Flood risk modelling: a crisis of confidence?

Transcription

1 Proceedings of ICE Civil Engineering 150 May 2002 Pages Paper Keywords design methods & aids; floods & floodworks; statistical analysis Flood risk modelling: a crisis of confidence? B. A. Hamer and R. Mocke Ben Hamer is a principal engineer with Halcrow Group in Swindon The flood events of October 2000 brought significant damage and disruption to wide swathes of the UK. Lessons learnt included the need to improve on the rather simplistic approaches to assessing flood risk that are commonly in use. Both the operating authorities and the insurance industry have sharpened their resolve to improve on flood risk assessment in order to ensure that limited resources are focused on the most effective means of mitigation.this paper describes the research work of the UK Environment Agency s primary consultancy on flood risk. It demonstrates the range of sophisticated tools currently available from broad-scale appraisals of assets at risk to site-specific flood propagation models and calls for civil engineers to put more confidence in their application. Richard Mocke is a chartered engineer with Halcrow Group in Swindon Civil engineers in England and Wales are all familiar with the images of severe flooding that followed the rainfall events of October 2000 (Figs 1 and 2). Behind these dramatic images lay significant losses and disruption to people across the country, affecting working and private lives in many towns and villages. What was most unusual about the floods in 2000 was perhaps the fact that their effects were so widespread across the country. It is widely acknowledged that there are insufficient funds available to undertake all those flood defence measures that may be justified on a straightforward economic basis (that is where the present value cost of defence works is less than the present value of the potential economic damages if no measures are taken). Accordingly, the government Fig. 1. Flooding at Yalding, Kent, in autumn 2000 Fig. 2. Flooding of the Rother Valley, South Yorkshire in autumn C I V I L E N G I N E E R I N G

2 FLOOD RISK MODELLING: A CRISIS OF CONFIDENCE? economic losses encountered in the recent flooding were often poorly predicted by the models adopted administers a priority score system, through the Department for Environment, Food and Rural Affairs (DEFRA), such that the limited budget may be focused in an appropriate manner. The priority of a scheme has, to date, been measured as a function of urgency, economic justification and other priorities that are a matter of policy such that urban areas have greater priority than rural locations. An updated version of this priority assessment system is now out for consultation within the industry. For this competition for funds to be an efficient and fair means of administering the flood defence budget, the assessment of urgency and economic justification clearly needs to be robust and consistent across England and Wales. Urgency is most commonly related to the probability of flooding, which, in turn, is related to the probability of defence collapse and the probability of the defences simply being overflowed, as was most commonly the case in The adequate assessment of these factors is highly dependent on the availability of data on physical parameters, such as level and flow in rivers, and wave characteristics and water levels around the coast. The past five to ten years have seen a slow but steady acknowledgement of the value of such data and, increasingly, these data are available at most sites of significant flood risk. Best practice for dealing with uncertainty The approach to the evaluation of the potential economic damages due to flooding should follow guidance provided by DEFRA. 1 Whereas this guidance sets out a consistent approach to the evaluation of national losses or economic damage, it cannot set out best practice in terms of flood risk analysis, which is a transient state. Operating authorities and their consultants are rightly left to ensure that best practice is applied at a scale that is consistent with the study site. The authors hope this and related papers in this special issue of Civil Engineering will assist in the function of spreading best practice and in communicating to the wider industry the type of tools that are now available. While the degree of uncertainty inherent in flood risk appraisal does not always justify the use of accurate flood propagation models, the authors do not consider that this level of uncertainty is a sound basis for adopting inaccurate flood modelling techniques, which may compound any errors. The increasing availability of lidar (light detection and ranging) data and other topographic datasets make the construction of digital terrain models for interrogation within geographic information systems (GIS) a more reliable alternative for these straightforward cases. The flood risk models adopted by the insurance industry tend to focus on the stochastic assessment of failure probability, which is the subject area for which insurers have some historic data. However, flood propagation is not addressed on the whole, as borne out by the fact that the economic losses encountered in the recent flooding were often poorly predicted by the models adopted. It is against the background of an increasing need for a rigorous analysis of flood risk, both to focus the spending on flood defence measures and to determine insurance levels, that this paper has been prepared. The authors describe the work of the authors consultancy firm Halcrow from national to local level for DEFRA, for the British insurance industry, and as the Environment Agency s lead supplier of consultancy services. Broad-scale appraisal of flood risk At a national level, Halcrow has undertaken an appraisal of the economic value of assets at risk from flooding throughout England and Wales on behalf of DEFRA. The objectives of this study were to determine the approximate value of the assets at risk from flooding, and their spatial distribution within the country. This information was then linked to a broad assessment of the current state of flood defences in order to determine the level of expenditure that would be required to maintain the current defence standard and to determine the additional cost that would be required to meet the indicative standards set out in the DEFRA project appraisal guidance. 1 Whereas the ideal approach would be to analyse the discharge of water over or through a defence and onto the floodplain for a series of flood events with specified return periods, there are several limitations at this national scale. Most notably, good quality data for topographic mapping are not consistently available, the probability and magnitude of hydraulic events are not well established at a national level (results are generally site specific and of varying quality), and up-to-date information relating to defence condition and defence standard is sparse. Against this background, a pragmatic approach to flood risk appraisal was warranted, using a range of data in a defined order of preference. Flood risk data were extracted by preference from section 105 modelling (so called as it is carried out by the Environment Agency under section 105 of the Water Resources Act 1991), which define flood risk areas for events up to the 200 year return period. As an aside, these maps are available through the Environment Agency s website and might form a useful basis for conveyancing searches. Where section 105 mapping had not been completed, then historical flood mapping was adopted, with specific events recorded for estimated return periods. Where no such mapping was available, the results of a national study by the Institute of Hydrology were used, providing a coarse level of analysis. Flood mapping reveals 0.6 billion national risk The large-scale use of GIS was shown to have major advantages over traditional mapping techniques in the flexibility that was provided to the project team, as they refined the approach to the economic appraisal. Some of the key study results are as follows. If there were no flood defences in England and Wales, annual average damages of 2.8 billion are predicted. If current defence standards are maintained, annual average damages will reduce to approximately 0.6 billion a year. C I V I L E N G I N E E R I N G 31

