.

Yates, M.V., W.A. Jury, S.R. Yates, D.L. Anderson, L.M. Stark and P. Sherblom (2000). Measurement of virus and indicator survival and transport in the subsurface.

Page 130

AWWA Research Foundation, Denver, USA.

Yates, M.V. (1985). Septic tank density and groundwater contamination. Ground Wat. 23:586-591.

Yates, M.V. and S.R. Yates (1989). Septic tank setback distances: a way to minimise virus contamination of drinking water. Ground Wat. 27:202-208.

Page 131

Paper III Designing on-site sewage disposal systems to protect public health

Published as:

Charles, K. J., J. F. Schijven, C. Ferguson, D. J. Roser, D. A. Deere and N. J. Ashbolt (2003). Designing on-site sewage disposal systems to protect public health. On-site '03 Future directions for on-site systems: Best management practice, Armidale, Lanfax Labs. 101-108.

Author Contribution Role (all authors helped with reviewing and (%) editing) Charles, K 85 Research and writing Schijven, J 2.5 Ferguson, C 2.5 Roser, D 2.5 Co-supervisor Ashbolt, N 2.5 Supervisor Deere, D 2.5 Industry advisor McGuinness, R. 2.5 Industry advisor

Page 132

Abstract Contamination of surface water by human enteric viruses from on-site sewage treatment and disposal systems is a major concern for the protection of public health. Limited information is available on virus concentrations in septic tanks, however, calculations indicate that maximum concentrations could exceed 1010 viruses per litre for short periods when there is an infected resident in the household. Virus removal from wastewater in septic tanks, primarily due to sedimentation with attached particles, is less than one log10. Hence, land application systems and subsequent effluent transport in buffer distances still need to achieve large reductions of viruses to protect water quality.

The present knowledge of virus removal mechanisms can be practically applied to critical control points in design and management of on-site systems, many of which are already stipulated in the guidelines, to maximise virus inactivation and minimise virus transport.

Introduction The performance and sustainability of on-site sewage treatment and disposal systems have frequently been assessed as poor compared to centralised sewage systems. This bad reputation has been reinforced by incidents such as the Hepatitis A outbreak at Wallis Lakes in 1996, as well as many incidents of contaminated drinking water supply wells overseas. Hence, on-site systems have become a scapegoat whenever water quality problems occur.

The primary potential threat to public health posed by on-site systems is from viruses, due to their small size allowing easy transport and their robustness in the environment. The recognition of this threat has led increasingly to research being focused on the transport and fate of viruses. The on-site industry is not alone in identifying these concerns and hence increasing the research, with many other water and wastewater treatment disciplines also identifying this issue. The research includes the removal of viruses from water and wastewater through to treatment systems, disinfection systems and transport through soil in land application systems, sand filters and buffer distances.

It is important that this insight is applied within the on-site industry. From design to council inspections the potential for public health impacts from on-site systems needs to

Page 133

be assessed, not just with respect to faecal coliforms, but with respect to pathogenic microorganisms such as viruses.

Decisions about protecting drinking water quality implies a need to deal with the question of whether or not it is necessary to sewer a community where there is concern about water quality impacts. For example, Sydney Catchment Authority (SCA) manage the Sydney drinking water catchments, in which it has been estimated that there are over 18,000 on-site sewage treatment and disposal systems, approximately 72% of which are septic tanks and 24% are aerated wastewater treatment systems (Charles et al., 2001). The poor performance of these systems has been widely reported (Charles et al., 2001) and, as such, has been the focus of management reforms, however the impact of these failures is not properly quantified. While the causes of failure are varied, appropriate design and regular maintenance address many of the risks associated with sewage disposal.

The constituents of sewage that are of greatest concern in the drinking water catchments are phosphorus, nitrogen and pathogens. The pathogens group includes human enteric viruses, Cryptosporidium, Giardia, pathogenic bacteria and emerging pathogens. While SCA seek to minimise the transfer of all contaminants to waterways, relative inputs of nutrients and larger microbial pathogens (bacteria and protozoa) from other diffuse sources (agriculture and dairying) are significant in the Sydney catchments, and hence it is the most mobile pathogens, human enteric viruses, that are the main contaminant of concern from on-site systems. ((Charles et al., 2002); Ferguson et al., 2003)

Human enteric viruses are viruses that enter the human gastro-intestinal tract, reproduce in the throat and intestine and are excreted in faeces. They most commonly cause gastroenteritis in infected people, however they may also cause severe illnesses such as meningitis, encephalitis, paralytic poliomyelitis, myocarditis, hepatitis and/or conjunctivitis. Common waterborne viruses that can infect humans are enteroviruses (such as ECHO, polio and coxsackie), rotaviruses, human caliciviruses (such as the Noroviruses), Hepatitis A and E viruses, astroviruses and adenoviruses. These viruses differ in size, robustness and surface charge, and hence, may differ in their transport and fate in the environment.

Bacteriophages (phages) are viruses that infect bacteria and are harmless to humans.

Page 134

They have similar properties to the human enteric viruses, as well as similar variability between different phages, and hence are often used as model viruses in experimental situations. They are easy to propagate in high concentrations and are easy to measure. These properties make phages appropriate for use as tracers or model viruses in studies on the fate and subsurface transport of viruses.

This paper provides a review of literature on virus transport and fate in the environment as it relates to on-site sewage disposal, particularly in Sydney’s drinking water catchments. Specifically, it investigates the role of moisture content, organic matter, loading duration, rainfall and point of disposal on virus attenuation to inform the design and placement of sewage disposal areas.

Additionally, this paper assesses potential critical control points for minimising risk that may be applicable to larger developments and/or sensitive sites. And the potential implications of these management practices on the performance of the disposal areas with respect to hydraulics, phosphorus removal and nitrogen removal.

Virus Numbers in Sewage and Septic Tanks Few studies have been undertaken to quantify the numbers of human enteric viruses present in septic tanks as a result of an infection within the household. Deborde et al. ((Deborde et al., 1998b) reported ranges of 0.26-4.4 viruses L-1 in school septic tank systems serving 350 people. Lewis and Stark ((Lewis et al., 1993) reported ranges of 0.07 viruses L-1 to greater than 59 L-1 for household septic tanks, with viruses detectable in effluent for up to 137 days following an infection. In centralised sewage systems, virus concentrations of 104-105 viruses L-1 of raw sewage ((Gerba, 2000a)a) have been reported.

However, it is possible that the concentrations of viruses in septic tanks may be much higher than reported due to variability in rates of excretion and duration of infection, with concentrations of up to 107 viruses L-1 predicted by Canter and Knox (Canter et al., 1985). For example, an infected individual can shed between 102 viruses per gram of stool during coxsackie and ECHO virus infections, which may continue for two weeks to four months, but up to 1012 for rotavirus infections, which may continue two to three months (Gerba, 2000b)b). Norovirus (Norwalk-like virus) infections cause shedding for seven to 14 days with a peak of up to 109 viruses per gram (Meschke, 2001).

Page 135

Based on what is known of virus shedding, the peak concentrations (Table 1) were calculated using the following assumptions:

• A septic tank of 3000 L receiving 1000 L per day of wastewater;

• One infected individual excreting 150g per day of faeces (fromFeachem et al., 1983);

• A log10 triangular distribution of virus load defined by the peak load (at time = 33% of duration) and duration.

Table 1 Virus concentrations calculations (dilution only, no removal)

Duration Geometric mean Concentration of Excretion influent Peak in effluent infection Peak concentration concentration in 30d after Data (days) (viruses g-1) (L-1) effluent (L-1) peak (L-1) Norwalk Average 7 106 2 x 102 9 x 103 5 x 10-2 Maximum 14 109 1 x 104 2 x 107 1 x 102 Rotavirus Average 30 1010 7 x 104 5 x 108 5 x 103 Maximum 90 1012 9 x 105 8 x 1010 3 x 106

These results do not include potential virus inactivation or removal within the septic tank. Virus removal from wastewater in septic tanks, primarily due to sedimentation with attached particles, is less than one log10 (Faulkner et al., 2002). Removal in secondary treatment systems is expected to be slightly higher than septic tanks with an additional 1-2 log10 by disinfection (Charles et al., 2003b).

Factors Affecting Virus Transport and Survival Several reviews have been published on the transport and fate of viruses in soil and groundwater (Yates et al., 1988); (Schijven et al., 2000); (Jin et al., 2002), with many attempts have been made to assess and model the main virus removal mechanisms – inactivation and adsorption – however, the conditions under which these models are appropriate require better definition. For example, they depend on the virus as removal rates of viruses vary greatly depending on virus properties (e.g. size, surface charge, stability) and soil/aquifer properties (e.g. porosity, surface charge, antagonistic microbial activity).

There are many conditions that affect the fate and transport of viruses. In sewage land

Page 136

application systems there are a number of conditions that can be manipulated to maximise virus removal through inactivation and minimising virus transport including: soil moisture content, hydraulic conditions, organic matter, and salt concentrations/rainfall. While different viruses do respond differently to these conditions, there a number of generalisations that can be made (Table 2).

Table 2 Factors affecting virus transport and fate that can be manipulated by land application area design

Factor Survival Migration Moisture Inactivation decreases with Transport increases with content increasing moisture content increasing moisture content Salt Increasing concentration enhances concentration virus adsorption (conversely rain water reduces ionic strength and desorbs viruses) Organic matter Organic matter decreases Organic matter may reduce inactivation adsorption Hydraulic Virus migration increases with conditions increasing hydraulic loads and flow rates

Soil moisture content Transport in unsaturated soil plays an important role in virus removal from land-applied sewage, accounting for significantly higher removal than the saturated zone. This is due to greater adsorption, through greater contact with the soil, and greater inactivation, due to desiccation and surface tension stresses. For example, Powelson and Gerba (Powelson et al., 1994) reported that unsaturated conditions resulted in an average removal coefficient more than three times higher than saturated conditions.

Salt concentration / rainfall Virus attachment to soil is proportional to the salt concentration or ionic strength of the soil water, and detachment is enhanced if conditions are reversed. Sewage generally has a high ionic strength which enhances virus attachment, however, a decrease in ionic strength can result in the viruses desorbing. One of the common causes of this reduction in ionic strength is rainfall.

Another consideration with rainfall is the generation of runoff from the sewage land

Page 137

application area.

Organic matter The composition of the sewage will affect the transport and fate of viruses. For example, both dissolved organic matter and detergents (anionic surfactants) in the effluent decrease attachment of viruses to soil. Organic matter, both in the soil and in the effluent, reduces adsorption due to either complexing with and/or by competing with it (Powelson et al., 1991) and can counteract the effect of higher salt content present in sewage.

While reducing organic matter increases virus adsorption, the reduction achieved from primary to secondary treated effluent is not necessarily sufficient to affect virus transport (Gerba et al., 1978), however the reduction achieved from secondary to tertiary treatment has been reported to increase virus adsorption (Dizer et al., 1984). Additionally, dissolved organic carbon and anionic surfactants have been reported to decrease the inactivation rates of phage (Ryan et al., 2002).

Where there are expected to be an abundance of attachment sites, such as in clay soils, the organic load may not be relevant (Gerba et al., 1978).

Hydraulic conditions The loading rate and duration of effluent application will affect the transport of viruses. The loading rate relates to the moisture content effects described above, with greater loading rates increasing the degree of saturation. Additionally, prolonged application of effluent can result in a decrease in virus adsorption (Schaub et al., 1982).

Design Criteria The present knowledge of virus removal mechanisms can be practically applied to design and management of on-site sewage land application systems to maximise the removal of viruses from sewage through inactivation and minimise virus transport through increasing adsorption (Table 3).

Page 138

Table 3 Conditions and actions for improved virus management for on-site systems

Condition Action Main Effect Moisture Increase trench surface area; Decreasing saturation increases content Reduce irrigation loading rate; virus inactivation Improve distribution system to ensure uniform distribution of effluent Salt Use run-off mitigation measures Run-off mitigation reduces the concentration/ around disposal areas wash-off of viruses and reduces rainfall remobilisation of adsorbed viruses Organic Increase treatment to remove Decreasing organic matter matter organic matter prior to land increases adsorption application Hydraulic Use alternative disposal systems, Mitigates the reduction in virus conditions such as alternating between two adsorption capacity of the soil or more trenches, or moving the experienced with prolonged sprinkler system regularly sewage application

Soil moisture content The design of subsurface land application areas is commonly based on the capacity of the soil to absorb the effluent. However, during the wetter and cooler periods of the year, the water content of the soil is close to or at saturation. Similarly, surface land application areas are designed not to be waterlogged.

To maximise virus removal, the size of effluent land-application areas needs to be increased, so that, the moisture content is minimised, and rarely, if ever, achieves saturation. Increasing the size of trenches and improving the distribution systems within them will decrease the loading rate and decrease the degree of saturation underneath the trench. Similarly, lowering loading rates in irrigation systems will maximise virus removal in the unsaturated zone.

While secondary treatment does provide greater virus reduction than primary treatment, the combination of insufficient data on virus removal and poor performance/maintenance of AWTS (Charles et al., 2003) makes it unreasonable to consider the surface spray irrigation of effluent acceptable in areas where runoff is an issue, or where aerosols may contact or infect people. Additionally, increased loading rates of secondary treated effluent in subsurface land application areas are not advisable

Page 139

as they will increase transport and survival of viruses.

The current critical control point used in this industry to assess soil moisture content is the presence of waterlogging in land application areas. A better appropriate critical control point would be the installation of observation wells within subsurface land application areas to assess the ponding of water in trenches or high water tables.

The increase in area for land application would also facilitate the removal of phosphorus through increased adsorption through contact with soil, and would assist the nitrification of the effluent through improved aeration.

Salt concentration/rainfall The higher salt concentration experienced in effluent land-application areas is expected to increase virus attachment in that area. However, this attachment is subject to variation with changing salt concentrations. For example, rain can elute viruses from the soil transporting them further down the soil column and into the groundwater. Run-off mitigation measures will limit the amount of additional water entering the land application area, and reduce the wash-off of viruses and other contaminants from surface disposal area or where there is surfacing from subsurface disposal areas.

The current critical control point of diverting run-off away from the land application area needs to be more widely and more routinely adopted and enforced.

Organic matter Increasing treatment to a tertiary level should not only achieve greater removal of viruses and nutrients, it will enable more effective treatment of the effluent by the land application system. Particularly, the reduction in organic matter will facilitate greater adsorption of viruses to the soil. And while land application systems do provide treatment, the additional treatment is most beneficial before release to unsaturated soil.

For organic matter, a critical control point for tertiary treatment similar to the current one for secondary treatment (i.e. biochemical oxygen demand and suspended solids removal requirements) could be adopted.

Hydraulic conditions As the duration of sewage application affects the virus removal properties of the soil,

Page 140

the use of multiple disposal areas may increase the life-expectancy of on-site systems.

The use of alternating disposal areas in land application systems is not a new concept, but it is also not a widely implemented one through design or management. Similarly, the concept of moving sprinklers is well known but difficult to enforce. Hence, these critical control points need to be upgraded to ensure that each development has two land application systems that can be alternated with ease, with each land application system designed to receive the full effluent load such that the land area is doubled.

The effect of doubling the land application area should also increase the removal of phosphorus through adsorption, as well as increasing the removal of both nitrogen and phosphorus through increased plant uptake.

Discussion These methods described here for maximising the removal and minimising the transport of viruses are based on the best available research in the field. As many as possible of these methods need to be adopted at each site as a multiple barrier system to prevent water contamination, that in turn is part of a multiple barrier system to prevent drinking water contamination that may include sewage treatment, buffer distances between sewage disposal and waterways, drinking water treatment and disinfection.

Similarly to those critical control points discussed above, buffer distances can maximise virus removal and minimise virus transport through provision of increased filtration and sedimentation from surface runoff, and increased adsorption and travel time (decay) in groundwater.

Soil type is naturally a consideration in system design and in virus removal. However, the methods for increasing virus removal discussed here are applicable to all soil types.

The results of laboratory experiments often overestimate the virus removal experienced in the field. One of the reasons for this is that virus removal decreases with distance (Schijven, 2001). Hence, it is important that results are adopted into guidelines and regulations conservatively.

There are a number of concerns about the use of surface spray irrigation as an effluent land application method. Surface application does increase virus inactivation due to

Page 141

exposure to sunlight, as well as increasing nutrient uptake by plants, however, run-off of pathogens and nutrients is a concern. However, surface application can also increase public health risks through direct human contact with sewage and aerosols.

Conclusion Critical control points for the management of on-site sewage treatment and disposal systems are important for implementing a multiple barrier system to protecting water quality, the environment and public health. The current guidelines and standards do provide several critical control points for maximising virus inactivation and minimising virus transport. However, these need to be upgraded to be more effective and more readily enforceable, including:

• Installation of monitoring bores in subsurface land application areas to assess ponding in trenches and soil moisture;

• Run-off diversion;

• More stringent guidelines for biochemical oxygen demand and suspended solids; and

• Installation of two active land application areas for each on-site treatment system.

References Canter, L. W. and R. C. Knox (1985). Septic tank system effects on ground water quality. Chelsea, Michigan, Lewis Publishers.

Charles, K., N. Ashbolt, et al. (2001). On-site sewage management, AS/NZS 1547 and the Sydney catchment area. On-Site '01 Advancing On-site Wastewater Systems: Design & Maintenance, Armidale, University of New England.

Charles, K. J., N. J. Ashbolt, et al. (2002). Impacts of centralised versus decentralised sewage systems on water quality in Sydney's drinking water catchments. Small Water & Wastewater Treatment Plants, Istanbul, IWA.

Charles, K. J., N. J. Ashbolt, et al. (2003). Disinfection in Aerated Wastewater

Page 142

Treatment Systems. Ozwater, Perth.

Deborde, D. C., W. W. Woessner, et al. (1998). "Virus occurrence and transport in a school septic system and unconfined aquifer." Groundwater 36: 825-834.

Dizer, H., A. Nasser, et al. (1984). "Penetration of different human pathogenic viruses into sand columns percolated with distilled water, groundwater, or wastewater." Applied & Environmental Microbiology 47(2): 409-415.

Feachem, R. G., D. J. Bradley, et al. (1983). Sanitation and disease health aspects of excreta and wastewater management. New York, World Bank/John Wiley & Sons.

Gerba, C. (2000a). Domestic wastes and waste treatment. Environmental Microbiology. R. M. Maier, I. L. Pepper and C. P. Gerba. San Diego, Academic Press: 505-534.

Gerba, C. P. (2000b). "Assessment of enteric pathogen shedding by bathers during recreational activity and its impact on water quality." Quantitative Microbiology 2(1): 55-68.

Gerba, C. P. and J. C. Lance (1978). "Poliovirus removal from primary and secondary sewage effluent by soil filtration." Applied & Environmental Microbiology 36(2): 247- 251.

Ferguson, C., N. Altavilla, et al. (2003). "Prioritizing watershed pathogen research." Journal of the American Water Works Association 95(2): 92-102.

Jin, Y. and M. Flury (2002). "Fate and Transport of Viruses in Porous Media." Advances in Agronomy 77: 39-102.

Lewis, A. L. and L. Stark (1993). Florida OSDS Research Project. Viral Study Summary. Florida, Tampa Branch Laboratory/Virology.

Meschke, J. S. (2001). Comparative adsorption, persistence and mobility of norwalk virus, poliovirus type 1 and F+RNA coliphages in soil and groundwater. Environmental Science and Engineering, School of Public Health. Chapel Hill, University of North Carolina: 274.

Powelson, D. K. and C. P. Gerba (1994). "Virus removal from sewage effluents during

Page 143

saturated and unsaturated flow through soil columns." Water Research 28(10): 2175- 2181.

Powelson, D. K., J. R. Simpson, et al. (1991). "Effects of organic matter on virus transport in unsaturated flow." Applied & Environmental Microbiology 57(8): 2192- 2196.

Ryan, J. N., R. W. Harvey, et al. (2002). "Field and Lab investigations of inactivation of PRD1 and mS2 attached to iron oxide coated quartz sand." Environmental Science & Technology 36: 2403-2412.

Schaub, S. A., H. T. Bausum, et al. (1982). "Fate of virus in wastewater applied to slow- infiltration land treatment systems." Applied & Environmental Microbiology 44(2): 383-394.

Schijven, J. F. (2001). Virus Removal from Groundwater by Soil Passage: Modeling, Field and Laboratory Experiments. D. Amsterdam, Delft Technology University.

Schijven, J. F. and S. M. Hassanizadeh (2000). "Removal of viruses by soil passage: Overview of modeling, processes, and parameters." Critical Reviews in Environmental Science and Technology 30(1): 49-127.

US-EPA (2002). On-site Wastewater Treatment Systems Manual, US EPA, Washington.

Yates, M. V. and S. R. Yates (1988). "Modelling microbial fate in the subsurface environment." CRC Critical Reviews in Environmental Control 17(4): 307-344.

Page 144

Paper IV Centralised versus decentralised sewage systems: a comparison of pathogen and nutrient loads released into Sydney’s drinking water catchments

Charles, K. J., N. J. Ashbolt, C. Ferguson, D. J. Roser, R. McGuinness and D. A. Deere (2004). "Impacts of centralised versus decentralised sewage systems on water quality in Sydney's drinking water catchments." Water Science & Technology: 48(11-12):53-60.

Author Contribution Role (all authors helped with reviewing and (%) editing) Charles, K 75 Research and writing Ashbolt, N 5 Supervisor. Paper development Ferguson, C 5 Roser, D 5 Co-supervisor, Paper development McGuinness, R. 5 Industry advisor Deere, D 5 Industry advisor, Paper development

Page 145

Abstract Data collected from centralised and decentralised sewage treatment plants throughout Sydney’s drinking water catchments was used to calculate the relative catchment loads of Cryptosporidium, enteric viruses, nitrogen and phosphorus for an initial screening assessment. Loads were assessed at median and 90 percentile values for expected and worst-cases scenarios. The expected scenario in the Sydney drinking water catchments is that decentralised systems (servicing 32,800 people) provide similar total loads to centralised systems (serving 70% of the catchment population) for total phosphorus (37,090 kg.y-1), Cryptosporidium (1011 oocysts.y-1) and enteric viruses (9.1 x 1013 y-1), but higher loads of total nitrogen (237,610 vs. 136,740 kg.y-1). Decentralised systems, however, were predicted to have higher loads in the worst-case scenario with 620,620 kg.y-1 TN, 82,040 kg.y-1 TP, 7.3 x 1013 Cryptosporidium oocysts.y-1 and 9 x 1015 enteric viruses per year. Greater load variability was experienced with decentralised systems, which presumably reflects less reliability in their current operation and maintenance. Overall, catchment water quality is therefore not only affected by sewage disposal methods, but also failure issues. Decentralised system disposal to land may afford a degree of mitigation that can be enhanced, if the degree of failure is reduced.

Introduction The debate over whether centralised or decentralised sewage systems are ‘better’ has many facets, including establishment and ongoing maintenance costs, levels of maintenance, effluent quality and environmental and social impacts. The relative importance of each of these depends on local priorities. In Sydney’s drinking water catchments the priority, in light of the water quality Cryptosporidium incidents of 1998 (McClellan, 1998), is for drinking water quality protection from faecal contamination.

Centralised systems provide a reliable and high level of sewage treatment, however, treated effluent is commonly discharged directly into catchment source waters. Decentralised systems are often poorly designed and maintained, thereby providing a highly variable level of sewage treatment. Nonetheless, dispersed disposal to soil can provide additional treatment thereby buffering the impact of the effluent on waterways.

The majority of faecal contamination data for Sydney and other catchments is limited to thermotolerant (faecal) coliform data, although great efforts have been made to collect

Page 146

data on Cryptosporidium oocysts, Giardia cysts and enteric viruses in recent years. Neither of these datasets alone, however, are sufficient to estimate the contributions from human versus other sources of faecal pollution (Ashbolt et al., 2001; Ashbolt et al., 2002).

This paper reports research undertaken to assess potential pathogen and nutrient loads from centralised and decentralised systems within the Sydney drinking water catchments. The aim of this research is to provide data to aid catchment management and to prioritise faecal load rectification projects.

Methods Sewage treatment plant (STP) nutrient data, as collated from a variety of project reports, was sourced from CH2M HILL (2001). Samples for pathogen analysis (Griffith et al., 2002) were collected from the licence compliance point of the twelve STPs. Twenty nine STP effluent samples were concentrated for the analyses of Cryptosporidium oocysts using flatbed filtration followed by immunomagnetic separation and detected using immunofluorescent antibody assay as described for Method 1623 (US-EPA, 1999). Recovery of oocysts was determined by the incorporation of ColorSeed™ (Texas Red labelled, flow cytometric counted 100±1 oocysts, BTF Pty. Ltd., Sydney) to all samples. Virus particles (from 1 L of sewage) were concentrated with 8% polyethylene glycol (PEG-6000) and the concentrate resuspended in 40 mL Minimal Essential Medium. Concentrated virus samples were inoculated onto three cell lines (Hep-2,

LLC-MK2 and MRC-5) and based on cytopathic effects, identified as reoviruses, adenoviruses or enteroviruses (Sydney Water method EPWI503G). Small round structured viruses (SRSV’s), including Norwalk-like viruses (NLV or noroviruses), were analysed using reverse transcriptase-polymerase chain reaction (RT-PCR) (Griffith et al., 2002). Septic tank supernatant samples were collected from the outlet of 40 septic tanks at residential and community or multi-user sites (SCA, unpublished data). The samples were analysed for a range of constituents including: Total Nitrogen (TN) and Total Phosphorus (TP) (as per Standard Methods [APHA, 1995]), Cryptosporidium oocysts (Method 1623 [US-EPA, 1999]) and enteric viruses (Sydney Water method EPWI503G; Griffith et al., 2002).

Two scenarios were examined for contaminant loads. The expected nutrient load from

Page 147

the STPs was calculated as the average dry weather flow (ADWF), converted to litres per year, multiplied by the average (or 50th percentile) concentration. The worst case load was calculated using the 90th percentile or expected maximum (Mt Victoria, Blackheath, Braemar, Marulan, Goulburn and Braidwood) for each STP. The Cryptosporidium load was calculated as the ADWF, converted to litres per year, multiplied by the method recovery-adjusted concentration of Cryptosporidium per litre. The expected and worst case scenarios represent the ADWF multiplied by the median and maximum concentrations respectively for each STP. Marulan was assumed to have equivalent concentrations of enteric viruses and Cryptosporidium to Goulburn STP. The expected virus load was calculated based on a 17% incidence rate of 105 per litre being removed with 30% efficiency within the STP, and a background concentration of 102 per litre for the remainder of the time. The worst case load was calculated on a 100% incidence rate of 105 per litre.

Expected and worst case scenarios for decentralised systems were calculated from system flow, converted to litres per year, multiplied by the average or 90th percentile nutrient concentration for that system type to provide the expected and worst case values respectively. Pathogen loads were calculated similarly to nutrient loads, however incidence rates of 8% and 50% were assumed for Cryptosporidium and enteric viruses respectively, based on the occurrence data.

Page 148

Results and Discussion

Centralised and Decentralised Sewage Systems in Sydney’s Drinking Water Catchments Sydney’s drinking water, which provides for over 4 million customers, is sourced from 16 000 km2 of mixed land-use catchments. The predominant land uses are bushland (50.5%) and unfertilised grazing (42.5%), with the remaining 7% consisting of sewered urban, unsewered urban and peri urban, orchards, intensive vegetable growing, fertilised grazing and industrial (Swanson, 2001). While the catchment areas surrounding dams are generally protected from development, the catchments hold a permanent human population of approximately 109 400; over 50% of which are concentrated in two of the twelve Local Government Areas (LGAs), Wingecarribee (39 500) and Goulburn (20 450).

Within the catchments, 70% of the population are serviced by centralised sewage. This includes eleven centralised sewage treatment plants (Table 1), as well as one centralised septic tank effluent treatment plant (Marulan) which was included with STPs for the purposes of this paper. The majority of these systems discharge effluent into the waterways within the catchments. The exceptions are Goulburn and Marulan that irrigate the effluent. A further five centralised STPs are proposed with a combined capacity to serve 7,120 equivalent persons. The remaining catchment population (approximately 32,800) is serviced by decentralised sewage treatment systems, primarily single household systems.

Page 149

Table 1 Existing Sewage Treatment Plants in Sydney’s catchments

STPs (LGA) Equivalent Average Dry Weather Flow population (ADWF) kL/day Lithgow (Lithgow) 16560 4400 Wallerawang (Lithgow) 2100 530 Mt Victoria (Blue Mountains) 1000 300 Blackheath (Blue Mountains) 5300 1600 Braemar (Wingecarribee) 9000 2250 Bowral (Wingecarribee) 9975 2500 Moss Vale (Wingecarribee) 6500 1600 Bundanoon (Wingecarribee) 2040 210 Berrima (Wingecarribee) 1480 370 Marulan (Mulwaree) 450 110 Goulburn (Goulburn) 28000 6800 Braidwood (Tallaganda) 960 240 Total 83365 20910

Decentralised sewage systems in the catchments are currently managed on an individual level, overseen by the local government (Local Government (Approvals) Regulation 1999). There are 18,465 decentralised systems (Table 2), with the majority in the Wollondilly and Wingecarribee local government areas. The current treatment options include:

• septic tanks, which provide primary sedimentation, usually with effluent disposal via an absorption trench or field;

• Aerated Wastewater Treatment Systems (AWTS), which provide secondary treatment and disinfection, usually with effluent disposal via spray irrigation;

• pump-out systems, where sewage is stored on-site then collected for treatment and disposal at a centralised facility; and

• a small selection of alternative systems including sand mounds and composting toilets.

Page 150

Table 2 Decentralised sewage systems in the catchments (Charles et al., 2001)

System Type Number of systems Percent of Total Septic tank 13 500 73 AWTS 3 744 20 Pump out 1 168 6 Other1 53 <1 Total 18 465 100 1 includes composting toilets, sand mounds and sand filters.

The other form of sewage treatment not covered by these options is package sewage treatment plants. At present there are approximately 125 small package sewage treatment plants in the catchments, however there is insufficient data to consider them further in this evaluation and hence, the population that would be served is considered under the other decentralised sewage options discussed.

Centralised Sewage System Performance Concentrations of TN and TP in the effluents discharged from the twelve STPs examined are provided in Table 3. There were significant (p<0.05) differences between some of the plant effluents for both TN and TP, presumably reflecting the different efficacies of the plants rather than inherent differences in the domestic sewage being treated. Goulburn and Marulan have higher nutrient concentrations as the effluent from these plants is irrigated rather than being discharged to waterways.

