Household experiences of flooding in Brisbane and Ipswich, Queensland

Transcription

1 Record 2016/30 ecat Household experiences of flooding in Brisbane and Ipswich, Queensland Results of Geoscience Australia surveys following flooding in South East Queensland in 2011 and 2013 Shelby Canterford, Valdis Juskevics, Miriam Middelmann-Fernandes, Martin Wehner and Won Oh APPLYING GEOSCIENCE TO AUSTRALIA S MOST IMPORTANT CHALLENGES

2

3 Household experiences of flooding in Brisbane and Ipswich, Queensland Results of Geoscience Australia surveys following flooding in South East Queensland in 2011 and 2013 GEOSCIENCE AUSTRALIA RECORD 2016/30 Shelby Canterford, Valdis Juskevics, Miriam Middelmann-Fernandes, Martin Wehner and Won Oh

4 Department of Industry, Innovation and Science Minister for Resources and Northern Australia: Senator the Hon Matthew Canavan Assistant Minister for Industry, Innovation and Science: The Hon Craig Laundy MP Secretary: Ms Glenys Beauchamp PSM Geoscience Australia Chief Executive Officer: Dr Chris Pigram This paper is published with the permission of the CEO, Geoscience Australia Commonwealth of Australia (Geoscience Australia) 2016 With the exception of the Commonwealth Coat of Arms and where otherwise noted, this product is provided under a Creative Commons Attribution 4.0 International Licence. ( Geoscience Australia has tried to make the information in this product as accurate as possible. However, it does not guarantee that the information is totally accurate or complete. Therefore, you should not solely rely on this information when making a commercial decision. Geoscience Australia is committed to providing web accessible content wherever possible. If you are having difficulties with accessing this document please clientservices@ga.gov.au. ISSN X (PDF) ISSN (Print) ISBN (PDF) ISBN (Print) ecat Bibliographic reference: Canterford, S., Juskevics, V., Middelmann-Fernandes, M., Wehner, M. and Oh, W., Household experiences of flooding in Brisbane and Ipswich, Queensland: Results of Geoscience Australia surveys following flooding in South East Queensland in 2011 and Record 2016/30. Geoscience Australia, Canberra.

5 Contents Executive Summary... 1 Brisbane, Ipswich and the flood events... 1 The surveys... 1 Warnings and risk perception... 2 Preparation, evacuation and moving back home... 3 Damage and repair... 4 Financial implications... 5 Medical impacts, mental health and support... 5 Discussion and conclusions... 6 Warnings and risk perception... 6 Mental health... 6 Financial impacts... 7 Friends, family and neighbours... 7 Returning to unrepaired homes... 7 Towards a framework of household resilience... 8 Conclusion Introduction Brisbane, Ipswich and flooding Historical flood events Floods Floods The surveys Survey design Survey sample Warnings and risk perception Warnings, warning relevance and sources Interpreting warnings Risk perception Chapter 4 summary Preparation, evacuation and moving back home Preparing for the floods Evacuation and relocation Returning home Chapter 5 Summary Damage and repair Flood damage Highlighted suburbs Fig Tree Pocket and Rocklea Disruptions to the household Repair and rebuild Highlighted suburbs Fig Tree Pocket and Rocklea Deterioration after repair Mitigating against future flooding...52 Household experiences of flooding in Brisbane and Ipswich iii

6 6.6 February 2013 follow-up field survey Chapter 6 summary Financial implications Extra costs and finance Working following the flood Property values Analysis of residential sales data Chapter 7 summary Medical impacts, mental health and support Medical impacts Subjective well-being Mental health Clinical diagnoses and treatment Relationships Insurance Other sources of stress and concern Positive impacts Survey Support Gratitude Chapter 8 summary Discussion and conclusions Warnings and risk perception Mental health Financial impacts Friends, family and neighbours Returning to unrepaired homes Towards a framework of household resilience Conclusion...85 Abbreviations and Acronyms...86 References...87 iv Household experiences of flooding in Brisbane and Ipswich

7 Executive Summary In January 2011 extreme flooding occurred in the south-east corner of Queensland resulting in over 17,000 homes inundated in the cities of Brisbane and Ipswich. In the weeks following the floods, Geoscience Australia (GA) conducted a field survey to assess the scale of the damage on individual homes. The engineering survey was supplemented with GA s first social survey, which was then repeated in 2013 following another flood event. This report presents the results of the social survey investigating the impacts of the 2011 and 2013 floods on flooded households in the Brisbane and Ipswich City Council areas. Brisbane, Ipswich and the flood events Brisbane is the capital of Queensland and in 2011 was home to 1.09 million people (ABS 2012). Ipswich is part of the Greater Brisbane area, lying approximately 35km to the south-west, and had a population of 172,000 (ibid). Both have experienced strong population growth over a number of decades, exposing an increasing number of people to natural hazards. Brisbane and Ipswich have both had a history of severe flooding. The first flood event was the result of an exceptional rain event in the second week of January 2011 following record levels of rain during December that saturated catchments and caused flooding (Executive Summary Figure 1) in the Brisbane and Bremer Rivers (BOM 2011). On the 13th of January 2011, the Brisbane and Bremer Rivers peaked at 4.46m (Brisbane City gauge) and 19.4m (Ipswich gauge) respectively. Around 17,100 properties were inundated over the course of three days; 3,000 in Ipswich and 14,100 in Brisbane (Queensland Floods Commission of Inquiry 2011). Many of the tributary creeks also flooded due to inflow or by being unable to drain into the flooded rivers. The second flood event occurred after Tropical Cyclone Oswald formed on 21 st January 2013 from a tropical low in the Gulf of Carpentaria (BOM 2013). The cyclone itself was very weak and made landfall as a Category 1 cyclone only 12 hours after being named. However, broader weather conditions at the time forced the resulting low south along the Queensland coast, causing extensive flooding in many communities in Queensland and New South Wales. Very heavy rainfall, that reached the second highest on record for two, three and four day totals, fell in the Brisbane River catchment between the 26th and 28th of January. Record levels of flooding were recorded in Lockyer Creek, and the Bremer River reached 13.9m at Ipswich. The Brisbane River reached a minor flood peak of 2.3m at Brisbane City (BOM 2014) and many of the creeks also experienced flooding. The surveys Following the flooding of the Brisbane-Bremer River catchment in 2011 a postal survey instrument was designed to examine the impact of flooding on the community, and was then replicated with minor amendments for the flooding in The first survey of the Brisbane and Ipswich floods in January 2011 was undertaken in April and May of The development of the survey instrument was the result of a collaboration between GA and a New Zealand Government agency, the National Institute of Wind and Atmospheric Research (NIWA). It was designed to provide input into a range of work programs at GA and NIWA, mainly the social, economic and engineering programs. The second survey was triggered by the smaller flood event in January 2013 which affected Brisbane and Ipswich. Household experiences of flooding in Brisbane and Ipswich 1

8 This later survey was undertaken in October and November The first survey drew on NIWA s prior survey experience and based the format and some questions on their own template. GA activity leaders were then consulted to determine their needs, and the resulting questions incorporated into the instrument. Executive Summary Figure 1 Locations of responding households for the 2011 and 2013 Brisbane and Ipswich surveys with the 2011 flood extent. Source: State of Queensland (Department of Natural Resources and Mines) 2012, The households invited to participate in the first survey were drawn from two earlier field survey activities conducted in Brisbane and Ipswich. Of the 5000 households invited to participate, 1267 responses were received, representing a 26% response rate. Of these respondents, 722 indicated they were willing to participate in further research and were mailed the second survey. A higher response rate was achieved in the second survey with 61% responding. The spatial distribution of affected households who responded to the surveys is mapped in Executive Summary Figure 1. Warnings and risk perception Warnings are an extremely important aspect of household preparedness for an impending flood event. While warnings are issued for most events, the effectiveness of reaching the target households can be variable. More than half of respondents did not hear a warning prior to the 2011 floods, and another 10% heard a warning but thought it was not relevant to them (Executive Summary Table 1). Two-thirds 2 Household experiences of flooding in Brisbane and Ipswich

9 of respondents received a warning prior to the 2013 floods, even though this was a much smaller event. There were no differences in households receiving a warning by location, with neighbours having very different experiences. In 2011 households received warnings most commonly from television, radio and friends, family or neighbours. In 2013 websites replaced friends, family or neighbours as a common warning source. For households that received a warning from one source only in 2011, most received the warning from friends, family or neighbours, but were also least likely to think the warnings were relevant. Respondents who heard warnings from three or more sources were most likely to think that the warnings were relevant to them. The number of households who received a warning increased between the 2011 and 2013 households, showing that experience is important, however experience did not always influence warning perception during the 2011 event. Executive Summary Table 1 Warning receipt and warning relevance, proportion of responding households, 2011 and 2013 floods. Warning receipt and relevance 2011 flood (%) (n=1237) 2013 flood (%) (n=407) Received a warning and the warning was relevant Received a warning but the warning was not relevant 9 20 Didn t receive a warning Percentages may not add to 100 due to rounding. The reasons why warnings were not relevant shows that the warnings were considered too general and respondents were unable to relate them to their circumstances. Warnings that included flood heights at a specific gauge or did not mention particular suburbs were not seen as relevant. Other respondents had previous advice or understanding that their home would not flood. This research shows the understanding of risk needs to be explored further. Many households were unaware that their homes could potentially flood, and so did not see the warnings as relevant. Respondents commented that they purchased or built above the 100 year average recurrence interval or 1974 flood level and were surprised to find that their home flooded during the 2011 event. Respondent s comments indicated that they did not realise that the broader understanding of flood levels can change with new modelling or analysis of historical events, and that changes upstream may change their risk. The answers to the warnings questions indicate that respondents have a much more complex view on what a warning looks and sounds like than expected. This view cannot be explained by the data collected in either survey. The complexity may be a result of the long length of time between the flood events and when the surveys were conducted, a time where memories are lost, altered and augmented. Preparation, evacuation and moving back home Adequate preparation and being aware of the potential danger can, to some extent, minimise the damage caused by floods. Some preparation activities must commence long before a flood is approaching, particularly mitigation activities and purchasing insurance. When the flood is imminent potential damage to the home cannot be avoided, however household actions can reduce the amount of damage to furniture and personal possessions. Household experiences of flooding in Brisbane and Ipswich 3

10 Many households were unaware of the risk of flooding before the 2011 floods were imminent, and so most preparation actions focused on flood response activities such as lifting furniture. The majority of households lifted furniture and personal possessions in 2011 with fewer removing items. However many households found this ineffective given the height of the water. Others regretted undertaking activities like sandbagging rather than saving more items, so in 2013 many more households removed items. For many households who responded to the 2013 survey, the experience and lessons learned from the 2011 flood influenced their behaviour in the later flood. The responses suggest that the experience gained in 2011 may have reduced the vulnerability of households to potential flood damage in When initially evacuating most households stayed with family or friends. Many households stayed with family or friends longer term, but private rental accommodation was also used by some. The average time to return back home was 3 months but the median time was only 3 weeks. Respondents whose homes were still to be repaired at the time of the survey returned home earlier than those whose homes were complete, indicating households were returning to homes still needing repair. The height of overfloor inundation, delays in finding tradesmen, having children aged under 5 and having insurance were all found to increase the length of time to return home. Households that included someone who volunteered in any organisation in the last year decreased the time to return home. Damage and repair The most noticeable impact of disasters is the damage to homes. Disaster are typically measured in the number of homes and other buildings impacted and the associated costs of repair and rebuild. Other items are also lost that mean more to households than their dollar value would imply. These personal items are a major part of the loss that households feel. Prior to the 2011 floods 71% of respondents thought that they were covered by flood insurance. However only 32% of all households had their claim fully paid, and a further 13% received only a partial payment. On average respondents received a damage bill of $140,500 of which $72,700 was paid from their own pockets. Extrapolating to all 17,100 homes the total damage bill was $2.4billion with the costs equally shared by households and insurance companies. Not all damage can be measured as dollar values, with irreplaceable items such as photographs, personal records and collections also lost. Items normally considered replaceable, such as books, were often frequently noted. Respondents noted the sentimental value of these items as they are associated with particular people and events. On average it took 95 days for repairs or rebuilding to start, with delays associated with the insurance companies or a lack of tradesmen. At the time of the first survey, conducted 15 months after the 2011 floods, three-quarters of households had completed repair or rebuild, with 4 per cent not started. The majority of respondents repaired their homes like-for-like with no mitigation for future flooding. Some respondents that were insured noted that the insurance companies would not pay for any changes to the home, while others found the costs of mitigation too high. More than a third of respondents noted deterioration such as warped flooring or mould following the repair of the home. The deterioration was greatest in households that left the frames to dry for less than 90 days and homes that received overfloor inundation of 0.15m or less. The impacts of the floods were not distributed evenly across the suburbs. An analysis of two areas within the suburbs of Rocklea and Fig Tree Pocket showed that the damage costs were higher in Fig Tree Pocket than in Rocklea. The increased costs are related to the higher median house price in Fig Tree Pocket rather than access to insurance or levels of flooding, which were similar for the two areas. Further, looking at the median weekly income for each area shows that Rocklea households 4 Household experiences of flooding in Brisbane and Ipswich