3 HAMER AND MOCKE Fig. 3. Coastal flooding at Sea Palling, Norfolk, in 1953 the area has recently been used to develop a new dynamic flood model If indicative standards are met, annual average damages will be closer to 0.2 billion a year. Clearly, the economic case for flood defence is strong, and the report 2 demonstrated that flood defence spending needed to be increase by 80% in order to maintain the status quo. Since the issue of the initial report in 2000, flood defence spending has increased with further increases anticipated. 3 As topographic data become increasingly available, and once an accurate record of flood defence assets is established and maintained, then it will be possible to undertake a more accurate appraisal of flood risk areas using the sort of tools described in the remainder of this paper. Until that time, the use of best available data at an appropriate level of analysis provides a robust tool for high-level decision-making. Assessing risk at post code level Further broad-scale appraisals are under consideration by the insurance and reinsurance industry, with the objective of evaluating flood risk on a spatial scale similar to the post code system or other manageable system for the application of insurance levies. From the consultancy s discussions with key industry players, it is evident that insurers have already invested substantial sums in obtaining topographic datasets and in improvements to flood risk models. The authors have demonstrated the viability of setting up large-scale flood propagation models to satisfy this objective, based on the Isis river-modelling suite. Such models can be built readily to provide the level of definition of the floodplain that is appropriate to the desired output typically, spatial resolution to the post code level is required for insurers. The model works by providing a representation of water flow between a series of interconnected reservoirs, with defined flow characteristics within them and across their boundaries. The method is described in more detail below, with reference to two case studies. Strategic level appraisal of flood risk At a strategic or scheme level, the viability of any proposed flood defence works is heavily reliant upon a robust evaluation of potential flood damage, both in the event that no works are implemented (the do nothing approach) and also to determine the residual damages associated with various intervention measures. Development of flood models offers more confidence Active development of flood propagation models in order to improve on the confidence and repeatability of predicting flood limits and flood depths, both in the fluvial and coastal environments, has been undertaken. In particular, the traditional role of river flow modelling to consider flow across the flood risk zone, taking account of time-dependent factors, such as varying water level (through the development of a floodplain unit incorporating floodplain roughness) or breach development (through the development of a new unit for this purpose simulating the spatial and temporal development of a breach in a flood defence). The improved representation of land drainage and hydrodynamic features (roads, embankments, culverts, outfalls, etc.) are also included within the flood model. In order to determine economic damages from potential failure of sea defences, a hydrodynamic model was established for a flood risk area in Norfolk. 4 The village of Sea Palling is one of several settlements that lie within a 6000 ha floodplain, which is protected by 14 km of sea defences. The village has encountered serious flooding, as shown in Fig C I V I L E N G I N E E R I N G