Cryptosporidium oocysts were detected in 76% (N=29) of samples. After correcting for oocyst recovery, the maximum count was in excess of 290 oocysts per litre, with a median of 0.7 oocysts per litre.

Page 151

Table 3 Nutrient concentrations in Sydney drinking water catchment STP effluents (mg/L)

STPs TN TP 50th 90th 50th 90th percentile percentile percentile percentile Lithgow (N=7) 8.2 9.5 4 6.1 Wallerawang (N=4) 12.8 16.3 8.4 9.2 Mt Victoria (N~28) 13.6 14.9* 0.3 0.45* Blackheath (N~28) 35.6 35.2* 7.5 7.5* Braemar (Design) <8 <0.3 Bowral (N=28) 16.3 21.2 0.73 1.04 Moss Vale (N=28) 5.6 6.7 0.19 0.34 Bundanoon (N=15) 5.9 7.9 0.29 0.38 Berrima (N=15) 2.5 4.7 0.1 0.16 Marulan (N=36) 10.5** 37.5** 11.5 14 Goulburn (unknown) 30 9 Braidwood (N=1) 2.8** 8.7 * mean; ** NH4

Thirteen STP effluent samples were analysed for enteric viruses by cell culture, with adenovirus present in 2 samples, reovirus in four and enterovirus in one. Hence, a total of 17% of STP samples were positive for the three groups of infectious enteric viruses examined. Samples were collected from April to July 2002, and would need to be collected over a few years to account for likely seasonal variability (Vantarakis and Papapetropoulou, 1999). While no attempt was made to quantify the enteric viruses, typical ranges in secondary effluents are considered to be 30-70% (Long & Ashbolt, 1994) of the total 104-105 viruses per L of raw sewage (Gerba, 2000a). Noting however, that the non-culturable NLV may be the most numerous group expected in sewage, with Lodder et al. (1999) reporting up to 107 RNA-containing particles per litre.

Page 152

Centralised Sewage System Loads The expected (median) loads of TN and TP from Sydney’s catchment STPs were 136,740 and 37,091 kg.y-1 respectively. The worst case scenario loads (90 percentiles) of TN and TP were 175,934 and 44,577 kg.y-1 respectively. The Cryptosporidium loads (oocysts.y-1) from Sydney’s catchment STPs were 1.0 x 1011 for the expected scenario and 2.4 x 1011 for the worst case scenario. The expected viral load released into the catchments was 9.1 x 1013 and for the worst case load, 5.3 x 1014 viruses per year.

Decentralised Sewage System Performance Nutrients concentrations in decentralised systems are summarised in Table 4, excluding values from one septic tank site due to the extremely high values – 1,000 mg/L TN and 460 mg/L TP.

Table 4 Nutrient concentration averages (mg/L) in catchment decentralised systems (SCA, unpublished data)

Analyte Septic tanks AWTS Other Septic tank All Residentia Community 90th percentile l TP 23 24 17 16 0.08 54 TN 170 178 143 39 23 410

Cryptosporidium was detected in 2 out of 24 samples (8%), with counts of 230 and 510,000 oocysts L-1. Non detects were assumed to represent none present rather than below detection, due to the sporadic nature of infection and an expected disease occurrence of 0.1-1.0% in the adult communities. In contrast, enteric viruses were detected in 50% of samples (N=34) for these decentralised systems, the majority (14) of which were NLV. Reovirus and enterovirus were also detected.

The concentrations of enteric viruses in decentralised systems are heavily dependant upon community infection and excretion rates, for which there is relatively little data. From the literature, septic tanks may be positive for enteroviruses 12% of the time and contain some 0.26-4.4 viruses.L-1 (N=16, Deborde et al., 1998), and during infection in the household, may contain up to 107 viruses.L-1 (Canter & Knox, 1985). The lower concentrations reported are not expected to be peak concentrations, as virus concentration per gram of stool has been reported to vary depending on the stage of

Page 153

infection and the virus type, ranging from 102 during coxsackie and echo virus infections and up to 1012 for rotavirus infections (Gerba, 2000b).

Decentralised Sewage System Loads To estimate nutrient and pathogen loadings from decentralised systems, a daily water flow rate of 150 litres per person.day was assumed (Department of Local Government, 1998). Furthermore, each system was assumed to serve 1.8 people, equivalent to a total of 30% of the catchment population. Expected loads included no discharge from pump- out systems as any loads to the catchments from a properly operated pump-out should be included in the consideration of STPs. It is also assumed that the failures that are widely reported for decentralised systems (Charles et al., 2001) were accounted for by the random sampling methodology employed. The worst case scenario included poorly managed AWTS and alternative systems, and pump out tanks that were illegally discharged to the environment. For nutrients, these systems were considered to display equivalent performance to acceptably functioning septic tanks, whereas for pathogens the worst-case performance was considered to be equal for all systems.

The expected concentration of Cryptosporidium was taken as 230 oocysts per litre and the worst case as 510,000 oocysts per litre. The expected concentration of viruses was taken as 105 per litre, corresponding to that used for sewage above, and the worst case as 107 from Canter & Knox (1985). No losses for soil sorption/inactivation were included.

Centralised Versus Decentralised Systems Overall expected and worst case estimates of nutrient and pathogen loadings from centralised and decentralised systems within the Sydney drinking water catchments are provided in Table 5. It is important to note that the Cryptosporidium method used did not consider viability status.

Page 154

Table 5 Estimated total annual loads of nutrients and pathogens from sewage systems within Sydney’s drinking water catchments

System Flows Scenario TN TP Cryptosporidiu Enteric m viruses ML.y- kg/y Kg/y oocysts/y viruses/y 1 Centralised 7632 Expected 136,740 37,090 1.0 x 1011 9.1 x 1013 Worst 175,930 44,580 2.4 x 1011 5.3 x case 1014 Decentralise 1797 Expected 237,610 36,040 3.1 x 1010 8.4 x d 1013 Worst 620,620 82,040 7.3 x 1013 9.0 x case 1015

Using first pass assumptions, we have compared loadings likely to be released from centralised and decentralised sewage systems. Generally the expected loads to the catchments from decentralised systems were equal to those from centralised systems, excepting TN, while the worst-case scenario loads were greater from decentralised than from centralised systems in all cases. In assessing the implications for catchment management there are a number of factors that need to be considered in the interpretation of these loads.

Firstly, there are limitations to the data. The loads of pathogens from centralised and decentralised systems alike depend on the prevalence of pathogens in the community. The limited duration of the sampling programs may have underestimated the impact of seasonal variability on Cryptosporidium oocyst concentrations and occurrence. For example Medema et al. (2001) reported concentrations of Cryptosporidium of 17 and 250 oocysts per litre (geometric mean) for effluent from two centralised STPs, with occurrence rates of 80% and 100% and troughs experienced in August/September.

Secondly, the results represent the loads to the catchment as a whole and do not consider the impact of these systems on local water quality, and thus, do not account for contaminant loss or inactivation during overland transport nor pathogen viability. The literature (CH2MHill, 2001) indicated that centralised treatment systems generally provide greater attenuation of pathogens than decentralised treatment systems, however, proper disposal of decentralised treatment system effluent to soil should greatly enhance

Page 155

pathogen removal (outperforming centralised systems with chlorine, UV or ozone disinfection). In this case study, ten of the twelve centralised STPs disposed of effluent directly to surface water, and hence there is no mitigation of the contaminant loads in 67% of the flow. The remaining two disposed of effluent via irrigation providing an avenue for further treatment and load reduction through nutrient-uptake and sorption, and pathogen sorption, desiccation and UV-inactivation. Similarly, well designed and managed decentralised systems, that are situated appropriately, based on soil type and topography, can provide this additional treatment.

Appropriate design and maintenance is critical to the proper operation of both centralised and decentralised sewage systems. In centralised systems exfiltration and sewer overflows will result in increased loads of raw sewage on the catchment and potentially to catchment water quality, however centralised management mitigates these risks to a certain extent. Similarly, in decentralised systems poor design, including inappropriate siting, and poor maintenance can lead to direct impacts on water quality. Centralised management of decentralised systems would mitigate the risks associated with poor performance, reducing the uncertainty evident in the range between the loads for the expected and worst case scenarios.

The issue of mitigation by soil treatment of decentralised system effluent is critical to the quantification of water quality impacts. Research in this area is continuing in the Sydney drinking water catchments.

Other land uses within the catchment also require consideration to appropriately use this data to manage the catchments and prioritise research. For example, the major land uses in Sydney’s catchments: bushland and unimproved grazing, have nitrogen generation rates of 1.3 and 5.3 kg.ha-1.y-1 respectively, and phosphorus generation rates of 0.1 and 1.5 kg.ha-1.y-1 respectively (Swanson, 2001). While sewered and unsewered areas have higher nutrient release rates (nitrogen 7.5 and 12.5; phosphorus 1.2 and 1.8 kg.ha-1.y-1 , respectively), the relatively small areas of the catchment with these landuse types, reduces their relative impact.

This scope of this research has been limited to the loads from operational sewage systems and hence, sustainability in terms of life cycle assessment including chemical inputs, construction and longevity is not considered here.

Page 156

Conclusions Based on available data, the expected scenario in the Sydney drinking water catchments is that decentralised systems have similar total loads to centralised systems for phosphorus (36,040 kilograms per year for decentralised systems versus 37,090 for centralised systems), Cryptosporidium oocysts (3.1 x 1010 per year for decentralised systems versus 1011 for centralised systems) and enteric viruses (8.4 x 1013 per year for decentralised systems versus 9.1 x 1013 for centralised systems), but higher loads for nitrogen and in the worst-case scenario. The implication of these results for catchment water quality are affected by the disposal methods and failure issues. Decentralised system disposal to land may afford a degree of mitigation that can be enhanced, if the degree of failure is reduced. Reducing failure rates is reliant upon adequate management, maintenance and design, based on soil type and topography. The high loads associated with the worst-case scenario indicate the impacts of decentralised systems failing due to the simplistic nature of the treatment system and poor management.

Acknowledgements The authors would like to gratefully acknowledge the work of Ecowise Environmental Pty Ltd (in particular Declan Page), Australian Water Technologies Pty Ltd and the Australian Water Quality Centre in the collection and analysis of samples. This work is funded by the Sydney Catchment Authority as part of their on-site sewage system research.

References APHA (1995). Standard Methods for the Examination of Water and Wastewater. Washington D.C., American Public Health Association.

Ashbolt N.J., Grabow W.O.K. and Snozzi M. (2001). Indicators of microbial water quality. In Fewtrell, L. and Bartram, J. (ed.), Water Quality: Guidelines, Standards and Health. Risk assessment and management for water-related infectious disease. IWA Publishing, London pp:289-316.

Ashbolt N., Roser D., Leeming R., Stevens M., Laver P., Magyar C., Wade A., Grooby W. and Steffensen D. (2002). Dry weather quality of protected versus developed surface

Page 157

water catchments - pathogen data and management. Oral presentation at the 3rd World Water Congress, Melbourne, Australia, International Water Association, London (CD- ROM).

Canter L.W. and Knox R.C. (1985). Septic tank system effects on ground water quality. Lewis Publishers, Chelsea, Mich..

CH2M HILL (2001). Review of Catchment Sewerage Needs. Prepared for Sydney Catchment Authority.CH2M HILL Australia.

Charles K., Ashbolt N., Roser D., Deere D., and McGuinness R. (2001) Australasian Standards for on-site sewage management: Implications for nutrient and pathogen pollution in the Sydney drinking water catchments. Water (Aust.) 28(8):58-64. 2001.

Department of Local Government (1998). Environment and Health Protection Guidelines: On-site Sewage Management for Single Households. Department of Local Government, NSW.

Deborde D. C., Woessner W. W., Lauerman B. and Ball P. N. (1998). Virus occurrence and transport in a school septic system and unconfined aquifer. Ground Water 36 (5), 825-834.

Gerba C. (2000a). Domestic wastes and waste treatment. In Maier, R. M., Pepper, I. L. and Gerba, C. P. (ed.), Environ. Microbiol. Academic Press, San Diego, p. 505-534.

Gerba C. (2000b). Assessment of Enteric Pathogen Shedding by Bathers during Recreational Activity and its Impact on Water Quality. Quantitative Microbiology 2, 55-68, 2000.

Griffith M., Cox P., Ault G., Warnecke M., Angles M., Prassad R., Logan M., 2002. Spatial Variation of Pathogens within Sydney’s Water Supply Catchments. Report prepared by AWT PL for SCA.

Lodder W. J., Vinj J., van de Heide R., de Roda Husman A. M., Leenen E. J. T. M. and Koopmans M. P. G. (1999). Molecular detection of Norwalk-like caliciviruses in sewage. Appl. Environ. Microbiol. 65, 5624-5627.

Page 158

Long J. and Ashbolt N. J. (1994). Microbiological Quality of Sewage Treatment Plant Effluents. AWT Science & Environment report number 94/123. Sydney Water Corporation, Sydney.

McClellan, P. (1998). Sydney Water Inquiry Fifth Report, Final Report Volume 2. Sydney: Premier's Department, NSW Government. http://www.premiers.nsw.gov.au/our_library/archives/sydwater/5threp/r5toc.htm

Medema G. J., Ketelaars H.A.M. and Hoogenboezem W. (eds) (2001). Cryptosporidium and Giardia: occurrence in sewage, manure and surface water. Association of River Waterworks – RIWA, the Netherlands.

Swanson P. (2001). Contaminant Budgeting – Estimating the likely water quality effects of existing and proposed land-uses. Draft Report. Prepared for the Sydney Catchment Authority. Australian Water Technologies, Sydney.

US-EPA. (1999). Method 1623 - Cryptosporidium and Giardia in Water by Filtration/IMS/IFA EPA-821-R99-006. Office of Water, United States Environment Protection Agency.

Vantarakis A. and Papapetropoulou M. (1999). Detection of enteroviruses, adenoviruses and hepatitis A viruses in raw sewage and treated effluents by nested-PCR. Wat. Air Soil Pollut. 114 (1-2), 85-93.

Page 159

Paper V Effluent quality from 200 on-site sewage systems: Design values for guidelines

Charles, K. J., N. J. Ashbolt, D. J. Roser, R. McGuinness and D. A. Deere (2005). "Effluent quality from 200 on-site sewage systems: design values for guidelines." Water Science & Technology 51(10): 163-169.

Author Contribution Role (all authors helped with reviewing and (%) editing) Charles, K 70 Research, analysis and writing. Monitoring plan development. Ashbolt, N 5 Supervisor. Monitoring plan development. Roser, D 10 Co-supervisor, Paper development. Monitoring plan development. McGuinness, R. 5 Industry advisor. Monitoring plan development. Deere, D 10 Industry advisor. Monitoring plan development. Other Monitoring was undertaken by consultants AWT and Ecowise

Page 160

Abstract The quality of effluent from an on-site sewage treatment system is a critical factor in designing the disposal area, and hence, ensuring the sustained performance of the system. Contaminant concentrations in effluent are typically specified in regulatory guidelines or standards; however the accuracy of these guideline values are brought into question due to the poor performance of septic tanks and high failure rates of disposal systems reported here and elsewhere. Results from studies of septic tank effluent quality indicated that the effluent is of poorer quality than currently suggested by guidelines. Aerated wastewater treatment systems were found to perform to accreditation guidelines; however insufficient nutrient data is presently available to assess nutrient loads. It is proposed that the 80th percentile of system performance be adopted as the design value for sizing effluent disposal areas to minimise failure associated with overloading. For septic tanks this equates to 660 mg.L-1 SS, 330 mg.L-1 BOD, 250 mg.L-1 TN and 36 mg.L-1 TP.

Introduction On-site sewage system performance is a function of the type of treatment and disposal system, with effluent quality being one of the primary factors in disposal system design and performance. Effluent disposal system design is typically based on hydraulic load and treatment level, as defined by Suspended Solids (SS), Biochemical Oxygen Demand (BOD), thermotolerant coliforms and nutrient loads (Total Phosphorus, TP, and Total Nitrogen, TN). In NSW, effluent must be treated to a secondary level (BOD <20 and SS <30 mg.L-1) and disinfected prior to surface disposal (NSW Health, 1998).

The hydraulic load for a system is derived from the daily flow guidelines based on household population and water saving devices. Nutrient loads are a function of guideline effluent quality values for different systems and the hydraulic load. These effluent quality values are based on the ‘expected’ concentrations in effluent. Underestimation of the contaminant loads and concentrations used for designing the effluent disposal area can potentially lead to the failure of the disposal system. High levels of SS and BOD may affect the infiltration capacity in subsurface disposal areas through the build up of solids resulting in hydraulic failure and effluent surfacing, as well as causing odour problems in surface disposal areas. Nitrogen and phosphorus levels in excess of design values may impact on vegetation causing native vegetation

Page 161

stress and promoting weed growth.

Values of effluent quality are typically taken from regulatory guidelines or standards, however as few effluent quality studies have been published, the information on which to base guidelines is limited. Effluent quality is a function of treatment system design, maintenance and the composition of the influent. Quality will vary over time due to diurnal fluctuations in household activity, time since last pump-out or maintenance visit, advances in on-site treatment technology and the changing nature of household chemical/detergent use. The current NSW guidelines (NSW Department of Local Government, 1998; NSW Health, 1998) differ from reported effluent quality from published research for both septic tanks (Table 1) and secondary treatment systems or aerated wastewater treatment systems (AWTS) (Table 2). In particular, several studies report poorer septic tank effluent quality than the NSW guidelines, the implication of which is an increase in the likelihood of disposal system failure. Similarly, studies of AWTS have reported high levels of non-compliance with the criteria for surface disposal.

For sustainable on-site sewage disposal, appropriate design values for effluent quality are required. This may require updating existing guidelines and approaches. For example, designs based on the average loads will potentially result in failure in approximately 50 % of systems due to underestimation of the solids and nutrient loading. Hence for design purposes, the average and/or median are not considered to be the most appropriate numbers. To increase the sustainability of on-site sewage disposal areas it may be better to adopt the 80th percentile value, as recommended in the ANZECC (2001) Water Quality Guidelines for trigger values. This design value for septic effluent should aid in reducing disposal system failure associated with contaminant overloading.

Page 162

Table 1 Performance and guideline values for septic tanks

Value SS BOD TN TP Thermotolerant Guideline coliforms s (mg.L- (mg.L-1) (mg.L- (mg.L- (cfu.100mL-1) 1) 1) 1) International Average a 94 44** 8.6 3.7 x 104 USA Range a 46-156 19- 7-17 4 x 103–3 x 105 USA 53** Range b 50-100 140-200 40-100 5-15 106 – 108 USA Average c 165 280 92* 10.5 1.6 x 105 85th 250 350 105* 14 2.6 x 106 percentile c Average d 54 158 55 15 105 Range d 11-695 20-480 10-125 4-90 5 - 107 Range e 44-54 129-147 41-49 12-14 3 x 105–1 x 106 Australian Expected f 50 150 50-60 10 -15 105 – 107 NSW Average g 32 175 Average h 72* 16 Range i 17- 22-2133 106 – 108 6970 Range j 40-250 120-280 30-60 10-20 2 x 103–3 x 106 Average k 448.5 365.7 75.7 21.3 * Ammonia; ** TKN; a Anderson et al. (1994); b USEPA (2002); c Viraraghavan (1976); d Otis and Boyle (1976); e USEPA (1978) (cited in Brouwer (1983)); f NSW Department of Local Government (1998); g Bruty and Mann (1977) (cited in Brouwer (1983)); h Victoria Department of Health (1976) (cited in Brouwer (1983)); i Khalife and Dharmappa (1996); j Martens (1996); k Sydney Water (1999)

This paper reports the results of sampling and analyses of sufficient numbers of samples for 80th percentile estimates from septic tank effluents and AWTS in NSW. The results are compared to the values currently used in guidelines and reported in literature.

Page 163

Table 2 Performance and guidelines for AWTS

Value SS BOD TN TP Thermotoleran Free t Coliforms Residual Chlorine (mg.L- (mg.L- (mg.L- (mg.L- (cfu.100mL-1) (mg.L-1) 1) 1) 1) 1) Guidelines Average a 30 20 30 Maximum a 45 30 100 0.2-2.0 Average b 30 20 10 Maximum b 45 30 20 (in 80%) >0.5* Study Results Average c 45 39 6.3 x 104 0.5 80th percentile 56 61.5 2.6 x 104 0.6 c Average d 11 3 19 Range d 1-41 1-9 1-160 Range e 5-100 5-50 25-60 4-10 103 – 104 Average f 53 47 38 35 104 Range f 4-252 0-208 16-78 7-140 1-105 * Total Chlorine, a NSW Health (1998); b AS/NZS 1547:2000; c Coote (1995); d Marzella et al. (1995); e USEPA (2002); f Otis and Boyle (1976)

Methods Different sampling projects have been undertaken by the Sydney Catchment Authority (SCA) relating to on-site sewage treatment and disposal systems. Data from these projects has been consolidated for the purposes of characterising wastewater from on- site sewage treatment systems to aid in design and planning. Systems assessed included:

• AWTS with chlorine or ultraviolet disinfection (models approved by NSW Health);

• Septic tanks, more than 10 years old, serving single households; and

• Septic tanks, more than 10 years old, serving community facilities.

In total, 48 septic tanks and 143 AWTS were sampled over six separate projects. Septic tanks were sampled from the exit T-junction of the tank unless stated otherwise. AWTS were sampled immediately prior to disinfection (for BOD and SS) and from the irrigation system (for thermotolerant coliforms and free chlorine) in accordance with the NSW Health (1998) sampling criteria. All analyses were by Standard Methods (APHA,

Page 164

1995) by a NATA-approved contract laboratory.

Effluent quality sampling project details Grab samples of septic tanks and AWTS. Grab samples were taken from a total of 46 systems, including 35 residential septic tanks, nine communal septic tanks and two AWTS (unpublished data, SCA). The ages of the systems ranged from ten years to greater than fifty years with an average of 29. One grab sample was taken from each system and analysed for a broad range of constituents to enable characterisation of effluents. No selection criteria other than householder approval were used.

Routine monitoring of a communal septic tank. A baffled septic tank servicing four houses, an office and facilities at two picnic areas was sampled fortnightly for twelve months (24 samples) (unpublished data, SCA). Effluent was sampled from a storage reservoir located after the septic tank and a pump well. The system was one year old when sampling commenced. Over the sampling period the hydraulic load was consistently below the design capacity.

Routine monitoring of two residential septic tanks. Two septic tanks were sampled over one year (unpublished data, SCA). Both systems were 10-20 years old and served households of four people. System A had 10 samples. System B had 6 samples. Both septic tanks required maintenance, however the results post-pumpout were not available at the time of this paper.

AWTS disinfection study. Grab samples were taken from the clarification chamber and after disinfection from a total of 21 AWTS (Charles et al., 2003) in accordance with NSW Health (1998) guidelines. Ten systems had chlorine tablet disinfection and the remaining eleven had ultraviolet disinfection. One system of each type of disinfection was sampled over 4 days. All systems were aged between 6 months and 2 years and had undergone maintenance in the previous 3 months in accordance with the guidelines and appropriate to manufacturers specifications.

Short-term routine monitoring of two residential systems. One septic tank and one AWTS were sampled over 6 weeks (Australian Water Technologies, 2002). Both systems were over 10 years old. The AWTS was sampled 12 times from the clarification chamber. The AWTS was not maintained and contained no chlorine tablets

Page 165

in the disinfection system. Samples were taken from the distribution box as direct sampling access to the septic tank was not available. Effluent was only present on one occasion.

NSW Health AWTS Anniversary testing. De-identified data on the performance of 120 AWTS from anniversary testing (NSW Health, 1998) was supplied by NSW Health.

Statistical Methods Grab sample and average time-series data was combined into a single database for analysis. Outliers were assessed, and case summaries were undertaken in SPSS (V11.5.2.1, SPSS Inc.). Percentiles and the goodness of fit for lognormal distributions were calculated in @Risk (V4.5, Palisade Corp.) add-on to ExcelTM. Extreme outliers were excluded from case summaries and lognormal distributions but included in percentile calculations.

Results and Discussion Similarly to the effluent quality samples previously reported in literature (Table 1), the results indicate that effluent from septic tanks is of much poorer quality than represented guideline and literature values (Table 3). Furthermore, the results indicate that the current guidelines in NSW only represent the top 13% of systems for SS and the top 33% for BOD. While these values do not directly relate to disposal system design they do indicate problems with system performance that have the potential to impact on disposal system performance. For nutrients, the top performing 23% and 40% of systems for TN and TP respectively are addressed. The implications are that up to 87% of disposal system failure may be related to overloading of solids, with up to 77% inadequately designed for nutrient uptake. Hence, system failure should not be a surprise, even with moderate maintenance, but will also depend on hydraulic loads, as discussed below.

Residential septic tanks had higher averages for all analytes compared to communal systems, however, the system exhibiting the worst effluent quality was a communal system. Results for the two types of systems were not found to be significantly different and were pooled for subsequent analysis.

Page 166

Table 3 Septic tank effluent quality

SS BOD TN TP Thermotolerant coliforms (mg.L-1) (mg.L-1) (mg.L-1) (mg.L-1) (cfu.100mL-1) Guideline values* 50 150 50-60 10 – 15 105 – 107 Equivalent 13% 33% 15-23% 25-40% 32-93% percentile Average 379 224 160 21 106 20th percentile 56 74 59 9.2 104 80th percentile 660 330 250 36 106 Number of 43 43 45 46 38 systems * NSW Department of Local Government, 1998 (Expected quality)

Overall it is contaminant loads, rather than the concentrations, that will increase the potential for disposal system failure, such that, if hydraulic flows are appropriately conservative they may balance out the underestimation of concentrations. Limited hydraulic flow data was available from these studies to enable comparison of contaminant loads with those predicted by the guidelines, however, flow data was available (Table 4) from one residential septic tank serving a house of four people. Nitrogen, which was below the average concentration found in septic tanks, resulted in loads in excess of those predicted by the NSW guidelines (NSW Department of Local Government, 1998). Phosphorus, which was approximately equal to the average concentration found in septic tanks, resulted in loads in excess of those predicted by the Australian Standard (AS/NZS 1547:2000) but within the range of those predicted by the NSW guidelines. Hence, use of the current NSW guidelines for designing effluent disposal areas are likely to result in under predictions of the loads as well as concentrations of nutrients in approximately 50% of systems.

Page 167

Table 4 Comparison of actual septic tank effluent nutrient loadings with guideline predictions

System Average flow Concentrations Site loads TN TP TN TP (L.d-1) (mg.L- (mg.L- (g.d-1) (g.d-1) 1) 1) Residential septic 526 (132 L per person) 123 22 65 12 results Guideline predictions 720 (180 L per person)* 50-60 10-15 36-43 7-11 900 (300 L per 50-60 10-15 45-54 9-14 bedroom)** * AS/NZS 1547:2000; ** NSW Department of Local Government (1998)

Thermotolerant coliforms were the parameter closest to guideline limits for septic tanks. The expected thermotolerant coliform concentration affects effluent disposal design through the regulations for subsurface and surface disposal. If the aim of including the thermotolerant coliform concentration is to indicate public health impacts from pathogens potentially in the effluent, then it is recommended that the high level of thermotolerant coliforms be retained. This is due to the higher inactivation rates of coliforms than pathogens, as well as the greater potential movement of pathogens, particularly enteric viruses, as they are smaller than coliforms. It is not recommended that thermotolerant coliforms are used as the sole indicator of public health risk, but that consideration of a range of index microorganisms (Ashbolt et al., 2001) such as coliphages and Clostridium perfringens be included.

AWTS performance (Table 5) was found to be comparable to the NSW Health Accreditation requirements (NSW Health, 1998). The performance of these systems is much improved compared to the study undertaken by Coote (1995) which reported only 5% level of compliance. From the present study, the equivalent level of compliance is 42%, but with the changes in guidelines to account for grab sampling the actual compliance is 49%. Free chlorine concentrations alone account for 22% of non- compliance.

Page 168

Table 5 AWTS effluent quality

Reference SS BOD Thermotolerant Free coliforms Chlorine (mg.L-1) (mg.L-1) (cfu.100mL-1) (mg.L-1) Guideline values* 30 (45)** 20 (30) 10 (100) 0.2-2.0 Equivalent percentiles 85% (94) 85% (89) 67% (76) 54-91% Average 18 11 6 200 0.5 20th percentile 5 2 1 0.1 80th percentile 27 15 162 0.5 Number of systems 141 140 140 119 * NSW Health, 1998; ** Upper limit for grab samples

If the 80th percentile value for effluent quality was selected it would result in large increases in the design loads as illustrated in Table 5. Nitrogen would increase approximately five-fold from 56-60 mg.L-1 to 250 mg.L-1. Phosphorus would increase from 10-15 mg.L-1 to 36 mg.L-1. Effluent disposal areas would have a corresponding increase which would aid in reducing disposal system failure attributable to contaminant overloading to approximately 20%, with further reductions in hydraulic load failure provided by conservative design flows. Similarly to these design value for septic tanks, it is recommended that, due to the high values of nutrients experienced in septic tanks, the design values for nutrients in AWTS effluent be reviewed. Lognormal distributions of parameters in septic tank effluent were fitted (Table 6) to provide additional information on the quality of effluent for system design and management.

Table 6 Lognormal distribution parameters and goodness-of-fit for septic tank effluent

Parameter SS TP TN Log10 thermotolerant coliforms* (mg.L-1) (mg.L-1) (mg.L-1) (log10 cfu.100mL-1) µ; σ 411; 884 21.8; 18.5 163;159 5.2; 1.2 χ2 3.3 1.3 1.0 8.7 * Normal distribution

Four systems were extreme outliers that were excluded from the statistical analysis. Two septic tanks were excluded due to high SS (5,800 and 29,000 mg.L-1), the latter of which, a communal system, was also an outlier for TP (460 mg.L-1), TN (1000 mg.L-1) and BOD (1400 mg.L-1). Two AWTS were excluded, one due to high SS (726 mg.L-1) and BOD (160 mg.L-1) and one due to high free chlorine (5 mg.L-1).

Page 169

The primary limitations of this research have been the lack of information on newer septic tanks with baffles, and data on nutrients in AWTS. To address this, an additional 50 septic tanks (baffled, 2-5 years old, serving single households) and 50 AWTS (selection equivalent to NSW Health anniversary testing) are proposed to be tested. It is anticipated that the performance of baffled septic tanks will be superior to those tested to date, and hence, the recommended 80th percentile values will decrease. Based on the data collected during these studies, the nutrient concentrations of AWTS are expected to be higher than the current guidelines, however additional data is required.