11 would find it harder to pay for the repair from household budgets. Households in Rocklea would need 18 months of income to pay the repair if they could dedicate their entire income, while households in Fig Tree Pocket would repay the cost in a year. This highlights the problems with measuring the severity of a disaster by damage costs alone. Financial implications Even respondents who were fully covered by insurance found they were required to find additional sources of funds to pay for items such as temporary accommodation, increased utility charges (mainly from excess water used to clean the home), cleaning and increased transport costs to get to and from work (due to road closures). Damage to swimming pools, gardens and fences were not usually covered by insurance and resources had to be found to repair and replace these items. Out-of-pocket expenses were mainly funded by household savings in 2011 however government assistance was also utilised. In 2013 expenses were mostly funded by savings alone. Reconstruction expenses were also funded by savings, but nearly half of the respondents also utilised insurance claims and government assistance. Two-thirds of households had one or more persons unable to work following the flood. On average household income was reduced for a period of 6 months in 2011, reducing to 19 days following the 2013 floods. Homeowners were concerned that the drop in resale value of their home, prevented them from moving to a less flood-prone location. Analysis of individual sales show that while there was a significant drop following the 2011 floods, values had begun to recover by the end of By the end of 2014 median sales prices had dropped universally, even for not flooded areas, indicating that sales prices were under the influence of broader market conditions. Medical impacts, mental health and support Physical health and mental health are closely related. Stress and poor mental health can result in poor physical health. Conversely people who suffer from health problems are more likely to have declining mental health. It is sometimes difficult to determine which came first, the physical or mental health problems. A quarter of respondents reported a householder requiring medical treatment following the 2011 floods, with some acquiring long term health conditions that were still evident at the 2013 survey. Poor mental health is seen as a consequence of disasters, but it can also have broader implications. Poor mental health can slow recovery following a disaster and have financial implications for those who are unable to work. When poor mental health exists prior to a disaster, it may cause people to react inappropriately when warnings are issued. Declining mental health was reported by many respondents as a major impact of the floods that was, at times, seen to be of greater impact than the damage to homes. Some respondents were receiving treatment such as anti-depressants, and others contemplated suicide. In the 2013 survey respondents indicated that the 2011 floods were still impacting on their mental health, with 72% indicating they became nervous or anxious when it rained. Some respondents reported that the flood was a galvanising event, from which they had learnt to appreciate their friends and community better. Practical and emotional support from friends, family, neighbours and even work colleagues was a common theme throughout the surveys, highlighting the importance of personal networks and community. Respondents had mixed feelings about the mud army. On the one hand they were grateful for the help provided by strangers. On the other the volunteers were over-enthusiastic and threw things out indiscriminately. Household experiences of flooding in Brisbane and Ipswich 5

12 Discussion and conclusions The factors that influence a household s resilience are numerous and complex. The survey instrument used was quite broad in nature in an attempt to capture some of this complexity. While to some extent GA has been successful in doing so, the results cannot be organised into a single finding, but instead are grouped into five main themes. The surveys have also uncovered a number of topics that require further investigation. The topics discussed within the main themes are some of the key research needs identified by GA, and do not reflect all of the gaps in the large and complex body of household resilience research. Warnings and risk perception Understanding why households acted the way they did requires an understanding of their risk perception and whether they received a warning. Risk perception is the understanding of the likelihood of a disaster and how it will impact. Risk perception influences how warnings are interpreted and a household s subsequent behaviour. Households felt that they were not given adequate and timely warning of the floods. Yet many of these households did receive a warning, but it was not recognised or, in some cases, ignored. Some respondents felt that the warnings were not intended for them because they did not think they could flood, that is, their risk perception was not appropriate for their situation. Their cultural understanding of the behaviour of the river, the effectiveness of land use planning, and the role of Wivenhoe Dam was that flooding did not occur in their area. Household risk perception must be appropriate before a flood is imminent, and changing the wording of a warning is unlikely to change a household s behaviour. Respondents received warnings from a range of sources, including from friends, family and neighbours, with all of them considered both effective and unhelpful by different respondents. Those who received warnings from multiple sources were more likely to perceive warnings as relevant than those who received them from one source. Warning effectiveness can be increased by recognising that households will seek confirmation or more information. Relevant authorities can facilitate this by including links or references to other sources when warnings are issued. Social connections also have an important role to play in warnings and should not be ignored. The clarity of warnings was an issue for many respondents. Warnings need to contain enough information for households to be able to relate them to their personal circumstances. To improve the clarity of warnings respondents suggested introducing flood height markers on homes, or at points within suburbs, and maps presenting inundation and depth according to a range of gauge heights. Mental health Mental health issues are a major impact of disasters as well as a factor that makes households more vulnerable. The GA surveys did not explicitly ask about mental health, preferring to use an alternative measure, subjective wellbeing. Despite this, the impacts on mental health were commonly reported by respondents. Poor mental health had far greater and longer lasting impact on respondents than the damage to the home. For some, these impacts persisted in the 2013 survey, even though the home was repaired and not inundated in As a result of the stress of the floods some families split, household members became sick, others needed psychological treatment or were unable to work, and some contemplated suicide. 6 Household experiences of flooding in Brisbane and Ipswich

13 Some households experienced positive impacts from the floods. These households gained an appreciation for, and grew closer to their families and friends. Some respondents felt a new sense of community spirit, and others became less attached to their possessions, recognising that these are replaceable, while people are not. Financial impacts Quantifying the full costs of disasters, direct and indirect, will allow more appropriate economic models to be built when assessing the costs and benefits of mitigation and resilience building activities. The GA surveys did not quantify all the costs of the impacts that households experienced, however has collected information on the range of impacts to indicate the type of costs incurred. Households experienced a range of financial impacts, even if they were insured and, in the case of the 2013 floods, even if they were not flooded. The 2013 event saw many households prepare for the possibility of flooding. Even though most of these households were not inundated, they incurred costs such as increased living expenses and storage. The longer term impacts of using financial reserves to fund recovery were not studied, but some households indicated that their lifestyle was impacted with retirement or holiday plans put on hold. Understanding how households change their spending patterns after the rebuilding is complete, and how this impacts on household members, is an area that needs further research. Friends, family and neighbours Friends, family and neighbours provided a common thread throughout the surveys, and had a role to play in all the main findings. Having social networks that can be called on in times of trouble are a key feature of social capital. The OECD (2001, p.41) defines social capital as: "networks, together with shared norms, values and understandings which facilitate cooperation within or among groups. Unlike physical capital, social capital is not depleted when it is used, rather it is enhanced. Family groups usually help each other, but groups of friends and neighbours extend the networks to others who can help together. The power of social capital allows strangers to cooperate to help others, as seen by the mud army. Networks are strengthened, and in another time of trouble, previously affected households are more likely to reciprocate the kindness previously shown to them. Social capital provides opportunities to minimise the impact and costs of disasters. This has been realised in the costs of managing evacuation centres, with few households requiring a place of refuge as they stay with families or friends. The role of friends and family in warning households and encouraging them to evacuate provides an opportunity for risk awareness campaigns. There are also opportunities to be gained by building cohesive communities and neighbourhoods that can support each other. Returning to unrepaired homes One of the biggest shortcomings of this research is that GA did not gain an adequate understanding of households returning home to unrepaired homes. From the data and the concluding comments it is clear that some households moved back into damaged and flood affected homes as a forced coping strategy. What is not clear from the surveys is how effective this is in allowing households to maintain their lifestyles, though some comments show it is an ineffective strategy. The numbers of households moving back home early, and the amount of work still required, needs to be matched to health, mental health and more detailed financial measures to understand the full impacts of this strategy. Household experiences of flooding in Brisbane and Ipswich 7

14 Towards a framework of household resilience The main themes presented here are forming the foundation of a framework to understand household s resilience to disasters and to guide future research. This framework is still in development but is evolving around four key themes: Social capital Health and mental health Economic resources Risk perception. The fifth main theme presented in the previous paragraph, returning to unrepaired homes, highlights the difficulties in assuming that households have recovered by simple measures such as returning to the flooded home. The time to return to home is still a possible measure of the resilience of a community, however it needs to be qualified by the state of the home. Further, issues such as poor mental health, lack of access to support, and financial impacts need to be factored into any outcomes measure. Conclusion This report has presented the results of research conducted by GA in Brisbane and Ipswich following the 2011 and 2013 floods. These surveys have been successful in providing insights into household responses to and recovery from disaster. The results of these surveys have already fed into the national review of warnings, local government planning processes, and GA engineering loss models. The survey results have also been used as the basis of a framework for household resilience at GA, which in turn is being used to guide future research. The research presented here is not an endpoint, and much work remains to be done to understand household resilience to disasters in order to reduce the adverse impact of future disasters on people. 8 Household experiences of flooding in Brisbane and Ipswich

15 1 Introduction In January 2011 extreme flooding occurred in the south-east corner of Queensland. The cities of Brisbane and Ipswich saw over homes inundated. In the weeks following the floods, Geoscience Australia (GA) conducted a field survey to assess the scale of the damage on individual homes. GA has conducted many of these surveys since 2001, looking at hazards such as cyclones, tornadoes, earthquake, bushfire and flood. However in 2011 the Brisbane and Ipswich survey, along with its engineering purposes, was the start of a framework for GA s first social survey to follow. Geoscience Australia (GA) is an Australian Government listed entity within the Industry, Innovation and Science portfolio. GA is the geoscience advisor to the Australian Government on matters of national importance, and custodian of the geographical and geological data and knowledge of the nation. By understanding the geological history and geospatial aspects of our continent, we build our knowledge and capability to prepare us for the future. GA delivers a wide range of products that address important and significant national issues to assist government and the community to make informed decisions about the use of natural resources, the management of the environment, and community safety. Geoscience Australia provides applied research and information for the Australian community in six key strategic priority areas, to obtain sustainable economic, social and environmental benefits: Building Australia's Resource Wealth Ensuring Australia's Community Safety Securing Australia's Water Resources Managing Australia's Marine Jurisdictions Providing Fundamental Geographic Information Maintaining Geoscience Knowledge and Capability The second key strategic priority area, ensuring Australia s community safety, relates to GA s role in supporting Australia's capability to manage the impact of natural hazards. Natural hazards, particularly floods, cyclones, bushfires and earthquakes, have a significant impact on the economy, the environment and society. The direct costs of building damage, infrastructure impacts and business losses are significant, as are the Government's outlays in relief and recovery payments. Australia's ability to effectively mitigate the impacts of natural hazards and disasters is contingent upon the availability of information on the incidence, impact and mitigation opportunities for specific hazards along with early warning and advice. Geoscience Australia supports the Government through research to develop an understanding of natural hazards and community resilience. Drawing on its research, GA provides authoritative and independent information and advice that can inform decisions on preparation and response to severe hazard events. This report presents the results of GA s first social survey investigating the impacts of the 2011 floods on households in Brisbane and Ipswich combined with the results of a follow up survey conducted following the 2013 floods in the same area. The data from these surveys has been used separately in the development of GA s building vulnerability models, to guide early development of a framework for household resilience and to shape future research directions. Some of the data tables have been used Household experiences of flooding in Brisbane and Ipswich 9

16 by local governments to guide their emergency plans. An earlier version of Chapter 4: Warnings was published by GA (Canterford and Juskevics 2014) to contribute to the National Review of Warnings conducted by Emergency Management Victoria with funding from the Commonwealth Government s National Emergency Management Project (NEMP) Grants Program (EMV 2014). Chapter 2 discusses the Brisbane and Ipswich City Council areas and the history of flooding in the region. Chapter 3 details the survey method and outlines how the survey instrument was constructed. Chapters 4 to 8 discusses the results of the surveys, grouped by theme. The initial survey covered a broad range of topics. These topics are grouped together into themes, with the themes determined partly by common concerns in responses, but are also grouped in rough order of the sequence of events from warnings to long term impacts. These themes are being used as a starting point to construct a framework for understanding household resilience and to guide future research. Chapter 4 looks at household risk perception prior to the floods, and the warnings received when the 2011 floods were imminent. Warnings are an important part of protecting life and allowing households to protect their possessions. However merely issuing warnings is not sufficient; households need to understand the relevance of the warnings to their situation. Risk perception refers to the household s understanding of the likelihood, severity and impacts of a given risk. Understanding households risk perception gives insights into how households behave before, during and after a disaster. Following the receipt of a warning, households will then prepare for and respond to the onset of flooding. Chapter 5 investigates how and when households prepared for the floods. Where households evacuated to and when they returned to their homes is also discussed. Chapter 6 discusses the damage caused by the floods and how households have repaired their homes. A particular focus of this chapter is the actions that households have taken to mitigate any future flooding. Disasters have costs beyond the insurance costs often reported in the media. Chapter 7 discusses some of the broader costs of the floods to households, including employment impacts, and how households have financed those costs. One of the costs of a disaster is the reduction in the value of an affected property. Chapter 7 concludes with an analysis of house sales data to measure this impact. The final analysis chapter is focused on the health effects of the floods, discussing medical impacts, mental health and social support. Declining mental and physical health can be a major impact of disasters that is far greater than the cost of damage to the home. The GA surveys have not directly measured mental health, however some respondents have provided concluding comments that has allowed some analysis of respondent s health. The concluding chapter brings together some of the key findings of the surveys grouped into themes. The conclusions discuss how these themes are important to household resilience and how they are being applied in future research conducted by GA. 10 Household experiences of flooding in Brisbane and Ipswich

17 2 Brisbane, Ipswich and flooding The survey data presented in this report is collected from within the flood footprint covered by the urban areas of Brisbane and Ipswich City Councils in South-East Queensland. Brisbane is the capital of Queensland and in 2011 was home to 1.09 million people (ABS 2013a). Ipswich is part of the Greater Brisbane area, lying approximately 35km to the south-west, and has a population of 172,000 (ibid). Both Brisbane and Ipswich have experienced strong population growth over a number of decades (Figure 2.1), exposing an increasing number of people to natural hazards. Figure 2.1 Historical Census population of the Brisbane Greater Capital City Statistical Area, which includes Brisbane and Ipswich, as well as Logan, Redlands and Moreton Bay local government areas. Source: Australian Bureau of Statistics, Australian Historical Population Statistics, 2014, Cat no While Brisbane and Ipswich are affected by a number of natural hazards, flooding is the greatest natural hazard risk across the region (Granger and Hayne 2001). The Brisbane-Bremer River system which flows through Brisbane and Ipswich, has a catchment area of 13,750 square kilometres (BCC 1998), about half of which is below Wivenhoe Dam (BOM 2011). The Brisbane River has its headwaters in the Great Diving Range, flowing south-easterly to Wivenhoe Dam. From the dam, the Brisbane River continues south-east, before turning north-east and draining into Moreton Bay. The city of Brisbane developed along the lower reaches of the Brisbane River. The Brisbane River is tidal for 40-60km upstream from the river mouth (Bell & McEwan 2010). Two major tributaries flow into the Brisbane River below Wivenhoe Dam; Lockyer Creek and the Bremer River. Lockyer Creek is the largest of the sub-catchments of the Brisbane River (Middelmann, Harper and Lacey 2001) and joins the Brisbane River just south of Wivenhoe Dam. The Bremer River represents the southernmost corner of the Brisbane River catchment. It flows from the MacPherson Ranges in the south, through Ipswich, and into the Brisbane River near Moggill. The Brisbane-Bremer River catchment has a history of major floods (BOM 2011b). Large amounts of rainfall across the catchment are likely to cause major flooding in the Brisbane and Bremer Rivers. Household experiences of flooding in Brisbane and Ipswich 11