Conclusions The high rates of failure of effluent disposal areas (Charles et al., 2001) can potentially be attributed to the under estimation of contaminant concentrations and loads from on- site sewage treatment systems. Septic tank effluent quality was found to be poor compared to guidelines and current literature. While AWTS performance is comparable to the NSW Health Accreditation requirements, it is anticipated that due to the high nutrient concentrations found in septic tanks, high nutrient loads will also be present in AWTS. It would appear that a design guideline adopting the 80th percentile would respond to issues raised in this paper. Relevant design values for sizing a disposal area receiving septic tank effluent would be 660 mg.L-1 SS, 330 mg.L-1 BOD, 250 mg.L-1 TN and 36 mg.L-1 TP. Further analysis with additional sampling will offer greater confidence in these revised values. Additional sampling is being undertaken to address the limitations of this research, including sampling of baffled septic tanks and nutrient concentrations in AWTS effluent.

Acknowledgements The authors would like to acknowledge the work Danielle Baker and Ecowise Environmental, Christobel Ferguson, Mark Angles and Declan Page, and thank Neil Shaw and the NSW Department of Health for providing data on AWTS performance.

References Anderson, D. L., R. J. Otis, J. I. McNeillie and R. A. Apfel (1994). In-situ lysimeter investigation of pollutant attenuation in the vadose zone of a fine sand. In: On-Site Wastewater Treatment: Proceedings of the Seventh International Symposium on Individual and Small Community Sewage Systems, American Society of Agricultural

Page 170

Engineers, St. Joseph, MI, USA.

ANZECC (2001). Australian Water Quality Guidelines for Marine and Freshwaters, Australian and New Zealand Environment Conservation Council, Canberra, Australia.

AS/NZS 1547:2000 On-site domestic-wastewater management. (2000). Standards Australia/ Standards New Zealand, Homebush.

Ashbolt, N. J., W. O. K. Grabow, and M. Snozzi. 2001. Indicators of microbial water quality, p. 289-315. In: Water Quality: Guidelines, Standards and Health. Risk assessment and management for water-related infectious disease, Chapter 13, L. Fewtrell and J. Bartram (ed.), IWA Publishing, London.

Australian Water Technologies (2002). On-site Sewage Management System Pilot Investigation, Report prepared for Sydney Catchment Authority, Australian Water Technologies, Sydney, Australia.

Brouwer, J. (1983). Land Capability for Septic Tank Effluent Absorption Fields, Australian Water Resources Council Technical Paper No. 80, Australian Government Publishing Service, Canberra.

Charles, K., N. Ashbolt, D. Roser, D. Deere and R. McGuinness (2001). Australasian standards for on-site sewage management: Implications for nutrient and pathogen pollution in the Sydney drinking water catchments. Water (Australia) 28 (8), 58-64.

Charles, K. J., N. J. Ashbolt, D. A. Deere and D. J. Roser (2003). Disinfection in Aerated Wastewater Treatment Systems. In: Ozwater: Innovation in Water, AWA 20th Convention, 6-10 April 2003, Perth.

Coote, B. (1995). Aerated Septic Systems for Camden Council, AWT Report 95/194. Sydney Water Corp., Sydney.

Khalife, M. A. and H. B. Dharmappa (1996). Aerated septic tank systems: Field survey performance. Water (Aust.) September/October, 25-28.

Martens, D. M. (1996). Runoff in Sydney's unsewered urban areas. PhD thesis, Department of Geography, University of Sydney.

Page 171

Marzella, M., B. Coote and P. Swanson (1995). Performance evaluation of on-site wastewater treatment in single households. Australian Water Technologies, Sydney.

NSW Department of Local Government (1998). Environment and Health Protection Guidelines: On-site Sewage Management for Single Households. Department of Local Government, NSW.

NSW Health (1998). Aerated Wastewater Treatment Systems (AWTS) Accreditation Guideline, September 1998, NSW Health Department, Sydney.

Otis, R.J. and W.C. Boyle (1976). Performance of single household treatment units. J. Environ. Eng. 102, 175-189.

Standard Methods for the Examination of Water and Wastewater (1995). 19th edn, American Public Health Association/American Water Works Association/Water Environment Federation, Washington DC, USA.

Sydney Water (1999). The Oaks, Oakdale and Belimbla Park Sewerage Scheme EIS. Sydney Water, Sydney, Australia.

US-EPA (2002). On-site Wastewater Treatment Systems Manual, US EPA, Washington.

Viraraghavan, T. (1976). Septic tank efficiency. J. Environ. Eng. 102, 505-508.

Page 172

Paper VI Disinfection Performance in Aerated Wastewater Treatment Systems

Charles, K. J., N. J. Ashbolt, D. Deere and D. Roser. (2003). Disinfection in Aerated Wastewater Treatment Systems. Ozwater: Innovation in Water, AWA 20th Convention, 6-10 April 2003 Perth, Western Australia, AWA.

Author Contribution Role (all authors helped with reviewing and (%) editing) Charles, K 70 Research, analysis and writing. Monitoring plan development. Ashbolt, N 10 Supervisor. Monitoring plan development. Deere, D 10 Industry advisor. Monitoring plan development. Roser, D 10 Co-supervisor, Paper development. Monitoring plan development. Other Monitoring was undertaken by consultants Ecowise

Page 173

Executive Summary Chlorination in aerated wastewater treatment systems (AWTS) has been reported to be unreliable, and to be ineffective in the inactivation of protozoa and have limited effect on enteric viruses. It may also produce byproducts, halogenated organics, which are harmful to humans and the environment. UV disinfection has gained popularity in municipal sewage treatment plants, in particular where the effluent is reused. It is more effective than chlorination in the inactivation of pathogens, however, the effectiveness may be limited by the turbidity and concentration of suspended solids in the effluent.

This paper reports the results of a pilot study undertaken to measure the removal efficiency of chlorine and UV disinfection systems with respect to bacteria and to predict the removal efficiencies for human enteric virus (based on somatic and MS-2 coliphage results) and Cryptosporidium oocysts (based on Clostridium perfringens spore results).

Two of 22 AWTS inspected were discovered to have a failure in the treatment system. One system had a mechanical failure in disinfection system. All systems tested satisfied the NSW Health requirements for BOD and SS. Chlorination achieved, on average, a

3.6 log10 reduction of thermotolerant coliforms and 3.7 log10 reduction of E. coli, however, it achieved a much lower reduction of other faecal indicators. AWTS with UV achieved a lower log10 reduction of thermotolerant coliforms, with 3 systems failing to meet the NSW Health criteria of 100 cfu.100mL-1. The reduction of enterococci and Clostridium perfringens was higher than that for chlorine disinfection. UV disinfection produced a greater reduction in MS2 (1.8 log10) than chlorine (1.0 log10).

Chlorine residuals were within acceptable limits (0.2-0.6 mg.L-1). Disinfection by- product formation was high with a mean increase of total trihalomethanes of over 28 mg.L-1 during chlorination.

Overall, the ultraviolet disinfection system displayed greater reduction of surrogates of viruses (MS2) and Cryptosporidium oocysts (C. perfringens spores) than measured with chlorine disinfection, and a lower potential toxicity.

Introduction Aerated wastewater treatment systems (AWTS) provide secondary treatment and

Page 174

disinfection to household wastewater prior to disposal on-site, commonly by surface spray irrigation. The most common configuration is activated sludge treatment and chlorine disinfection. Ultraviolet (UV) disinfection is also available. In NSW, these systems undergo accreditation by NSW Health to ensure that they can achieve suitable final effluent quality (Table 1). However, long-term performance does not always meet these criteria. Coote (1995) reported 95% of AWTS failing to conform to the performance criteria, and while management of these systems has improved since that report was released, performance failure is still common.

Table 1 Criteria for Surface Spray Irrigation of Effluent

Parameter NSW Australian & New Zealand Guidelines1 Standard2 Median Maximum Biochemical oxygen demand (mg.L-1) <20 ≤ 20 30 Suspended Solids (mg.L-1) <30 ≤ 30 45 Thermotolerant coliforms (cfu.100 mL-1) <30 ≤ 10 20 (in 4 of 5 samples) Free residual chlorine (mg.L-1) ≥ 0.2, ≤ 2 Total chlorine (mg.L-1) ≥ 0.5 1 NSW Department of Local Government (1998) 1 Standards Australia/Standards New Zealand (2000)

In the US (US-EPA, 2002) similar problems are being encountered with thermotolerant coliform concentrations of over 200 cfu per 100 mL in 93% of samples from similar aerated treatment systems with tablet chlorinators. Sixty eight percent of systems had no chlorine residual. Problems were cited as solids accumulation in the chlorinator, tablet caking, failure of the tablet to drop into the sleeve, and failure to maintain the tablet supply. No similar studies of UV performance associated with AWTS have been reported.

It is well known that coliform bacteria are more vulnerable to chlorine and many other disinfectants than enteric viruses and parasitic protozoa (Sobsey, 1989; Tyrrell et al., 1995; Tree et al., 1997; US-EPA, 1999). Interestingly, despite the very high resistance of Cryptosporidium to chlorination, it is readily inactivated by UV (Finch and Belosevic, 2001). Nonetheless, performance testing of AWTS is generally restricted to the regulatory criteria (Table 1), with inactivation of viral and protozoan pathogens, if

Page 175

considered at all, estimated from literature values. An additional concern with AWTS located in drinking water catchments is that of non-specific reactions between the disinfectant and natural organic matter (Singer, 1993). Some of these disinfection byproducts are potentially harmful to human and the receiving water ecology (Gray et al., 2001), and specific byproducts are now regulated in drinking water (Table 2).

UV disinfection has also been shown to produce disinfection byproducts, including acetaldehyde, formaldehyde, glyoxal and m-glyoxal, but only at dosages of greater than 900 (Linden et al., 1998). The recommended irradiation rate for AWTS is 60 mW.s.cm-2 (USEPA, 2002).

Table 2 Disinfection by-product regulations (mg.L-1)

Disinfection By- US Drinking Australian Drinking WHO Guidelines for product Water Rule?1 Water Guidelines2 drinking water quality3 Trihalomethanes 0.1 0.25 Chloroform 0.003 Chloroacetic Acid 1 Dichloroacetic Acid 0.05 Trichloroacetic Acid 0.1 Chloral hydrate 0.02 1 US-EPA (1998) 2 NH&MRC (1994) 3 WHO (2002)

The drinking water supply catchments for 4 million customers in the Sydney region contain 18,465 on-site sewage systems (Charles et al., 2001). The number of AWTS is currently 21%, however this is growing rapidly, with AWTS accounting for the majority of on-site sewage development approvals in the past 4 years. This paper reports the findings of a pilot study undertaken by the Sydney Catchment Authority (SCA) to assess the benefits and detriments of chlorination of AWTS effluent, and to compare it with the potential benefits of UV disinfection.

Page 176

Methodology

Site Selection Manufacturers of AWTS with chlorine disinfection were approached through the AWTS Manufacturers Association. The manufacturers of AWTS with UV disinfection were approached directly. Ten AWTS of each type were selected by Ecowise Environmental Pty Ltd (ECOWISE) and SCA from a list of fifteen sites supplied by the manufacturers. This ensured that the chosen systems sampled were at least one year of age and properly maintained. The time since last maintenance was up to 3 months.

Sampling Sampling and analysis was undertaken by ECOWISE. Pre-disinfection grab samples were taken immediately prior to disinfection, and post-disinfection grab samples were taken from the irrigation hose (UV) or pump well (Chlorine). The UV light was triggered 5 minutes prior to sampling to allow adequate time for the lamp to warm. Field sheets recording site-specific information, including the conditions of the chlorinator or UV lamp, were prepared for each site and photos taken. Samples were analysed in the field for pH, electrical conductivity (EC), temperature and free chlorine (Free Cl2) for chlorine disinfection systems. Samples were analysed in the laboratory for suspended solids (SS), biochemical oxygen demand (BOD), total organic carbon (TOC), ammonia, Thermotolerant coliforms (TCs), Escherichia coli (E. coli), enterococci (ENT), Clostridium perfringens (CP), and somatic coliphages by APHA (1995) NATA approved methods at ECOWISE, except that the CP method (AS/NZS 4276.17.1:2000) was not NATA approved. Samples were sent to Australian Water Quality Centre for laboratory analysis of trihalomethanes (THM), chlorohydrate, and chloroacetic acid (NATA approved methods).

One AWTS with UV disinfection and one AWTS with chlorine disinfection was selected for the experimental phage spiking. MS-2 bacteriophage (ATCC 15597-B1) (800-900mL aliquot) to provide approximately 108 pfu.mL-1 suspended in buffer was added to the septic tank of each system. Sampling of effluent pre- and post-disinfection was undertaken to assess the reduction of faecal microorganisms in each disinfection system. Samples were analysed in the field for pH, EC, temperature and, for chlorine disinfection systems, free chlorine. Samples were sent to Australian Microbiological

Page 177

Services (AMS, Sydney) for NATA-approved laboratory analysis of the MS-2 coliphage by the method of Havelaar and Hogeboom (1984). At the time of MS2 application a small bottle of the same suspension was placed in the ground close to the AWTS tested, to act as an on-site control, providing information as to the longevity of the coliphage at that site. A sub-sample of this control and effluent samples from pre- and post- disinfection were taken at each sampling event, and were assayed on the day of sampling. All samples were tested on the day of sampling.

Results and discussion Of the 22 AWTS that were inspected as part of this study, one UV lamp required replacing; all chlorine disinfection systems had a suitable number of chlorine tablets present; and 2 systems, one of each type, had blockages causing internal overflow and flooding.

Pre-disinfection effluent quality in the two disinfection systems (Table 3, Table 4) were generally similar. BOD and SS were not significantly different (<0.05). The average concentrations of BOD and SS in pre-disinfected effluent from AWTS with chlorination were above the guideline limits (Table 3), specifically, 3 systems were above the SS guideline and 4 above the BOD guideline. However, NSW Health (1998) stipulates that grab sample concentrations for BOD and SS (taken pre-disinfection) should be less than 30 and 45 mg.L-1 respectively, in which case the concentrations in pre-disinfected effluent from AWTS are acceptable. AWTS with UV were below the guideline limits for SS and BOD, with one system above the guideline limits for SS and three for BOD. Turbidity and TOC were lower in AWTS with UV than with chlorination, where as ammonia and enterococci were higher. Somatic phage was only detected in one sample, a pre-disinfected effluent sample from an AWTS with chlorination.

Page 178

Table 3 Effluent Quality Pre-Disinfection

Disinfection Turbidity* SS TOC BOD Ammonia NTU mg.L-1 mg.L-1 mg.L-1 mg.L-1 Chlorine Mean 31.3 36.8 15.7 22.8 6.1 (N=14) Standard 2.1 44.8 5.5 20.2 19.0 deviation Maximum 33.0 180 26 84 72 Ultraviolet Mean 6.3 10.3 10.0 13.2 18.0 (N=13) Standard 2.5 10.7 4.2 15.8 21.7 deviation Maximum 9.7 40 19 59 64 * Spiked systems only, N=4. Table 4 Microbial Effluent Quality Pre-Disinfection

Disinfection TCs E.coli Enterococci* CP cfu.100mL-1 cfu.100mL-1 cfu.100mL-1 cfu.100mL-1 Chlorine Geometric 7640 6918 53 284 Mean (N=14) Log10 0.8 0.8 0.5 1.1 Standard deviation Maximum 180000 140000 250 16000 Ultraviolet Geometric 4808 2277 10086 170 Mean (N=13) Log10 1.2 1.7 0.5 1.0 Standard deviation Maximum 2500000 2500000 6000 25000 * Spiked systems only, N=4.

Physical and chemical post-disinfection effluent quality (Table 5) were generally similar to pre-disinfection effluent quality. Turbidity in AWTS with chlorination was higher post-disinfection than pre-disinfection.

The post-disinfection microbiological effluent quality (Table 6) from AWTS with chlorination was within guideline limits on all samples. Chlorination achieved, on average, a 3.6 log10 reduction (Table 7) of thermotolerant coliforms and 3.7 log10 reduction of E. coli, however, it achieved a much lower reduction of other faecal indicators. AWTS with UV achieved a lower log10 reduction of thermotolerant coliforms, with 3 systems failing to meet the NSW Health (1998) criteria of 100 cfu.100mL-1. The reduction of enterococci was greater than that for chlorine disinfection.

Page 179

Table 5 Effluent quality post-disinfection

Disinfection Turbidity* SS TOC BOD Ammonia NTU mg.L-1 mg.L-1 mg.L-1 mg.L-1 Chlorine Mean 36.3 21.3 24.0 10.9 6.1 (N=14) Standard 1.0 13.7 7.8 11.3 18.5 deviation Maximum 37.0 45.0 34.0 39.0 70.0 Ultraviolet Mean 8.3 8.2 10.3 8.0 17.1 (N=13) Standard 3.9 9.9 4.4 12.4 21.5 deviation Maximum 14.0 38.0 18.0 47.0 64.0 * Spiked systems only, N=4.

Table 6 Microbial effluent quality post-disinfection

Disinfection TCs E. coli Enterococci* CP cfu.100mL-1 cfu.100mL-1 cfu.100mL-1 cfu.100mL-1 Chlorine Geometric 1.9 1.5 1.4 91.9 Mean (N=14) Log10 Standard 0.4 0.3 0.1 1.4 deviation Maximum 17 10 1.5 11000 Ultraviolet Geometic 20.3 19.9 10.5 38.8 Mean (N=13) Log10 Standard 1.5 1.5 0.4 1.3 deviation Maximum 33000 25000 44 5000 * Spiked systems only, N=4.

The spiked systems were each sampled on four occasions. Each system demonstrated satisfactory effluent quality over the sampling period. The AWTS with chlorination has BOD and SS concentrations slightly in excess of the guideline limits but within range of the NSW Health limits. Both systems produced consistent effluent quality. The AWTS with UV spike was delayed by one day to allow the system to recover from being flooded due to a blockage. Turbidity and SS levels were slightly higher on the first day of sampling, within guideline limits, and no further detrimental impact on effluent quality was experienced.

MS2 concentrations of 106 and 105 pfu.mL-1 were recorded in the effluent pre- disinfection in the AWTS with chlorination and UV respectively. There was no significant difference in the microbial removals between the chlorine and UV disinfection systems. The MS2 reduction was comparable with that reported by

Page 180

Humphries et al. (1998) for enteroviruses and faecal streptococci.

Table 7 Mean and decimal reduction of faecal indicators by Cl2 and UV

Disinfectant TCs E.coli Enterococci CP MS2 Chlorine Mean reduction 99.99% 99.99% 98.46% 10.70% 89.9% Log10 reduction 3.6 3.7 1.6 0.5 1.0 UV Mean reduction 98.60% 98.90% 99.89% 43.42% 92.2% Log10 reduction 2.4 2.1 3.0 0.6 1.8

The variability of free chlorine residual in both pre- and post-disinfection effluent was high with the average change being a decrease of 0.4 mg.L-1, due in part to a pre- disinfection concentration of 6.0 in one system excluding which an increase of 0.03 was experienced. The variations in the spiked system were between 0.1 and 0.4 pre- disinfection and 0.2 and 0.6 post-disinfection which is within the limits of the guidelines.

Performance of UV disinfection can be limited by turbidity and suspended solids shielding the pathogens, and also by biofouling or the aggregation of organic material on the UV lamp sheath reducing the irradiation of the wastewater. The UV lamps in this study displayed a range of biofouling densities or thicknesses. No correlation was evident between visible biofouling and performance. The UV system with the highest SS had the lowest reduction in microorganisms for all AWTS.

Disinfection byproducts were only measured in the chlorinated effluents (Table 8), and showed a marked increase between the pre-disinfection and post-disinfection effluents. Additionally, the post-disinfection effluent concentrations were some two orders of magnitude above recommended guidelines for drinking water (Table 2). The environmental impacts of both chlorine residuals and disinfection byproducts have resulted in an increase in the frequency of dechlorination of effluent from municipal wastewater treatment plants (US-EPA, 2000).

Page 181

Table 8 Concentrations of disinfection byproducts pre- and post-chlorination

Total Chloroform Chloral hydrate Chloral hydrate Trihalomethanes Trihalomethanes Chloroacetic Acid Chloroacetic Dichloroacetic Acid Dichloroacetic Trichloroacetic Acid Trichloroacetic

Pre- Mean 1.9 0.6 2.5 0.6 1.4 0.6 disinfection Standard deviation 0.40 0.40 0.00 0.45 1.48 0.40 Post- Mean 30.6 27.4 7.7 21.9 26.1 4.6 disinfection Standard deviation 39.8 35.6 16.6 15.5 29.9 6.2

Conclusions Overall, there were no significant difference between the performance of the ultraviolet and chlorine disinfection systems for surrogates of viruses (MS2) and Cryptosporidium oocysts (C. perfringens spores). UV disinfection did not meet the guideline concentrations for effluent thermotolerant coliforms routinely during the study. The added consideration that UV disinfection is reported not to produce by-products at the levels required to achieve this performance indicates that it may be a safer alternative to chlorine disinfection in systems with spray irrigation. Neither disinfection system can remove the risk to the householder or to the environment as effectively as the treatment afforded by a well designed sub-surface effluent application area, however, each of these alternatives are subject to failure due to poor operation and maintenance.

References APHA (1995). Standard Methods for the Examination of Water and Wastewater. 19th ed. (Series Eds.: Eaton, A.D., L.S. Clesceri and A.E. Greenberg.) American Public Health Association, Washington D.C.

Beavers, P., I. Tully and A. Woolley (1999). Performance evaluation of On-site Aerated Wastewater Treatment Systems. In: Proceedings of On-site '99 Conference: Making on- site wastewater systems work; 13th-15th July 1999, University of New England. (Ed.: Patterson, R.A.) Lanfax Laboratories, Armidale, pp. 45-52.

Page 182

Charles, K., N Ashbolt, D. Roser, D. Deere and R. McGuinness (2001). Australasian standards for on-site sewage management: Implications for nutrient and pathogen pollution in the Sydney drinking water catchments. Water Journal of the Australian Water & Wastewater Association.

Coote, B. (1995). Aerated Septic Systems for Camden Council. AWT Report 95/194. Sydney Water Corp., Sydney.

Craik, S.A., D. Weldon, G. Finch, J. Bolton and M. Belosevic (2001). Inactivation of Cryptosporidium parvum oocysts using medium- and low- pressure ultraviolet radiation. Wat. Res. 35:1387-1398;

Department of Local Government (1998). Environment and Health Protection Guidelines: On-site Sewage Management for Single Households. Department of Local Government, NSW.

Gray, S., Houlihan, J., Wiles, R. (2001). Consider the source – farm runoff, chlorination byproducts and human health. Environmental Working Group. Washington DC.

Finch, G.R. and M. Belosevic (2001). Controlling Giardia spp. and Cryptosporidium spp. in drinking water by microbial reduction processes. Can. J. Civ. Eng. 28 (Suppl 1), 67-80.

Havelaar, A. H., W. M. Pot-Hogeboom, et al. (1984). F-specific RNA bacteriophages in sewage: methodology and occurrence. Water Science and Technology 17: 645-655.

Humphrey, N.C., H. Merrett, D. Owens and C. Weatherley (1998). The Fate of Enteroviruses in Waste Water Treatment, including UV Disinfection. In: Disinfection '98 The latest trends in Wastewater Disinfection: Chlorination vs UV Disinfection. WEF, Baltimore, pp. 415-424.

Linden, K.G., G. Shin and M.D. Sobsey (2001). Comparative effectiveness of UV wavelengths for the inactivation of Cryptosporidium parvum oocysts in water. Wat. Sci. Tech. 43(12):171-174.

NH&MRC and ARMCANZ (1994). Austalian drinking water guidelines. Canberra, National Health and Medical Research Council, Agriculural and Resource Management

Page 183

Council of Australia and New Zealand: 248.

NSW Health (1998). Aerated Wastewater Treatment Systems (AWTS) Accreditation Guideline. September 1998. NSW Health Department, Sydney.

Singer, P. (1993 ). Formation and characterisation of disinfection byproducts. In: Craun, G. Safety of water disinfection – balancing chemical and microbial risks. International Life Sciences Institute. Washington DC. p201.

Sobsey, M.D. (1989). Inactivation of health-related microorganisms in water by disinfection processes. Wat. Sci. Tech. 21(3):179-195.

Standards Australia/Standards New Zealand (2000). AS/NZS 1547:2000 On-site domestic-wastewater management. Standards Australia, Homebush.

Tree, J. A., M. R. Adams and D. N. Lees. (1997). Virus inactivation during disinfection of wastewater by chlorination and UV irradiation and the efficacy of F+ bacteriophage as a 'viral indicator'. Water Science & Technology 35(11-12): 227-232.

Tyrrell, S. A., S. R. Rippey and Watkins D.W. (1995). Inactivation of bacterial and viral indicators in secondary sewage effluents, using chlorine and ozone. Water Research 29(11): 2483-2490.

US-EPA (2002). On-site wastewater treatment systems manual. EPA/625/R-00/008. Office of Water & Office of Research and Development. United States Environmental Protection Agency.

US-EPA (2000). Wastewater Technology Fact Sheet – Dechlorination. EPA 832-F-00- 022. Office of Water, United States Environmental Protection Agency.

US-EPA (1999). Alternative Disinfectants and Oxidants Guidance Manual EPA 815-R- 99-014. Office of Water, United States Environmental Protection Agency.

US-EPA (1998) Stage 1 disinfectants and disinfection byproducts rules. 63 FR 69399- 69476. United States Environmental Protection Agency.

World Health Organisation (2002). Water and sanitation. Guidelines for drinking water quality – chloramines.

Page 184

Acknowledgements The authors would like to acknowledge the work of Ecowise Environmental Pty Ltd in undertaking this project, as well as the efforts of Australian Microbiological Services and Australian Water Quality Centre in the analysis of samples.

Page 185

Paper VII Fate and transport of viruses during sewage treatment in a mound system

Charles, K. J., F. C. Souter, D. L. Baker, C. M. Davies, J. F. Schijven, D. J. Roser, D. A. Deere, P. K. Priscott and N. J. Ashbolt (2008). "Fate and transport of viruses during sewage treatment in a mound system." Water Research 42 (12): 3047. Author Contribution Role (all authors helped with reviewing and (%) editing) Charles, K 50 Experimental plan. Set up and running of column experiments. Phage analysis for column studies. Assistance in field trials and survival in amended soil Statistical analysis and modelling. Writing. Souter, F 10 Phage analysis in field study and survival in sewage and effluent. Baker, D 10 Field sampling. Davies, C 5 Assistance in laboratory column studies. Phage analysis for survival in amended soil. Schijven, J 10 Assistance with experimental design of survival studies and column studies and analysis. Roser, D 5 Co-supervisor. Assistance with experimental design of field studies. Deere, D 4 Industry advisor. Assistance with experimental design. Priscott, P 1 AMS labs manager. Ashbolt, N 5 Supervisor. Assistance with experimental design and analysis.

Page 186

Abstract Studies undertaken to assess the performance of filter materials to remove phosphorus in decentralised sewage systems have not reported on the broader performance of these systems. This study aimed to identify virus fate and transport mechanisms at the laboratory scale for comparison with field experiments on a mound system amended with blast furnace slag. Inactivation was a significant removal mechanism for MS2 bacteriophage, but not for PRD1 bacteriophage. Column studies identified rapid transport of PRD1. Laboratory studies predicted lower removal of PRD1 in a full scale system than was experienced in the field study, highlighting the importance of considering pH and flow rate in pathogen removal estimates. The results highlight the necessity for studying a range of organisms when assessing the potential for pathogen transport.

Introduction A broad range of treatment options are available for decentralised sewage systems, but high rates of failure of traditional septic tanks (USEPA, 2002) are discouraging innovation in favour of centralised sewerage installation. Uptake of alternative technologies has also been limited where the ability to provide centralised management is limited by geographic or economic constraints. A further concern to regulatory authorities is the increasing potential for failure as the complexity of systems and treatment processes increases, the belief that householders will not maintain them, and the greater human and environmental risks arising from such failures.

Soil mound systems are one type of low-maintenance system, and can be enhanced with various materials to facilitate phosphorus removal. These materials facilitate ion exchange and precipitation of phosphorus through provision of increased surface area, pH and chemicals such as aluminium, iron and calcium compounds. Amendment materials investigated for their nutrient removal capabilities have included red mud (Ho et al., 1991), blast furnace slag (Johansson, 1999), fly ash (Cheung et al., 2006), zeolite (Sakadevan et al., 1998) and lightweight expanded clay aggregate (Ausland et al., 2002). However, limited research has been undertaken on how these materials affect the fate and transport of other contaminants, including pathogens.

A previous study aimed to assess the performance of a sand mound amended with blast

Page 187

furnace slag (Charles et al., 2004b). Two sand mounds (138 m2 x 1 m high) were installed to treat septic tank effluent from four households and two public toilet blocks. Effluent flowed horizontally out from a central trench to the edge of the impermeable liner, with pH increasing with distance travelled due to the high pH of the slag. The horizontal flow path resulted in saturated conditions within the mound. Bacteriophage, viruses that infect bacteria, were used as models for human enteric virus behaviour in the environment. They span the range of shapes, sizes, surface charges and persistence exhibited by many human enteric viruses. Their physical characteristics and tolerance of wastewater treatments make them ideal models to illustrate the behaviour of enteric viruses in groundwater, soil or subsurface environments and for sewage treatment efficacy evaluation. Routine sampling results indicated high removal of thermotolerant coliforms (> 3.8 log10), however there was significant transport of somatic coliphages present in the sewage with a mean concentration in effluent of 15 pfu (plaque forming -1 units).mL , a reduction of 1.5 log10 units. Two bacteriophage were used for field tracer experiments. MS2 is an icosahedral phage with a diameter of 27 nm and an isoelectric point of 3.5 reported to have little or no adsorption in saturated sandy soils at pH 6 - 8 with low organic carbon content. PRD1 is an icosahedral phage with a diameter of 62 nm and an isoelectric point between 3 and 4. Field tracer experiments resulted in similar transport of PRD1 bacteriophage (2.9 log10 removal) compared to somatic coliphages, but over 5 log10 removal of MS2 bacteriophage.

Virus transport and fate in soils is predominantly a function of advection, inactivation, sorption and desorption (Schijven et al., 2000). While advection and desorption are dependant on the flow conditions, adsorption and inactivation depend on a range of additional factors. Virus adsorption is generally the most important process for attenuation and depends on the pH, soil type, organic matter content, ionic strength and flow rate. Inactivation and irreversible adsorption are required for virus removal. Fate and transport characteristics vary between enteric viruses (Schijven et al., 2000). Virus inactivation is dependant on the physical characteristics of the different viruses (e.g. surface chemistry and morphology) and environmental factors (e.g. temperature, microbial activity, pH and ammonia) (Schijven et al., 2000). Flow rate can affect the contact of viruses with the attachment sites, with increasing velocities reducing contact time and therefore attachment (Schijven et al., 2000). High ionic strength, such as septic

Page 188

tank effluent, favour virus adsorption; with low ionic strength waters, such as rainfall, able to remobilise attached viruses (Schijven et al., 2000).

It was hypothesised that inactivation was likely to be the main mechanism for virus removal as the conditions within the mound were generally favourable for virus transport (high pH), but unfavourable for virus survival (up to 25 °C and pH 11.0). Furthermore, ammonia is known to be virucidal above pH 8 (Ward et al., 1977).

This paper reports the results of laboratory experiments undertaken in collaboration with the Sydney Catchment Authority to assess the mechanisms of virus transport in an amended sand mound system. The first set of experiments aimed to investigate the role of inactivation in removal of viruses in the field experiments, under the neutral pH conditions in the influent and high pH conditions in the effluent. The second set of experiments aimed to quantify the transport and fate of viruses in a laboratory column under more adverse conditions for virus removal; specifically high flow rates and low ionic strength.

Materials and Methods

Bacteriophage Inactivation in Sewage and Effluent Ten glass containers (100 mL) were filled with wastewater from the mound, making five influent-effluent pairs. Physico-chemical analyses (pH, temperature and DO [dissolved oxygen]) were undertaken on two pairs one incubated in the dark at 4 °C, the other at ambient temperature (22 °C). Of the remaining three pairs (Table 1), one pair (SI22 and SE22) were sterilised by autoclaving at 121 °C for 15 minutes to reduce the influence of microbial activity within the sample. Suspensions of MS2 and PRD1 (106 pfu.mL-1) were added to each of the six containers. One unsterilised pair (I22, E22), with the sterilised pair, were stored in the dark at 22 °C to replicate conditions within the mound (Table 1). The remaining unsterilised pair (I4, E4) were stored at 4 °C providing a temperature control to assist in the identification of the impact of the amending material. The low temperature was also assumed to reduce the influence of microbial activity. All containers were sampled over a 28 day period, which corresponded with the time period of the previous field experiments.

Page 189

Table 1 Experimental conditions for bacteriophage inactivation in influent and effluent

Sample Code Sample pH Temperature Sterilised I4 Influent 8.1 4°C No I22 Influent 8.1 22°C No SI22 Influent 8.1 22°C Yes E4 Effluent 10.6 4°C No E22 Effluent 10.6 22°C No SE22 Effluent 10.6 22°C Yes

Independent duplicate 1 mL aliquots were sampled from each container with separate dilution series for each sample. Three dilutions were analysed for each sample. On three occasions, duplicate analyses of the sample were undertaken. Results are reported as the average of duplicate plates, from the appropriate dilution. Bacteriophage analyses were undertaken by AMS Laboratories, Sydney. The spiked MS2 coliphage (ATCC 15597- B1) was analysed by the method of Havelaar and Hogeboom (1984) using host E. coli (Migula) (ATCC 15597) (phage and host supplied by the American Type Culture Collection, Manassas, USA). The spiked PRD1 bacteriophage was analysed by the method for somatic coliphages (ISO 10705-2, 2000) using host Salmonella typhimurium L29 (phage and host supplied by Prof. C. Gerba, University of Arizona, Tucson, Arizona). Background concentrations were measured prior to each spike.

PRD1 Laboratory transport study Column experiments were undertaken using the amended soil with artificial rain water

(ARW, per L of deionised water: 4.07 mg NaNO3, 3.24 mg NaCl, 0.35 mg KCl, 1.65 mg CaCl2.2H2O, 2.98 mg MgSO4.7H2O, 3.41 mg NH4.2SO4) (Davies et al., 2004).. The column was constructed from 10.5 cm PVC (internal diameter), capped with a reducer and 5.0 cm internal diameter PVC cap, fitted with a brass nipple (2 cm long, 2 mm internal diameter). The inside of the column was spirally machine roughened to minimise soil-column interface flow and therefore limit edge effects. Silicone tubing (Masterflex HV-96410-16 Precision Silicone (platinum) Tubing, internal diameter 3.1 mm) was fitted to the brass tube. The base of the soil was supported with three layers of wire mesh (pore size 2 mm) overlaid with a 2 cm layer of 2 mm glass beads. Soil slurry was added to a depth of 50 cm during up-flow conditions in small amounts. Regular stirring and shaking aimed to achieve uniform packing. At the top of the column, the interface of the soil and the column was sealed with silicone sealant and a ring of

Page 190

aluminium foil to further limit the short-circuiting of water via the column edges. A soil-free control column was used as a control with the supporting layers of wire mesh and beads, but no soil.

Several pore volumes of ARW were fed (downwards) through the column prior to experimentation to equilibrate the column. Saturated flow conditions were used as flow in the field is horizontal, and therefore primarily saturated. A flow rate of 1.1 m.d-1 was used. The sodium chloride concentration of the ARW was increased by 2.9 g.L-1 for several pore volumes to enable determination of flow dispersion. The electrical conductivity (EC) was analysed with a Lab Analyser 440 (TPS Pty Ltd, Australia). After completion of the salt tracer, the ARW was inoculated with PRD1 (final concentration 104 per mL). PRD1 was grown up in host cells of Salmonella typhimurium strain LT2 (ATCC 19585). The column influent was inoculated with PRD1, and the concentration in column effluent was quantified periodically during and after the application of the inoculum to determine the removal of viruses by passage through soil. A minimum of triplicate samples were collected to determine the background and final effluent phage concentrations. Column experiments were undertaken at ambient temperature (approximately 20 ºC). At the completion of experiments, the column was destructively sampled to determine percent moisture determination by drying in pre-weighed crucibles at 105 °C for 48 hours (APHA, 1998).

Inactivation in the column material and ARW was studied. Seventy portions sieved amended mound soil (2 g) was weighed into 5 mL polyethylene vials, and an additional 70 vials used with only 2 mL of ARW. Soil was analysed using standard methods (APHA, 1998) for pH, grain size distribution, carbon content, total and exchangeable iron, bulk density and cation exchange capacity. Phosphorus sorption was analysed using the method from Rayment and Higginson (1992). Stock suspensions of PRD1 were diluted in sterile deionised water such that 0.1 mL of the diluted bacteriophage suspension could be added to achieve approximately 1 x 106 virions per vial. Sterilised (autoclaved) ARW was added to the soil microcosms to saturate each of the soils as equally as possible. Control vials of each soil type were left uninoculated to be used for moisture determinations; 0.1 mL of sterile deionised water was added in place of the inoculum. The vials were incubated at 20 °C in the dark. An ibutton™ miniature temperature probe (Maxim/Dallas Semiconducter Corp., Dallas, Texas, USA) taped to

Page 191

the inside of the jar was used to record the temperature inside the jar at 120 min intervals throughout the experiment. The soil was then inoculated with PRD1 and triplicate vials were randomly selected and withdrawn periodically from each sealed jar for determination of infectious PRD1. Hence, the microcosms were sampled destructively. In addition, duplicate vials were removed periodically from each jar for percent moisture determination by drying in pre-weighed crucibles at 105 °C for 48 hours (APHA, 1998).

Each 2 g of inoculated soil from sampled vials was washed into a 50-mL Falcon tube using 20 mL 3 % (w/v) beef extract solution (Straub et al., 1992) (pH 9). The soil slurry was then vortexed for 2 mins and shaken for 30 mins. After further vortexing for a few seconds the slurries were centrifuged for 15 mins at 2 500 g, after which 1 mL of the supernatant was withdrawn by pipette. This was diluted serially in sterile deionised water and assayed by the double agar layer technique (Adams, 1959). Concentrations of phages were expressed as plaque-forming units per vial.

A recovery control was prepared for each soil type by freshly inoculating 2 g of soil with approximately 1 x 106 PRD1 from a stock suspension. The phage was allowed to adsorb to the soil by storing at room temperature for two hours, before being processed as described above. The titre of the stock suspension used to inoculate the recovery controls was also determined. Percent recovery for each soil type was determined as the concentration of phage recovered divided by the concentration inoculated into the soil x 100. Phage concentrations in ARW were measured by directly diluting the contents of the microcosm vials with deionised water, and assaying by the double agar layer technique as above.

Statistical analyses For the inactivation studies, two inactivation models were applied: first-order and biphasic. First-order inactivation was modelled as C = C0.Exp(-µt) (Equation 1), where

C was the concentration of phages, C0 was the concentration of phages at time = 0, t was time, and the inactivation rate, µ, was considered to be a function of pH, temperature and/or microbial activity. Biphasic inactivation was modelled as C = C0

[α.Exp(-µ1.t) + (1-α).Exp(-µ2.t)] (Equation 2), α was the fraction of less stable viruses with a higher inactivation rate coefficient µ1, and (1- α) is the fraction of more stable

Page 192

viruses with a lower inactivation rate coefficient µ2 (Petterson et al., 2001c) . The inactivation rate coefficients were derived from the experimental data by fitting the inactivation models with a log-likelihood method previously described by Schijven et al. (2002a). Using the likelihood ratios test, the first-order and biphasic inactivation models were compared for each dataset, with significance defined by the χ2 test at p < 0.05 level. Outliers were identified as the residuals from fitting of first-order and biphasic models that were statistical outliers based on boxplot results using SPSS (version 11.5.2.1, SPSS Inc., 2003). They were defined as points where the residual from the fitted inactivation model was more than 1.5 times the inter quartile range above the third quartile. Outliers were excluded from final inactivation rate analyses to improve the statistical comparison of different experimental conditions and prevent the overestimation of the inactivation rate. Outliers only occurred in the initial two days of the studies, therefore the remaining data provided information on the longer term behaviour of the viruses. A maximum of one time point within the first two days of the study was excluded from five of the twelve studies. Two time points (t = 0,1) were excluded from one study.

Modelling of breakthrough curves was undertaken for the field experiments (Charles et al., 2004b) and laboratory experiments. HYDRUS-1D (United States Salinity Laboratory, Riverside) (Šimùnek et al., 1998) is commonly used in modelling micro- organism transport in porous media (Schijven et al., 2002c; Charles, 2007; Foppen et al., 2007). Conservative tracer breakthrough curves in the column and in the mound were fitted to the convection-dispersion solute transport equation using HYDRUS-1D to calculate porosity and dispersion within the columns. The fit was optimised using the Levenberg – Marquardt non-linear minimisation algorithm for least-squares solutions.

The relative concentrations from the phage breakthrough curves (C/C0) were fitted (using log resident concentrations) to the one-site kinetic adsorption equations using

HYDRUS-1D. For modelling purposes, inactivation of unattached viruses (µl) was assumed to equal the inactivation in ARW, and inactivation of attached viruses (µs) was assumed to equal inactivation in the soil-water microcosm. Virus transport coefficients were fitted from the column breakthrough curve to the one-site kinetic adsorption equations, and were compared with the results from the field study results.

Page 193

Results

Survival in sewage and effluent The sample chemistry was generally stable over the duration of the study. Samples had initial pH values of 8.1 and 10.6 for influent and effluent respectively. Final pH values (32 days) were 7.9 and 7.3 for influent at 4 °C and 22 °C respectively, and 10.8 and 10.5 for effluent at 4 °C and 22 °C respectively. An initial decrease in dissolved oxygen and slight pH variation was observed over the first 24 hours from sampling, with the sample stabilising after such time. Average initial concentrations, as quantified on day zero from the spiked sample containers, of 106 pfu.mL-1 of MS2 and PRD1 were achieved for all samples except unsterilised effluent at 22 °C where MS2 was 104 pfu.mL-1. The average temperatures were 4.7 °C (± 1.7 SD) and 22.8 °C (± 1.3 SD).

MS2 bacteriophage was very sensitive to the conditions within the mound: high pH, 22 °C, high microbial activity (Figure 1). Unsterilised effluent exhibited the most rapid inactivation with no phage detectable (<1 pfu.10 mL-1) after two days. PRD1 was more stable in the conditions in the mound (Table 2). Both MS2 and PRD1 exhibited an increased inactivation with increased temperature. Outliers were identified in six of the twelve studies which included t = 0 for PRD1 E4 and MS2 I4; t = 1 for PRD1 SI22 and I22; t = 2 for PRD1 SE22; and t = 0 and 1 for MS2 I22 (see Table 1 for sample codes). The biphasic inactivation model (Equation 2) was preferred for four of the studies (Table 2) based on the log-likelihood ratios test (significant at the χ2 95th percentile). PRD1 had significantly (p < 0.05) lower rates of inactivation than MS2 and the first- order model was preferred in all except for the case with highest inactivation, E22 (Figure 2).

Page 194

(a)

) 1.E+08 -1

1.E+05

1.E+02

Concentration (pfu.mL 1.E-01 01530 Time (d)

(b) )

-1 1.E+08

1.E+05

1.E+02

1.E-01 Concentration (pfu.mL Concentration 01530 Time (d)

Figure 1 MS2 data (circles), first-order model (continuous line) and biphasic model (dotted line) for

(a) influent and (b) effluent at 22 °C

Page 195

Table 2 Bacteriophage inactivation rate coefficients and likelihood ratios in sewage, ARW and

ARW with amended soil

2 Sample First- Biphasic model Likelihood ratio test (χ 0.95) order model

µ α µ1 µ2 Log Preferred likelihood model (day-1) (day-1) (day-1) ratio

MS2 I4 0.028 0 2.1 0.028 -1.6 x 10-14 First-order

I22 0.18 0.9956 0.23 0 11.4* Biphasic

SI22 0.52 0.9996 1.23 0.26 40.8* Biphasic

E4 11.3 Insufficient data - (First-order)

E22 > 8.48 Insufficient data - (First-order)

SE22 0.44 0.996 1.09 0.26 17.8* Biphasic

PRD1 I4 0.0012 0 31.7 0.001 -4.4 x 10-12 First-order 2

I22 0.018 0 15.4 0.018 -1.7 x 10-10 First-order

SI22 0.015 0 3.85 0.015 2.8 x 10-9 First-order

E4 0.049 1 0.049 3.3 x -1.6 x 10-6 First-order 10-5

E22 0.071 0.877 0.287 0 22.3* Biphasic

SE22 - 0 3.45 -2E- -4.9 x 10-8 First-order 0.00015 04

ARW 0.0283 0.014 0.161 0.002 13.2* Biphasic 2

ARW 0.0210 0 0.238 0.021 -4.0 x 10- First-order + soil 11

Page 196

)

-1 1.E+08

1.E+07

1.E+06

1.E+05 Concentration (pfu.mLConcentration 01530 Time (d)

Figure 2 PRD1 data (circles), first-order model (continuous line) and biphasic model (dotted line) effluent at 22 °C

The first-order model inactivation coefficient, µ, for MS2 (Table 3) varied significantly between each sample type, except for the sterilised influent/effluent pair. Inactivation was significantly more rapid (p < 0.05) in unsterilised effluent than in unsterilised influent, both at 4 °C and 22 °C. Temperature was a major factor for MS2 inactivation in influent, with µSI22 significantly (p < 0.05) greater than µI4. The MS2 biphasic models (Table 3) for sterilised influent and effluent were not significantly different, with the data able to be described with all common coefficients. However, both were significantly different from MS2 I22. PRD1 was more robust under the conditions of the study, and hence, comparison of the PRD1 first-order inactivation model coefficients (Table 3) elicited fewer significant factors. Nonetheless, sample type was significant (p < 0.05) for unsterilised samples with inactivation being more rapid in effluent than in influent at 4 °C and 22 °C. Temperature and microbial activity, separately, were only significant in effluent however combined were significant in influent but not effluent.

Page 197

Table 3 Comparisons of phage inactivation based on first-order inactivation model

Samples Variable* MS2 Likelihood ratios test PRD1 Likelihood ratios test (χ2 ) (χ2 ) 0.95 0.95 Result Common µ Result Common µ

I4 v I22 T, M µI22 > µI4 - µI22 > µI4 -

I4 v SI22 T µSI22 > µI4 - - 0.00719

I4 v E4 S µE4 > µI4 - µE4 > µI4 -

I4 v E22 T, S, M µE22 > µI4 - µE22 > µI4 -

I4 v SE22 T, S µSE22 > µI4 - - 0.000824

I22 v SI22 M µSI22 > µI22 - - 0.0170

I22 v E4 T, S, M µE4 > µI22 - µE4 > µI22 -

I22 v E22 S µE22 > µI22 - µE22 > µI22 -

I22 v SE22 S, M µSE22 > µI22 - µSE22 > µI22 -

SI22 v E4 T, S µE4 > µSI22 - µE4 > µSI22 -

SI22 v E22 S, M µE22 > µSI22 - µE22 > µSI22 - SI22 v SE22 S - 0.470** - 0.0100

E4 v E22 T, M (µE4 > - - 0.0495 + µE22)

E4 v SE22 T µE4 > µSE22 - µSE22 > µE4 -

E22 v SE22 M µE22 > µSE22 - µSE22 > µE22 - *T= temperature; M= microbial activity; S = sample;**common biphasic model also; + limited data

Survival in amended soil The amending material resulted in increased pH and phosphorus sorption capacity as well as in calcium and cation exchange capacity (Table 4). The average temperature within the microcosm jars was 19.5 °C (SD ± 0.77 ºC) for the duration of the experiment (131 days). The moisture content was 17.7 ± 1.0 %. The efficiency of the phage recovery method for the soil microcosms was 56 ± 25 %, and was not affected by time. Percent recovery was not used to adjust the phage concentrations. PRD1 inactivation (Table 2; Figure 3) in the soil/water matrix was significantly lower than inactivation in ARW alone. The biphasic model provided a better fit in the ARW only.

Page 198

Table 4 Physicochemical properties of amended soil used for laboratory experiments

Parameter Slag Amended Sand Sieved Column sand material Texture Sand Loamy Sand Clayey sand Clayey sand d50 (mm) 0.63 0.43 0.38 0.50 pH 9.4 9.2 5.4 8.7

Bulk Density (g.cm-3) 1.6 1.4 1.35 1.68

Electrical conductivity 1.2 0.5 <0.1 - (d.m-1) (1:5 water)

Exchange Capacity (cmol.kg-1)

Aluminium 0 0.03 2.82 0

Calcium 41.55 13.99 0.67 18.4

Cation 42.36 14.59 4.29 18.9

Magnesium 0.36 0.28 0.59 0.3

Phosphorus sorption 24 000 15 000 1 700 - capacity (kg/hectare)

% Carbon (% dry weight) 0.4 0.4 0.4 0.3

Page 199

(a) )

-1 1.E+07 1.E+06 1.E+05 1.E+04 1.E+03 1.E+02 1.E+01

Concentration (pfu.mLConcentration 1.E+00 0 15 30 45 60 75 90 105 120 135 150 Time (d)

(b) )

-1 1.E+07 1.E+06 1.E+05 1.E+04 1.E+03 1.E+02 1.E+01

Concentration (pfu.mLConcentration 1.E+00 0 15 30 45 60 75 90 105 120 135 Time (d)

Figure 3 PRD1 inactivation data (circles) with error bars indicating standard deviation, linear model (continuous line) and biphasic model (dotted line) for (a) Artificial rain water; (b) amended soil

Page 200

Transport in columns Salt tracer recover was 100%, providing a good fit of the salt tracer breakthrough curve. The porosity used in the model was based on the total porosity of the column (including the pore space in the cap), with the range calculated from minimum to maximum water content. Therefore the porosity in the model was higher than the water content of the soil. The modelled porosity was 0.233 ± 0.021 (standard error), with a dispersivity 0.042 ± 0.015 m. The values of porosity and dispersivity were then used in the fitting of a one-site kinetic model in HYDRUS-1D to the breakthrough curve data for PRD1, to estimate the attachment and detachment coefficients, katt and kdet respectively.

Dispersivity and µs were varied where acceptable fits were not possible fitting just katt and kdet.

PRD1 breakthrough was rapid, occurring after 0.34 pore volumes (Figure 4). Over the

50 cm of transport in the columns, there was a 0.36 log10 reduction in PRD1 concentration, calculated as the log10 of the maximum effluent concentration divided by the maximum influent concentration. The total load recovered of inoculated phage after transport through the column was 59 % (0.39 log10), which is comparable to the reduction in concentration. Sampling was stopped prematurely due to difficulties with the inocula. This resulted in limitations for fitting and high standard errors.

1.E+00

1.E-01

1.E-02

Phage Concentration C/Co Concentration Phage 1.E-03 01234 Time (days)

Figure 4 PRD1 bacteriophage breakthrough curve data (circles), model PRD1 input (grey line) and the fitted one-site kinetic model (solid line)

Page 201

-1 Modelling of the virus breakthrough curve resulted in a katt of 11.1 ± 8.24 .d , kdet 0.91 ± 2.72 .d-1 and an R2 of 0.910. The attached inactivation rate for the model was 0.831 ±

1.17 ln units per day. The standard errors for katt and kdet were high, which needs to be considered in the application of the results to the field scale. Inactivation was observed in the phage inocula (Figure 4), and was included in the breakthrough curve modelling. Inactivation in the inoculated ARW fed into the columns where full breakthrough curves were obtained resulted in the overestimation of PRD1 removal (Table 2).

Extrapolation of the HYDRUS-1D model from the laboratory column experiments to the field scale predicted a reduction in the concentration of PRD1 of 0.53 log10 over 1.4 m transport (0.73 log10 over 1.9 m transport), compared with the 2.9 log10 reduction experienced in the field mound studied.

Discussion

Inactivation The inactivation rates in this study were comparable to previously reported groundwater studies (Logsdon, 1994; Schijven et al., 2000). Temperature is considered one of the most important factors in, and is negatively correlated with, virus survival. In the current study, increasing temperature significantly increased MS2 inactivation in influent, which was consistent with previous reports that MS2 is more stable at temperatures of less than 10 °C (Schijven et al., 2000; Schaper et al., 2002). For PRD1, which has been reported to remain relatively stable at temperatures up to ambient temperature (23 °C) (Schijven et al., 2000), temperature was only a significant factor in the more adverse high pH conditions of the effluent. Human enteric virus survival has generally been considered to be greatest near neutral pH, yet these viruses are stable within a range of pH 3-10 (Gerba et al., 1996).

Rapid inactivation of MS2 was experienced in unsterilised effluent (pH 10.6) in this study, with 4.9 log10 and >3.6 log10 reduction in one day at 4 °C and 22 °C respectively, compared with no detectable inactivation in unsterilised influent (pH 8.1). However, the effluent samples also had the lowest initial concentrations (8.6 x 105 and 2.5 x 104 pfu.mL-1), compared to a mean initial concentration in the other samples of 1.8 x 106 pfu.mL-1, which may have resulted from rapid inactivation in the time (two hours) between introducing the phage to the sample and sampling than could not be measured

Page 202

by this experiment but assumed given that the same inoculum was used. Such rapid rates of MS2 inactivation have been reported at high pH, with greater than 4 log10 reduction in 2 hours at pH 11 - 11.5 (Logsdon, 1994). While MS2 is more resistant to ammonia than other F-RNA coliphages (Schaper et al., 2002), higher temperatures and ammonia may have a synergistic effect resulting in an increased rate of inactivation. The rapid inactivation of MS2 in the effluent sampled from the mound indicates that inactivation was likely to be a significant removal mechanism, with a 6 log10 reduction predicted in less than 48 hours based on the first-order inactivation rate.

PRD1 was more robust under the mound conditions than MS2, with inactivation therefore predicted to account for only 0.6 log10 reduction over 21 days in the mound. Rotavirus is considered to be one of the more resistant human enteric viruses (Höglund et al., 2002a). However, PRD1 was more resistant to high pH than has been reported for simian rotavirus, SA11 (Estes et al., 1979), and reovirus (Ward et al., 1977). Inactivation at high pH in wastewater, however, is confounded by the presence of ammonia and its greater virucidal effect on viruses (Ward et al., 1977). For both model viruses examined, pH was the most significant factor (Table 3), with the inactivation rates in effluent consistently greater than those in influent regardless of temperature or sterilisation. Therefore, the ability to modify pH within sewage treatment systems, such as amended mounds, can potentially increase inactivation of human enteric viruses. However, it is important to recognise that the efficacy of pH is dependent on the type of virus.

While pH was assumed to be the main difference between the influent and effluent samples, the transport of the influent through the high pH sand mound would have resulted in changes to the composition of the wastewater that were not quantified here. For example, the effect of microbial activity in effluent may have been less than for influent: temperature was a significant factor in PRD1 survival in influent but not in effluent. Ammonia may have contributed to the increased inactivation in effluent as it is known to be virucidal above pH 8 as discussed above. While total nitrogen decreased in the sewage system, the proportion of ammonia increased from 89 % to 92 %, and the calculated concentration (at 25 °C) of free ammonia increased with pH from 5 mg.L-1 in influent (pH 8.1) to 53 mg.L-1 in the effluent (pH 10.6).

Page 203

The increased inactivation of MS2 in sterilised influent may indicate changes in chemistry during the autoclaving process, as the elimination of microbial activity would generally be expected to decrease the inactivation rate. Similarly the decreased inactivation with increasing temperature experienced in sterilised effluent at 22 °C compared to the unsterilised effluent at 4 °C, where no significant microbial activity was assumed, indicated that autoclaving affected the matrix chemistry. Unfortunately, due to low influent flow at the field site additional samples for chemical analyses were not available.

Outliers were defined for this study by both the first-order and biphasic models. Commonly in inactivation studies, when the second sample is taken shortly after the first, there is an apparent increase in concentration (e.g. Yates et al., 1985). It is hypothesised that this is due to disaggregation in the inactivation media. Inclusion of this first data point value in the determination of the inactivation rate coefficient will underestimate the initial inactivation. Hence, sampling time points were included at one and two days, providing a more reliable value of the actual virus load at the time of inoculation.

Inactivation rates are quantified to allow the comparison and extrapolation of experimental data sets. These are used for use in management, such as in risk assessment, or in the calculation of the contribution of inactivation to the removal of culturable viruses in experiments. Inactivation has typically been modelled using a first- order model, which assumes a log-first-order decay. However, in many studies, the rate of inactivation decreases over the duration of the study. In these cases, application of the first-order model overestimates the long-term inactivation rate. Alternative inactivation models have been developed to address this issue. Cerf (1977) developed the biphasic model used here, where there are two distinct subpopulations with separate inactivation rates. This model has been employed in the food industry (Geeraerd et al., 2005), and in previous environmental survival of viruses (Petterson et al., 2001a). First-order inactivation was experienced under a variety of conditions including high temperature and high pH, however in all cases first-order inactivation was preferred where µ < 0.05.d-1. Above this value, the biphasic model was the preferred model (where sufficient data was available). For the four cases where the biphasic model was preferred, the long-term inactivation rates varied from 0.00 to 0.26.d-1 for both phages. Where the

Page 204

first-order model was preferred, a maximum of 0.6 log10 reduction was experienced compared to a minimum 0.8 log10 reduction before the lower of the biphasic inactivation rates dominated. It is therefore conceivable that the biphasic model would become preferable in more scenarios with increased time and inactivation.

The implications of biphasic decay are that over time the inactivation rate decreases, and this long term inactivation rate is therefore overestimated by a linear model. In the mound investigated, the timescale for a virus to pass through the system was approximately 20 – 30 days. Over this timescale, there was limited biphasic behaviour exhibited, and due to the short duration and the nature of the project, there was not sufficient confidence in the biphasic data to extrapolate the results. However, there was evidence that there were small fractions of viruses, in this case 0.4 to 0.04 % MS2 and 12 % PRD1, that were considerably more robust in these conditions. It has been theorised that these robust viruses may represent a more resistant subpopulation (Petterson et al., 2001b). It is not known is how these more robust viruses will behave in the environment when they are released from the sand mound, and therefore how to assess the risks of such enteric viruses as they are transported through the system and potentially ingested.

Column and field studies Virus inactivation is usually considered to be insignificant in soil column transport experiments due to the small timescales involved. However, as breakthrough was monitored for up to six weeks in the test columns, phage inactivation was considered to be influential. Inactivation in soil-water microcosms was assumed to represent phage inactivation in the attached phase, µs. However, the method does not differentiate between attached and free viruses. The amended soil had a lower rate of removal (attachment) within the column and therefore it is likely they had lower rates of attachment within the microcosms. However, quartz sand has been reported to have a virus sorption capacity of 2.2 x 1012 pfu.kg-1 (Moore et al., 1981) which far exceeds the initial virus load applied (106 pfu to 1.3 g-1 soil dry weight).

Column experiments using repacked soil are commonly used to characterise virus transport in soil (e.g. Sobsey et al., 1995), with results generally considered to be representative of field scale. Column experiments with amended sand (Ho et al., 1991)

Page 205

have demonstrated higher removal of E. coli and poliovirus in sand amended with red- mud than unamended sands, with greater than 7 log10 removal of both organisms in 65 cm columns of amended sand. However, field verification has not been reported, nor virus removal in other media used in decentralised sewage systems. Laboratory studies aimed to provide quantitative data on virus fate and transport in a mound under conditions anticipated to maximise transport: saturated, low retention time, rain water. Previous studies of virus transport in sand have found similarly rapid transport of viruses (McKay et al., 1993; Schijven et al., 2002b). Physical heterogeneities, such as roots, macropores and rock fractures, can provide preferential water flow paths (McKay et al., 1993). Soil columns by their nature introduce paths for preferential transport along the soil-column interface, however as repacked column experiments aim to establish the adsorption capacity of the soil and its influence on virus transport and fate, every attempt is made to minimise their influence. The use of rainwater was assumed to provide a worst-case scenario for transport due to low ionic strength resulting in low attachment. However, in sewage there is the potential for viruses to attach to colloids which may decrease inactivation and facilitate transport (Jin et al., 2002). Organic matter, including surfactants in the water have been reported to decrease adsorption due to competition for attachment sites and are also reported to increase desorption of viruses (Schijven et al., 2000).