18 Average catchment rainfalls in excess of mm in 48 hours may result in moderate to major flooding (BOM 2011; BOM 2011a). When the Brisbane River is in major flood, backwater flooding can also cause flooding in Ipswich (Middelmann, Harper and Lacey 2001). Flooding in Brisbane more frequently occurs as a result of the flooding caused by local creeks, including the Oxley and Bulimba Creeks on the south side, and Kedron Brook, Moggill, Enoggera, Cabbage Tree and Downfall-Nundah Creeks in the northern and western suburbs (BOM 2011). Widespread flooding in Ipswich can result when heavy rainfall occurs in the Bremer-Warrill headwaters (Middelmann, Harper and Lacey 2001). Localised flooding in Ipswich can be caused by the Deebing, Bundamba and Woogaroo Creeks (BOM 2011a). Rainfalls in excess of 35mm in three hours may result in minor flooding in the most of the creeks (BOM 2011; BOM 2011a). Average catchment rainfalls in excess of 100mm in three hours may result in major flooding. The Oxley Creek catchment is more sensitive, and average catchment rainfalls in excess of 100mm in 6-12 hours may result in major flooding (BOM 2011). 2.1 Historical flood events Figure 2.2 shows the annual maximum flood peak for the Brisbane River at the Brisbane City gauge (BOM 2011). Figure 2.3 shows the river levels for the Bremer River at the Ipswich gauge (BOM 2011a). Both rivers have exceeded the major flood level many times since the 1820s. Figure 2.2 Historical highest annual flood peak in the Brisbane River at Brisbane City gauge, 1841 to Source: Geoscience Australia presentation of Bureau of Meteorology data. Particularly severe or notable floods have occurred in Brisbane on four occasions prior to The first recorded severe flood occurred in 1841 and remains the Brisbane River's highest flood in 12 Household experiences of flooding in Brisbane and Ipswich

19 recorded history peaking at 8.43m at the City gauge (BOM 2011b). In 1893 a flood known as the Great Flood or the Black February flood peaked at 8.35m. Nearly a third of all households in Brisbane were made homeless as a result of the floodwaters (BOM 2011). This flood was followed by a second peak of 8.09m a fortnight later. The fourth major flood event occurred in 1974, and while smaller in extent than the floods of 1841 and 1893, caused more damage due to the greatly increased population in the region (BOM 1974). In total 8,500 houses were flooded, with some washed away completely, and 14 people died. Floods greater than the 2011 flood event were also recorded in 1844, 1890 and However, the impact of the 2011 floods was much greater due to the larger population. Figure 2.3 Historical highest annual flood peak in the Bremer River at the Ipswich gauge, 1840 to Source: Geoscience Australia presentation of Bureau of Meteorology The Bremer River has flooded frequently with annual flood peaks exceeding the major flood level of 11.7m at the Ipswich gauge (also known as David Trumpy Bridge) 22 times since 1840 (Figure 2.3). Some areas and homes can flood at levels as low as 7.4m (BOM 2011a). The highest recorded flood was in 1893 which peaked at 23.47m at the Ipswich gauge. In the following century, the Bremer River peaked at 20.7m in the 1974 floods. These two events illustrate that extreme flooding levels can occur in the Bremer River. Both events coincided with flooding in Brisbane and were probably influenced by backwater from the Brisbane River (Babister 2011). Wivenhoe Dam was designed following the 1974 floods to provide both a secure water supply to the region and flood mitigation capacity. Completed in 1984, the dam has a water supply capacity of 1.17 million ML and a flood mitigation capacity of 1.97 million ML (DEWS 2014). Because of the extra capacity, the dam's normal operating capacity is 100%, and capacities up to 200% occur when the dam has been used for flood mitigation. The flood mitigation capacity is designed to reduce the 1974 flood level by 2m, however the dam does not eliminate the possibility of floods to Brisbane (SEQWater 2009). Wivenhoe Dam has no ability to reduce flooding in Ipswich except where the Bremer River Household experiences of flooding in Brisbane and Ipswich 13

20 floods as a result of backflow from the Brisbane River. Brisbane may also experience inundation as a result of flooding in the Bremer River and Lockyer Creek which flow into the Brisbane River below Wivenhoe Dam. Wivenhoe Dam is a zoned earth and rock fill embankment dam. This type of dam structure is not able to cope with overtopping, so water releases at times of high inflows are required to guarantee dam safety (DEWS 2014). These releases may also inundate properties Floods Figure 2.4 Flood footprint for the Bremer and Brisbane River systems, Source: State of Queensland (Department of Natural Resources and Mines) 2012, After record levels of rain during December 2010 that saturated catchments, an exceptional rain event in the second week of January 2011 caused flooding (Figure 2.4) in the Brisbane and Bremer Rivers (BOM 2011). On 13 January 2011, the Brisbane and Bremer Rivers peaked at 4.46m (Brisbane City gauge) and 19.4m (Ipswich gauge) respectively. Around 17,100 properties were inundated over the course of three days; 3,000 in Ipswich and 14,100 in Brisbane (Queensland Floods Commission of Inquiry 2011). Many of the tributary creeks also flooded through inflow or by being unable to drain into the flooded rivers. According to the GA survey qualitative data, many residents also stated that they were flooded from surcharging stormwater drains before they were flooded from the direction of the Brisbane River or nearby creeks. 14 Household experiences of flooding in Brisbane and Ipswich

21 Floods Tropical Cyclone Oswald formed on January 21 from a tropical low in the Gulf of Carpentaria (BOM 2013). The cyclone itself was very weak and made landfall as a Category 1 cyclone only 12 hours after being named. However, broader weather conditions at the time forced the resulting low south along the Queensland coast, causing extensive flooding in many communities in Queensland and New South Wales. Very heavy rainfall, that reached the second highest on record for two, three and four day totals fell in the Brisbane River catchment between the 26th and 28th of January. Record levels of flooding were recorded in Lockyer Creek, and the Bremer River reached 13.9m at Ipswich. The Brisbane River reached a minor flood peak of 2.3m at Brisbane City (BOM 2014) and many of the creeks also experienced flooding. Household experiences of flooding in Brisbane and Ipswich 15

22 3 The surveys Following the flooding of the Brisbane-Bremer River catchment in 2011 a postal survey instrument was designed to look at the impact of flooding on the community. The survey instrument was then replicated with minor amendments for the flooding in The first survey of the Brisbane and Ipswich floods in January 2011 was undertaken in April and May of The development of the survey instrument was the result of a collaboration between GA and the New Zealand Government agency, the National Institute of Wind and Atmospheric Research (NIWA). The second survey was triggered by a smaller flood event in January 2013 which affected Brisbane and Ipswich. This later survey was undertaken in October and November The second survey instrument was designed to link to the first survey and sought to identify if household reactions to the second flood event in 2013 had changed from that in the earlier event in The postal questionnaire approach was selected for two reasons. Firstly, the breadth of information required to serve both GA s and NIWA s needs made an interview approach too long and cumbersome. The second reason related to the need to have statistically significant results for the engineering component. The cost of such a large sample made conducting interviews unviable. 3.1 Survey design The first survey instrument was designed to provide input into a range of work programs at GA and NIWA, mainly the social, economic and engineering programs. The survey drew on NIWA s prior survey experience and based the format and some questions on their own template. GA activity leaders were then consulted to determine their needs, and the resulting questions incorporated into the instrument. Copies of the survey instruments can be obtained by contacting Geoscience Australia. Section A of the survey covered subjective well-being, and is taken from the subjective component of the adult version of the Comprehensive Quality of Life Scale (ComQOL) (Cummins 1997). Subjective well-being is an individual s own assessment of their quality of life and has long been linked to mental health (Keyes 2006). The ComQOL is included at the start of the survey as the responses to this section can be influenced by any questions that may precede it. GA does not have any experience or qualifications in assessments of mental health, however mental health has an important role in resilience to disasters. Subjective well-being was selected as an alternative measure to represent the mental health of survey respondents. The ComQOL is a freely available instrument designed to be self-administered by the general population (Cummins 1997), so was considered ideal for use in the GA surveys. The full ComQOL instrument contains objective and subjective axes, each covering the seven domains of: material wellbeing, health, productivity, intimacy, safety, place in community, and emotional well-being. The two axes measure different aspects of well-being and are not normally correlated with each other. The GA survey instrument is a lengthy instrument, and well-being was not the main priority for collection, so the available space for well-being questions was limited. The ComQOL is estimated to take minutes to complete, with the subjective axis estimated to take only a third to a quarter of that time at about 5 minutes, making it the easier of the two to accommodate. Subjective well-being in its various forms is commonly assessed in disaster literature (for example Rehdanz et al 2015; Tiefenbach and Kohlbacher 2014; Weinberg and Cummins 2012). ComQOL is an older index that has been replaced by the Personal Well-being Index. Cummins (2002) outlines the reasons for the change, but also notes 16 Household experiences of flooding in Brisbane and Ipswich

23 that data generated from the use of ComQOL remain valid and reliable measures of life quality. As such the results presented in this report are considered valid, but should be interpreted in that context. Section B contains background demographic and household questions that are aligned to GA's vulnerability indicators (Canterford 2009) with three additional questions. The first question relates to whether there were enough motor vehicles to evacuate everyone, to test the differences between having no motor vehicles or not enough motor vehicles. The second question asked whether anyone in the household volunteered with an emergency service organisation to test the significance of this kind of activity to risk perception. The final question asked whether the household had previously been impacted by a disaster to test the impact of experience. Section C asks about flood impacts including damage, cleaning and repair, evacuation, insurance, risk perception, risk communication and damage costs. The risk communication section includes questions that were collected as background information for the development of the Australian Flood Risk Information Portal as well as the general research program. Questions about damage and costs were designed to aid the development of engineering vulnerability models at GA (detailed in Wehner et al, in preparation) as well as cost models at NIWA. The final section includes questions about employment and income impacts, utility disruption and health. These questions are mainly intended to inform economic loss model development at GA and NIWA, however they are also relevant to household impacts. At the end of the formal survey questions, respondents were asked if they wished to provide supporting commentary about their survey responses or to provide additional information about the household s experiences during the floods. In the 2011 survey 59% of respondents provided additional comments and in the 2013 survey 51% of respondents provided comments. Only one page was provided for comments but respondents were encouraged to attach separate pages if they needed more room. Respondents were also asked if they would be willing to participate in follow-up research. The primary purpose of the second survey was to replicate the first survey instrument in order to enable comparisons between household responses to the two flood events. However, in order to improve the quality of responses, a number of small changes were implemented to question wording and question order. In particular, the questions on warnings and understanding risk were modified in the second survey as it was clear from the responses to the first survey that these questions were not interpreted as intended. 3.2 Survey sample The sample of households invited to participate in the first survey was drawn from two earlier field survey activities conducted in Brisbane and Ipswich. The first of these was a Rapid Damage Assessment (RDA) of community needs in the immediate flood aftermath conducted by the Queensland Government in January The second was a survey of building damage conducted by GA over January and February 2011, which collected detailed, spatially coded information on flood damage and included photos of the damaged houses. All households whose building had been surveyed in the latter survey were invited to participate in the postal survey (777 households). From the RDA dataset, a further 4223 affected households were randomly selected and invited to participate. A quarter of the households invited to participate from the GA field survey responded, allowing this data to be linked to the original damage data. Of the 5000 households invited to participate, 1267 responses were received, representing a 26% response rate. The spatial distribution of affected households who responded is mapped in Figure 3.1. Nine per cent Household experiences of flooding in Brisbane and Ipswich 17

24 (462) of surveys were returned unopened. These may indicate addresses where households had not returned to their homes at the time of the survey, or rental homes that have not yet been re-let. The demographics of the survey respondents were compared to the 2011 Census to assess how representative the sample is to the general population. Only demographics that were collected on a household basis were able to be compared, as shown in Table 3.1. The Census data used is from the first release of the 2011 Census of Population and Housing (ABS 2012). The Census data was obtained as part of an information consultancy with the Australian Bureau of Statistics, and represents the number of dwellings that contain a person of a particular characteristic. The Census statistics were compiled by aggregating all Statistical Area 1 polygons that contained a survey respondent. Table 3.1 Survey demographics compared to Census data, proportion of households or respondents, 2011 floods. Characteristic 2011 Census (% households) 2011 Survey (% respondents) Home owned (with or without mortgage) Household has no access to a motor vehicle Household includes a child under Household residents are all/includes someone age 65 or over Household includes someone who speaks English not well or not at all No one in the household has completed year 12 or higher Household includes someone who has a need for assistance with self-care activities or communication Household includes someone who volunteered in the last year Source: ABS The survey question is measured differently to the Census data, in that the Census is measuring households where everyone is aged 65 or over, whereas the survey asked if anyone was aged 65 or over. The sampling technique used for data collection entailed sending a survey form to an address that was flooded. This meant that only people who were still living in the dwelling, or had mail redirections, received a survey form. The consequence of this sampling technique was that renters, who had generally moved to a new home elsewhere, were not captured. This is apparent with 92% of the survey respondents owning the dwelling, with or without a mortgage, compared to 64% in the Census. This was a known weakness of the sampling technique at the time of the survey, but there was no opportunity to correct it. Over-sampling owner-occupied dwellings has had consequences to many of the other demographic variables. More than twice as many respondent households contain someone aged 65 or over when compared to the Census, at 32% and 14% respectively. However the Census data measures where all members of the household are aged over 65, not just some or one, so the sample difference is likely to be not as great. When compared to the Census the survey sample also included more people with a need for assistance with self-care activities or communication (7% compared to 11%). There are also more households that include people who have lower levels of English (4% compared to 6%), however it should be noted that there are measurement differences that may have an impact here, where the Census identifies people who speak English not well or not at all, and the survey asks if the household includes someone who does not have sufficient English skills to understand an emergency warning. Conversely the survey has fewer households with no access to a vehicle (9% in the Census compared to 5% in the sample) and more households that volunteered in the last 12 months (34% 18 Household experiences of flooding in Brisbane and Ipswich