Removal within the amended sand column was greater than in natural sandy loams (Charles, 2007) although inactivation in this soil was lower. The amended soil was not preconditioned and therefore was expected to have little microbial activity, and hence low inactivation (Jin et al., 2002). Inactivation of PRD1 in the soil-water matrix without preconditioning was lower than in effluent from the soil matrix. Virus removal within the sewage treatment mound system that utilises this soil was greater in the field experiments (1.1 log10 in 0.7 m) than experienced in the column experiments carried out in the present study (0.36 log10 in 0.5 m).

The differences between these column experiments and the on-site system scenario that will affect the up-scaling of these results to the field situation are:

Page 206

• Rainwater, as used in the columns, will result in decreased attachment compared to sewage, however colloids/organics in the sewage may also facilitate virus transport and survival;

• Heterogeneity in the field may increase travel velocity compared to the repacked columns, thereby decreasing attachment, and may affect removal over the site; and

• Lack of microbial activity in the microcosms, if extrapolated, may underestimate inactivation in the field.

Implications for household systems The mound system received flows averaging one third of the design flow during the field study. This low hydraulic load was anticipated to result in an over estimation of the performance of the system, with microorganism inactivation and attachment expected to decrease during increased flows. For a single household system, this degree of under- loading is equivalent to household occupancy rates of two to three people living in a four to five bedroom house, assuming the system is conservatively designed for dual occupancy of each bedroom. Therefore the higher loading rates that would be expected at houses with higher occupancy rates may result in decreased performance of the system.

Extrapolation of the column study model to the field-scale provided a prediction of virus transport in a less well established system and under higher flow rates. At the reduced detention time modelled, seven days to maximum breakthrough compared to over 20 in the field, the removal was only 0.73 log10 over the 1.9 m width of the system. It is anticipated that this would generally be an underestimate. In an established, well constructed system, the development of a clogging layer would retard virus transport. The high degree of homogeneity in the systems would be expected to increase virus removal, although poor construction could lead to areas without sufficient amending material, and therefore variable pathogen and nutrient removal. However under other conditions, such as prior to clogging layer development and under high flow, this may be representative of transport.

The high pH environment within the mound strongly affected many of the indicator

Page 207

organisms studied. It is not known how long this virucidal high pH may persist, but as it is also important for phosphorus removal it is a key management parameter. Furthermore, high pH is possibly more important than the availability of sorption sites as calcium leaching is expected to result in a decrease in phosphorus sorption.

In the mound system studied, the laboratory estimate of the inactivation rate indicates that inactivation is an important removal mechanism. Therefore these results would not necessarily hold for a similar design with a lower pH. Overall, the amended material sand mound provided greater removal of enteric virus models from sewage than reported for other sewage treatment systems including septic tanks (0.6 log10), aerated treatment units (1 log10) , and disinfection systems (1 - 1.8 log10 for MS2) (Charles, 2007).

Conclusions

• The primary mechanisms for virus removal varied significantly depending on the susceptibility of the individual virus to the system chemistry. MS2 removal as with thermotolerant coliforms, was rapid in the mound. Inactivation studies

for MS2 suggested that inactivation was the primary mechanism a 6 log10 reduction in less than 48 hours in effluent. Similarly, removal of PRD1 and somatic coliphages was lower in the mound, and inactivation of PRD1 was lower. These results highlight the necessity for studying a range of organisms when assessing the potential for pathogen transport.

• The application of the amending material to mound systems at the individual household scale requires considerations of the likely flow rates and pH of the system. Considerations for implementing amended soil mound technology should also include the life expectancy of the system and management protocols for decommissioning and/or replacement.

Acknowledgements The authors would like to acknowledge the assistance of Robert McGuinness and Christobel Ferguson of the Sydney Catchment Authority in the development and management of this work. The project was funded by the Sydney Catchment Authority as part of their commitment to improve decentralised sewage management within

Page 208

Sydney’s drinking water supply catchments.

References Adams, M. H. (1959). Bacteriophages. New York, Interscience Publishers Inc.

APHA (1998). Standard Methods for the Examination of Water and Wastewater. Washington D.C., American Public Health Association.

Ausland, G., Stevik, T. K., Hanssen, J. F., Kohler, J. C. and Jenssen, P. D. (2002) Intermittent filtration of wastewater - removal of fecal coliforms and fecal streptococci. Water Research 36(14), 3507-3516.

Cerf, O. (1977) Tailing of survival curves of bacterial spores. Journal of Applied Bacteriology 42(1), 1-19.

Charles, K. J. (2007). Quantitative Microbial Risk Assessment: a catchment management tool to delineate buffer distances for on-site sewage treatment and disposal systems in Sydney’s drinking water catchments. Civil and Environmental Engineering. Sydney, University of New South Wales. PhD.

Charles, K. J., Schijven, J. F., Baker, D., Roser, D. J., Deere, D. A. and Ashbolt, N. J. (2004). Transport and fate of nutrients and pathogens during sewage treatment in a mound system. On-site Wastewater Treatment X: Proceedings of the Tenth National Symposium on Decentralised Sewage Treatment and Disposal Systems. Sacramento, USA, American Society for Agricultural Engineers: 460 - 469.

Cheung, K. C. and Venkitachalam, T. H. (2006) Kinetic studies on phosphorus sorption by selected soil amendments for septic tank effluent renovation. Environmental Geochemistry and Health 28(1-2), 121-31.

Davies, C. M., Ferguson, C. M., Kaucner, C., Altavilla, N., Deere, D. A. and Ashbolt, N. J. (2004) Dispersion and transport of Cryptosporidium oocysts from fecal pats under simulated rainfall events. Applied and Environmental Microbiology 70(2), 1151-1159.

Estes, M. K., Graham, D. Y., Smith, E. M. and Gerba, C. P. (1979) Rotavirus stability and inactivation. Journal of General Virology 43(2), 403-409.

Page 209

Foppen, J. W., van Herwerden, M. and Schijven, J. (2007) Measuring and modelling straining of Escherichia coli in saturated porous media. J Contam Hydrol 93(1-4), 236- 54.

Geeraerd, A. H., Valdramidis, V. P. and Van Impe, J. F. (2005) GInaFiT, a freeware tool to assess non-log-linear microbial survivor curves. International Journal for Food Microbiology 102(1), 95-105.

Gerba, C. P., Rose, J. B., Haas, C. N. and Crabtree, K. D. (1996) Waterborne rotavirus: A risk assessment. Water Research 30(12), 2929-2940.

Havelaar, A. H. and Hogeboom, W. M. (1984) A method for the enumeration of male- specific bacteriophages. Journal of Applied Bacteriology 56, 439-447.

Ho, G. E., Gibbs, R. A. and Mathew, K. (1991) Bacteria and virus removal from secondary effluent in sand and red mud columns. Water Science and Technology 23(1- 3), 261-270.

Höglund, C., Ashbolt, N. J., Stenström, T. A. and Svensson, L. (2002) Viral persistence in source-separated human urine. Advances in Environmental Research 6(3), 265-275.

ISO 10705-2 (2000). Water quality - Detection and enumeration of bacteriophages. Part 2: Enumeration of somatic coliphages. Geneva, International Organization for Standardization.

Jin, Y. and Flury, M. (2002) Fate and transport of viruses in porous media. Advances in Agronomy 77, 39-102.

Johansson, L. (1999) Blast furnace slag as phosphorus sorbents - column studies. The Science of the Total Environment 229, 89-97.

Logsdon, G. S. (1994). The removal and disinfection efficiency of lime softening processes for Giardia and Viruses. Denver, Co., American Water Works Association Research Foundation and American Water Works Association.

Page 210

McKay, L. D., Cherry, J. A., Bales, R. C., Yahya, M. T. and Gerba, C. P. (1993) A field example of bacteriophage as tracers of fracture flow. Environmental Science and Technology 27(6), 1075-1079.

Moore, R. S., Taylor, D. H., Sturman, L. S., Reddy, M. M. and Fuhs, G. W. (1981) Poliovirus adsorption by 34 minerals and soils. Applied and Environmental Microbiology 42(6), 963-975.

Petterson, S. R., Teunis, P. F. M. and Ashbolt, N. (2001) Modeling virus inactivation on salad crops using microbial count data. Risk Analysis 21(6), 1097-1107.

Rayment, G. E. and Higginson, F. R. (1992). Australian Laboratory Handbook of Soil and Water Chemical Methods. Port Melbourne, Reed International Books Australia P/L, trading as Inkata Press.

Sakadevan, K. and Bavor, H. J. (1998) Phosphate adsorption characteristics of soils, slags and zeolite to be used as substrates in constructed wetland systems. Water Research 32(2), 393-399.

Schaper, M., Duran, A. E. and Jofre, J. (2002) Comparative resistance of phage isolates of four genotypes of F-specific RNA bacteriophages to various inactivation processes. Applied and Environmental Microbiology 68(8), 3702-3707.

Schijven, J. F. and Hassanizadeh, S. M. (2000) Removal of viruses by soil passage: Overview of modeling, processes, and parameters. Critical Reviews in Environmental Science and Technology 30(1), 49-127.

Schijven, J. F., Hassanizadeh, S. M. and de Bruin, H. A. M. (2002) Column experiments to study nonlinear removal of bacteriophages by passage thorugh saturated dune sand. Journal of Contaminant Hydrology 58, 243-259.

Schijven, J. F. and Simunek, J. (2002) Kinetic modeling of virus transport at the field scale. Journal of Contaminant Hydrology 55(1-2), 113.

Šimùnek, J., Šejna, M. and van Genuchten, M. T. (1998). The HYDRUS-1D software package for simulating the one-dimensional movement of water, heat and multiple

Page 211

solutes in variably saturated media. Golden, Colorado, International Ground Water Modelling Centre, Colorado School of Mines.

Sobsey, M. D., Hall, R. M. and Hazard, R. L. (1995) Comparative reductions of hepatitis A virus, enteroviruses and coliphage MS2 in miniature soil columns. Water Science and Technology 31(5-6), 203-209.

Straub, T. M., Pepper, I. L. and Gerba, C. P. (1992) Persistence of viruses in desert soils amended with anaerobically digested sewage sludge. Applied and Environmental Microbiology 58, 636-641.

USEPA (2002). On-site Wastewater Treatment Systems Manual. US EPA, Washington.

Ward, R. L. and Ashley, C. S. (1977) Identification of the virucidal agent in wastewater sludge. Applied and Environmental Microbiology 33(4), 860-864.

Yates, M. V., Gerba, C. P. and Kelley, L. M. (1985) Virus persistence in groundwater. Applied and Environmental Microbiology 49(4), 778-781.

Page 212

Paper VIII Virus fate and transport from onsite sewage systems in Sydney drinking water catchments: laboratory and field studies

Charles, K. J., C. M. Davies, D. L. Baker, C. J. Charles, F. C. Souter, N. J. Ashbolt, J. F. Schijven, P. K. Priscott and D. A. Deere (Submitted). "Virus fate and transport from on- site sewage systems in Sydney drinking water catchments: laboratory and field studies." Submitted to Environmental Science and Technology.

Author Contribution Role (all authors helped with reviewing and (%) editing) Charles, K 50 Experimental plan. Set up and running of column experiments. Phage analysis for column studies. Assistance in field trials and survival in soil. Statistical analysis and modelling. Writing. Davies, C 10 Set up and phage analysis for survival studies. Assistance in laboratory column studies. Baker, D 10 Field sampling. Charles, C 5 Column development. Souter, F 5 Phage analysis in field study. Ashbolt, N 7.5 Supervisor. Assistance with experimental design and analysis. Schijven, J 7.5 Assistance with experimental design of column studies and analysis. Priscott, P 1 AMS labs manager. Deere, D 4 Industry advisor. Assistance with experimental design.

Page 213

Abstract Human enteric viruses are persistent and highly mobile in the environment. Onsite sewage systems are a potential source of virus contamination of water resources used for drinking or recreational purposes. The aim of this project was to assess the potential for the transport of effluent, and in particular viruses, from onsite sewage systems in Sydney’s drinking water catchment. Laboratory and field experiments were undertaken using five soils and two bacteriophages (MS2 and PRD1) to quantify inactivation, and transport at the column and field scale. Survival of PRD1 in water and soil-water matrices was generally longer than for MS2. For both phages removal was greatest, but highly variable, in clayey soils, with between 1.06 to > 3.00 log10 PRD1 reduction and

1.21 to 1.32 log10 MS2 reduction within 0.5 m. Sandy loam soils exhibited much lower removal of between 0.00 and 0.23 log10 PRD1 reduction and 0.14 log10 MS2 reduction. The column breakthrough curves were fitted to a one-site kinetic attachment model to determine the transport coefficients. Of the two field experiment sites, only one had detectable breakthrough, where virus removal was lower than in the column experiments, highlighting the potential for overestimation of the removal at field-scale when extrapolating from column experiments.

Introduction In centralised sewage systems, the concentrations of infectious enteric viruses in sewage have been reported at about 104 - 105 .L-1 (Gerba, 2000a), although concentrations of total norovirus particles of up to 107 .L-1 have been detected by RT-PCR (Lodder et al., 2005). Concentrations in decentralised or onsite sewage systems such as septic tanks are expected to reach a greater maximum than in centralised sewage systems due to lower dilution. Virus removal in onsite sewage systems through sedimentation, biological treatment and disinfection are likely to be in the order of 1 to 3 log10 (Charles, 2007) resulting in high concentrations viruses being released into the environment via effluent disposal systems such as absorption trenches or spray irrigation, potentially contaminating groundwater or surface water drinking water resources.

Virus transport in soil is influenced by a range of factors including soil properties, heterogeneity, flow and water quality (Schijven et al., 2000). The transport of viruses at

Page 214

a particular site will be a function of effluent quality (e.g. pH, ionic strength, organic content, solids content, surfactants), and the soil (e.g. structure, pH, surface charge, organic matter), as well as hydrology and climate factors. Rain events have a significant effect on the transport of viruses from effluent disposal areas. In Australia, unsewered urban areas have been reported to significantly impact surface water quality as a result of rainfall events, increasing microbial loads by a factor of between 18 for Cryptosporidium and 42 for Campylobacter (Roser et al., 2007). This is due to a combination of reduction in effluent disposal system performance through increased hydraulic loading, pathogens are highly mobile in overland runoff across grassed or bare soil surfaces (Ferguson et al., 2007), and rainfall will increase the subsurface transport of pathogens, particularly viruses. Increasing soil saturation reduces the removal of viruses by reducing inactivation and decreasing attachment at the air-water interface (Jin et al., 2000). Increased flow velocity can affect the contact of viruses with the attachment sites, with increasing velocities reducing contact time and therefore attachment (Schijven, 2001). Matrices of higher ionic strengths, such as septic tank effluent, favour virus adsorption. However, the low ionic strength of rainfall can remobilise attached viruses (Schijven, 2001).

The aim of this project was to assess the potential for the transport of effluent, and in particular viruses, from onsite sewage systems. Virus transport is commonly quantified in column experiments using repacked soil (Schijven et al., 2002a; Van Cuyk et al., 2004), with results generally considered to be representative of field-scale. Repacked soil columns provide a relatively easy and rapid method of for quantifying transport, which enables easy comparison between soils. However, due to considerations of heterogeneity there are limitations on the extrapolation to the field-scale (Schijven, 2001). For this project, repacked soil column experiments were supplemented with field experiments to enable greater comparison between soil types, and improve the extrapolation of results to the field-scale. Laboratory experiments used rainwater, with rainfall simulations used in the field studies, to study virus transport during the periods of greatest transport.

Materials and Methods The study area was the surface drinking water supply catchment that provides Sydney, a major Australian city, previously described by Charles et al. (2003a). The inland

Page 215

catchment covers 16,000 km2 of mixed land-uses, with an average annual rainfall ranging from 650 mm to 1,665 mm, and regional lithology that comprises metamorphic (21%), quartz sandstone (16%), granite (11%) and lithic sandstone (9%). Five sites were included in the study. Two of which were selected for field studies including routine analyses over 12 months, and a one-off inoculation with PRD1 and rhodamine. Soils from all five sites were used in the laboratory studies.

Bacteriophages PRD1 and MS2 were used as model viruses to simulate the fate and transport of human enteric viruses that may be found in septic tank effluent (Ferguson et al., 2003). The double-stranded DNA bacteriophage PRD1, approximately 63 nm in size and with an isoelectric point (pI) of 4.2, was considered similar to adenoviruses (double-stranded DNA, 60-80 nm in size) and rotaviruses (70 nm in size) in terms of transport characteristics. PRD1 was employed in both laboratory and field experiments due to its stability at ambient field temperatures (Schijven, 2001). The F-specific coliphage MS2, approximately 26 nm in size and with a pI of 3.9, has similarities with Hepatitis A virus (27 nm, pI 2.8), noroviruses (25 nm) and reoviruses (75 nm, pI 3.9) (Schijven et al., 2000). MS2 is less robust than PRD1 at temperatures above 15 °C but is considered more mobile in groundwater due to its smaller size, hence it was used in column experiments for comparison with PRD1 transport. Stock suspensions of PRD1 and MS2 were prepared as follows: PRD1 and MS2 were grown up in host cells of Salmonella typhimurium strain LT2 (ATCC 19585), and Salmonella typhimurium strain WG49 (ISO 10705-2, 2000), respectively. The stock virus suspensions were diluted in sterile deionised water to give approximately 109 virions.mL-1.

Laboratory experiments Soils were collected from depths of 30-60 cm below the surface, approximately the depth of subsurface wastewater disposal systems, and designated Soil 1 to 5. The soils were dry sieved through a 1.2 mm (pore size) screen to remove rocks and roots, enabling greater uniformity in the soil column packing to be achieved. Soils were analysed using standard methods (APHA, 1998) for pH, grain size distribution, carbon content, total and exchangeable iron, bulk density and cation exchange capacity.

Page 216

Inactivation experiments Seventy portions of each of the sieved soils (2 g) were weighed into 5 mL polyethylene vials, and an additional 70 vials were used without soil but with 2 mL of ARW added. Stock suspensions of PRD1 or MS2 were diluted in sterile deionised water such that 0.1 mL of the diluted bacteriophage suspension could be added to achieve approximate 1 x 106 virions per vial. A predetermined volume of sterilised (autoclaved) ARW was added to the soil microcosms to saturate each of the soils as equally as possible. A number of control vials of each soil type were also left uninoculated to be used for moisture determinations. To these vials, 0.1 mL of sterile deionised water was added in place of the inoculum.

Salt solutions (approximately 250 mL) containing 0.08 M NaCl were placed into the bottom of sealable jars (capacity approximately 2 L) to simulate soil matrix potential of field capacity. Wire mesh discs were used to elevate the vials above the level of the salt solution. The vials, with caps loosened, were placed in airtight sealed jars, which were incubated at 20 °C in the dark. All five soils were inoculated with PRD1 but only the microcosms for ARW and Soils 1 to 3 were inoculated with MS2. Three replicate inoculated vials for PRD1 or MS2 were withdrawn periodically from each sealed jar for determination of infective PRD1 or MS2. The microcosms were sampled destructively. In addition, duplicate vials were removed periodically from each jar for percent moisture determination by drying in pre-weighed crucibles at 105 °C for 48 hours (APHA, 1998).

Column experiments Six columns were constructed from 10.5 cm PVC pipe (internal diameter). The inside of the columns were roughened. Silicone tubing (Masterflex HV-96410-16 Precision Silicone [platinum] Tubing, internal diameter 3.1 mm) was fitted to the brass tube, and sealed with silicone. The base of the soil was supported with three layers of wire mesh (pore size 2 mm) overlayed with a 2 cm layer of 2 mm glass beads. Soil slurry was added to a depth of 50 cm during up-flow conditions with regular stirring and shaking. The upper interface of the soil and the column was sealed with silicone sealant and a ring of aluminium foil to limit short-circuiting of water. One column was used as a control with the supporting layers of wire mesh and beads, but no soil.

Page 217

Artificial rainwater (ARW) (Davies et al., 2006) was used for the experiments (.L-1 of deionised water: 4.07 mg NaNO3, 3.24 mg NaCl, 0.35 mg KCl, 1.65 mg CaCl2.2H2O,

2.98 mg MgSO4.7H2O, 3.41 mg NH4.2SO4). Several pore volumes of ARW were fed (downwards) through the column prior to experimentation to equilibrate the column. Saturated flow conditions were used. Flow rates were representative of the field hydraulic capacity, or the maximum of the column hydraulic conductivity. Sodium chloride concentration of the ARW was increased (by 2.9 g.L-1 for the control and Soils 1 and 4, and by 1.5 g.L-1 for Soils 2, 3 and 5) for several pore volumes to enable determination of flow dispersion. The electrical conductivity (EC) was analysed with a Lab Analyser 440 (TPS Pty Ltd, Australia). After completion of the salt tracer, the ARW was inoculated with PRD1 (all columns) and MS2 (control and Soils 1-3), fed through the column, and the phage concentrations in column effluent were quantified periodically during and after the application of the inoculum to determine the removal of viruses by passage through soil. A minimum of triplicate samples were collected to determine the background and final effluent phage concentrations. Column experiments were undertaken at ambient temperature (approximately 20ºC).

Enumeration of bacteriophages Each 2 g of inoculated soil from sampled vials was washed into a 50-mL Falcon tube using 20 mL 3% (w/v) beef extract solution (Straub et al., 1992) (pH 9) and vortexing. The soil slurry was then vortexed for 2 mins and shaken for 30 mins. After further vortexing for a few seconds the slurries were centrifuged for 15 mins at 2500 g, after which 1 mL of the supernatant was withdrawn by pipette. This was diluted serially in sterile deionised water and assayed by the double agar layer technique (Adams, 1959). Concentrations of phages were expressed as plaque-forming units (pfu) per vial.

A recovery control was prepared for each soil type by freshly inoculating 2 g of soil with approximately 1 x 106 PRD1 or MS2 virions from a stock suspension. The phage was allowed to adsorb to the soil by storing at room temperature for 2 hours, before being processed as described above. The titre of the stock suspension used to inoculate the recovery controls was also determined. Percent recovery for each soil type was determined as the concentration of phage recovered divided by the concentration inoculated into the soil x 100. Phage concentrations in ARW were measured by directly

Page 218

diluting the contents of the microcosm vials with deionised water, and assaying by the double agar layer technique as above. Recovery was not used to adjust data.

Decommissioning of soil columns When PRD1 and/or MS2 were no longer detected in the column effluent, the columns were sectioned and sampled to assess the porosity and packing of the column (all columns), and the phage concentrations through the column (Soils 1, 2, 3 and 5) were determined. To enable this, any pooled water was pumped off the top of the column. Silicone sealant was removed from the surface of the column and the cap was removed. The column was inverted, and the soil allowed to slide out of the PVC pipe. The soil was cut at various depths as it was extracted to allow for sampling at 5-6 depths. To facilitate extraction of the clayey Soils (3 and 5), the PVC column was sawn off at the required lengths. Soil was extracted from the columns in triplicate 2 g portions for phage analyses, and in triplicate 5 g portions for soil moisture analysis.

Field experiments

Set up Two households (site 3 and site 5) within the drinking water catchment were used as experimental sites. Participation in the study was voluntary. Monitoring wells were installed in July/August 2002 up-gradient of the absorption trench, as a control, and along a transect down-gradient at approximately 1m, 2m, 5m, 10m and 20m from the disposal area (see Supporting Information). Additional wells were located at the property boundary. Multiple soil samples were taken from three wells per site, classified for texture and analysed for electrical conductivity (1:5 water), phosphorus (P) content, Bray exchangeable phosphorus, pH (water), cation exchange capacity, exchange capacities for aluminium, calcium, magnesium, potassium and sodium, carbon content and grain size sieve analysis using standard methods (APHA, 1998). Surface infiltration rates and saturated hydraulic conductivities were measured using a double ring falling head permeameter and constant head permeameter, respectively. Online monitoring equipment was installed to monitor the water depth, pH, conductivity and temperature (Greenspan CDTP300, Tyco) in the boreholes and septic tanks.

Page 219

Sample collection Septic tank effluent and groundwater were sampled fortnightly initially, using a peristaltic pump and clean, sterilised tubing (vinyl and Teflon, ISCO). Indoor and outdoor water usage was recorded at each sampling visit. Septic tank effluent was sampled from the tank outlet. Water depth was measured prior to the borehole being purged. Water was then allowed to return to the boreholes before they were sampled. Specific conductivity, temperature and pH were measured onsite. Samples were analysed for alkalinity; chloride; sulphate; dissolved calcium, magnesium and sodium; total, oxidised and Kjeldahl nitrogen; ortho- and total phosphate, thermotolerant coliforms, enterococci, Clostridium perfringens and somatic phage by standard methods (APHA, 1998).

Rainfall simulation In order to simulate rainfall conditions, circular sprinklers were installed to supply microfiltered, UV and ozone disinfected reclaimed sewage effluent over a 10m by 10m area spanning 5m either side of the central boreholes and 10m along the down-gradient transect, starting up-slope of the absorption trench. Instrumentation enabled water supply to be regulated by soil moisture at a depth of 1m. The rainfall simulation was run for 14 days to achieve the required soil saturation prior to addition of tracers to the septic tank, during which period background concentrations were analysed. The tracers included conservative tracer Rhodamine WT and viral tracer bacteriophage PRD1. Simultaneous to the inoculation of the septic tank, phage inocula controls were buried 5 cm below the soil surface. These controls were sampled periodically throughout the field sampling to assess the contribution of temperature related decay. Samples of septic tank effluent and groundwater were taken over three months and analysed for the tracers and routine analytes listed above. Rhodamine was assayed in the field by hand held fluorometer. PRD1 samples were stored at 4 °C for transport to the laboratory, and analysed within 24 hours based on the F-specific Bacteriophage method (Havelaar et al., 1984).

Quality control included blank and duplicate samples, and was undertaken on routine, rhodamine and phage samples.

Page 220

Modelling and parameter estimation Analysis was undertaken to determine the phage inactivation rates for use in modelling fate and transport of viruses, using linear and biphasic inactivation models. Linear inactivation was modelled as C = C0.Exp(-µt) (Equation 1), where C is the concentration of phages, C0 is the initial phage concentration, t is time, and µ is the inactivation rate coefficient. Biphasic inactivation was modelled as C = C0 [α.Exp(-µ1.t)

+ (1-α).Exp(-µ2.t)] (Equation 2), α is the susceptible fraction of viruses, and µ1 and µ2 are inactivation rates for the different fractions (Petterson et al., 2001c). A log- likelihood method was used to fit the inactivation rate models to the data and compare the linear and biphasic models previously described by Schijven et al. (2002b). Outliers were confirmed using SPSS (version 11.5.2.1, SPSS Inc., 2003) based on the residuals from the linear or biphasic models.

Salt breakthrough curves were fitted to the convection-dispersion solute transport equation using HYDRUS-1D (Šimùnek et al., 1998) to calculate porosity dispersion within the columns. The fit was optimised using the Levenberg – Marquardt non-linear minimisation algorithm (Marquardt, 1963) for least-squares solutions. The relative concentrations from the phage breakthrough curves (C/C0) were fitted (using log resident concentrations) to the one-site kinetic adsorption equations (Schijven, 2001), also using HYDRUS-1D. For modelling purposes, µl was assumed to equal the inactivation in ARW, and µs was assumed to equal inactivation in the soil-water microcosm.

Summary statistics for the routine field water quality data were prepared using SPSS (version 11.5.2.1, SPSS Inc., 2003). Removal was calculated based on the difference between effluent and borehole pairs. Borehole water depth data and rhodamine breakthrough data were fitted to the convection-dispersion equation using HYDRUS-2D (Šimùnek et al., 1999) to calculate porosity and dispersivity. Virus transport coefficients were fitted from the breakthrough curve to the one-site kinetic adsorption equations, and were compared with the predictions from the parameters derived from the column experiments.

Page 221

Results The texture of the soils used for laboratory experiments ranged from sandy loam (Soils 1 and 4), to loam (Soil 5) and clay loam (Soils 2 and 3) (see Supporting Information). The recoveries in the inactivation experiments ranged from 40 % to 61 % for MS2 and 56 % to 80 % for PRD1, and were not affected by time. Percent recoveries were not used to adjust the phage concentrations. The inactivation rates for PRD1 were significantly lower than those for MS2 in Soils 1 and 2, but not significantly different in ARW and Soil 3 based on the log-likelihood ratios test at the χ2 95th percentile. Virus inactivation in soil/water matrices was significantly greater than inactivation in ARW alone, which may relate to the soil biota that were present in the soils sampled from field sites but absent from the sterile ARW. This biotic effect was further supported by the presence of bacterial growth on the double agar layer plates for the PRD1 assay for decommissioned soil column samples and for microcosm soils towards the end of the experiment when lower dilutions were required. In particular, the final sample (t = 110 days) for Soil 3 was affected by bacterial growth, and was excluded from analyses (obscured by background bacterial growth). PRD1 inactivation in Soils 3, 4 and 5 were not significantly different (see Supporting Information). The biphasic model provided a better fit for MS2 in all microcosms and for PRD1 in the ARW and Soil 4 microcosms. The lower inactivation rate coefficient from the binomial model was 93 % of the linear inactivation rate (R2 = 0.997) for the microcosms where the biphasic model was assumed.

Column porosity, including pore space in column design, and dispersivity were calculated from the salt tracer breakthrough, and used to model the virus breakthrough curve data. For these analyses, µl was assumed to be equal to the linear inactivation coefficient from the ARW microcosm. The soil-ARW microcosms inactivation coefficients were used for the inactivation rate of viruses attached to soils, µs. Phage inocula were 104.mL-1 PRD1 and 102 to 104.mL-1 MS2 and the total load recovered of inoculated phage after transport through the columns was calculated. The breakthrough concentrations were generally highest in sandy soils (soil 1 and 4) and for PRD1 (Table

1). The log10 reduction in concentration was calculated as the log10 of the maximum effluent concentration divided by the maximum influent concentration. Breakthrough was most rapid in sandy soils.

Page 222

Table 1 Column phage tracer experiments: Flow rates and log10 reductions in phage concentrations Soil Flow Water Log10 reduction in Pore volumes to Content ± concentration continuous breakthrough SD

m/d PRD1 MS2 PRD1 MS2

ARW control 1.8 - -0.21 - - -

Soil 1 2.1 0.19 ± 0.03 -0.06 0.14 0.16 0.24

Soil 2 0.4 0.38 ± 0.07 1.94 1.32 27.4 29.7

Soil 3 0.05 0.39 ± 0.06 1.06 1.21 0.99 0.28

Soil 4 0.8 0.21 a 0.23 - - -

Soil 5 0.2 0.31 ± 0.03 3.00 - >70 - a Insufficient samples to calculate standard deviation. ARW – artificial rain water

Rapid breakthrough of MS2 in Soil 3 may indicate short circuiting. Full breakthrough curves were not achieved in all soils (Figure 1, Figure 2). High phage retention in Soils 2 & 5 limited breakthrough. Breakthrough was modelled using HYDRUS-1D (Table 2).