25 compared to 39% in the sample). Despite the apparent older age profile, the number of households that contain children aged under five is similar to the Census, with both at 12%. The number of households where no one has completed year 12 or higher is also comparable at 12%. Figure 3.1 Locations of responding households for the 2011 and 2013 Brisbane and Ipswich surveys with the 2011 flood extent. Source: State of Queensland (Department of Natural Resources and Mines) 2012, All of the 722 respondents that indicated in the first survey that they would be willing to participate in further research were invited to participate in the 2013 survey. Unsurprising, a higher response rate was received in the second survey at 61%, equating to 440 responses (mapped in Figure 3.1). A pre-approach letter was sent before the first survey form was mailed, to give households advance warning of the upcoming survey. Both surveys were sent out as a paper-based questionnaires, with participants able to either respond via the paper copy (using the reply paid envelope provided), or electronically via the online survey tool Survey Monkey. Households were given up to four weeks to respond, however a few households reported that the forms did not arrive until the week of the due date. A few days after the due date, a reminder letter was sent to households that had not returned their forms. Data entry and analysis was performed at GA by GA staff, many of whom contributed to the design of the survey instrument. Tables were prepared in Microsoft Excel, statistical analysis was conducted in the R statistical package, and spatial analysis was conducted in ArcGis. Household experiences of flooding in Brisbane and Ipswich 19

26 4 Warnings and risk perception Warnings are an extremely important aspect of household preparedness for an impending flood event. While warnings are issued for most events, the effectiveness of reaching the target households can be variable. This chapter presents the results relating to warnings prior to the 2011 and 2013 floods. How the warnings were received are presented along with their perceived relevance by survey respondents. Questions relating to warnings are best asked shortly after an event when details and memories are not altered with subsequent events, media reports or discussions with other residents. As both household surveys were conducted months after the respective 2011 and 2013 events 1, the results need to be approached with the understanding that people responded with what information they remembered from the event. While some details may have been missed, and others added, the questions on warnings can still provide important insights into the respondents behaviour. The responses to the questions about warnings and their relevance were at times inconsistent across questions. Some respondents answered that they didn t receive a warning in the initial question, but then would answer subsequent questions (such as relevance and sources) as if they had received a warning (for an example contrast Table 4.1 and Table 4.2). This anomaly may be a result of respondents changing their thinking on warnings as the survey questions progressed, particularly once a list of warning sources was presented to select from. The inconsistencies may also be due to the vagaries of human memory, as mentioned before, or may represent a more complex view of warnings that was not able to be captured by the survey. Without understanding the reason for the inconsistencies, no attempt has been made to balance responses when presenting the data, and so each table should be interpreted individually. 4.1 Warnings, warning relevance and sources Respondents were asked whether they had received a warning prior to the 2011 and 2013 floods. Just over half of respondents (54%) indicated that they didn t receive a warning at all in advance of the 2011 floods (Table 4.1). This was supported by many in their concluding comments at the end of the survey. We were not given any warning water was coming. We had to make our own decision to evacuate premises. (Corinda, 2011 survey) Even after the flood the Courier Mail did not list our suburb as being flood affected. We were not given any warning. No one came to our street, no police, no SES, no council workers. (Westlake, 2011 survey) Respondents were then asked whether they thought the warnings received were relevant to their house. Table 4.1 combines the responses to both questions. 1 The first household survey was sent out 15 months after the 2011 floods. The second household survey was sent out 9 months after the 2013 floods. The surveys were conducted a long time after the flood events because the survey instruments were designed to measure other variables, for example length of time to move back into the dwelling, and these requirements outweighed the advantage of obtaining better data on warnings for the purposes of the GA research. 20 Household experiences of flooding in Brisbane and Ipswich

27 Table 4.1 Warning receipt and relevance, proportion of responding households, 2011 and 2013 floods. Warning receipt and relevance 2011 flood (%) (n=1237) 2013 flood (%) (n=407) Received a warning and the warning was relevant Received a warning but the warning was not relevant 9 20 Didn t receive a warning Percentages may not add to 100 due to rounding. The results of the survey highlight that even when a warning was received, some respondents were not able to relate the warnings to their own situation. Questions relating to the relevance of the warnings, and the comments provided by survey respondents, highlight that warning recognition was low during the 2011 floods. Our neighbour who had been in his house in 1974 said how high the flood had come then (over the roof) and I had friends had told me lots of stories about the 1974 floods (I arrived in Brisbane 20 years ago). It was however obvious that friends who had not had access to this information and were much more recent arrivals, were even less prepared for the flood than we were. I guess the thinking is either it's all unreal or it can't happen to me. They left the decision to evacuate very late indeed and saved very little. Unless you have had some experience it is difficult to take the warnings seriously, and it's difficult to be really aware what they mean and the implications of what can happen. (Graceville, 2011 survey) The experience of flooding can make householders more aware in future events. In 2011, more than a third of respondents (38%) indicated that they had received a warning and thought it was relevant. A further 9% acknowledged that they had received a warning but that they had not realised at the time the warning s relevance to them. In contrast 67% of respondents heard a warning in the 2013 floods, despite most respondents not needing to be concerned due to the smaller extent of the event. Many respondents to the 2013 survey (47%) treated the warnings seriously, even if the warnings weren t intended to be relevant to them, and prepared for flooding 2. There were very stark differences between the amount and quality of the information provided by local and state government during the 2011 and 2013 flood events. The performance of both levels of government in 2011 was appalling and frankly shameful, whereas in 2013, starting with early releases from Wivenhoe and continuing with regular, location specific flood risk information, we were well-informed and reassured by government agencies. (Fig Tree Pocket, 2013 survey) Respondents received warnings from a range of sources (Table 4.2). Three warning sources dominated in the 2011 floods, including television (49%), friends, family or neighbours (47%) and the radio (38%). In the 2013 floods, all of these more traditional warning sources remained important. However the use of the internet for obtaining information on warnings almost tripled from 15% in 2011 to 42% in 2013, making the internet the second most common source of accessing warnings in Network congestion and disruption to the electricity supply prevented some respondents from accessing the internet, so internet use is likely to have been higher if the internet was available. 2 See Chapter 5 for the data on preparing for the impending flood. Household experiences of flooding in Brisbane and Ipswich 21

28 Table 4.2 Sources of warnings, proportion of responding households, 2011 and 2013 floods. Warning source 2011 flood (%) (n=1215) 2013 flood (%) (n=415) Television Radio Newspaper 9 15 Website Social media 4 7 Mail/ 2 0 Emergency SMS Friends, family or neighbours Police, SES or Council visit 9 2 Self-observation 4 2 No warning Respondents may have provided multiple responses so percentages may not add to 100. The changes in proportions between different warning sources between the two flood events most likely reflect the increased awareness of flooding after the 2011 floods. As respondents become more aware of the prospect of flooding, they are more likely to actively seek out official sources of flood information, making respondents less reliant on friends, family and neighbours as a warning source. This increased awareness occurred even though the 2013 event did not impact on the majority of the respondents, and so warnings were not necessarily relevant to them. Table 4.3 Source of warning, proportion of respondents who obtained a warning from only one source, 2011 flood. Warning source 2011 flood (%) (n=362) Television 23 Radio 12 Newspaper 0 Website 2 Social media 1 Mail/ 2 Emergency SMS 6 Friends, family or neighbours 40 Police, SES or Council visit 6 Self-observation 4 Percentages may not add to 100 due to rounding. In 2013, the proportion of respondents receiving warnings from an emergency SMS more than doubled (10% in 2011 to 23% in 2013). Unlike the internet, where users must actively seek information on warnings (implying some awareness of the possibility of flooding prior to accessing the internet), 22 Household experiences of flooding in Brisbane and Ipswich

29 SMS is a push technology, where respondents directly receive warning messages. These two warning sources are complementary, with the SMS bringing households attention to the possibility of an event, and the internet supplying detailed information about the event. Two-thirds of respondents received warnings from two or more sources in both 2011 and 2013 (66% and 60% respectively). Of the respondents that said that they heard a warning from only one source in 2011, 40% heard from friends, family or neighbours (Table 4.3) with television the next most common warning source (23%). Both surveys asked respondents which warning source they found the most helpful (Table 4.4). In 2011, the most helpful were identified as friends, family and neighbours and the television (both at 44%), followed by the radio (37%). In 2013, websites were identified as the most helpful (43%). Other sources were less helpful in 2013 than 2011, with friends, family and neighbours dropping the most in popularity (falling from 44% in 2011 to 10% in 2013). Table 4.4 Most helpful warning source, proportion of responding households, 2011 and 2013 floods. Warning source 2011 flood (%) (n=977) 2013 flood (%) (n=281) Television Radio Newspaper 4 2 Website Social media 4 4 Mail, 2 0 Emergency SMS 9 14 Friends, family or neighbours Police, SES or Council visit 5 1 Self-observation 3 1 Respondents may have provided multiple responses so percentages may not add to 100. Table 4.5 Relevance of warnings, proportion of households by the number of warning sources, 2011 flood. Number of warning sources Warnings relevant (%) One (n=362) 37 Two (n=333) 50 Three (n=205) 62 Four (n=101) 65 Five (n=45) 64 The number of warnings received also made a difference to how households perceived the warning relevance. Respondents who received warnings from one source only were not likely to perceive the warnings as relevant, whereas respondents who received warnings from three sources were more likely to perceive those warnings as relevant. There was little change in the relevance of warnings for additional warning sources (Table 4.5). Alternatively, some respondents that thought an initial warning was relevant may be more likely to seek confirmation of that warning from additional sources. Household experiences of flooding in Brisbane and Ipswich 23

30 4.2 Interpreting warnings Some respondents reported hearing warnings, but didn t think those warnings were relevant to their home (Table 4.1). Respondents were asked to comment on why they thought that warnings were not relevant, with the text responses compiled in Table 4.6. Table 4.6 Reasons why warnings were not relevant to the home, proportion of responding households, 2011 flood. Why warnings not relevant 2011 flood (%) (n=348) Respondent s area not mentioned 9 Flood height level related to different area 4 Timing of expected flooding optimistic 2 Advice home will not flood (e.g. council maps, above Q100 1 ) 17 Did not expect home could flood 2 21 Thought Wivenhoe Dam would prevent flooding of buildings 7 No mention of stormwater inundation 1 Warnings too general 31 Warnings too late 13 Other 1 1 Q100 refers to the 100 year average return interval flood. This is the flood level used by the Brisbane City Council (2013) and Ipswich City Council (2013) for planning purposes, where dwellings should be above the Q100 level. A flood of this magnitude can be expected to occur once in every 100 years on average. This type of flood can also be expressed as the 1% annual exceedance probability, where there is a 1% chance of occurring in any one year. 2 Including thought media claims exaggerated. Respondents may have provided multiple responses so percentages may not add to 100. The most common reason given as to why the warnings were not relevant was that the warnings received were too general (31%). This is related to two other reasons given including that the respondent s area was not mentioned (9%) and that the flood levels related to other areas (4%). Respondents commented that nearby areas were mentioned in warnings or were visited by emergency service personnel, but as their area was not they assumed that their properties would not be flood affected. Other respondents were confused about how the warnings, given in terms of height levels at a particular gauge, related to the level of water at their home. The warnings on the BCC [Brisbane City Council s] website, radio, TV etc. were not adequate enough to determine if our house was at risk. Saying x metres at y gauge means nothing. Similar to colouring my house blue on a map. We did not know what depth of water would inundate the property. (Fairfield, 2011 survey) We need much more accurate ways of letting us tell how individual properties are at risk, e.g. how to translate heights at the City Gauge into heights at our property, and functioning websites where continually updated information is available, with degree of uncertainty quoted where that exists. And better predictions overall (from modelling under various scenarios). (Graceville, 2011 survey) 24 Household experiences of flooding in Brisbane and Ipswich

31 The second most frequent reason was that respondents did not expect that their home would flood (21%). As the flood levels for 2011 were forecast to be lower than in 1974, many households believed that their home was not at risk. Therefore, it came as a great surprise to respondents that had not flooded in 1974, when their home flooded in The commentary collected as part of the survey showed many respondents had consulted flood maps, either at the time of home purchase or online before the flood, and had determined their home was not at risk of flooding (17%). Some respondents wrote that as their house was built above the Q100 level they had not expected they would be flooded. Receiving misleading information from council at time of purchase has greatly impacted us. We would not have bought this property had we been accurately informed of the flood risk and we now owe the bank more than recent property valuation. (Karalee, 2011 survey) Others complained that the warnings were too late (13%). Related to this was the complaint that the timing of the expected flooding was optimistic (2%). The warning were disgustingly inadequate. The water rose so fast when we woke up on Wed 12 January we could not walk out of our house I had to kayak out. We asked police in the area at 11pm on Tues 11 January whether we should evacuate now and they indicated it was not necessary it appears when the waters rose in the middle of the night they just departed a good old fashioned loud hailer would have been helpful. (Auchenflower, 2011 survey) Some of the respondents who commented on the warnings coming too late were affected by stormwater flooding, and felt that there had be no information about stormwater. Evacuation warning came too late. Did not receive any notification to evacuate. Flooding originated through stormwater drains during the night and no warning of this was provided. (West End, 2011 survey). It is a common belief in many floodplains where flood control has been undertaken that the flood controls will prevent the disaster (Birkland et al 2003). The GA surveys received many responses relating to Wivenhoe Dam. Wivenhoe Dam was built on the Brisbane River following the 1974 floods for both flood mitigation and water supply reasons. The belief that the cause of the flooding was the result of the mismanagement of Wivenhoe Dam was widespread. While only 7% of the comments related to Wivenhoe Dam, many of these comments were the most passionate. G day, I am uneducated, but I know why Rocklea flooded. One week before the flood, the fools who control the gates on the dam were skiting that the dam was 200% full. Then it rained, and the imbeciles opened the gates wide enough to wash away 6 acres of land on the way to meet the high tide coming up the Brisbane River. It wasn t raining here and the water came up 9ft deep in my yard The Wivenhoe Dam was built to stop the flooding not to fill up to 200%. The flood was caused by human error. (Rocklea, 2011 survey) Amazed and horrified by lack of common sense & incompetence of those responsible for control of Wivenhoe and Somerset Dams. No one listened to local farmers and those knowledgeable of the facts of the 1974 flooding not remembering Wivenhoe was built for flood protection for Brisbane which is not possible if full. (Chelmer, 2011 survey) Household experiences of flooding in Brisbane and Ipswich 25

32 The role of Wivenhoe Dam in the 2011 event was reinforced in some respondent s mind by the 2013 floods, with many saying that flooding would not occur again because the dam was managed more effectively. This response was given by 18% of those who responded to the relevance question and 4% of all respondents to the 2013 survey. Some respondents acknowledged that the warnings were there, but were not always heard. We live in an area badly affected in the 1974 floods. Most residents are aware of this. When the Oxley Creek merged into the park we knew the water would continue to rise. Some of our neighbours clearly could/would not believe this and therefore took no precautions. Others were able to move all their possessions to higher ground. The warnings were there, some chose not to listen. (Sherwood, 2011 survey) The 2013 flood was a much smaller event than 2011, with far fewer properties flood affected. Table 4.1 shows that a greater proportion of respondents in 2013 (67%) indicated that they heard a warning than in 2011 (47%). The greater awareness of flood warnings in the 2013 event is likely to have come from the 2011 flood experience. However it would seem only the recent flood made a difference to receiving warnings. The 2011 survey asked respondents if they had experienced any kind of disaster previously. Only small differences were found in those who had previously experienced a flood or not and whether they received warnings. Obviously having experienced a flood does not necessarily increase awareness, and other factors also play a part. Long periods of time since that previous flood may also make it less likely for warnings to be perceived as relevant for households. Analysis of the short answer reasons why the warnings were not relevant from the 2013 floods also highlighted respondent s greater understanding of the threat. A number of respondents commented that all the available information from flood maps and warnings showed the flood levels were much lower than those experienced in 2011 and that they would not flood. The smaller amount of rainfall received also reassured some respondents. Another group of respondents stated that they received no warnings, which may have been appropriate for many households given that 2013 was a much smaller event. However five respondents who had water in their home stated they received no warnings. In 2013, respondents expressed very different viewpoints on warning clarity, which can largely be grouped into three categories: Warnings were just as confusing as 2011, with no understanding of the relationship of river height at gauge to water level at their home, or what a negative height at their home meant. Warnings were clearer, with local landmarks used as reference points. Warnings were exaggerated and caused unnecessary panic, particularly the warnings from the media. Some respondents looked at maps online showing the 2011 flood levels without realising that they were historical, and thought those levels were also expected in Of the respondents whose homes were flooded in 2013, some noted that they received confusing and inaccurate information, with the water reaching higher levels than forecast. A spatial analysis of the responses to the questions on warnings given following the 2011 floods showed no clear pattern. Notably, respondents that received warnings were often near neighbours of respondents who did not receive a warning, as illustrated in Figure Household experiences of flooding in Brisbane and Ipswich

33 Figure 4.1 Selection of respondents who received or did not receive warnings highlighting the proximity of those who received warnings to those who did not. 4.3 Risk perception A prior understanding that a home may potentially flood is important to being receptive to a warning. A person s understanding of the likelihood of flooding and how it will impact them is known as risk perception. Forty-six per cent of respondents did not know that their house was at risk of flooding before the 2011 floods. The 2011 survey instrument provided a free text space for respondents to detail any actions they took when they learnt of the risk (presented in Table 5.2 in the next chapter). Around a third of respondents took protective actions such as modifying the house and purchasing insurance showing they were aware that flooding was possible. Half of the respondents listed actions such as lifting or removing contents on learning of their risk. About a quarter of respondents reported that there was no time for actions or did not know to take actions. Respondents were asked to identify where they received information on their flood risk from, and where they had expected to find this risk information (Table 4.7). The majority of respondents (88%) reported they expected the Council to provide information on their flood risk, though only 43% reported actually finding the information from the Council. More than half of respondents expected to find the information from the state government (54%) but only about a quarter (11%) said that they actually received the information here. The number of respondents who obtained information on risk from the media was comparable to their expectations (39%). Insurance companies and emergency services were also expected to provide flood risk information but few reported actually receiving flood risk Household experiences of flooding in Brisbane and Ipswich 27

34 information from these sources. Neighbours were a common source of information (39%), highlighting the important role of flood experienced community members in sharing their knowledge with their community. Table 4.7 Actual and expected sources of information for flood risk, proportion of responding households, 2011 flood. Source Actually found information (%) (n=1121) Expected to find information (%) (n=1223) Neighbour Family, friends 6 1 Council State government Federal government 1 11 Insurer 3 25 Emergency services 8 28 Media including the internet Community group 1 6 Observation 3 - History 8 - Real estate agent, lawyer, or bank at time of purchase - 2 Previous experience, local knowledge - 1 Other 4 2 Respondents may have provided multiple responses so percentages may not add to no data, these sources were responses to the other option in the respective survey questions. Respondents were asked how they would like information on their flood risk to be presented. Of the 858 responses, 31% related to warnings immediately before an event. Of these, broadcast media such as radio and TV, was preferred by the majority of respondents, with phone and SMS the second most preferred method. A number of comments about honesty, accuracy and timeliness came from respondents who had consulted flood maps online in the lead-up to the 2011 event, or at the time of purchase, and found that their home was not in an area that was expected to flood. The subset of responses relating solely to the presentation of risk, before a flood is expected, are categorised in Table 4.8. The majority of respondents would like detailed reports including maps (42%) or letters, pamphlets or newsletters sent to them (21%). Twenty-three per cent of respondents wanted the information to be available online via interactive mapping or an address search. Other suggestions included making it compulsory to be provided at the time of purchase, building approval or rental (13%) and for flood height levels to be marked on street lights or poles, on homes, or on signs in each suburb (10%). Regular updates to take into account changes in modelling were also mentioned by some respondents. A few respondents felt that the current arrangements were adequate, however it was not clear if this was the case before the 2011 floods, or if they were referring to updated information available afterwards. 28 Household experiences of flooding in Brisbane and Ipswich

35 Table 4.8 Preferred way for information on flood risk to be presented, proportion of responding households, 2011 flood. Presentation medium 2011 flood (%) (n=494) Advice at time of purchase, building approval or rental 13 Flood height levels on home, or poles in street/suburb 10 Reports including maps 42 Rates notice/valuation notice 4 Internet maps or online address search 23 Mail, letter box drop or newsletters 21 Insurance companies 3 People may have multiple provided responses so percentages may not add to 100. The results presented in Table 4.8 hide the diversity of views held by respondents, often in opposition to each other, and as noted by some, a one size fits all approach is unlikely to address all needs. Most respondents wanted to see maps in some form, however some respondents wanted to see those maps in a physical report or booklet that they can file away, while others wanted them online to be searched when needed. The amount of detail requested ranged from detailed statistical reports with the modelling methods specified, to simple and easy to understand pamphlets of one page. Another contrast relates to the availability of the information, with many respondents noting that risk information should be publicly available online for all homes. Meanwhile, some preferred the information to be kept private, but provided to home owners on request or at time of purchase or rental. The preferred content of the risk information was also noted by some respondents, with information on the historical events of 1893, 1974 and 2011 requested. Some respondents wanted to know the depth of any potential flooding on their property so they could prepare appropriately. Information that contained risk relating to Q100 or recurrence intervals were not seen as helpful. Respondents preferred information, including maps, to relate to various flood heights on a relevant gauge. The overriding theme for the responses on content was the information needed to be detailed spatially or personalised so that respondents could interpret the impact of any flood on their own property. 4.4 Chapter 4 summary The answers to the warnings questions indicate that respondents have much more complex views on what a warning looks and sounds like than expected. This view cannot be explained by the data collected in either survey. The complexity may be a result of the long length of time between the flood events and when the surveys were conducted, a time where memories are lost, altered and augmented. The number of households who received a warning increased between the 2011 and 2013 households, showing that experience is important, however experience did not always influence warning perception during the 2011 event. While television and radio were important sources of warnings, it was found that friends, family or neighbours were more important for those who received warnings from only one source. The 2013 flood event saw online sources of information become a primary source of warnings, but few suggested online sources for presentation of their risk following the 2011 floods. Household experiences of flooding in Brisbane and Ipswich 29

36 Households who received warnings from two or more sources were more likely to see the warnings as relevant to them compared to households that had one warning source. Increasing the number of warning sources to four or greater did not increase perceived relevance. Many households found the warnings too general or confusing to relate to their own homes. Some households had received previous advice or expectations that the dwelling would not flood, including that received at the time of purchase or construction of their dwelling, and so believed the warnings were not meant for them. Respondents suggested a variety of ways they preferred their risk to be presented, sometimes in opposition to each other. Some preferred paper reports to be kept handy when needed, while others preferred online options to be searched from anywhere. Whether flood risk should be kept private to the household or made public was also an issue for discussion. Other suggestions included flood height markers on street poles and noting the gauge height beyond which the home becomes flooded on the meter box. This research shows the understanding of risk needs to be explored further. Many households were unaware that their homes could potentially flood, and so did not see the warnings as relevant. Respondents commented that they purchased or built above the Q100 or 1974 flood level and were surprised to find that they flooded during the 2011 event. Respondent s comments indicated that they did not realise that new modelling and changes upstream may change their risk. 30 Household experiences of flooding in Brisbane and Ipswich

37 5 Preparation, evacuation and moving back home This chapter provides information on how and when householders prepared for the imminent flooding and what they did to mitigate any potential flood damage. It also examines where respondents went if they evacuated and how long they were away from their homes. Being adequately prepared and being aware of the potential danger can, to some extent, minimise the damage caused by floods. Some preparation activities must commence long before a flood is approaching, particularly mitigation activities and purchasing insurance. When the flood is imminent potential damage to the home cannot be avoided, however household actions can reduce the amount of damage to furniture and personal possessions. Understanding evacuation behaviour, including the length of time away from home, is necessary for emergency managers to plan for future events. However evacuation behaviour is also an indicator of household s resilience, by protecting themselves and their possessions and through the use of social networks for evacuation. The length of time households are away from their homes can also be seen as an indicator of resilience. Households with lower levels of damage can return home quicker, but households with higher levels of assistance can also recover much quicker. 5.1 Preparing for the floods On Monday 10 January 2011 at pm, the then Queensland Premier Anna Bligh stated at a joint Emergency Services media conference, I have to stress that we are now in unprecedented territory (Department of the Premier and Cabinet 2011). Premier Bligh highlighted the need for households and businesses to undertake every possible precaution over the next 24 to 48 hours. The flood peaked at 4.46m at Port Gauge on the Brisbane River at 3.00am on Thursday 13 January In Ipswich the flood peaked at 19.4m at the Ipswich (David Trumpy Bridge) gauge on the Bremer River two days earlier, at 1.30pm on Tuesday 11 January In 2013 official warnings, including flood maps, were issued on Sunday 27 January, however warnings of the potential for flooding were issued the day before, and extreme flooding occurring further north gave some residents early indications. The flood peaked at Port Gauge at 2.3m on Tuesday 29 January 2013 and at 14m on Monday 28 January at Ipswich. Table 5.1 and Figure 5.1 shows that on the day following the 2011 media conference a significant proportion of households (42%) responded to the Premier s warning and commenced flood preparations. This is in stark contrast to householder actions in the previous days. In the 2013 floods, considerably more respondents commenced flood preparations earlier than in 2011, indicating that past experience had made a positive impact on the recognition of warnings and household behaviour. Household experiences of flooding in Brisbane and Ipswich 31

38 Table 5.1 Day households commenced flood preparations, proportion of responding households, 2011 and 2013 floods flood 2011 flood (%) (n=1037) 2013 flood 2013 flood (%) (n=367) Didn t prepare 5 Didn t prepare 13 Wednesday 5 January 2011 or earlier 5 Monday 21 January 2013 or earlier 14 Thursday 6 January Tuesday 22 January Friday 7 January Wednesday 23 January Saturday 8 January Thursday 24 January Sunday 9 January Friday 25 January Monday 10 January Saturday 26 January Tuesday 11 January Sunday 27 January Wednesday 12 January Monday 28 January Thursday 13 January 2011 (flood peak - Brisbane) 9 Tuesday 29 January 2013 (flood peak - Brisbane) 2 After Thursday 13 January After Tuesday 29 January Percentages may not add to 100 due to rounding. Figure 5.1 Day of preparation commencement in relation to peak flood, proportion of responding households, 2011 and 2013 floods. 32 Household experiences of flooding in Brisbane and Ipswich

39 Figure 5.2 Day of preparation commencement, proportion of Ipswich households, 2011 flood. Figure 5.3 Day of preparation commencement, proportion of Brisbane households, 2011 flood. Figure 5.2 and Figure 5.3 show the proportion of respondents who commenced preparing themselves and their home for the impending flooding in Ipswich and Brisbane. The flood peaked at David Trumpy Bridge two days before it peaked at Port Gauge, on the day of the Premier s press conference, so most Ipswich respondents (42%) only commenced preparation on the day of peak flood (Figure 5.2). In comparison, most Brisbane respondents (45%) commenced their preparation two days before the flood peaked at Port Gauge (Figure 5.3). The number of Ipswich households starting commencement after peak flood seems high and can t be explained by other data. A possible explanation may be that some Ipswich households have had the date of peak flood in Brisbane memorised from repeated Household experiences of flooding in Brisbane and Ipswich 33

40 media reports, which may neglect that the flooding occurred in Ipswich earlier. Therefore, some households have responded according to the date of peak flood as they have remembered it. If this explanation is the case, the data shows the difficulties of studying reactions to floods so long after the event, as already discussed in Chapter 4. Respondents were asked what actions they took to protect the home or to reduce the risk when it was known there was a risk of flooding. The intention of this question was to identify actions households had taken before the 2011 floods were forecast. In analysing the responses however, it is evident that nearly half of the respondents did not know that their home was at risk of flooding until the 2011 floods. Most of the actions reported related to the immediate flood rather than longer term actions prior to any event occurring (Table 5.2). Table 5.2 Actions taken to protect the home or reduce risk when learnt of risk, proportion of responding households, 2011 flood. Preparation actions 2011 flood (%) (n=825) Modified the house 9 Investigated options for flood insurance 21 Lifted furniture and personal possessions 1 34 Removed furniture and personal possessions 1 15 Obtained more information about the risk 8 Too late to do anything 13 Did not think the home would flood 10 Other 9 1 The intention of this question was to determine actions taken long before floods were imminent, and not in response to an event, so the number of these responses are in conflict with the following Table 5.3 which captures the 2011 flood event. Respondents may have provided multiple responses so percentages may not add to 100. A fifth of respondents investigated options for insurance, particularly for those who were unable to modify their homes, however for some it either wasn t available or wasn t affordable. Only nine per cent of respondents noted that they had modified their house. Some of these respondents had built or raised their homes above the 1974 and Q100 flood levels with the expectation that this would protect them from flooding. A handful of respondents commented that they had installed water resistant materials in the lower level of their home to minimise any future potential flood impact. Households were asked what actions they took to prepare for the floods, with the intention that this question would identify what actions were taken immediately before the 2011 floods. Table 5.3 shows that in 2011, 81% of households lifted their furniture and personal possessions, and 14% of households removed their possessions. In the flood event two years later, over half (54%) of households lifted their furniture and personal possessions and 40% removed their possessions, even though many households were not expected to flood in This increase in response could be attributed to negative experiences associated with the 2011 floods. Numerous respondents commented that their experience in 2011 cast doubt in their minds whether to believe the authorities when they indicated that they would not be flooded in Household experiences of flooding in Brisbane and Ipswich