The standard errors for katt and kdet were high, which should be considered in the application of the results to the field scale. Inactivation was observed in the phage inocula and was included in the breakthrough curve modelling. For Soil 3, the inactivation of attached phage, µs, was found to be different from that predicted by the microcosms.

Page 223

(a) (d)

1.E+00 1.E+00

1.E-01 1.E-01

1.E-02 1.E-02 1.E-03 Phage Concentration C/Co 00.511.522.533.5 1.E-03

Time (days) Phage Concentration C/Co 0 0.2 0.4 0.6 0.8 1 Time (days)

(b) (e)

1.E+01 1.E+01

1.E+00 1.E-01

1.E-01 1.E-03

1.E-02 1.E-05 1.E-03 Phage Breakthrough C/Co 0 102030405060 Time (days) 1.E-04

Phage Breakthrough C/Co 1.E-05 0 5 10 15 20 25 Time (days)

(c)

1.E+00

1.E-01

1.E-02

1.E-03

1.E-04

Phage Breakthrough C/Co Phage 1.E-05 0 1020304050 Time (days)

Figure 1 PRD1 bacteriophage breakthrough curve data (circles), modelled influent concentration of PRD1 (grey line) and the fitted one-site kinetic model (solid line) (Schijven, 2001) (a) Soil 1; (b) Soil 2; (c) Soil 3; (d) Soil 4; and (e) Soil 5

Page 224

(a) (c)

1.E+00 1.E+00

1.E-01 1.E-01 1.E-02

1.E-02 1.E-03

1.E-04 1.E-03

Phage Concentration C/Co 1.E-05

00.511.52Phage Breakthrough C/Co Time (days) 0204060 Time (days) (b)

1.E+01

1.E+00

1.E-01

1.E-02

1.E-03 Phage Breakthrough C/Co Breakthrough Phage

1.E-04 0 5 10 15 20 25 Time (days)

Figure 2 MS2 bacteriophage breakthrough curve data (circles), modelled influent concentration of MS2 (grey line) and the fitted one-site kinetic model (solid line) (Schijven, 2001) (a) Soil 1; (b) Soil 2; and (c) Soil 3

Page 225

Table 2 One-site kinetics parameters from fitting column breakthrough curves (± standard error) 2 Soil Phage Dispersivity µs katt kdet R

[m] [.d-1] [.d-1] [.d-1]

1 MS2 0.149 ± 0.08 0.246 1.32 ± 2.54 2.29 ± 5.90 0.694

1 PRD1 0.05 ± 0.083 0.124 7.30 ± 5.87 1.64 ±1.06 0.437

3 MS2 0.066 ± 0.138 ± 0.71 ± 0.13 0.0067 ± 0.687 0.023 0.057 0.0092

3 PRD1 0.066 * 0.129 ± 6.40 ± 2.03 0.82± 0.44 0.504 0.023

* Dispersivity for PRD1 assumed to equal that for MS2.

Destructive sampling of the columns after the study identified uniform soil packing based on visual inspection, with a general decline in water content with depth (see Supporting Information). There was greater retention and survival of PRD1 than MS2 within the columns (Figure 3). Soil 1, sampled 27 days after discontinuation of the phage inoculum, had final effluent concentrations of 29 pfu.mL-1 PRD1 and below detection for MS2. The greatest retention of PRD1 was on the surface of column for Soils 1 & 5, indicating that straining may have had a role in virus retention (Bradford et al., 2003). Soil 2 had a final effluent concentration of 46 pfu.mL-1 PRD1 and 6 pfu.mL-1 MS2, and due to the high retention in the column was still being spiked immediately prior to decommissioning. The 50 cm sample from this column is potentially cross- contaminated due to the difficulty extracting it. Soil 3 (decommissioned 28 days after spike stopped) had a low final MS2 concentration (2 pfu.10mL-1) and minimal retention of MS2 within the column. PRD1 retention for Soil 3 (final concentration 29 pfu.mL-1) was only slightly lower than for Soil 5 which had limited breakthrough and a final concentration 1 pfu.mL-1 (inoculum applied right up to column decommissioning). The low concentration of phage at depth in Soil 5 indicates that there may be a virus sorption

Page 226

front between 30 cm and 40 cm, and therefore breakthrough would not occur until the front had reached the bottom of the column.

(a) (c)

0 0 10 5 10 20 15 30 20 25 40 30 Column depth(cm) Column depth Column (cm) 50 35 40 -1 0 1 2 -10123 Concentration (log10 per gram dry weight) Concentration (log10 per gram dry weight)

(b) (d)

0 0

10 10

20 20

30 30

Column Depth (cm) 40 40 Column Depth (cm) 50 50 -101234 -10123 Concentration (Log10 per gram dry weight) Concentration (Log10 per gram dry weight)

Figure 3 PRD1 (•) and MS2 (∆) concentrations with depth in decommissioned columns (a) soil 1; (b) soil 2; (c) soil 3 and (d) soil 5

At the field sites, septic tanks were pumped out prior to the rainfall simulation. Soil depth at each of the experimental sites was limited, as is the case in much of the catchments, with average well depths of 1.5 m at each site (Site 3 range: 0.8 m to 3.6 m; Site 5: 0.7 m to 3.0 m). At Site 3, pH generally increased with distance down gradient and depth, from 6.3 to 8.7. There was minimal variation in pH (range 5.2 – 5.6) across Site 5. Total iron varied from 8,200 to 46,000 mg.kg-1 and aluminium exchange capacity from 7.2 to 35.6 cmol.kg-1. Borehole 1 (BH1) was the deepest at 3.0 m, with

Page 227

increased sand content below 2 m. Routine groundwater (and effluent) samples were limited to ten occasions at Site 3 and nine occasions at Site 5 due to drought (Table 3). At Site 3, temperature decreased from 19.6 °C (± 0.02 SD) to 11.3 °C (± 0.06 SD), and pH increased from 6.5 ± 0.2 in the septic tank to 6.9 ± 0.1 in BH2 to 9.5 ± 0.4 in BH4. The average indoor household water usage over this period was 540 L.d-1. The rainfall simulation contributed an average of 3.3 mm.d-1 to the absorption trench and the area immediately down slope (~440 m2). PRD1 concentrations in the septic tank at site 3 prior to inoculation were 1.5 x 102 pfu.mL-1 and below detection in groundwater. The septic tank inoculation achieved concentrations of 9.7 mg.L-1 Rhodamine and 4.1 x 106 pfu.mL-1 PRD1. Rhodamine breakthrough into the groundwater was limited (see Supporting Information). PRD1 breakthrough in groundwater was rapid (Figure 4), although at a 5 log10 reduction. At Site 5, temperature and pH were consistent at 21.4 °C (± 0.8 SD) and 7.4 ± 0.3, respectively. PRD1 was not detected in the septic tank, dam (see Supporting Information) or groundwater prior to inoculation. Initial attempts to inoculate the septic tank reverted to below detection within 24 hours. Survival studies indicated that overnight inactivation was insignificant. A second series of inoculation, daily doses of one litre of 109 PRD1 pfu.mL-1 for three days, resulted in a more sustained concentration in the septic tank of 107 pfu.mL-1. However, no breakthrough of rhodamine nor PRD1 were detected in groundwater over 49 days. PRD1 was detected in the dam on three occasions, and while concentrations of the phage were low (<10 pfu.mL-1) they were confirmed in duplicate samples.

Page 228

Table 3 Field results (average ± SD) for nutrient and microbial concentrations across the field sites from routine sampling and analysis (Site 3 N = 10, Site 5 N=9, except where * only 2 samples). ND = None detected over all sampling events. Where microbes were detected once or more the half the limit of detection was used for non detects in calculations. Analyte Units Site Septic Tank 1m 5m 10m 20m Total Nitrogen mg.L-1 3 123 ± 33 45 ± 21 57 ± 34 19 ± 4.1 0.7 ± 0.7 5 156 ± 155 40 ± 25 * 19 ± 10 22 ± 11 10 ± 11 Total mg.L-1 3 22 ± 9.8 0.2 ± 0.3 0.1 ± 0.2 0.02 ± 0 0.2 ± 0.1 Phosphorus 5 45 ± 62 0.9 ± 0.9 * 0.4 ± 0.2 0.4 ± 0.3 1.5 ± 1.0 Thermotolerant log10 3 5.0 ± 1.3 1.3 ± 1.1 0.3 ± 0.3 0.3 ± 0.5 0.4 ± 0.6 -1 Coliforms cfu.100mL 5 5.6 ± 0.4 ND ND 0.3 ± 0.5 0.5 ± 0.6 Clostridium log10 3 4.0 ± 0.6 0.4 ± 0.9 ND ND ND -1 perfringens cfu.100mL 5 4.3 ± 1.4 ND ND 0.4 ± 0.8 0.3 ± 0.5 spores Somatic Phages log10 3 5.0 ± 1.1 ± 1.0 0.7 ± 0.1 ND ND -1 pfu.100mL 5 4.6 ± 0.6 ND ND ND 1.3 ± 0.6

1.E+07

1.E+05

1.E+03

PRD1 concentration (pfu/ml) concentration PRD1 1.E+01 0 1020304050607080 Time (days)

Figure 4 PRD1 breakthrough at Site 3 in the septic tank effluent (z) and 1 m downslope of the absorption trench at BH2 ({).

Discussion Virus inactivation is usually considered to be insignificant in soil column transport experiments (Schijven, 2001) due to the small time scales involved. However, as breakthrough was monitored for up to six weeks in the test columns, phage inactivation was considered to be significantly influential. The inactivation rates for PRD1 in Soils 3 and 4 (0.061 d-1) were similar to those reported for a loam soil from the same -1 geographical area (0.053 d [converted from log10 to ln]) (Davies et al., 2006).

Page 229

However Soil 5, classified as a loam, which was sampled from a lower depth on the same site as the clay loam used in the study reported by Davies et al. (2006), had a -1 -1 lower inactivation rate for PRD1 (0.083 d compared to 0.16 d [converted from log10 to ln]). The inactivation in sandy loams was comparable to that reported by Blanc and Nasser (1996) at 23 ºC in a sand-groundwater matrix and in a loamy sand-groundwater matrix. For PRD1, inactivation was slightly lower on average that that suggested by the temperature dependent model suggested by Harvey and Ryan (2004). Similarly for MS2, a model proposed by Gerba et al. (1991) predicted slightly higher inactivation on average.

Inactivation in soil-water microcosms was assumed to represent phage inactivation in the attached phase, µs. However, the method does not differentiate between attached and free viruses. For Soils 2, 3 and 5, which had higher rates of removal in the columns, it can be assumed that there were higher rates of attachment in the soils and therefore that the majority of the phages within the microcosms were attached. Soils 1 & 4 had lower rates of removal within the columns and therefore it is likely they had lower rates of attachment within the microcosms. However, quartz sand has been reported to have a virus sorption capacity of 2.2 x 1012 pfu.kg-1 (Moore et al., 1981), which far exceeds the initial virus load applied (106 pfu to 1.3 g-1 soil dry weight); indicating other factors were also important. Models of virus transport in the environment typically include only linear inactivation. Furthermore the inactivation rate coefficients are commonly derived over short time periods, such that initial high rates of inactivation can overestimate the longer term virus decay. The similarities between the linear inactivation rate coefficient and the minimum biphasic model inactivation rate coefficient suggest that the microcosms were sampled over sufficient time to minimise this effect, and that the fitting of ln C0 simultaneously with µ provides a better estimate of long-term inactivation. However, it is recommended that the minimum biphasic inactivation rate be used where conservative estimates of long-term virus survival are required.

Previous studies of virus transport in sand have found similarly rapid transport of viruses (Powelson et al., 1994; Schijven et al., 2002b). Of the sandy loams, the greatest removal was in Soil 4, which was the finer of the two soils. Soil 4 had higher aluminium exchange capacity and lower pH, which have been correlated with virus adsorption (Goyal et al., 1979). The clay loam and loam soils had much greater removal of phage

Page 230

than the sandy soils, with high variability between them. Clayey soils have been reported to have high absorption capacities through batch experiments (Goyal et al., 1979). However, limited transport studies have been undertaken. Of the natural clayey soils, the order of removal was Soil 5 (loam) > Soil 2 (clay loam) > Soil 3 (clay loam). Similarly to the sandy loams, this order relates to the order of aluminium exchange capacity and pH. Soils 2 & 3 resulted in very different breakthrough times, with phage in Soil 2 (at a higher flow rate than in Soil 3) not breaking through until more than 25 pore volumes after Soil 3, but with only 1 Log10 reduction in load recovery in Soil 2 (noting 0.4 % recovery for PRD1 and MS2) compared to Soil 3 (3.3 % recovery PRD1, 1.7 % for MS2). MS2 removal in the columns was greater than PRD1 in Soils 1 & 3, but less than PRD1 in Soil 2. This may be due to a number of factors: pH was lowest in Soil 2, and aluminium exchange capacity and total iron were highest in Soil 2. The low pH may have had a greater effect on MS2 due to its lower pI.

Rapid transport of viruses in soil has been attributed to the formation finger-flow pathways in water repellent sand (i.e. fingers of saturated flow conditions) and to physical heterogeneities that enable bypass flow through macropores and fractures in the soil (2001). Chemical and biological heterogeneity can also affect virus adsorption and transport (Rehmann et al., 1999). The use of sieved, disturbed soil columns reduces the impact of heterogeneity, and therefore, introduces several limitations to the direct application of the results to the field-scale. Sieving the soil increased the soil homogeneity, limiting the effects of soil structure and reducing hydraulic conductivity. Hence, repacked column experiments were expected to overestimate virus removal due to increased adsorption from greater soil-virus interactions, and these limitations should be considered when extrapolating results to field scale. The experimental design aimed to minimise any preferential flow paths along the soil-column interface, however, the decline in moisture content in the columns may indicate short-circuiting occurred, and hence, not all the adsorption capacity of the soil was utilised. This may account for why there was no particular relationship evident between soil properties and phage removal. Furthermore, the physicochemical heterogeneity can result in apparent reductions in removal with distance (Schijven et al., 2000). This heterogeneity may be accentuated by the injection or infiltration of sewage into soil through development of an organic matter gradient. Other potential causes of non-linear removal are variability between

Page 231

subpopulations of viruses (e.g. variability in surface charge/coat); or colloid assisted transport. The non-linear relationship is expected to affect the scaling up of results from the column to the field-scale, however, this is expected to be mostly addressed by consideration of site heterogeneity.

The breakthrough curves differed between different columns. This may in part be explained by different durations of phage application, e.g. see Figures 1 and 2, and the decay during the latter. The differences in µs between Soil 3 microcosm and column experiments may be due to the increased shear forces exerted on the phage by the flow (Schijven, 2001).

While the experiment aimed to address virus fate and transport during land application of effluent from an onsite sewage system, artificial rainwater was used in the experiment instead of sewage. This provided not only a water of uniform constitution, but probably also a worse-case scenario for virus fate and transport based on the assumption that lower ionic strength increases transport more than organic constituents in sewage. In addition to the differences in heterogeneity discussed above, two differences between these column experiments and the onsite effluent disposal field scenario that affect the up-scaling were:

• Rainwater, as used in the columns, resulted in decreased attachment compared to sewage due to lower ionic strength resulting reducing attachment. However, colloids and surfactants in the sewage may also facilitate virus transport and survival (Schijven et al., 2000; Jin et al., 2002); and

• Lack of microbial activity in the ARW microcosms may have led to underestimation of virus inactivation in the field.

There were several limitations with the field experiments, including difficulties obtaining sufficient sites, interference from domestic animals and variable hydro- geological conditions. Nonetheless, field experiments aid in overcoming these upscaling issues. The results from the column experiments were compared with those of the field experiments, with the aim of extrapolating to virus transport in soils at a catchment scale for application to the delineation of setback distances for septic tanks. The removal of PRD1 in the homogenous repacked soil columns (1 log10 reduction over the

Page 232

0.5 m column) would overestimate the reduction experienced in the field (3.8 log10 reduction in 3 m and >6.6 log10 in 7 m), despite the use of conditions conducive to virus transport in the column experiments. The decreased removal in the field was assumed to be due to a combination of heterogeneity, the effluent quality, and the preconditioning of the site with effluent and viruses. Factors such as dilution, dispersion, organic matter and a biomat in the absorption trench, which would increase virus reduction compared to the column experiment, were apparently less significant than the factors that increased transport listed above.

At Site 5, there was no apparent relationship between nitrogen and microbial contamination (Table 3). Due to the lack of virus transport this was assumed to suggest microbial contamination from other sources such as animals, the dam or neighbouring septic tanks, although it is possible that effluent bypassed the first two boreholes. At this location the water flow was through a shale layer overlain by a loam, it was the clay loam that was studied in the column experiments. The low pH of the loam (4.7 in the column, 5.3 in the field) will have contributed to the high removal of phage during transport through the column, and restricted vertical movement in the field from the absorption trench to the shale layer below. Furthermore, Davies et al. (2006) reported high rates of inactivation of PRD1 in soil from the same site.

The potential for the transport of effluent from onsite sewage systems in the Sydney drinking water catchment area varies greatly between the different soils in the catchments. Contaminants are known to be mobile in the more sandy soils, as was experienced in the current experiments with viruses in Soils 1 & 4. In the soils with higher clay contents, despite the similarity in the patterns of nutrient transport across the two field sites, there was high variability in the behaviour of virus fate and transport in these soils with greater removal of viruses in Soil 5 than in Soil 3. While the clayey soils afforded a high degree of removal in column experiments than the sandy soils, the results for Soil 3 highlighted that significant virus transport in clays is possible, especially considering low rate of inactivation in this soil.

Further modelling work will be undertaken to extrapolate these results to the catchment scale, with careful consideration of the limitations, including the potential for repacked

Page 233

columns to overestimate the degree of virus removal, and the recognition of the high variability between soils.

References Adams, M. H. (1959). Bacteriophages. New York, Interscience Publishers Inc.

APHA (1998). Standard Methods for the Examination of Water and Wastewater. Washington D.C., American Public Health Association.

Blanc, R. and A. Nasser (1996). "Effect of effluent quality and temperature on the persistence of viruses in soil." Water Science and Technology 33(10-11): 237-242.

Bradford, S. A., J. Šimùnek, M. Bettehar, M. T. van Genuchten and S. R. Yates (2003). "Modelling colloid attachment, straining, and exclusion in saturated porous media." Environmental Science and Technology 37: 2242-2250.

Charles, K., D. Roser, N. Ashbolt, D. Deere and R. McGuinness (2003). "Buffer distances for on-site sewage systems in Sydney's drinking water catchments." Water Science and Technology 47(7-8): 183-189.

Charles, K. J. (2007). Quantitative Microbial Risk Assessment: a catchment management tool to delineate buffer distances for on-site sewage treatment and disposal systems in Sydney’s drinking water catchments. Civil and Environmental Engineering, University of New South Wales. PhD.

Davies, C. M., M. R. Logan, V. J. Rothwell, M. Krogh, C. M. Ferguson, K. Charles, D. A. Deere and N. J. Ashbolt (2006). "Soil inactivation of DNA viruses in septic seepage." J Appl Microbiol 100(2): 365-74.

Ferguson, C., C. M. Davies, C. Kaucner, M. Krogh, J. Rodehutskors, D. Deere and N. J. Ashbolt (2007). "Field scale quantification of microbial transport from bovine faeces under simulated rainfall events." Journal of Water and Health 5(1): 83-95.

Ferguson, C. M., A. M. de Roda Husman, N. Altavilla, D. Deere and N. J. Ashbolt (2003). "Fate and transport of surface water pathogens in watersheds." Critical Reviews in Environmental Science and Technology 33(3): 299-361.

Page 234

Gerba, C. (2000). Domestic wastes and waste treatment. Environmental Microbiology. R. M. Maier, I. L. Pepper and C. P. Gerba. San Diego, Academic Press: 505-534.

Gerba, C. P., M. V. Yates and S. R. Yates (1991). Quantitation of Factors Controlling Viral and Bacterial Transport in the Subsurface. Modelling the Environmental Fate of Microorganisms. C. J. Hurst. Washington DC, USA, American Society for Microbiology: 77-88.

Goyal, S. M. and C. P. Gerba (1979). "Comparative adsorption of human enteroviruses, simian rotavirus, and selected bacteriophages to soils." Applied and Environmental Microbiology 38(2): 241-247.

Harvey, R. W. and J. N. Ryan (2004). "Use of PRD1 bacteriophage in groundwater viral transport, inactivation, and attachment studies." FEMS Microbiology Ecology 49: 3-16.

Havelaar, A. H. and W. M. Hogeboom (1984). "A method for the enumeration of male- specific bacteriophages in sewage." Journal of Applied Bacteriology 56: 439-447.

ISO 10705-2 (2000). Water quality - Detection and enumeration of bacteriophages. Part 2: Enumeration of somatic coliphages. Geneva, International Organization for Standardization.

Jin, Y., Y. J. Chu and Y. S. Li (2000). "Virus removal and transport in saturated and unsaturated sand columns." Journal of Contaminant Hydrology 43(2): 111-128.

Jin, Y. and M. Flury (2002). "Fate and transport of viruses in porous media." Advances in Agronomy 77: 39-102.

Lodder, W. J. and A. M. de Roda Husman (2005). "Presence of noroviruses and other enteric viruses in sewage and surface waters in The Netherlands." Applied and Environmental Microbiology 71(3): 1453-1461.

Marquardt, D. W. (1963). "An algorithm for least-squares estimation of nonlinear parameters." Journal of the Society for Industrial and Applied Mathematics 11(431- 441).

McLeod, M., J. Aislabie, J. Smith, R. Fraser, A. Roberts and M. Taylor (2001). "Viral

Page 235

and chemical tracer movement through contrasting soils." Journal of Environmental Quality 30: 2134-2140.

Moore, R. S., D. H. Taylor, L. S. Sturman, M. M. Reddy and G. W. Fuhs (1981). "Poliovirus adsorption by 34 minerals and soils." Applied and Environmental Microbiology 42(6): 963-975.

Petterson, S. R. and N. J. Ashbolt (2001). "Viral risks associated with wastewater reuse: Modelling virus persistence on wastewater irrigated salad crops." Water Science and Technology 43(12): 23-226.

Powelson, D. K. and C. P. Gerba (1994). "Virus removal from sewage effluents during saturated and unsaturated flow through soil columns." Water Research 28(10): 2175- 2181.

Rehmann, L. L. C., C. Welty and R. W. Harvey (1999). "Stochastic analysis of virus transport in aquifers." Water Resources Research 35(7): 1987-2006.

Roser, D. J. and N. J. Ashbolt (2007). Source Water Quality Assessment and the Management of Pathogens in Surface Catchments and Aquifers. Research Report 29. Bolivar, CRC for Water Quality and Treatment.

Schijven, J. F. (2001). Virus removal from groundwater by soil passage: Modeling, field and laboratory experiments. Delft, University of Delft, The Netherlands.

Schijven, J. F. and S. M. Hassanizadeh (2000). "Removal of viruses by soil passage: Overview of modeling, processes, and parameters." Critical Reviews in Environmental Science and Technology 30(1): 49-127.

Schijven, J. F., S. M. Hassanizadeh and H. A. M. de Bruin (2002). "Column experiments to study nonlinear removal of bacteriophages by passage thorugh saturated dune sand." Journal of Contaminant Hydrology 58: 243-259.

Šimùnek, J., M. Šejna and M. T. van Genuchten (1998). The HYDRUS-1D software package for simulating the one-dimensional movement of water, heat and multiple solutes in variably saturated media. Golden, Colorado, International Ground Water Modelling Centre, Colorado School of Mines.

Page 236

Šimùnek, J., M. Šejna and M. T. van Genuchten (1999). The HYDRUS-2D software package for simulating two-dimensional movement of water, heat and multiple solutes in variably saturated media. Golden, Colorado, International Ground Water Modelling Centre, Colorado School of Mines.

Straub, T. M., I. L. Pepper and C. P. Gerba (1992). "Persistence of viruses in desert soils amended with anaerobically digested sewage sludge." Applied and Environmental Microbiology 58: 636-641.

Van Cuyk, S. and R. Siegrist (2004). Fate of viruses in the infiltrative surface zone of systems that rely on soil treatment for wastewater renovation. On-Site Wastewater Treatment X: Proceedings of the tenth national symposium on individual and community sewage systems. St Joseph, Michigan, American Society for Agricultural Engineers.

Supporting information

Page 237

House

BH1 Septic

Trench

BH2

BH3

Slope

BH4

BH5 Property boundary

BH6 BH7 BH8

Gully/creek

Figure S1 Site 3 layout, indicating the locations of the boreholes (BH 1 - 8)

Page 238

BH1 House Septic

Trench

BH2 Slope

BH3

Approximate Scale 5 m BH4

Dam

BH5

Figure S2 Site 5 layout, indicating the locations of the boreholes (BH 1 - 5)

Page 239

35 2 mm 30 1 mm 25 0.6 mm 20 0.355 mm 15 0.2121 mm 10 0.125 mm

Percent retained (%) 5 0.075 mm 0

1 2 3 4 5 , , , 3 , 5 ld mn mn eld e lu lu Fi Fi Column, Column, Column, Co Co Soil

Figure S3 Soil grain size distribution (percent retained by sieve size)

(a) (c) ) ) -1 1.E+07 -1 1.E+07 1.E+06 1.E+06 1.E+05 1.E+05 1.E+04 1.E+04 1.E+03 1.E+03 1.E+02 1.E+02 1.E+01 1.E+01

Concentration (pfu.mL 1.E+00 Concentration (pfu.mL 1.E+00 0 15 30 45 60 75 90 105 120 135 0 153045607590 Time (d) Time (d)

(b) )

-1 1.E+07 1.E+06 1.E+05 1.E+04 1.E+03 1.E+02 1.E+01

Concentration (pfu.mL 1.E+00 0 153045607590 Time (d)

Page 240

(d) (f) ) ) -1 -1 1.E+07 1.E+07 1.E+06 1.E+06 1.E+05 1.E+05 1.E+04 1.E+04 1.E+03 1.E+03 1.E+02 1.E+02 1.E+01 1.E+01

Concentration (pfu.mLConcentration 1.E+00

Concentration (pfu.mLConcentration 1.E+00 0 153045607590105120 0 153045607590105120 Time (d) Time (d)

(e) )

-1 1.E+07 1.E+06 1.E+05 1.E+04 1.E+03 1.E+02 1.E+01

Concentration (pfu.mL Concentration 1.E+00 0 153045607590105 Time (d)

Figure S4 PRD1 inactivation data (circles) with error bars indicating standard deviation, linear model (continuous line) and biphasic model (dotted line) for (a) Artificial rain water; (b) Soil 1; (c) Soil 2; (d) Soil 3; (e) Soil 4; and (f) Soil 5

Page 241

(a) (c) ) )

-1 1.E+07 -1 1.E+07 1.E+06 1.E+06 1.E+05 1.E+05 1.E+04 1.E+04 1.E+03 1.E+03 1.E+02 1.E+02 1.E+01 1.E+01

Concentration (pfu.mL 1.E+00 Concentration (pfu.mL 1.E+00 0 15 30 45 60 75 90 105 120 135 015 Time (d) Time (d)

(b) (d) ) )

-1 1.E+07 -1 1.E+07 1.E+06 1.E+06 1.E+05 1.E+05 1.E+04 1.E+04 1.E+03 1.E+03 1.E+02 1.E+02 1.E+01 1.E+01

Concentration (pfu.mL 1.E+00 Concentration (pfu.mL 1.E+00 0 15304560 0 153045607590 Time (d) Time (d)

Figure S5 MS2 inactivation data (circles) with error bars indicating standard deviation, fitted linear model (continuous line) and fitted biphasic model (dotted line) for (a) Artificial rain water; (b) Soil 1; (c) Soil 2; and (d) Soil 3

50%

40%

30% Soil 5

Moisture content 20% Soil 3 Soil 2 Soil 1 10% 0 1020304050 Soil depth (cm)

Figure S6 Moisture content profiles in soil columns

Page 242

(a) (c)

0 0 10 5 10 20 15 30 20 25 40 30 Column depth (cm) Column depth Column (cm) 50 35 40 -1 0 1 2 -10123 Concentration (log10 per gram dry weight) Concentration (log10 per gram dry weight)

(b) (d)

0 0

10 10

20 20

30 30

Column Depth Depth (cm) Column 40 40 Column Depth (cm) 50 50 -101234 -10123 Concentration (Log10 per gram dry weight) Concentration (Log10 per gram dry weight)

Figure S7 PRD1 (•) and MS2 (∆) concentrations with depth in decommissioned columns (a) soil 1; (b) soil 2; (c) soil 3 and (d) soil 5

Page 243

1.E+07

1.E+05

1.E+03

PRD1 concentration (pfu/ml) concentration PRD1 1.E+01 0 1020304050607080 Time (days)

Figure S8 PRD1 breakthrough at Site 3 in the septic tank effluent (z) and 1 m downslope of the absorption trench at BH2 ({).