41 We looked at the maps [ ] and I personally felt the Experts would not get it wrong again, they said peak flood level would be +/- 2 metres lower than I was against evacuating from the house preferring to raise furniture off the floor. My wife panicked so we emptied the house helped by friends. The water authorities were correct and the level was +/- 2 metres lower than Mind you when the water level got to our driveway I started to have some doubts. (Kenmore, 2013 survey) The house did not flood because Wivenhoe dam stopped releasing water when Bremer River peaked. This did not happen in Know what to do if floods now move everything upstairs. Disagreement in family whether was going to flood or not. Compromise spent all day moving furniture only to have to move it all back again later in week. Stayed up all night 28/29th watching in river. Did not trust council saying we would not flood after But would in future. (Jindalee, 2013 survey) Table 5.3 Actions undertaken to prepare for the flood, proportion of responding households, 2011 and 2013 floods. Preparation actions 2011 flood (%) (n=1186) 2013 flood (%) (n=406) Sandbags 2 14 Investigated options/obtained flood insurance Lifted furniture and personal possessions Removed furniture and personal possessions Stocked supplies Obtained more information about the risk Evacuated 4 na Too late to do anything 4 7 Did not think the home would flood 3 33 Household members were not home at the time 1 3 Other 3 1 Respondents may have provided multiple responses so percentages may not add to 100. A large number of respondents however did appreciate that the forecast 2013 floods posed a much lesser risk than the 2011 floods, reflected in the 30 percentage point increase in the proportion of households who did not think it would flood in A few respondents commented on the effectiveness of their preparation prior to the floods. Sand bags were commonly seen as not useful, and the items that households did not save were also regretted. Prepared sandbags and sandbagged the house - a waste of time. Prepared things that I could pack things in if needed. Was clear in my mind re what needed to be removed from the house and what could ultimately be replaced if it came to that. HOWEVER - my judgement was wrong. I only came to understand the impact and what was important afterwards. (Kenmore, 2011 flood) Household experiences of flooding in Brisbane and Ipswich 35

42 5.2 Evacuation and relocation In 2011 most (94%) households evacuated their homes. Households who evacuated were asked where the household initially evacuated to, and where members of the household stayed the longest. In a small number of cases, not all household members evacuated to the same location. Staying with family or friends were the most common responses, with 90% of households choosing these evacuation locations following the 2011 event as their initial place of stay. Table 5.4 shows that the proportion of households initially staying with family, relatives or friends was similar in 2011 and in Generally households did not go to evacuation centres, with only 2% of respondents choosing to go to a centre in The mix of accommodation locations for the longest period (Table 5.5) shows different patterns to the initial evacuation locations (Table 5.4). While only 1% of households stayed in private rental accommodation immediately following the 2011 event, 21% of respondents rented longer term. This reduced to 6% following the 2013 floods, and is reflective of the smaller geographical extent and impact of the event. The importance of family and friends following both events is highlighted again in Table 5.5, which identifies family or friends as the place where flood impacted respondents resided the longest. Table 5.4 Initial place of stay, proportion of evacuated households, 2011 and 2013 floods. Initial place of stay 2011 flood (%) (n=1147) 2013 flood (%) (n=65) Family/relatives Friends Other Respondents may have provided multiple responses so percentages may not add to 100. Table 5.5 Longest place of stay, proportion of evacuated households, 2011 and 2013 floods. Longest place of stay 2011 flood (%) (n=1163) 2013 flood (%) (n=64) Family/relatives Friends Private rental 21 6 Other 9 11 Respondents may have provided multiple responses so percentages may not add to 100. Many respondents commented on the emotional and financial impacts posed by the need to find alternative short and longer term accommodation. The recovery period has taken a long time - we would never have imagined that nearly 18 months after the event we still wouldn't be back in our home. The daily frustrations have been enormous - we are just about to move for the 5th time since the floods due to rental properties being sold or owners having changed circumstances. This has resulted in a constant change of personal details with utility companies, phone companies, government organisations etc. Our energy levels and capacity to cope have been severely challenged over this time. (Fig Tree Pocket, 2011 survey) 36 Household experiences of flooding in Brisbane and Ipswich

43 5.3 Returning home Following the 2011 floods, households in Brisbane and Ipswich lived away from their homes for an average of 95 days. In 2013 the average time away fell to 17 days, reflecting the reduced impact from the 2013 flood. After the 2011 flood, 1% of households did not return to their home, with nearly all of these home owners. Some may have returned initially, but have since moved. Families are still missing from my street as their houses are not repaired. Some sold up and had to accept severely reduced price for their house. I will most probably do the same when I get the house up to scratch. (Anonymous, 2011 survey) In 2013 no household reported needing to find a new permanent place to live as a result of flooding. However some respondents saw the 2013 flood as the last straw, even if they were not directly impacted, and decided to move despite the costs. After January 2013 we decided to move house and did so in April This was mainly due to stress and risk of flooding again so soon after 2011.Our old flooded (2011) property is now being rented out (Jindalee, 2013 survey) My wife had significant trauma responses following 2013 floods, and this forced us to move house. This has been a huge burden on our family emotionally, practically and financially. (Rocklea, 2013 survey) While the average time away from home following the 2011 floods was 3 months, the median time away was only 21 days. A quarter of households reported that they had returned home a week after the floods receded. A further 2% of respondents had not returned home at the time of the survey. Looking at the length of time taken to return home in relation to other questions showed that households with insurance returned later than those without. Households whose homes were still being repaired lived away from their home on average 38 days fewer than those whose homes were completely repaired at the time of the survey. This indicates that some households are returning to live in their homes when they have not been repaired. Figure 5.4 combines data on the repair or rebuild state of the home at the time of the survey 3 with data about insurance claims. It shows that for those whose claims were successful, the length of time to return home was similar regardless of the state of the home. Similarly, where claims were found to be not successful, there was only marginal difference in the length of time to return home. Overall, respondents returned home much quicker if their claims were unsuccessful than if their claims were successful. 3 The repair or rebuild state of the home is discussed in section 6.3. Household experiences of flooding in Brisbane and Ipswich 37

44 Figure 5.4 The average length of time to return home in days following the 2011 floods, by the level of insurance and the completion state of the home at the time of the survey. An insurance claim is considered successful if 90% or more of the damage was paid, and partially successful if more than 10% but less than 90% of the damage was paid. This data shows that respondents were returning to incomplete homes, in some cases even if their insurance claim was successful. The consequences of living in an unrepaired home is apparent from the concluding comments respondents supplied. My home is still in the repair phase since the 2011 flood. We had to knock down the back room and laundry and only this year have built a deck in its place. It is still unfinished. I feel sort of trapped in my home because of the lack of free space. We lost wardrobes and drawers and clothing and household items are stacked in plastic boxes all around the house. I literally have only one pathway or lane through my lounge room and my two bedrooms (I feel like a dog that has made a path up and down the fence line). (Tennyson, 2011 survey) For households with a partial insurance payment there were large differences in the time to return home, according to whether the home was finished repair or not. However the length of time to return home was not related to how much was paid. No other information is available to explain the difference. The reason why people with insurance take longer to return home was not evident from the data, though the effect was found to be significant in a more detailed analysis. However it is possible to speculate why this is happening based on the responses in the concluding comments. One possible explanation is that people with insurance spend time waiting for the insurance company to assess the damage and organise appropriate tradespeople. Another is that many people with insurance receive rental subsidies for temporary housing therefore can ensure that their home is repaired fully before returning home. 38 Household experiences of flooding in Brisbane and Ipswich

45 The detailed analysis of the length of time to return home also highlighted other factors. Households with children aged under five were found to significantly take longer to move back home. Whether or not a household also had older children did not have an influence. This result seems obvious to parents; young children need more care and attention and are often at home with a parent reducing the amount of time a parent can spend on rebuilding activities. Meanwhile older children attend school during the day, allowing parents to meet with builders and be present at the damaged home. The height of the water significantly increased the time to return to home, as more repair work needed to be undertaken. Difficulty in finding tradesmen which then led to delays also significantly increased the time to return home. Households that contain someone who volunteered was the only characteristic found to have a positive influence and reduce the time to return home. It is unlikely that the act of volunteering in its own right has an influence, but rather that volunteering is a proxy indicator for social capital 4. Social capital supplies resources that households may call on to assist in difficult times. Households with high social capital may have greater access to willing hands to help or donate goods, making the recovery load lighter. See section 8.4 for a further discussion on social support. 5.4 Chapter 5 Summary Many households were unaware of the risk of flooding before the 2011 floods were imminent, and so most preparation actions focus on flood response activities such as lifting furniture. For many households who responded to the 2013 survey, the experience and lessons learned from the 2011 flood influenced their behaviour in the later flood. The responses suggest that the experience gained in 2011 may have reduced the vulnerability of households to potential flood damage in The majority of households lifted furniture and personal possessions in 2011, with fewer removing items. However many households found this not useful given the height of the water. Others regretted undertaking activities like sandbagging rather than saving more items, so in 2013 many more households removed items. When initially evacuating most households stayed with family or friends. Many households stayed with family or friends longer term, but private rental accommodation was also used by some. The average time to return home was 3 months but the median time was only 3 weeks. Respondents whose homes were still being repaired at the time of the survey returned home earlier than those whose homes were complete, indicating households were returning to homes still needing repair. The height of overfloor inundation, delays in finding tradesmen, having children aged under 5 and having insurance were all found to increase the length of time to return home. Households that contained someone who volunteered decreased the time to return home. 4 The OECD (2001, p.41) defines social capital as: "networks, together with shared norms, values and understandings which facilitate cooperation within or among groups. Household experiences of flooding in Brisbane and Ipswich 39

46 6 Damage and repair The most noticeable impact of many disasters, particularly as portrayed in the media, is the damage to homes. Disaster severity is typically measured in the number of homes and other buildings impacted and the associated costs of repair and rebuild. Other items are also lost that mean more to households than their dollar value would imply. These personal items are a major part of the feeling of loss that households experience. This chapter discusses these quantifiable and intangible direct costs as well as the repair process and its effectiveness. This chapter includes an analysis that contrasts two areas within the suburbs of Rocklea and Fig Tree Pocket. The two areas are bounded by two of Australian Bureau of Statistics (ABS) SA1s (Statistical Area 1) in Rocklea 5 and Fig Tree Pocket 6. Unless otherwise stated this chapter s reference to Rocklea and Fig Tree Pocket refer to the respective SA1 and not to the whole suburbs of Rocklea and Fig Tree Pocket. These two areas were chosen because of good response rates to the 2011 survey, similar levels of flooding, and because they provide a contrast of socio-economic status. According to the ABS Index of Relative Socio-Economic Advantage and Disadvantage (IRSAD) (ABS 2013), Rocklea is placed in the 3rd decile while Fig Tree Pocket is in the top decile. The IRSAD summarises information about the economic and social conditions of people and households within an area, including both relative advantage and disadvantage measures. A low decile indicates relatively greater disadvantage and a lack of advantage in general. For example, an area could have a low score if there are (among other things) many households with low incomes, or many people in unskilled occupations AND few households with high incomes, or few people in skilled occupations. A high score indicates a relative lack of disadvantage and greater advantage in general. 6.1 Flood damage Survey respondents were asked to estimate the maximum depth of flooding inside the home. Where they could not provide this estimate, they were asked to approximate if the water reached the level of household items: floor level; chair level; table top; top of a chest of drawers; top of a book shelf; or more than one floor/over the roof. In this report, data has been cross-classified by inundation depth cohort. The depth cohorts are grouped by representative damage as shown in Table 6.1, where the amount of damage increases as the depth increases. Where survey respondents had not provided any indication of the depth of overfloor inundation the data is included in the tables as depth not stated. Inundation depth and damage cost data was collected to improve the flood vulnerability models, developed by structural engineers in Geoscience Australia for different building types. These models are the subject of another report (Wehner et al, in preparation) and are not discussed here. The 2011 flood survey asked questions about the cost of the building and contents damage, along with the amount covered by insurance. Insurance is a common strategy for households to mitigate a range of risks. However insurance must be obtained in advance, at a time when there is no immediate threat of disaster, therefore requiring pre-planning. At the time of the 2011 floods some insurers provided flood insurance only for certain types of flooding. While flooding from storm water or burst pipes was often covered, riverine flooding was often not. Many households did not understand the 5 SA1 number SA1 number Household experiences of flooding in Brisbane and Ipswich

47 differences between the types of flooding and believed they were insured (Queensland Floods Commission of Inquiry 2012). Households who believe they have insurance tend to be less prepared (Paton, Bürgelt & Prior 2008), leaving a household who believed they were insured but did not actually have insurance especially vulnerable. Table 6.1 Maximum overfloor inundation depth cohorts and representative damage. Overfloor inundation depth (metres) Number of responses 2011 flood Representative damage 0.15m or less 66 External on-ground air-conditioning, external hot water system, floor coverings, skirting boards, cabinet kickboards, internal doors and full length curtains replaced m 0.7m 216 As above plus electrical outlets, on floor cabinetry, dishwasher, oven, cooktop, half height wall linings, external door and shower screen replaced. 0.71m 1.2m 187 As above plus full height wall linings, cornices, light switches, windows, curtains, all remaining cabinetry and fuse box replaced. 1.21m 2.4m 306 As above plus ceilings, insulation, light fittings, extraction fans, rangehood and internal air-conditioning replaced. More than 2.4m 352 Inundation over roof one storey all remaining non-structural parts replaced. Inundation of second storey. Depth not stated 151 Not applicable Flooding from overwhelmed storm water drains was a factor in some areas during the 2011 flood, with some homes inundated by storm water before later being inundated further by the rising rivers. Because of this, insurance payments may cover only part of the damage; the part that is assessed to have been caused by storm water or damage from other means (Queensland Floods Commission of Inquiry 2012). Many households received no payment or a token compassionate benefit. Some policies do cover riverine flooding, and these households were covered for all of the damage, including rental subsidies for temporary housing. In June 2012 the Australian Government introduced regulations for insurance companies to use a standard definition for flood to avoid future confusion about coverage. Prior to the 2011 floods, 71% of respondents thought they had full coverage for flooding. However, only 32% of all respondents had their claim fully paid, creating a large difference between what people thought they were insured for and what was actually covered. A further 13% had partial claims paid, representing those properties where some of the damage was attributed to stormwater. Over half (55%) of the survey respondents, including those that did not try to claim, were left with no insurance payment. Of these, less than a third (29%) of respondents knew they were not covered for flood insurance and so had prepared accordingly. Table 6.2 shows the average dollar value of the building damage, insurance payment and shortfall by depth of overfloor inundation for the 2011 flood. Note that this data covers all responding households, including households that did not receive an insurance payment. Survey responses show that the average estimate of building damage ranged from $71,127 (for inundation of 0.15m or less) to $156,261 (where the inundation was 2.4m or more). The average damage to all buildings is estimated at $123,176 per home. The amount of damage covered by insurance claims differs by the depth of inundation. The insurance payment versus damage costs ratio sits at 49% for the lowest level of inundation (0.15m or less), 66% for 0.15m to 0.7m of inundation, and 54%-58% for higher inundation depths. Presumably households with lower levels of inundation were less likely to make a claim, however it is not clear why the 0.15m to 0.7m inundation depth cohort have a higher proportion paid than the higher inundation depths. Household experiences of flooding in Brisbane and Ipswich 41

48 Table 6.2 Average value of building damage, insurance payment and shortfall by depth of overfloor inundation, all responding households, 2011 flood. Overfloor inundation depth (m) Average estimate of building damage ($) Average insurance payment ($) Average shortfall 1 ($) Insurance/ Damage ratio (%) 0.15m or less 71,127 34,923 36, m 0.7m 106,886 70,845 39, m 1.2m 109,135 62,672 48, m 2.4m 121,914 70,787 56, More than 2.4m 156,261 84,876 72, Depth not stated 88,420 62,573 31, Overall average 123,176 71,733 54, Average shortfall is not necessarily the difference between the damage estimate and the insurance payment as some respondents did not answer all questions. Damage to contents also increased with inundation depth. Table 6.3 shows that the average value of contents damage ranged from $33,445 for inundation of 0.15m or less to $60,303 where the inundation was 2.4m or more, representing an almost doubling in cost. The average estimate of damage to all contents is estimated to be $47,662 per home. The ratio of insurance claim paid to the damage costs shows a similar pattern to the building damage. Table 6.3 Average value of contents damage, insurance payment and shortfall by depth of inundation, all responding households, 2011 flood. Overfloor Inundation Depth (metres) Average estimate of contents damage ($) Average insurance payment ($) Average shortfall 1 ($) Insurance/ Damage ratio (%) 0.15m or less 33,445 13,296 21, m 0.7m 36,187 21,260 16, m 1.2m 46,440 22,620 24, m 2.4m 50,513 24,225 28, More than 2.4m 60,303 30,353 32, Depth not stated 20,309 13,607 8, Overall average 47,662 24,118 25, Average shortfall is not necessarily the difference between the damage estimate and the insurance payment as some respondents did not answer all questions. Table 6.4 combines the data from the previous two tables to show the average shortfall for the 2011 floods per home for building and contents damage at the different inundation depth cohorts. Figure 6.1 summarises the insurance payment and shortfall to give the total damage costs. On average, respondents had to find an additional $72,697 to repair and replace the damage caused to building and contents. 42 Household experiences of flooding in Brisbane and Ipswich

49 Table 6.4 Combined average value of building and contents damage, insurance payment and shortfall by depth of inundation, all responding households, 2011 flood. Overfloor Inundation Depth (metres) Combined average estimate of damage ($) Combined average insurance payment ($) Combined average shortfall 1 ($) Insurance/ Damage ratio (%) 0.15m or less 87,622 38,735 46, m 0.7m 135,321 84,554 50, m 1.2m 143,014 77,582 66, m 2.4m 164,103 88,941 79, More than 2.4m 204, ,750 97, Depth not stated 80,288 57,736 30, Overall average 157,812 87,018 72, Average shortfall is not necessarily the difference between the damage estimate and the insurance payment as some respondents did not answer all questions. Figure 6.1 Total average value of building and contents damage, insurance payment and shortfall by depth of inundation, all responding households, 2011 flood. By extrapolating the overall average insurance payment from this survey ($87,018) to the 17,100 (Queensland Floods Commission of Inquiry 2011) homes inundated in the Brisbane and Ipswich region, total residential insurance losses for Brisbane and Ipswich can be estimated at $1.5 billion. This compares well to Insurance Council of Australia (ICA) data that shows a total insurance payment of $2.4 billion for residential and commercial losses for the Queensland floods (Insurance Council of Australia 2014). The ICA estimate is expected to be higher as it includes all losses resulting from the flooding in Queensland between 21 December 2010 and 14 January 2011 (i.e. includes the severe flood damage in the Lockyer Valley and elsewhere). The ICA estimate also includes payments made to businesses for losses to their premises including contents, and business interruption costs. From the household survey results the total damage to residential property (including contents) in Brisbane and Ipswich following the 2011 floods has been estimated at $2.7 billion. The total shortfall, once insurance payments had been made, has been estimated to be $1.2 billion. Household experiences of flooding in Brisbane and Ipswich 43

50 Table 6.5 Comparison of damage costs insurance payment and shortfall for responding (flooded) households, 2011 and 2013 floods. Average costs Building damage ($) Contents damage ($) Outbuildings/landscape damage ($) Damage costs: 2011 flood 123,176 47,662 na Insurance payment: 2011 flood 71,733 24,118 na Shortfall: 2011 flood 54,524 25,515 na Damage costs: 2013 flood 121,229 44,923 34,315 Insurance payment: 2013 flood 86,800 36,333 33,342 Shortfall: 2013 flood 34,464 11,385 3,538 na: Not applicable; This information was not collected in the survey. Note: The 2013 figures are based on 14 responses only so 2013 data should be interpreted with caution. Table 6.5 shows a comparison of costs relating to property damage in the 2011 and 2013 floods. Note that the 2013 figures are based on 14 responses only, as very few of the 2011 flood survey respondents were flooded in 2013, so the data should be interpreted with caution. Damage costs relating to outbuildings and landscaping such as gardens and pools were not directly collected for the 2011 survey, though many respondents provided evidence of the damage in their concluding comments. Very expensive to repair garden, clear debris etc. Mostly not covered by insurance though we paid an amount for landscaping in premiums. (Chelmer, 2011 survey) We estimate that we were $100,000 out of pocket minimum even with insurance cover. Insurance did not cover 100% contents (house did not cover landscaping). (Graceville, 2011 survey) Not all damage can be measured by dollar values, so respondents were asked to provide details of the irreplaceable contents that had been lost. Along with items considered irreplaceable, such as photos or artworks, many respondents also noted items that are generally considered replaceable, such as book collections and furniture. Some respondents noted the sentimental value of these items beyond their replacement costs, as they were gifts (such as wedding gifts) or were bought at particular stages of their lives and so had associated memories. For these respondents, the irreplaceable part of the item is not the piece of furniture per se, but the memories attached to that particular couch, or television. Table 6.6 shows for the 2011 floods that photographs were reported as being the highest loss (65%) followed by personal records (60%) and books (58%). These items were all suggested examples as part of the survey question so their prominence is expected. Antiques, custom made furniture and collections such as stamps or war memorabilia ere reported as lost by 12% of respondents. Personal items, which included more tangible items such as jewellery as well as more sentimental items such as children s artwork, trophies and travel memorabilia, were reported as lost by 14% of respondents. 44 Household experiences of flooding in Brisbane and Ipswich

51 Table 6.6 Lost irreplaceable contents, proportion of households reporting contents, 2011 flood. Irreplaceable contents 2011 flood (%) (n=867) Photographs 65 Personal records 60 Books 58 Artworks 28 Business or work related documents 2 Antiques, custom furniture or collections 12 Music, CDs and DVDs 7 Computers, hard drives and technology 3 Personal items 14 Furniture, tools or clothing 11 Everything 4 Other 1 Respondents may have provided multiple responses so percentages may not add to Highlighted suburbs Fig Tree Pocket and Rocklea In 2010, the median house prices for the suburbs of Fig Tree Pocket and Rocklea were $790,000 and $359,000 respectively. Therefore, it is to be expected that for the similar depths of over floor flooding in both areas the average estimate of building damage in Fig Tree Pocket would be higher in dollar terms than in Rocklea. The following table (Table 6.7) compares the average estimate of damage to buildings and contents in the two study areas. Data for the total survey population is included for comparative purposes. On average, the estimate of building damage in Rocklea was $114,880. This was $8,300 less than the survey average and $50,000 less than the average building damage in Fig Tree Pocket. For contents damage, the average estimate of damage in Rocklea was $44,760. This was $2,600 less than the survey average and $13,100 less than the average contents damage in Fig Tree Pocket. While the shortfall differed in absolute costs, the proportion of building damage costs covered by insurance was slightly lower in Rocklea (50%) than Fig Tree Pocket (52%) and both were lower than the total survey average (55%). Table 6.7 Comparison of damage costs and insurance payment, 2011 flood. Average costs Rocklea ($) (n=25) Fig Tree Pocket ($) (n=30) All households 2011 flood ($) Estimate of building damage 114, , ,176 Insurance payment for buildings 57,913 84,821 71,733 Shortfall 1 55,504 83,500 54,524 Estimate of contents damage 44,760 57,844 47,662 Insurance payment for contents 20,655 25,497 24,118 Shortfall 27,157 30,987 25,515 1 Average shortfall is not necessarily the difference between the damage estimate and the insurance payment as some respondents did not answer all questions. Household experiences of flooding in Brisbane and Ipswich 45

52 This data highlights the problems with quantifying the severity of a disaster solely by damage costs. By using a damage costs approach, the flood impacted Fig Tree Pocket far greater than Rocklea. However households in Rocklea may also have less capacity to pay the damage. The 2011 median household income in Rocklea was $1,066 per week compared to $2,045 in Fig Tree Pocket (ABS 2012). Dividing the total shortfall by the median weekly income shows that if households were able to dedicate their entire income to paying the shortfall, Rocklea households would repay the shortfall after 18 months, while households in Fig Tree Pocket could repay the shortfall in one year. In addition, the median monthly mortgage payment in Rocklea is higher than Fig Tree Pocket, putting even greater financial strain on Rocklea households. 6.2 Disruptions to the household Households surveyed after the 2011 and 2013 floods were asked if they experienced any disruptions caused by supply problems with utilities, medical and food supplies and with access to their homes. The following tables show the extent of the disruption. Table 6.8 shows the proportion of households that experienced some form of disruption by disruption source. The extent of the disruption following the 2011 and 2013 floods are shown in Table 6.9 and Table 6.10 respectively. Table 6.8 Disruptions experienced, proportion of responding households, 2011 and 2013 floods. Disruption 2011 flood (%) (number of responding households) 2013 flood (%) (number of responding households) Electricity supply 97 (n=1215) 31 (n=347) Water supply 53 (n=1136) 4 (n=324) Gas supply 36 (n=995) 3 (n=292) Landline telephone access 77 (n=1145) 13 (n=329) Mobile phone access 25 (n=1067) 7 (n=319) Food and or medical supplies 37 (n=1090) 4 (n=322) Road access 79 (n=1152) 20 (n=335) Evacuation of the area 64 (n=1112) 6 (n=317) Apart from disruptions to the electricity supply and road access to their homes, households in 2013 experienced relatively little disruption to their everyday lives. In the much larger 2011 flood event, there were significant disruptions to utility, medical and food supplies, and road access to homes. Not only did households experience greater levels of disruption during the 2011 flood than households during the 2013 flood, in many cases they experienced the disruptions for longer periods. For example, in 2011 the electricity supply was disrupted for longer than one week for 79% of households (Table 6.9), compared with just 11% of households in 2013 (Table 6.10). 46 Household experiences of flooding in Brisbane and Ipswich

53 Table 6.9 Extent and type of disruption, proportion of households experiencing disruption, 2011 flood. Disruption Less than one day Two days to less than one week One to four weeks More than a month Electricity supply (n=1183) Water supply (n=605) Gas supply (n=362) Landline telephone access (n=877) Mobile phone access (n=264) Food and or medical supplies (n=403) Road access (n=913) Evacuation of the area (n=1112) Percentages may not add to 100 due to rounding. Table 6.10 Extent and type of disruption, proportion of households experiencing disruption, 2013 flood. Disruption Less than one day Two days to less than one week One to four weeks More than a month Electricity supply (n=107) Landline telephone access (n=44) Road access (n=66) Note: Other disrupted services are not included as the number of respondents reporting that type of disruption was too low. Percentages may not add to 100 due to rounding. 6.3 Repair and rebuild Table 6.11 shows the average time taken to start the repair or rebuilding of the damaged home for each depth cohort. On average it took 95 days for repair or rebuilding work to start, with depths less than 1.2m taking less time than depths of 1.2m or more. Table 6.11 Average time taken to the start of repair or rebuilding by inundation depth, 2011 flood. Overfloor inundation depth (metres) Average time taken (days) 0.15m or less (n=56) m 0.7m (n=189) m 1.2m (n=249) m 2.4m (n=197) 104 More than 2.4m (n=312) 102 Depth not stated (n=57) 97 Total (n=1060) 95 Nearly a quarter of respondents (23%) reported that they had experienced a shortage of tradespeople required to commence the repair and rebuilding work. For those who reported delays it took on average 148 days to engage a contractor. Some commented that they were able to get in quickly and organised their own tradespeople, at a significantly reduced cost than those employed by the Household experiences of flooding in Brisbane and Ipswich 47