Page 244

Table S1 Soil physicochemical characteristics

Soil 1 2 3 4 5 3F* 5F*

Texture Sandy Clay Clay Sandy Loam Clay Loam Loam Loam Loam Loam Loam d50 (mm) 0.28 <0.075 <0.075 0.20 <0.075 0.66 0.12 pH 5.7 5 6.5 5.4 4.7 7.8 ± 0.7 5.3 ± 0.1

Bulk 1.27 1.08 1.22 1.03 0.86 1.34 ± 0.2 1.23 ± 0.1 Density (g/cm3)

Total Iron 16,000 120,00 54,000 8,400 19,000 26,200 ± (mg.kg-1) 0 11,600

Exchange Capacity (cmol.kg-1)

Aluminium 0 2.5 0 0.1 3.1 0 ± 0 13.8 ± 8.0

Calcium 2.5 3.7 5.4 6.5 0.4 5.4 ± 3.1 0.66 ± 0.4

Cation 2.9 7.9 8.4 8.5 3.9 14.7 ± 5.8 15.3 ± 8.2

Magnesium 0.3 0.7 2.5 1 0.1 7.3 ± 2.9 0.3 ± 0.2

% Carbon 4.3 2.0 1.2 6.6 3.8 0.34 ± 0.2 0.92 ± 0.8 (% dry weight)

* F denotes field samples. Results are average ± standard deviation for depths greater than 0.1 m (N = 13 & N = 10 for Sites 3 & 5, respectively). Except d50 which is for 0.5 m at BH2

Page 245

Table S2 Bacteriophage inactivation rate coefficients and likelihood ratios

Sampl Average Recovery Linear Biphasic model Likelihood ratio test (χ2 ** e moisture model 0.95) **

(%) (%) µ α µ1 µ2 Likelihood Preferred (day-1) (day-1) (day-1) ratio model Bacteriophage MS2 ARW - - 0.0375 0.718 1.69 0.0322 4.9* Biphasic Soil 1 28.7±0.9 58±21 0.246 0.919 0.853 0.214 19.4* Biphasic Soil 2 41.3±0.3 40±14 0.837 0.582 1.686 0.79 8.0* Biphasic Soil 3 28.8±0.3 61±23 0.0606 0.726 0.429 0.0494 30.1* Biphasic Bacteriophage PRD1 ARW - - 0.0283 0.014 0.161 0.0022 13.2* Biphasic Soil 1 28.9±0.7 74±14 0.124 0.158 10.385 0.122 3.7 Linear Soil 2 45.0±0.4 60±18 0.0962 1 0.0962 0 -9.0 x 10-10 Linear Soil 3 30.9±0.8 80±13 0.0606 0.757 1.625 0.0464 4.0 Linear Soil 4 33.2±0.5 71±15 0.0615 0.952 0.207 0.0349 29.4 * Biphasic Soil 5 43.0±0.4 57±14 0.0833 0.713 3.168 0.0721 2.8 Linear * Likelihood ratio significant, therefore biphasic model best choice. If not ratio not significant linear model best choice (less parameters) ** ± standard deviation

Table S3 Salt tracer model parameter estimates (± standard error)

Soil Flow Column porosity Dispersivity R2

[-] [m]

Soil 1 2.08 0.262 ±0.067 0.149 ± 0.08 0.905

Soil 2 0.95 0.482 ±0.015 0.038 ±0.0067 0.904

Soil 3 0.26 0.415 ± 1.04 0.370 ± 1.20 0.872

Soil 4 1.66 0.242 ±0.031 0.068 ± 0.032 0.918

Soil 5 0.44 0.444 ± 0.28 0.548 ± 0.447 0.969 a Adsorption isotherm coefficient; b Molecular diffusion coefficient in free water

Page 246

Table S4 Field soil properties at 1m depth

Site Soil type Infiltration Hydraulic Conductivity cm.min-1 m.day-1

3 Clay Loam 0.25 0.4 5 Loam 0.5 0.1

Page 247

Paper IX Quantitative microbial risk assessment modelling to aid management of onsite sewage systems in Sydney’s drinking water catchment, Australia

Charles, K.J., and N.J. Ashbolt. Quantitative microbial risk assessment modelling to aid management of onsite sewage systems in Sydney’s drinking water catchment, Australia. Submitted to Journal of Water and Health

Author Contribution Role (all authors helped with reviewing and (%) editing) Charles, K 80 Development of model. Analysis and modelling. Writing. Ashbolt, N 20 Supervisor. Assistance with model development and analysis.

Page 248

Abstract Onsite sewage systems, such as septic tank-absorption trenches, are used in areas of Sydney’s drinking water catchments where sewer connection is unavailable. Management of these systems involves consideration of siting as well as design and maintenance, including buffer distances to separate systems from source water streams. A quantitative microbial risk assessment model was applied to identify tolerable buffer distances and to estimate the current level of risk they may pose. The model utilised local experimental data on system performance and virus transport. Current buffer distances were estimated to be 150 ± 132 m (44 % < 100 m). The model was used to assess management options, and to provide an individual site assessment tool. Proscribing a minimum buffer distance of 100 m reduced virus transport events, however, with no reduction in failure rate (sewage surfacing) the maximum breakthrough was comparable with the present situation. Increasing maintenance (no failure) had a similar result to a minimum buffer distance with reduction in virus transport events but little reduction in maximum breakthrough due to existing short buffer distances. Setting a specified log reduction for site assessment provided considerable reductions in maximum virus breakthroughs. The log reduction method was also considered easier for retrofitting than standardising buffer distances, and provides a method for prioritising maintenance.

Introduction Onsite sewage systems provide collection, treatment and disposal of household sewage where sewerage is not available. Septic tank-absorption trenches, which are the most common onsite sewage treatment system in New South Wales, Australia (Charles et al., 2001), provide primary sedimentation with the effluent disposed of via subsurface absorption trenches. Domestic Aerated Wastewater Treatment Systems (AWTS) provide secondary biological treatment with chlorine or ultraviolet (UV) disinfection. The higher level of treatment provided by AWTS allows for a range of disposal options such as surface spray irrigation and drip irrigation. Alternative systems under consideration or with few installations include amended material mound treatment systems, composting or vermiculture toilets, and various hybrids including greywater systems. Absorption trench failure in Australia is based on visual inspection, including excessive plant or weed growth, surface pooling of effluent or bogginess. O'Neill et al.

Page 249

(1993) reported visible surfacing effluent at over 40 % of absorption trenches. Thirty percent of absorption trenches had visible surface flow and 73 % performed unsatisfactorily in South Australia (Geary, 1992). In Queensland, 39 % of trenches had poor performance or surface seepage (Jelliffe et al., 1994). Recent studies of effluent quality within the study catchment area (Charles et al., 2005) reported 24 % visual failure of absorption trenches and 40 % failure of spray irrigation systems. Currently there is no standard risk framework to evaluate or assess potential pathogen impacts from onsite systems in Australia.

Human enteric viruses and other pathogens are excreted by infected humans in large numbers, and are transported away from the source to sewage treatment plants or discharged directly into the local environment from onsite sewage systems. Faecal contamination of surface and groundwater resources can result in large outbreaks of gastrointestinal disease, particularly where that water is used as a source for drinking water. A number of recent outbreaks have been linked to contamination of drinking water by septic seepage. In South Bass Island, Ohio in 2004, 1450 people became ill with nausea and diarrhoea, which was attributed to septic tanks contaminating drinking water wells (Fong et al., 2007). In 2001, an norovirus outbreak at a ski resort that affected 33 people was linked to contamination of the drinking water well with sewage from a septic tank located only 10.5 metres away (Anderson et al., 2003). An outbreak of hepatitis A virus in 1998 from the consumption of contaminated oysters (Conaty et al., 2000) was linked to poorly functioning septic tanks, and overall highlighted the issue in Australia of how septic tanks and other onsite sewage systems are managed to protect water resources.

A key management tool for onsite systems is the use of setback or buffer distances. Buffer distances are designed to provide adequate transport distance and time for pathogen inactivation and sorption. In New York, a study undertaken to define septic system setback distances (Curry, 2000) quantified virus transport, only to comeback to the question of what is tolerable on a drinking water catchment basis. Previous methods of delineating buffer distances have adopted a 7 log10 reduction in virus concentration as acceptable (Yates et al., 1987a), based on the assumption that 104 viruses per L “of septic tank effluent travel through the soil and reach groundwater” and a target of no viruses detectable per 1000 L. However, is it still valid to assume that a 7 log10

Page 250

reduction in acceptable? Regli et al. (1991) suggested acceptable levels of viruses in drinking water between 2 x 10-7 and 2 x 10-3 in order to meet a 10-4 annual risk of infection. Limited data is available on virus concentrations in septic tank effluent (Charles et al., 2005). Furthermore, concentrations detected are typically below that of municipal sewage, ranging from 0.26 – 0.44 enteroviruses L-1 in a school septic system (Deborde et al., 1998b), 59 coxsackieviruses L-1 in a domestic septic tank (Lewis et al., 1993), and up to 148 enteroviruses L-1 (Vaughn et al., 1983) (all assayed by cell culture). In contrast, infectious virus concentrations in raw municipal sewage have been reported to contain enteroviruses ranging from an average 102 to 103 pfu.L-1 (Gantzer et al., 2002), to 105 pfu. L-1 (Dahling et al., 1989). Faulker et al. (2003) assumed a virus concentration of 105 .L-1 based on phage concentrations. Concentrations of noroviruses of up to 107 .L-1, and rotavirus up to 104 .L-1, have been detected in raw sewage by RT- PCR (Lodder et al., 1999), although the detection of nucleic material with this method may overestimate the number of infectious particles. However, these concentrations do indicate the scale of the reduction necessary to achieve the 10-4 tolerable risk.

Various mathematical models have been used to predict virus fate and transport in groundwater, with increasing complexity over time. Yates et al. (1986) described virus transport in terms of groundwater velocity and temperature dependent inactivation. Later models included consideration of virus sorption (Yates, 1995b; Schijven et al., 2002c), filtration (Pang et al., 2005) and fate and transport in the unsaturated zone (Yates et al., 1992). However, as models have increased in complexity, the amount of data required has also increased. The data required has included sorption and inactivation coefficients that will vary with soil type, temperature, pH, organic content, etc. (Schijven, 2001). However, site complexity and specificity of the data can restrict the application of current models. Therefore, a research priority identified at the National Research Needs Conference: Risk-based decision making for onsite wastewater treatment (Cliver, 2001) was to develop approaches to undertake a risk assessment approach to aid management of septic systems.

Microbial risk assessment is a formal methodology that enables variability and uncertainty to be built into a model describing pathogen impacts. Examples include the work of Faulkner et al. (2003) who employed a risk assessment methodology to calculate the probability of transport of viruses through an unsaturated soil zone, based

Page 251

on the assumption of no effective removal in the saturated zone. More recently, Schijven et al. (2006) employed a virus risk assessment methodology to calculate the setback distance around groundwater wells in the Netherlands required to achieve an individual risk level of less than 10-4 infections per year. Schijven’s work included a worst case scenario of aquifer inactivation as the sole mechanism for removal.

This paper provides details of the development of a methodology for delineation of buffer distances for onsite sewage systems. The methodology combines subsurface virus transport models and a model for transport of viruses in surface runoff with available data on onsite system performance in a Quantitative Microbial Risk Assessment (QMRA) framework. The model was applied to the Warragamba catchment in Sydney’s drinking water supply catchments, and specifically looks at the buffer distances required between effluent disposal and surface water. Furthermore, the model was refined to provide an individual site assessment tool for systems within the catchments.

Methodology

Model development The model was based on the work of Charles et al. (2003a; 2004a) and is divided into the following sections: catchment statistics; human virus excretion; onsite sewage treatment and effluent disposal system performances; virus transport across the buffer in unsaturated soil, saturated soil and surface runoff; virus transport and catchment survival, river transport and other sources; water treatment; and consumption. Model assumptions and inputs are summarised in Table 1.

Page 252

Table 1 Summary of assumptions and data QMRA calculations

Item Value or Distribution Source & assumptions [truncation]

Household

Infection rate Uniform (0.0142, 0.028) Reported rates of illness of 0.71-1.4 % (Mead et al., 1999; Wheeler et al., 1999) with morbidity 50 % (Haas et al., 1999)

12 Virus excretion rate Triangular (4, 7, 9) High excretion rates of up to 10 -1 (Log10 .g ) (Gerba, 2000b) have been reported, however lower triangular distribution assumed due to high sensitivity of the model to excretion rates

Faeces excretion rate Uniform (100, 200) (Feachem et al., 1983) (g.d-1)

Duration of shedding Lognormal (6, 3) [2, 23] Limits from (Gerba, 2000b), (d) lognormal distribution selected based on Petterson et al. (2001c).

Time dependent Exponential (0.1), Based on assumption that virus shedding (-) [0.000000001, 1] shedding will commonly peak within the first few days of infection, but may continue at a lower rate for up to several months.

Household population Poisson (Average house Australian 2001 census data, (ABS, population) 2002)

System performance

Page 253

Item Value or Distribution Source & assumptions [truncation]

Septic removal (Log10 ) Lognormal (0.6, 0.1) Virus removal ~ 75 % (Payment et [0,1] al., 1986; Rao et al., 1986)

AWTS removal (Log10) Lognormal (1, 0.1) [0,] Virus removal ~ 90 % (USEPA, 2002)

Disinfection removal Uniform (1.1, 1.8) From field studies (Charles et al.,

(Log10) 2003b)

Amended material sand Normal (1.7, 0.4) From laboratory and field studies mound (Charles et al., 2008)

Pump out removal All Assumes working appropriately, with all sewage removed for off- site treatment

Septic failure Discrete: 10 % failure Assumed failure rate, with no virus removal during failure

AWTS failure Discrete: 9.5 % failure From field studies (Charles et al., 2003b) 2 of 21 had complete failure of AWTS, no virus removal assumed. Disinfection removal not added on if AWTS is failing. A total of 49 % sampled failed to comply with

Disinfection failure Discrete: 8.3 % failure Based on mechanical failure from field studies (2 of 24 systems) (Charles et al., 2003b).

Page 254

Item Value or Distribution Source & assumptions [truncation]

Amended material sand Zero No data available. Assumption mound failure based on the fact that the systems are relatively new.

Pump out failure Discrete: 10 % failure Includes accidental overflow and illegal discharge. Connectivity assumed.

Disposal system failure Discrete: 1 % Total failure 40 % (O'Neill et al., connectivity, 39 % 1993). Connectivity implies no performance failure removal in buffer zone (Ferguson, 2005). Performance failure assumed to result in transport via surface runoff.

Buffer distance removal

Removal by infiltration Uniform(0.3, 0.8) Assumed equal to infiltration of

(RInf) Cryptosporidium (Davies et al., 2003)

Removal efficacy (Ex) Normal (0.8, 0.2)

Wet weather (Prain) Binomial (1, P) Based on historic SCA data for number of wet days. Wet weather assumes all water from failing septics and AWTS goes to runoff. Removes top 1m of unsaturated transport. Antecedent dry period Binomial (1, P) Based on historic SCA rainfall data

(Pantec)

Page 255

Item Value or Distribution Source & assumptions [truncation]

Duration of antecedent Normal ( µ, σ ) Based on historic SCA rainfall data dry period (tantec, days)

Inactivation during C = Co [α.exp(-µ1t)+(1- α = 99.88 %, h1 = 2.48, and h2 = antecedent dry period α).exp(µ2t)] 0.51, t = tantec (Petterson et al., 2001c)

Catchment fate & transport

Inactivation in stream Uniform(0.016, 0.031) Adenovirus used as model virus due to its persistence in the environment (Enriquez et al., 1995) STP loads QADWF x 365 x LP Lp was calculated as per LH in Equation 1, but using the population served by the STP. The Average Dry Weather Flow

(QADWF) for each STP was taken from CH2M HILL (2001).

Concentration of τ is the fraction of catchment flow Cc = τ Lcatchment / QAnnual viruses at catchment off that goes to drinking water, and take (Cc) QAnnual is the annual flow from catchment that goes to drinking water [ML.y-1].

Treatment & Distribution

Chemical precipitation Triangular (0.3, 1.3, 1.6) (Westrell, 2004)

(T1)

GAC filtration (T2) Uniform (0.7, 1.2) (Westrell, 2004)

Postchlorination (T3) Triangular (1.5, 2.0, 3.0) (Westrell, 2004)

Page 256

Item Value or Distribution Source & assumptions [truncation]

Distribution (TD) 0.5 Log10 Distribution was assumed to contribute viruses to the water supply due to accumulation in biofilms within pipes (Petterson et al., 2000) Concentration of Log Cdw = Log Cc - T1 - viruses in drinking T2 - T3 + TD water (Cdw)

Consumption

Drinking water Lognormal (1.1, 0.63) (Robertson et al., 2000) consumed (Vdw, L. -1 -1 person . d )

Probability of infection 1 – ( 1 + D/ β)-α Beta Poisson model (Teunis et al., (Pinf) 2000) with α=0.253 and β=0.422 as described for rotavirus (Ward et al., 1986)

Catchment statistics The location of onsite sewage systems within the Warragamba catchment were derived from a combination of digital mapping and local council surveys. Spatial data was accessed through the Sydney Catchment Authority’s (SCA) GIS (Geographic Information Systems) database and was used to define catchment characteristics. The locations of potential systems were defined as the centroid of cadastral parcels. Based on system location, the following parameters were collated from SCA GIS layers: distance to the nearest waterway, soil type, catchment, subcatchment and land-use. Based on the soil type, ranges of soil depth, slope and permeability were extracted and used in uniform distributions. Centroids were excluded from the risk assessment that met the following criteria: land-use equal to transport or water; centroid located within a National Park or within the designated Warragamba or Fitzroy Falls special areas; lot

Page 257

size < 500 m2 for all land-uses or < 1000 m2 for non-urban land-uses. Furthermore, land-uses were associated with a percentage of developed cadastral parcels (i.e. with an onsite sewage system).

Intensive agriculture was assumed to have a higher rate of occupancy than non-intensive agriculture, plantations or orchards. Mining sites were assumed to have complied with an Environmental Impact Statement (EIS) and therefore to have wastewater treatment systems onsite that performed significantly better than the domestic onsite systems. The latter were not included in the risk assessment. The GIS land-use layer identified 10 golf courses, which were assumed to have approximately two onsite systems each. Centralised wastewater sewers were assumed to be treated at well maintained sewage treatment plants (STPs), with effluent discharged to the catchments.

Vegetation designated as wetlands were assumed to be protected and therefore to not contribute sewage to the catchments. Half of the remaining vegetated areas were assumed to contribute sewage to the catchments (note: these areas do not include National Parks).

A survey of local government authorities (Charles et al., 2001) was used to provide information for the relative proportions of onsite sewage treatment system types. It also reported that the predominant forms of sewage disposal or land application were from AWTS with spray irrigation or for all other systems, subsurface absorption trenches. The type of disposal system was therefore assumed based on the treatment system type, such that all septic tank and alternative system effluent was disposed via absorption trenches, and AWTS effluent via spray irrigation.

Virus excretion The household model component contained information on virus generation based on community infection and excretion rates, and household population size. Virus loading is a function of the rate of infection of individuals and the excretion of the viruses during infection. For this QMRA rotavirus was selected as the ‘worst-case’ due to the high rates of infection and the high rates of excretion from infected individuals (Gerba, 2000b; Nwachuku et al., 2006). Additionally, rotavirus is the leading cause of gastroenteritis in children under five, and is attributed to half of all hospitalisations in Australia for gastroenteritis in children under five (Carlin et al., 1998). The household

Page 258

-1 virus load Lh (viruses.d ) was calculated by:

Lv = LnL fih .10.. λ (Equation 1)

-1 -1 Where Lf was the load of faeces per person per household (g faeces.person .d ) and Lv -1 was the peak virus load per gram of faeces (log10 viruses.g ). A virus excretion distribution λ was used based on the assumption that the peak of virus excretion during infection is for a limited period, and extended periods of shedding are likely to be at lower concentrations.

Onsite sewage system The fate of viruses within the treatment system was based on the type and performance of the system as described in Table 1. Due to limited data, all alternative systems were assumed to be amended sand mounds. The type of disposal system was assumed based on the treatment system type, such that all septic tank and alternative system effluent was disposed via absorption trenches, and AWTS effluent via spray irrigation. This then defined the effluent pathway.

Absorption trenches were assumed to result in discharge to the unsaturated zone at a depth of 0.5 m with vertical transport to groundwater or a confining layer, then horizontal transport to a stream or groundwater bore. Surface spray irrigation was assumed to discharge on the surface and follow the same pathway as absorption trench effluent in dry weather. During dry weather, viruses were assumed to accumulate on the ground surface, but were subject to inactivation. During wet weather, surface viruses that accumulated during the antecedent dry period, and remained viable, were assumed to be available to be mobilised in surface runoff.

In the model, disposal system failure assumed effluent bypassed the unsaturated zone in dry weather, and absorption trench effluent goes to runoff in wet weather. Of the total failure, 2.5 % were assumed to result in connectivity between the disposal system and the waterway. Connectivity is defined as where sewage disposal is effectively direct to waterways, thereby voiding all buffer distance removal mechanisms. It may occur naturally, such as rock fissures that result in rapid transport to waterways, accidentally, such as spray irrigation into drains, and anecdotal evidence suggests this may arise from deliberate tampering with systems.

Page 259

Buffer distance transport Transport and removal of viruses in the subsurface was based on column and field experiments undertaken using soils from the catchments (Charles et al., Submitted). Darcy’s law was used to calculate the groundwater flow velocity for various slope and soil conditions. Virus transport was modelled using HYDRUS-1D (Šimùnek et al., 1998) in a 150 m section of soil, with a constant flux input equal to the groundwater velocity. Dilution effects were assumed to be negligible as the model was load based.

The results were then converted to a log10 - removal with distance such that: R = asat . x

+ bsat (Equation 2), where R was the log10 removal of viruses and x was distance (m), d and asat = c . l (Equation 3) where l is slope. The derived values of the constants are provided in Table 2, slope and distance were based on the GIS database.

Table 2 Calculated virus transport and fate coefficients based on Paper VIII

Soil Saturated soil Unsaturated soil Overland flow

bsat * c d bunsat c d a b c Sandy -0.0041 0.0092 -0.70 0.035 0.13 -0.67 loam 0.04 -0.25 0.124 Clay 0.23 0.028 -0.57 0.011 0.32 -0.65 loam

*As the variation resulting from bsat was small, an average value for each soil was assumed.

Similarly, unsaturated transport was modelled with HYDRUS-1D (Šimùnek et al., 1998) over the maximum soil thickness for a range of effluent application rates. The same virus parameters were used as for the saturated transport modelling, however two virus adsorption sites were assumed to account for additional attachment at the air- water-interface (Thompson et al., 1998). The model for unsaturated soil removal was as for Equation 2, with

d aunsat = c . q (Equation 4)

Page 260

where q was flow (effluent land application rate and rainfall). The assumptions for the unsaturated soil transport component of the model included: the slopes as defined and based on a soil database (SCA), not on location specific data; the slope of the groundwater was assumed to be equal to the slope of the surface; and the effluent application rate was not considered significant in the groundwater flow scenario.

A model of virus removal in surface runoff was developed based on the results of Davies et al. (2003) and Ferguson et al. (2007), using the average for bare and vegetated soil results (Figure 1). It was assumed for the overland flow model that as the slope decreased, the removal of viruses increased, such that at zero slope removal was complete. Similarly, as the slope increased, removal was assumed to decrease. The removal in overland flow was calculated as the sum of removal in runoff and infiltration. The model of surface runoff was based on fitting linear and log-linear models to virus removal (Figure 2) (Ferguson et al., 2007). In order to address variability, a triangular distribution was applied, where the most likely value was the removal predicted by the linear model, and the minimum and maximum were defined by the linear model ± the difference between the general linear model and the linear model for the bare-plot data (22 %), such that

b b b Roverland flow = Triangular (0.78 a . x . l , a . x . l + c, 1.22 a . x . l ) + LOSMS.RInf (Equation 5).

In addition to the removal equations, a degree of efficacy was incorporated to address the decrease in removal experienced at field-scale due to subsurface heterogeneities such as fractures and macropores (Taylor et al., 2004), as well as surface runoff areas that were concave or firmly packed. The build-up of viruses during an antecedent dry period was defined by the virus load multiplied by the duration of shedding, with biphasic inactivation.

Page 261

(a)

0.5

) 0.4 10 0.3

0.2

Removal (LogRemoval 0.1

0 02.557.510 Distance (m)

(b)

6 ) 10

4

2 Removal (LogRemoval

0 0 50 100 150 200 Distance (m)

Figure 1 Comparison of linear (∆) and logarithmic (○) models of virus removal in overland flow over (a) experimental scale with original data points (▲ data for removal on bare ground and ● on vegetated ground [Ferguson et al., 2007]) and (b) extrapolated to field-scale

Page 262

14 12 )

10 10 8 6 4 Removal (log 2 0 0 0.2 0.4 0.6 0.8 1 Slope (%)

Figure 2 Comparison of predicted log10 removal of viruses in overland flow over 100 m by slope based on the linear (∆) and logarithmic (○) models

Catchment fate and transport Transport in waterways within the catchment was assumed to remove viruses by inactivation only. Sedimentation and resuspension was not included in the model. The removal was calculated as the inactivation rate (µ) times the travel time from the drainage unit outlet to the offtake, including both stream and reservoir transport. Reservoir travel time was assumed to be equivalent to the stream travel time, equal to travel distance divided by stream velocity.

Due to the low inactivation rate assumed for viruses in stream water, the transport times were not particularly sensitive parameters. The transport times for viruses in waterways in the catchments were derived from historical flow data. The loads of viruses from STPs were also considered, calculated as for onsite sewage systems based on the population served (CH2M HILL, 2001), with removal assumed to be equivalent to used for drinking water treatment, below. The annual load to drinking water from a catchment was calculated as the sum of the site loads minus the removal attributable to the water treatment and distribution.

Treatment, Distribution and Ingestion The water filtration plants in the region include screening; four stage coagulation: with

Page 263

ferric chloride as the primary coagulant supplemented with polydadmac cationic polymer, lime water pH correction and polyacrylamide filter aid addition; filtration on sand or dual media (anthracite and fine grained sand); and chlorination/chloramination followed by fluoridation (Mazounie et al., 2000). Westrell (2003) used distributions to describe performance of similar water treatment systems, which were adopted for use in this QMRA (Table 1).

Consumption of cold tap water has been reported to vary between 0.1 and 1.55 L per day (Mons et al., 2007). In Australia, consumption has been reported to vary between cities, with considerably higher rates of non-consumption in Adelaide than Melbourne theorised to be due to historically lower aesthetic quality of the Adelaide drinking water (Mons et al., 2007). The model did not include consideration of non-consumers. The total dose (D) was calculated as: D = Cdrinkwater . Vdrinkwater . 365 (Equation 6).

Application to catchment management The tolerable annual load of viruses in catchment source water was estimated by undertaking the section of the QMRA from the reservoir off-take to the consumer usgni different concentrations of viruses in the raw water. The tolerable load was defined by the virus at which the rate of infection in consumers was less than the tolerable per person annual infection risk of 10-4, such that the rate of infection in consumers was assumed to be 10-4 infections per person per year, at the 80th percentile. This model forms the basis for assessing the tolerable virus loads from each subcatchment. However, it is infeasible for this model to be used on a site by site basis for development applications. Hence, a log10 reduction approach was adopted for assessing individual houses, with the QMRA used to assess the impact of this log reduction requirement.

The log reduction approach proposed for development assessment followed three basic principles:

• Each stage of treatment and transport can be assumed to provide a reduction in the virus load;

• Log reductions can be added; and

Page 264

• A uniform log reduction for a catchment provides adequate safeguard for catchment water supplies.

The log reduction for a system was the sum of the log reductions for the treatment system, disinfection system, disposal system, and buffer distance. Log reductions were based on the average values adopted in the QMRA (Table 1).

A number of scenarios (Table 3) were investigated. The baseline scenario included all system types with their associated failure rates. An outbreak scenario was modelled as a doubling of the community infection rate. Management scenarios were modelled to predict and compare the impact of the 100 m buffer distance, the impact of failure and the impact from a single type of system. The log reduction method for site assessment was also investigated. This latter scenario considered the implementation of a 7 log10 reduction guideline, such that the sum of log virus removal in the treatment system, disposal system and buffer was a minimum of 7 log10. The implementation of the log reduction guideline was also modelled with the inclusion of disposal system failure.

Table 3 Summary of scenarios

Scenario Description

1 Normal Baseline: all system types with their associated failure rates

2 Double population Total population in each catchment doubles

3 Outbreak An outbreak scenario was modelled as a doubling of the community infection rate

4 Minimum 100 m setback All setback distances less than 100 m are increased to 100 m

5 Minimum 100 m buffer with All setback distances less than 100 m are double population increased to 100 m, and the total population in each catchment doubles

Page 265

6 All AWTS All systems are assumed to be AWTS

7 All AWTS with double All systems are assumed to be AWTS, and the population total population in each catchment doubles

8 All septics All systems are assumed to be septic tanks with absorption trenches

9 All amended material soil All systems are assumed to be amended material mounds soil mounds

10 No failure All system types are included, but failure rates are assumed to be 0 for all systems

11 Minimum 7 log10 reduction All systems with buffers which provide less than

7 log10 reduction are increased to provide 7 log10 removal

12 Double population 7 log10 All systems with buffers which provide less than

reduction 7 log10 reduction are increased to provide 7 log10 removal, and the total population in each catchment doubles

13 Minimum 7 log10 reduction All systems with buffers which provide less than

with disposal failure 7 log10 reduction are increased to provide 7 log10 removal,

14 All 7 log reduction total All systems provide exactly 7 log10 reduction

Sensitivity assessment was undertaken as part of the model development to assess the relative significant of the model parameters. It was undertaken on a single system basis by section, with all parameters in one section compared with each other, but only the most sensitive from each section compared across the model. The modelling and sensitivity analyses were undertaken in @Risk V4.5 (Palisade Corp. USA) within Microsoft ExcelTM 2003, using 10,000 iterations.

Page 266

Results The distributions that contribute the most sensitive input into estimating total virus loads were the maximum virus excretion rate, followed by the excreted virus concentration. The virus loading component was therefore revised to include an excretion rate that was a function of duration. The most sensitive parameter was the maximum virus excretion rate, followed by load over time distribution and infection duration. Of the parameters affecting flow in soil, the coefficients for unsaturated flow had the most significant impact. In wet or dry conditions, the greatest sensitivity was in order: buffer distances, slope, saturated soil removal, unsaturated soil removal (for dry conditions), conservative and maximum flows, and failure of removal in the saturated zone. System type differences resulted in average loads from sites with septic tanks that th were 0.9 log10 higher than for sites with AWTS (>2 log10 at the 95 percentile), 1.9 log10 higher than for sites with a septic pump out, and 0.8 log10 higher than for alternative systems. The relationship between setback distance and the Pinf is provided in Figure 3.

Page 267

50

r 0 -50 -100 -150 -200 removal in buffe in removal 10 -250 Log -300 0 200 400 600 800 1000 BufferDistance distance (m) (m)

(a)

(b)

-9.0 -10.0 -11.0 Pinf

10 -12.0

Log -13.0 -14.0 -15.0 0 200 400 600 800 1000 Buffer distance (m) Distance (m)

Figure 3 Relationship between buffer distance and the average (a) virus removal th th and (b) log10 Pinf (▲ and ∆ are the 5 and 95 percentiles, respectively)

Initial sub-catchment-by-sub-catchment assessments highlighted the Blue Mountains subcatchment as being a primary source of viruses. The Blue Mountains have a high

Page 268

proportion of sandy soils which, based on experimental results (Charles et al., Submitted), have potential for rapid transport of viruses. Due to the deep, permeable sandstone geology of the region, it was assumed that the distance travelled in groundwater was nominally ten times the surface buffer distance.