54 insurance companies. Other respondents complained that builders and tradesman lost interest once they learnt that it was not an insurance claim and provided excessive quotes. Respondents were asked about the state of rebuilding or repair of their homes and Table 6.12 shows most households had completed the repairs at the time of the survey. A small proportion (1%) advised that they had moved to another home. Table 6.12 State of repair or rebuild at time of survey (April-May 2012), proportion of responding households, 2011 flood. State of repair or rebuild 2011 flood (%) (n=1183) Completed repair 64 Completed rebuild 11 Started repair 15 Started rebuild 6 No repair or rebuild started 4 Have moved to another home 1 Percentages may not add to 100 due to rounding. Some respondents commented on the delays caused by working through their insurance company. Having experienced the January floods is traumatic enough however, having to deal with insurance companies and incompetent assessors e.g. losing initial contents claim, not returning phone calls, no process for structural damage to our house was evident, lengthy process to approve claim took 7 months to approve building insurance really affected our recovery. We are still completing our final repairs nearly 17 months on. Most tradespeople have been really helpful. (Graceville, 2011 survey) Others complained of the poor quality of work. Then the insurance company took 6 months to fix our house only because the tradesmen they employed were incompetent example door knobs on upside down everything had to be done twice or three times each. (Churchill, 2011 survey) The delay to the rebuild was because the 2nd company/builder never came back to us with the changes, even after 6 calls from ourselves and insurance company. Very stressful 16 months we are exhausted, we had a minimal social life. Worked full time at our jobs and then every night and all weekend worked on the house i.e. claims, scope of work, changes to house, fixing poor workmanship from trades people, taking on extra work ourselves to save money. (Corinda, 2011 survey) Reconstruction and repair work was conducted by a variety of sources. After the 2011 floods insurance companies (41%) and private contractors (56%) undertook most of the reconstruction works (Table 6.13). However, friends (37%) and family (42%) also made a significant contribution to the rebuilding effort. A small number of households reported that when there was a builder or tradesperson in the household they were not permitted, by the insurance company delegate, to 48 Household experiences of flooding in Brisbane and Ipswich

55 undertake any of the reconstruction work. This caused additional stress, both emotional and financial, in the household. A number of those not covered by insurance commented on the good quality of the volunteer tradesmen. Community assistance and volunteer response was overwhelming and life changing. Repairs were done at cost for materials and labour by building company using apprentices and trades assistants. Good low cost response. (Anonymous, 2011 survey) Table 6.13 Types of helper to repair or rebuild the home, proportion of households with helpers, 2011 flood. Type of helper 2011 flood (%) (n=1113) Friends 37 Family 42 Volunteers 11 Community associations 3 Insurance company 41 Private contractors 56 Others 3 Respondents may have provided multiple responses so percentages may not add to Highlighted suburbs Fig Tree Pocket and Rocklea There are clear differences in the state of the rebuilding or repair of homes between Rocklea and Fig Tree Pocket (Table 6.14). Nearly three quarters (71%) of households in Fig Tree Pocket reported they had completed repairing their damaged home, which is higher than the survey population (64%). Fewer homes in Rocklea (48%) had completed repair at the time of the survey. Table 6.14 Current state of repair or rebuild at time of survey, selected Rocklea and Fig Tree Pocket areas, 2011 flood. State of repair Rocklea (%) (n=31) Fig Tree Pocket (%) (n=34) All households (%) (n=1183) Completed repair Completed rebuild Started repair Started rebuild No repair or rebuild started Have moved to another home Percentages may not add to 100 due to rounding. There were some differences in who helped with the repair and rebuild between the two suburbs. The use of private contractors provided the strongest difference, with much higher usage in Fig Tree Pocket households (84%) than Rocklea (38%) despite similar rates of insurance coverage. This may Household experiences of flooding in Brisbane and Ipswich 49

56 be reflective of the greater ability to pay private contractors by Fig Tree Pocket owners than in Rocklea or the general survey population as a whole. 6.4 Deterioration after repair Following a flood event it is necessary to allow the frame of a home to dry completely before replacing internal fittings to prevent future structural problems. When GA conducted the initial damage survey in the weeks following the 2011 flood it was found that some households had replaced or were in the process of replacing wall linings. The survey of the 2011 floods included questions about the length of time the house structure was left to dry and any deterioration after repair to examine the effect of replacing linings early. While there is no definitive way to know how long an individual home needs to dry, the Hawkesbury- Nepean Floodplain Management Steering Committee (2006 p. 67) recommended that the frame of the house be left to dry for between 9-22 weeks. To make allowances for the warmer weather in Brisbane, a correction factor of 0.5 was also recommended (ibid p.68). If this calculation is applied, homes in Brisbane should be left to dry for between one and three months (31-90 days). Table 6.15 shows that on average households left the frame to dry for 11 weeks (78 days) following the 2011 floods, with the length of drying time increasing with depth. Table 6.15 Average time taken to dry the frame after the removal of wet wall linings by inundation depth, 2011 flood. Overfloor inundation depth (metres) Average drying time (days) 0.15m or less (n=37) m 0.7m (n=169) m 1.2m (n=146) m 2.4m (n=234) 80 More than 2.4m (n=272) 92 Depth not stated (n=21) 99 Total (n=879) 78 Table 6.16 Length of time wall frames are left to dry out, proportion of reporting households, 2011 flood. Time taken for frames left to dry out 2011 flood (%) (n=879) 7 days or less 12 8 days to 30 days days to 60 days days to 90 days days to 150 days 8 More than 150 days 20 Percentages may not add to 100 due to rounding. Table 6.16 shows the proportion of households that left wall frames to dry out, grouped into critical time periods. While the recommended drying time covers a broad time period, it is clear that some 50 Household experiences of flooding in Brisbane and Ipswich

57 households had placed new linings on the walls before the frame had an opportunity to completely dry. Forty per cent of households left their wall frames to dry for less than the minimum recommended time of 30 days and 33% within the recommended 31 to 90 days timeframe. The remaining 28% of households dried the wall frames for longer periods with 20% leaving the wall frames to dry for more than 5 months (150 days). More than a third (36%) of all households reported a deterioration of some kind since repairing or rebuilding the home. The appearance of mould or dampness (Table 6.17) rated the highest type of deterioration in all levels of flood inundation. These types of problems can have implications for the health of residents: The gastro suffered by our family was an issue which was not conclusively resolved yet seemed linked to the post-flood mould on our house. (Sherwood, 2011 survey) Table 6.17 Households identifying deterioration in the home during and after reconstruction, proportion of households with deterioration, 2011 flood. Type of deterioration 2011 flood (%) (n=452) Mould/dampness 37 Rotting timbers 14 Warped timbers 25 Warped flooring 24 Repair or replacement of stumps, damaged foundations or ground movement 11 Smell 2 External walls or structures crumbling or cracking 3 Cracks or movement in walls, cabinetry or tiles 15 Jammed doors or windows 5 Bubbling or peeling paint 6 Corrosion 2 Water stains reappearing 2 Ants and termites 2 Other 4 Respondents may have provided multiple responses so percentages may not add to 100. Warped timbers and warped flooring were also common problems, reported by a quarter of households. Warped flooring was reported more frequently by those with overfloor inundation of 0.15m or less (32%) when compared with 24% for all respondents. There is a high probability that inadequate drying of wall and floor timbers contributed to this deterioration. One respondent commented that they only noticed rising damp on doors some 16 months after the 2011 floods. A small proportion of households reported finding the presence of termites in the wall framing thus exacerbating the repair process. The termites may have been present prior to the floods, however they are also attracted to damp timbers, so a flooded house waiting for repair makes an attractive home for termite colonies. Household experiences of flooding in Brisbane and Ipswich 51

58 We have not finished repairing yard from 2011 floods and fences continue to deteriorate. Also any metal that was covered by flood water in 2011 is rusting although it appeared to have been clean and unaffected immediately after 2011 floods e.g. nails in buildings. (Tennyson, 2013 survey) Generally the proportion of households reporting deterioration was consistent across the depth cohorts. However half (52%) of the households that had a water depth of 0.15m or less reported deterioration. While information on the extent of repairs was not collected, it is possible that this difference is caused by households not understanding the extent of damage caused by low levels of overfloor flooding (e.g. damage to cabinet kickboards and skirting boards), and so did not make adequate repairs to their home. Combining the deterioration data with the time left to dry shows that households were more likely to report issues such as mould, dampness and warped timbers and flooring if the home was left to dry for 90 days or less (Table 6.18), though the effect is not large. Issues such as ground movement and damage to foundations or stumps did not differ according to the length of time dried. This was also true for cracking or movement in walls, cabinetry and tiles. This kind of damage can be caused either by the frame and cabinetry warping, or by movement in the foundations and stumps. Table 6.18 Households identifying deterioration in the home during and after reconstruction by drying time, proportion of households that noted a drying time, 2011 flood. Time taken for frames left to dry out Mould/ dampness (%) Warped timbers (%) Warped flooring (%) Any kind of deterioration (%) 7 days or less (n=103) days to 30 days (n=252) days to 60 days (n=173) days to 90 days (n=117) days to 150 days (n=67) More than 150 days (n=181) Mitigating against future flooding Disasters can be seen as an opportunity to make households more resilient, by repairing or rebuilding homes so they are more able to withstand future events. Table 6.19 shows 60% of respondents to the 2011 survey reported that no changes were made in repairing their home, that is, the damage was repaired like-for-like. Responses to the 2013 survey show that a similar proportion of homes were again repaired like-for-like, though the number of households inundated was very small. From this it can be inferred that should another similar flood occur sometime in the future, the same level of damage will again be experienced by the majority of households in the Brisbane and Ipswich area. Following the 2011 floods, 5% of respondents elevated the home. When broken down by water depth cohort, 9% of respondents who had 0.15m or less of inundation and 8% of those with 2.4m or more said they had elevated their home. In contrast only 1% of households that had overfloor inundation between 0.7m and 1.2m elevated the home. This pattern was also reflected in respondents who had changed the use of lower level rooms ranging from 15% and 14% (for inundation of 0.15m or less and 2.4m or more respectively) to 6% (for inundation between 0.7m and 1.2m). 52 Household experiences of flooding in Brisbane and Ipswich

59 Table 6.19 Mitigation actions to reduce the impact of future flooding, proportion of responding households, 2011 flood. Mitigation actions 2011 flood (%) (n=1176) Elevating the home 5 Installing water resistant linings 11 Replacing existing flooring with water resistant materials 19 Changing the use of lower level rooms 12 Raising electrical outlets 8 Installing non-absorbent insulation 2 Raising air-conditioning units/hot water systems 1 Demolishing existing building 1 No changes made 60 Other 5 Respondents may have provided multiple responses so percentages may not add to 100. In the survey following the 2013 floods, households were asked if any previous measures taken after the 2011 floods (like raising the home or installing water resistant materials), avoided or substantially reduced the damage sustained in the 2013 flood. Eight per cent of respondents reported that the mitigating actions taken post 2011 directly resulted in their home sustaining little or no damage. Some households gave reasons why they repaired like-for-like, either as part of the other category free text option, or in the concluding comments at the end of the survey. For many respondents that were insured, there was little option to make modifications as the insurance companies would not pay for the extra costs. If we had wanted to return our house to a pre-flood standard, we would have been covered by our insurance payout. We decided to change a lot in our house and used the disaster as a positive. (Tennyson, 2011 survey) Other respondents noted they wanted to be able to elevate their home but the costs were too high. Some of these respondents said they had plans to elevate at a later stage. There were some differences across areas, highlighted in our case study suburbs. In Rocklea the proportion of households who repaired like-for-like was lower at 45%, while in Fig Tree Pocket over three quarters (77%) of households reported repairing their home like-for-like. Households in Rocklea, therefore, have undertaken more actions to mitigate against future flooding than the households in Fig Tree Pocket. 6.6 February 2013 follow-up field survey In February 2013 a follow-up field survey was conducted to investigate the nature of repair work in homes which experienced extensive overfloor flooding in Houses that were subjected to 2m or more inundation above floor level were visually assessed from the street as to the nature of the repair work. The survey team visited 850 houses. Table 6.20 shows the results of that field survey. Based on Household experiences of flooding in Brisbane and Ipswich 53

60 what could be seen from the street, the vast majority of the 850 houses (90%) were repaired like-forlike. However, there was no way to tell if any changes had been made inside the home to mitigate against future flooding. Table 6.20 Repair strategy following 2m or greater inundation over floor level in 2011 flood, proportion of homes surveyed in February Repair strategy February 2013 observation (%) Property still abandoned 0.5 House under construction (replacement or repair) 1.1 Land reclaimed as open space 0.4 Existing house raised (usually on steel columns) 3.5 House replaced and elevated on mound 0.1 House replaced or repaired as before flood 90.2 House replaced with different house but not higher 0.4 The following photos (Figure 6.2 to Figure 6.5) show examples of work undertaken to mitigate against future flooding. Figure 6.2 Three new homes under construction which replaced two low set Queenslanders. The top of the steel columns are 6m above ground level. 54 Household experiences of flooding in Brisbane and Ipswich

61 Figure 6.3 This photo from the initial 2011 damage survey shows an extensively damaged low set house. This house has been replaced by the house in Figure 6.4. Figure 6.4 An example of a new elevated construction. Household experiences of flooding in Brisbane and Ipswich 55

62 Figure 6.5 An existing low set house that has been raised above flood height level. 6.7 Chapter 6 summary Prior to the 2011 floods 71% of respondents thought that they were covered by flood insurance. However only 32% of all households had their claim fully paid, and a further 13% received a partial payment. On average respondents incurred a damage bill of $140,500 of which $72,700 was paid directly from their own pockets. Extrapolating to all 17,100 homes, the total damage bill was $2.7billion with the costs shared by households and insurance companies. Not all damage can be measured as dollar values, with irreplaceable items such as photographs, personal records and collections also lost. Items normally considered replaceable, such as books, were frequently noted. Respondents noted the sentimental value of these items as they are associated with particular people and events. On average it took 95 days for repairs or rebuilding to start, with delays caused by the insurance company or a lack of tradesmen. At the time of the 2011 survey, 16 months after the 2011 flood event, three-quarters of households had completed repair or rebuild, with 4 per cent not started. More than a third of respondents noted a deterioration such as warped flooring or mould following the repair of the home. The deterioration was greatest in households that left the frames to dry for less than 90 days and homes that received overfloor inundation of 0.15m or less. 60% of respondents repaired their homes like-for-like with no mitigation for future flooding. Some respondents that were insured noted that the insurance companies would not pay for any changes to the home, while others found the costs too high. 56 Household experiences of flooding in Brisbane and Ipswich