A total of 18 149 onsite systems (28 595 cadastral lots) in the catchments were identified, with the majority in the Wollondilly sub-catchment (11 308 systems on 15 766 cadastral lots). The Blue Mountains sub-catchment had a total of 5 684 cadastral lots with 3 190 systems, the Mulwaree sub-catchment 4 484 cadastral lots, and approximately 2 400 systems, and the Wingecarribee sub-catchment 2 661 cadastral lots and 1 251 systems.

The calculated buffer distances, i.e. the shortest straight line distance between a centroid and a waterway, ranged from zero up to approximately 850 m in most sub-catchments, but up to 1 842 m in the Mulwaree sub-catchment. The average distance was shortest in the Blue Mountains sub-catchment with 112 ± 91 m compared to 154 ± 164 m in the Mulwaree, 165 ± 132 m in the Wingecarribee and 160 ± 132 m in the Wollondilly. In total, 44 % of existing systems had calculated buffer distances of less than 100 m.

The tolerable load (80th percentile) of viruses was calculated to be 108 virus per year in raw water and 104 viruses per year in treated drinking water. The statistics on the load of viruses to the catchment are provided in Table 4. The 80th percentile of predicted infections in Sydney’s drinking water consumers from virus contamination from sewage contamination in the drinking water supply was less than 10-4 infections per year (Table 5, Figure 4).

Page 269

Table 4 Log10 annual virus load from onsite sewage systems and centralised sewage treatment plants to the drinking water offtake at Warragamba Dam

Scenario 50 % 5 – 95 % 20 – 80 %

Septics only

Blue Mountains -1.5 nil – 6.1 -9.0 – 3.7

Mulwaree nil nil – 4.2 nil – -3.3

Wingecarribee nil nil – 1.6 nil – -16.7

Wollondilly 2.8 -23.0 – 7.7 -2.8 – 5.8

Whole catchment 4.3 -3.0 – 7.9 1.1 – 6.4

STPs 2.6 1.5 – 3.7 2.1 – 3.1

Total 4.4 2.1 – 7.9 2.8 – 6.4

Table 5 Log10 virus load in the treated drinking water supply, and the resulting log10 probability of infection within the population

Scenario 50th percentile 5 – 95th percentile 20 – 80th percentile

Septics load 0.6 -6.7 – 4.2 -2.6 – 2.7

Total load 0.7 -1.8 – 4.3 -0.8 – 2.7

Pinf Septics -6.7 -12.5 – -3.2 -9.5 – -4.7

Pinf Total -6.8 -9.2 – -3.1 -8.3 – -4.7

Page 270

0.250 20th percentile 80th percentile 0.200

0.150

0.100 Frequency 0.050

0.000 -16 -12 -8 -4 0 Log10 Pinf

Figure 4 Log10 Pinf for from drinking water ingestion based on virus input from onsite sewage systems with (black) and without (grey) the inputs from STPs (percentiles indicated are for with STP)

The impact of differing management scenarios on the log10 load of viruses from the Wingecarribee sub-catchment are provided in Table 6. The 80th percentile is not reported as there was no breakthrough at the 80th percentile for most scenarios. Scenario 4 (100 m buffer) and Scenario 10 (no failure) had the greatest reduction of virus load at the 95th percentile, down to zero virus load. However, these management options did not achieve the greatest reductions at the 99th percentile and of maximum loads. By th th comparison, the 7 log10 reduction resulted in lower loads at the 95 and 99 percentile and maximum. Scenario 14, with all systems having an equal 7 log10 reduction, had a higher average log10 load due to 68 % of cases having no virus transport. The impact of disposal failure remains significant with the 7 log10 reduction model, however, this does not necessarily represent failure with a system designed to meet the log reduction criteria.

Page 271

Table 6 Log10 virus load from Wingecarribee catchment, as modelled for the drinking water offtake at Warragamba, under different management scenarios

Scenario Mean SD 95th 99th Max %ile %ile

1 Normal -5.4 18.6 1.6 5.9 8.7

2 Double population -8.0 22.3 3.9 6.8 9.3

3 Outbreak: double rate of -8.0 22.7 4.2 6.7 8.8 infection

4 Minimum 100 m buffer -6.3 19.6 0 4.3 8.7

5 Minimum 100 m buffer with -8.7 22.8 3.0 6.2 8.9 double population

6 All AWTS -4.9 17.1 0.3 4.7 9.0

7 All AWTS with double -7.2 19.5 3.0 5.5 8.9 population

8 All septics -5.9 20.6 1.8 6.1 8.6

9 All amended material soil -6.6 21.4 0.9 5.1 8.1 mounds

10 No failure -8.2 24.6 0 4.4 7.8

11 Minimum 7 log10 reduction -5.1 19.0 0.9 2.0 3.4

12 Double population 7 log10 -7.2 21.5 1.5 2.3 3.6 reduction

13 Minimum 7 log10 reduction -4.9 19.3 1.9 5.2 10.8 with disposal failure

14 All 7 log reduction total 0.2 0.7 1.7 2.5 3.6

Page 272

A sensitivity assessment was undertaken for the log10 reduction method, based on a single system. The relationship between the log10 reduction value and the virus load from the site is illustrated in Figure 5. Without failure, the log reduction method reduced the higher loads from a site by the value used in the model. This was also the case in a catchment model. However, the differences in the probability of infection were considerably lower due to attenuation of viruses in the catchment and water treatment.

The difference between a 4 log10 and a 7 log10 buffer was a 1.0 log10 reduction in the th th virus load from the site at the 80 percentile (2.3 log10 at the 95 percentile), which corresponded with a 0.1 log10 reduction in the probability of infection (0.2 log10 at the th 95 percentile). The difference between a 7 log10 and an 11 log10 buffer resulted in a 2.8 th th log10 reduction in the virus load from the site at the 80 percentile (3.6 log10 at the 95 th percentile), with a 0.2 log10 reduction in the probability of infection (0.3 log10 at the 95 percentile).

10

5

0

-5 load from site from load

10 -10 Log -15 0 5 10 15

Minimum log10 reduction

Figure 5 Relationship between the log10 reduction value and the log10 load from a site (∆ is the 95th percentile)

In the Wingecarribee sub-catchment, in which only 16 % of centroids were on sandy soils, 100 m buffer reduced the loads, but did not reduce extreme events caused by virus to waterway connectivity. The log10 reduction method would ideally increase the treatment step, and reduce the number of barriers. By comparison, 47 % of centroids in

Page 273

the Blue Mountains sub-catchment were on sandy soils such that the 100 m buffer distance would provide only a 4.6 log10 reduction, compared to 10.7 log10 reduction in more clayey soils.

Discussion Due to the high concentrations of viruses in effluent, this parameter had the largest influence on the QMRA results. There was limited data available on the concentrations of viruses in septic tanks or other onsite sewage systems, and no information on the temporal distributions of viruses during an infection of an individual in a household. Lewis and Stark (1993) reported detections ranging from 0.07 viruses L-1 to greater than 59 L-1 for household septic tanks, with detections in effluent up to 137 days following an infection. Deborde et al. (1998b) reported concentrations of viruses in a school septic tank serving 350 people of 0.26 to 4.4 viruses L-1, although detection rates were 12 %, considerably higher than predicted for single household systems. These results are significantly lower than those reported for centralised systems which range from 104 – 105 per L (Gerba, 2000a).

The definition of what is a tolerable buffer distance adopted here was based on the assumption that a tolerable public health risk from a single source was equivalent to a 1 in 10 000 infection risk per year (Regli et al., 1991). Furthermore, there is debate as to whether the definition of tolerable should be the average (Haas, 1996) or the 95th percentile (de Roda Husman, pers. comm.). In this study we adopted the 80th percentile as the level of how many cases should comply with the criteria for what is tolerable. To estimate the 95th percentile with confidence requires considerably more data than is required that for an 80th percentile, so in part, what is considered tolerable relied on the precision available.

It is important to note that QMRA predicts the virus loads from the catchment at levels below that of current detection methods. Similarly the infection levels that are targeted are below the level of sensitivity for most epidemiological studies (e.g. illness rates of about 10%, Hellard et al., 2001). The tolerable concentrations of viruses in treated water (10-7 per L) calculated by the QMRA were comparable with that calculated in Regli et al. (1991).

The result from the QMRA predicted an average of 1.3 x 10-7 infections with the

Page 274

modelled virus due to Sydney drinking water contaminated with viruses discharged to the catchments from onsite systems.

Limited data was available on the performance of septic tanks with regard to virus removal, and no data was available on how the treatment performance of septic tanks affects virus removal. As viruses in sewage are generally associated with particles, the primary mechanism for removal of viruses in septic tanks was assumed to be physical settling of solids-associated viruses, and therefore would be expected to correlated with the improvement in effluent quality. The reported virus removal (Rao et al., 1986) is comparable to the reported solids removal of 70-80 % for primary wastewater (Ferguson et al., 2003). One of the few markers of septic tank failure is the build-up of solids. A properly maintained tank is expected to have a settled solids layer less than a third the depth of the tank, allowing adequate capacity for the tank to function. The build up of solids, and the associated reduction in capacity, will reduce the solids removal in the tank and eventually result in the overflow of settled solids. In the model, this performance failure is assumed to reduce virus removal to zero. The estimated failure of septic tanks (10 %) was considerably lower than the proportion of septic tanks that had suspended solids concentrations above the NSW guidelines in field studies (87

%) (Charles et al., 2005). Similarly to septic tanks, the assumed 1 log10 virus removal in AWTS corresponds with solids removal of 80 – 90 % in activated sludge treatment (Ferguson et al., 2003). Failure rates were based on field surveys (Charles et al., 2003b). As for septic tanks, the lack of data on influent concentrations or correlations between solids and virus removal performance resulted in the above assumption about complete failure. No data was available on connectivity in failure of disposal systems. In addition to natural or accidental connectivity, anecdotal evidence suggests this may arise from deliberate tampering with systems.

The use of GIS enabled assessment of the cumulative inputs across the catchments based on actual onsite and STP discharge locations within the catchments. Furthermore, it provided site specific data on slope, soil type and distance to waterways. Nonetheless, limited data was available on specific system locations, hence the use of the centroid of selected plots to estimate locations, based on land use as well as survey data on the number of systems in a local government area. In practice onsite sewage systems are typically placed close to the house, so the location will vary depending on the size of the

Page 275

lot. As the registration programme for onsite systems in NSW progresses, there will be an improvement in data on the location of systems, at least in terms of cadastral parcel, and for each system information will be available on type, age, size of household (number of bedrooms), and the age of the system.

Due to limitations with what was able to be undertaken with the GIS program, the distance between the onsite sewage system and a waterway was the shortest straight line distance . The surface travel buffer distance would be improved through delineation based on the downslope flow distance.

Buffer distance removal was modelled based on results from laboratory and field experiments for virus transport in soil and from overland flow. The removal of viruses in the homogenous repacked soil columns overestimated the reduction experienced in the field in the clay loam soil, but underestimated the measured removal in the loam where there were no virus transport detected in the field. The decreased removal in the field may have been due to a combination of heterogeneity, different effluent quality, and the preconditioning of the site with effluent and viruses. Factors such as dilution, dispersion, organic matter and a biomat in the absorption trench, would be expected to increase virus reduction compared to the column experiment. The low virus removal experienced in the sandy loam soil columns resulted in the need for the following assumption: that subsurface virus transport in areas with sandy soil was further than for vegetated surface transport.

While attempts were made in the QMRA to take account of site and soil variability across the catchments, the complexity of the systems means that it is impossible to thoroughly address this issue. Furthermore, heterogeneity in site and soil is likely to impact virus transport at a level that can not be addressed through the currently available GIS database from SCA.

While failure of onsite systems was considered, the contribution from centralised systems only considered normal treated discharges from sewage treatment plants. Additional impact would be expected from exfiltration, leakage and overflow from sewers to surface and groundwater, all subjects for future research.

To set an appropriate buffer distance the risk implications to the catchments need to be

Page 276

quantified along with buffer transport data integrated into the wider context of catchment management, to allow cumulative impact assessment. Furthermore, there is a practical need for the buffer distance to be easy to apply as a development assessment and risk assessment tool.

The QMRA model presented here provides an approach to quantify the risk of virus infection from Sydney’s drinking water. For the present situation, the model estimated a tolerable risk at the 80th percentile, with the potential for an unacceptable risk at the 95th percentile. With the limitations of the model, and the acceptable water quality of the catchments at present, this is not considered a significant concern. However, outbreaks associated with onsite sewage systems in Australia (Conaty et al., 2000) and overseas (Scandura et al., 1997) have highlighted the need for improved and informed management, particularly with increasing development pressure in unsewered areas.

A number of different management options were explored, and in particular, management options that are feasible tools to be used as part of assessing development applications. The management options included fixed buffer distances, adoption of one type of treatment system, reduction of system failure, and application of a fixed log reduction that includes treatment and buffer distance removal.

Buffer distances In the Wingecarribee, 39 % of existing systems had a buffer distance of less than 100 m. Increasing buffer distances up to a minimum of 100 m was estimated to achieved a raw th th water virus reduction of 1.6 log10 at the 95 and 99 percentile for the Wingecarribee sub-catchment. However, the failure rate for disposal systems was assumed to be the same, which resulted in high levels of transport in some cases. The key limitation of the uniform buffer distance approach would be expected to be in implementation due to the presence of existing systems within this buffer distance, with an alternative management approach required to address these existing systems, as well as existing sites which are not yet developed.

Preferred systems This management option assessed the difference in virus load to the catchment when only a single system is in use in the catchment. Septic tanks, AWTS and amended material sand mounds were considered. Of these, the greatest reduction at the 95th and

Page 277

99th percentiles was from AWTS. The amended material soil mound was assumed to have no complete failure; the broad distribution of treatment performance based on field and laboratory experiments was assumed to account for performance failure.

System failure Removal of all failure in system performance and buffer systems achieved a reduction th th of 1.6 log10 and 1.5 log10 at the 95 and 99 percentile, respectively, for the Wingecarribee sub-catchment. This was comparable with the reduction for the uniform buffer distance above, however, in this case, there was considerable transport from sites with short buffer distances. In terms of implementation, the registration system for onsite sewage systems in NSW (DLG, 1998) will improve system performance in general by systematically identifying failing systems and providing education or enforcement to reduce failure. However, the assumption of no failure is not realistic.

Log reduction The log reduction approach developed here provides for flexibility in buffer distances to deal with limitations on individual sites, including the siting of new systems and adoption of new technology. To improve the overall performance of onsite systems, an appropriate management system could be implemented on a site by site basis; for example, where there is uncertainty of the capability of the system to routinely achieve the calculated log reduction, regular maintenance requirements should be stipulated. Furthermore, the log reduction approach enables novel systems to be installed based on estimates of removal based on literature, with monitoring criteria built into development application approval. Similarly, as a conservative estimation of removal in the buffer is used, where there are better soils, the removal could be increased based on proper testing and management/controls. For example, management conditions should be stipulated based on how robust the system is, for example if there is high variability in system performance, or for novel systems where performance has not been well established, regular maintenance inspections can be a requirement of development approval. A site with clay loam soil and a 65 m buffer distance at a 10 % slope would provide, based on the groundwater transport model, 7 log10 removal in groundwater. For this site, the linear overland flow model would predict 4.6 log10 removal in runoff. Hence options for onsite systems might include an AWTS with spray irrigation, with a strict maintenance requirement, or subsurface disposal.

Page 278

The log reduction approach was specifically designed and modelled for Sydney’s drinking water catchments. Application to other catchments may require further intensive modelling to calculate the appropriate level of removal. The 7 log10 value for removal is in line with recommendations from the World Health Organization (WHO) (Yates et al., 1985). Further development of this approach could include assessment of log removals for alternative systems, and refinement of log removals for buffer distances under different conditions, as well as establishing guidelines for nutrient removal.

More discussion about the applicability and success of attempting to use QMRA to assess the influence of on-site systems on risk of infection.

Conclusions This paper presents a QMRA based approach to delineating tolerable buffer distances for onsite sewage systems in Sydney’s drinking water catchments. The model developed utilised local experimental data on system performance and the transport of viruses in the sub-surface as well as in surface runoff. A range of development assessment tools were considered. Fixed minium buffer distances reduced the number of occasions on which viruses were transported to waterways, however, with no reduction in onsite treatment failure rate, the maximum breakthrough was similar to the present situation. An increase in maintenance, resulting in an unrealistic zero failure rate, was forecast to have a similar result to fixed minimum buffer distances. Also resulting in reduction in the number of occasions on which viruses were transport to waterways, with little reduction in maximum breakthrough due to short buffer distances. Setting a specific log reduction site assessment achieved considerable reductions in maximum breakthrough, and was considered easier for retrofitting systems than standardising buffer distances, as well as providing a method for prioritising maintenance.

Page 279

References ABS, 2002 2001 Census of Population and Housing Basic Community Profile. , 20/11/03.

Anderson, A.D., Heryford, A.G., Sarisky, J.P., Higgins, C., Monroe, S.S., Beard, R.S., Newport, C.M., Cashdollar, J.L., Fout, G.S., Robbins, D.E., Seys, S.A., Musgrave, K.J., Medus, C., Vinje, J., Bresee, J.S., Mainzer, H.M. & Glass, R.I., 2003 A Waterborne Outbreak of Norwalk-Like Virus among Snowmobilers - Wyoming, 2001. Journal of Infectious Diseases, 187), 303.

Carlin, J.B., Chondros, P., Masendycz, P., Bugg, H., Bishop, R.F. & Barnes, G.L., 1998 Rotavirus infection and rates of hospitalisation for acute gastroenteritis in young children in Australia, 1993-1996. Medical Journal of Australia, 169(5), 252.

CH2M HILL, 2001 Review of Catchment Sewerage Needs. Prepared for Sydney Catchment Authority., Australia.

Charles, K., Ashbolt, N., Roser, D., Deere, D. & McGuinness, R., 2001 Australasian standards for on-site sewage management: Implications for nutrient and pathogen pollution in the Sydney drinking water catchments. Water Journal of the Australian Water & Wastewater Association, 28(2), 58.

Charles, K., Roser, D., Ashbolt, N., Deere, D. & McGuinness, R., 2003a Buffer distances for on-site sewage systems in Sydney's drinking water catchments. Water Science and Technology, 47(7-8), 183.

Charles, K.J., 2007 Quantitative Microbial Risk Assessment: a catchment management tool to delineate buffer distances for on-site sewage treatment and disposal systems in Sydney’s drinking water catchments. Civil and Environmental Engineering, PhD.

Charles, K.J. & Ashbolt, N.J., 2004 Quantitative Microbial Risk Assessment: a catchment management tool to delineate setback distances for septic systems. Young Researchers Conference 2004:139.

Charles, K.J., Ashbolt, N.J., Deere, D.A. & Roser, D.J., 2003b Disinfection in Aerated Wastewater Treatment Systems. Ozwater.

Page 280

Charles, K.J., Ashbolt, N.J., Roser, D.J., McGuinness, R. & Deere, D.A., 2005 Effluent quality from 200 on-site sewage systems: design values for guidelines. Water Science & Technology, 51(10), 163.

Cliver, D.O., 2001 Research Needs in Decentralised Wastewater Treatment and Management: Fate and Transport of Pathogens. National Research Needs Conference Proceedings: Risk-Based Decision Making for Onsite Wastewater Treatment.

Conaty, S., Bird, P., Bell, G., Kraa, E., Grohmann, G. & McAnulty, J.M., 2000 Hepatitis A in New South Wales, Australia, from consumption of oysters: the first reported outbreak. Epidemiology and Infection, 124(1), 121.

Curry, D.S., 2000 Final Report for the Septic Siting Study., New York.

Dahling, D.R., Safferman, R.S. & Wright, B.A., 1989 Isolation of enterovirus and reovirus from sewage and treated effluents in selected Puerto Rican communities. Applied and Environmental Microbiology, 55(2), 503.

Davies, C.M., Ferguson, C.M., Kaucner, C., Krogh, M., Altavilla, N., Deere, D.A. & Ashbolt, N.J., 2003 Dispersion and transport of Cryptosporidium oocysts from faecal pats under simulated rainfall events. IWA Health Related Microbiology.

Deborde, D.C., Woessner, W.W., Lauerman, B. & Ball, P., 1998a Coliphage prevalence in high school septic effluent and associated ground water. Water Research, 32(3781- 3785).

Deborde, D.C., Woessner, W.W., Lauerman, B. & Ball, P.N., 1998b Virus occurrence and transport in a school septic system and unconfined aquifer. Ground Water, 36(5), 825.

DLG, 1998 Environment and Health Protection Guidelines: On-site Sewage Management for Single Households, Sydney, NSW.

Enriquez, C.E., Hurst, C.J. & Gerba, C.P., 1995 Survival of the enteric adenoviruses 40 and 41 in tap, sea, and waste water. Water Research, 29(11), 2548.

Faulkner, B.R., Lyon, W.G., Khan, A.F. & Chattopadhyay, S., 2003 Modeling leaching

Page 281

of viruses by the Monte Carlo method. Water Research, 37), 4719.

Feachem, R.G., Bradley, D.J., Garelick, H. & Mara, D.D., 1983. Sanitation and disease health aspects of excreta and wastewater management. John Wiley & Sons (for The World Bank), Chichester, England.

Ferguson, C., 2005. Deterministic model of microbial sources, fate and transport: a quantitative tool for pathogen catchment budgeting. University of New South Wales - Centre for Water and Waste Technology, Sydney, Australia.

Ferguson, C., Davies, C.M., Kaucner, C., Krogh, M., Rodehutskors, J., Deere, D. & Ashbolt, N.J., 2007 Field scale quantification of microbial transport from bovine faeces under simulated rainfall events. Journal of Water and Health, 5(1), 83.

Ferguson, C.M., de Roda Husman, A.M., Altavilla, N., Deere, D. & Ashbolt, N.J., 2003 Fate and transport of surface water pathogens in watersheds. Critical Reviews in Environmental Science and Technology, 33(3), 299.

Fong, T.T., Mansfield, L.S., Wilson, D.L., Schwab, D.J., Molloy, S.L. & Rose, J.B., 2007 Massive microbiological groundwater contamination associated with a waterborne outbreak in Lake Erie, South Bass Island, Ohio. Environ Health Perspect, 115(6), 856.

Gantzer, C. & Schwartzbrod, L., 2002 Enteroviruses: Occurence and persistence in the environment. Encyclopedia of Environmental Microbiology.

Geary, P., 1992 Diffuse pollution from wastewater disposal in small unsewered communities. Aust. J. Soil & Wat. Conserv., 5(1), 28.

Gerba, C., 2000a Domestic wastes and waste treatment. Environmental Microbiology:505.

Gerba, C.P., 2000b Assessment of enteric pathogen shedding by bathers during recreational activity and its impact on water quality. Quantitative Microbiology, 2(1), 55.

Haas, C.N., 1996 How to average microbial densities to characterize risk. Water Research, 30(4), 1036.

Page 282

Haas, C.N., Rose, J.B. & Gerba, C.P., 1999. Quantitative Microbial Risk Assessment. John Wiley & Sons, Inc, New York.

Hellard, M., Sinclair, M.I., Forbes, A.B. & Fairley, C.K., 2001 A randomized blinded controlled trial investigating the gastrointestinal health effects of drinking water quality. Environmental Health Perspectives, 109(8), 773.

Jelliffe, P.A., Sabburg, G. & Wolff, J., 1994 Key factors in minimising water pollution from unsewered areas. AWWA 16th Federal Convention:85.

Lewis, A.L. & Stark, L., 1993 Florida OSDS research Project. Viral Study Summary., Florida.

Lodder, W.J., Vinj, J., van de Heide, R., de Roda Husman, A.M., Leenen, E.J.T.M. & Koopmans, M.P.G., 1999 Molecular detection of Norwalk-like caliciviruses in sewage. Applied and Environmental Microbiology, 65), 5624.

Mazounie, P., Bernazeau, F. & Alla, P., 2000 Removal of Cryptosporidium by high rate contact filtration: the performance of the Prospect water filtration plant during the Sydney water crisis. Water Science and Technology, 41(7), 93.

Mead, P.S., Slutsker, L., Dietz, V., McCaig, L.F., Bresee, J.S., Shapiro, C., Griffin, P.M. & Tauxe, R.V., 1999 Food-related illness and death in the United States. Emerging Infectious Diseases, 5(5), 607.

Mons, M.N., Van der Wielen, J.M.L., Blokker, E.J.M., Sinclair, M.L., Hulshof, K.F.A.M., Dangendorf, F., Hunter, P.R. & Medema, G.J., 2007 Estimation of the consumption of cold tap water for microbiological risk assessment: an overview of studies and statistical analysis of data. . Journal of Water and Health(in press).

Nwachuku, N. & Gerba, C.P., 2006 Health risks of enteric viral infections in children. Reviews in Environmental Contamination and Toxicology, 186), 1.

O'Neill, R.A., Roads, G.K. & Wiese, R.N., 1993 On-site waste water treatment and disposal in New South Wales, Sydney.

Pang, L., Close, M., Goltz, M., Noonan, M. & Sinton, L., 2005 Filtration and transport).

Page 283

Payment, P., Fortin, S. & Trudec, M., 1986 Elimination of human enteric viruses during conventional wastewater treatment by activated sludge. Canadian Journal of Microbiology, 32), 922.

Petterson, S.R. & Ashbolt, N.J., 2001 Viral risks associated with wastewater reuse: Modelling virus persistence on wastewater irrigated salad crops. Water Science and Technology, 43(12), 23.

Petterson, S.R., Ashbolt, N.J. & Sharma, A., 2001 Microbial risks from wastewater irrigation of salad crops: A screening-level risk assessment. Water Environment Research, 73(6), 667.

Petterson, S.R., Ashbolt, N.J. & Teunis, P.F.M., 2000 Viral risks associated with wastewater reuse: Modelling virus persistence on wastewater irrigated salad crops. 1st World Water Congress of the International Water Association - Health-Related Water Microbiology Symposium:107.

Rao, V.C., Lakhe, S.B., Waghmare, S.V. & Raman, V., 1986 Virus removal in primary setlling of raw sewage. Journal of Environmental Engineering, 107), 57.

Regli, S., Rose, J.B., Haas, C.N. & Gerba, C.P., 1991 Modeling the risk from Giardia and viruses in drinking water. Journal of American Water Works Association, 83(11), 76.

Robertson, B., Forbes, A., Sinclair, M. & Black, J., 2000 How well does a telephone questionnaire measure drinking water intake? Aust. N. Z. J. Public Health Care Manag Sci, 24(6), 619

Scandura, J.E. & Sobsey, M.D., 1997 Viral and bacterial contamination of groundwater from on-site sewage treatment systems. Water Science and Technology, 35(11-12), 141.

Schijven, J.F., 2001. Virus removal from groundwater by soil passage: Modeling, field and laboratory experiments. University of Delft, The Netherlands., Delft.

Schijven, J.F., Hassanizadeh, S.M. & de Bruin, R.H.A.M., 2002 Two-site kinetic modeling of bacteriophages transport through columns of saturated dune sand. Journal of Contaminant Hydrology, 57(3-4), 259.

Page 284

Schijven, J.F., Mulschlegel, J.H., Hassanizadeh, S.M., Teunis, P.F. & de Roda Husman, A.M., 2006 Determination of protection zones for Dutch groundwater wells against virus contamination--uncertainty and sensitivity analysis. Journal of Water and Health, 4(3), 297.

Šimùnek, J., Šejna, M. & van Genuchten, M.T., 1998 The HYDRUS-1D software package for simulating the one-dimensional movement of water, heat and multiple solutes in variably saturated media.

Taylor, R., Cronin, A., Pedley, S., Barker, J. & Atkinson, T., 2004 The implications of groundwater velocity variations on microbial tranpsort and wellhead prtection - review of field evidence. FEMS Microbiology Ecology, 49), 17.

Teunis, P.F.M. & Havelaar, A., 2000 The beta Poisson dose-response model is not a single hit model. Risk Analysis, 20(4), 511.

Thompson, S.S., Flury, M., Yates, M.V. & Jury, W.A., 1998 Role of the air-water-solid interface in bacteriophage sorption experiments. Applied and Environmental Microbiology, 64(1), 304.

USEPA, 2002 On-site Wastewater Treatment Systems Manual. EPA/625/R-00/008.

Vaughn, J.M., Landry, E.F. & Thomas, M.Z., 1983 Entrainment of viruses from septic tank leach fields through a shallow, sandy soil aquifer. Applied and Environmental Microbiology, 45(5), 1474.

Ward, R.L., Bernstein, D.I. & Young, E.C., 1986 Human rotavirus studies in volunteers: Determination of infectious dose and serological response to infection. Journal of Infectious Diseases, 154), 871.

Westrell, T., 2004 Microbial risk assessment and its implications for risk management in urban water systems. Department of Water and Environmental Studies.

Westrell, T., Bergstedt, O., Stenström, T.A. & Ashbolt, N.J., 2003 A theoretical approach to assess microbial risks due to failures in drinking water treatment. International Journal of Environmental Health Research, 13(2), 181.

Page 285

Wheeler, J.G., Sethi, D., Cowden, J.M., Wall, P.G., Rodrigues, L.C., Tompkins, D.S., Hudson, M.J. & Roderick, P.J., 1999 Study of infectious intestinal disease in England: rates in the community, presenting to general practice, and reported to national surveillance. The Infectious Intestinal Disease Study Executive. Bmj, 318(7190), 1046.

Yates, M.V., Gerba, C.P. & Kelley, L.M., 1985 Virus persistence in groundwater. Applied and Environmental Microbiology, 49(4), 778.

Yates, M.V. & Jury, W.A., 1995 On the Use of Virus Tranpsort Modeling for Determining Regulatory Compliance. Journal of Environmental Quality, 24), 1051.

Yates, M.V. & Ouyang, Y., 1992 VIRTUS, a model of virus transport in unsaturated soils. Applied and Environmental Microbiology, 58(5), 1609.

Yates, M.V. & Yates, S.R., 1987 A comparison of geostatistical methods for estimating virus inactivation rates in ground water. Water Research, 21), 1119.

Yates, M.V., Yates, S.R., Warrick, A.W. & Gerba, C.P., 1986 Use of geostatistics to predict virus decay rates for determination of septic tank setback distances. Applied and Environmental Microbiology, 52(3), 479.

Page 286