4 FLOOD RISK MODELLING: A CRISIS OF CONFIDENCE? the level method tends to significantly overestimate the flood extents by failing to account for any natural or man-made barriers to flow Fig. 4. User interface for the Isis flood propagation model results have proved to be consistent and repeatable, inspiring greater confidence An Isis flow model (user interface shown in Fig. 4) was established for the hinterland area at risk using a series of interconnected reservoir units linked by spills to represent any natural or man-made barriers, such as elevated roads or embankments. The results of this assessment were compared to two widely used, simplistic methods for evaluation of flood damage. These were termed the level method, whereby flood limits are determined by overlaying an extreme water level on land contours, and the volume method, whereby the volume of flow through a breach is calculated and then spread through the flood risk area. It was demonstrated that the level method tends to significantly overestimate the flood extents by failing to account for any natural or man-made barriers to flow. Determining the flood extents using the volume method can be arbitrary and often relies on a degree of judgement, although the method can be improved through the use of GIS. In either case, the method fails to consider the dynamic nature of flood propagation. The application of a dynamic flood model was shown to be technically preferable to the more simple approaches for the site, and showed reasonable agreement with historic records. The level method is expected to vastly overestimate economic damages in all cases, whereas the volume method has the potential for accurate predictions in sites where dynamic processes are not important, but with substantial variance where such factors are significant. Moreover, the dynamic model was consistent and repeatable, and sensitivities to changing assumptions on defence failure characteristics (e.g. breach width) could be tested efficiently through multiple runs once the model was set up. It was noted that the site was not particularly complex from a dynamic perspective, and Model of river system Flood event data Identify flood alleviation scheme options Geographic information system Digital terrain model Direct measurement at property threshold level Building floor area Building classification FHRC flood loss assessment information report Loss adjustor's report WINFAP HEH Statistical analysis Run ISIS - FEH Rainfall - Run off model Physical data existing conditions and flood alleviations scheme options Identify model node for each building Estimate of property threshold level Inflow hydrographs Set up ISIS - Flow Models of river system Run ISIS - Flow models Water level v Time at each model node Building database DEFRA ECOPAG 3 economic evaluation Calculate flood depth and duration for each building Visual presentation of damages Flood depth versus damages for each building classification Determine flood extent Economic evaluation Establish scheme costs Duration Level is exceeded at each model node Peak water level at each model node PV Costs PV damages PV damages awarded Net present value Average benefit cost ratio Incremental benefit cost ratio Fig. 5. Assessing the economic benefit of a flood scheme using the Isis flood model C I V I L E N G I N E E R I N G 33

5 HAMER AND MOCKE Key Building locations Flood cells Washlands storage Potential flow routes Dagenham incorporated into the model. The economic consequence of breaches at a range of locations can be evaluated to guide the preferred level of maintenance and to inform operations staff of those areas at particular risk of consequential flooding. Knowing the extent of flooding from a particular breach location can also have enormous benefit in guiding the extent of evacuation necessary when a breach is predicted. River Thames Motor works Motor works Beam River the differences between the methods are expected to be greater for sites with greater topographic variation. Better models ensure more costeffective solutions Having a fully reproducible model allows the efficient evaluation of various flood defence options in order to determine the most cost-effective solution. This concept was further developed on a fluvial project commissioned by the Environment Agency to consider the washlands area in Dagenham. 5 In this scheme, a flood control reservoir was to be linked to the tidal River Thames by way of a channel that passed through a high-value industrial area. The steps in the process of economic evaluation of flood risk are shown in Fig. 5. A flood model was established to describe the flood risk area as a series of interlinked channels and reservoir units (see Fig. 6) to represent the physical characteristics of the scheme. The model was calibrated successfully using a significant flood event in Running a series of flood event simulations for the present case confirmed that the system was failing (becoming inundated) more frequently than its design standard and a Motor works 0 m 1000 Fig. 6. Flood model for the Dagenham industrial area consists of a series of interlinked flood cells after calibration against a major 1998 event it enabled engineers to evaluate a wide range of flood defence options number of changes were proposed to improve the situation. The Isis flow model was used to predict the flood extents and depths as a result of these changes. These flood extents and depths were linked through to GIS-based flood mapping (using ArcView GIS) containing information on the location and threshold levels of economic assets in the flood risk areas. These were, in turn, linked to the DEFRA project appraisal guidance 1 spreadsheets used for the economic evaluation of flood defence options. The integration of these systems allowed for the quick and reproducible economic evaluation of a number of flood defence options. In this way, inefficient schemes were easily ruled out and the more efficient proposals could be enhanced to derive the preferred solution, to maximise the benefits and to deliver best value for the taxpayer. The establishment of flood models at a strategic level has a number of other advantages and functionality beyond the immediate aims and objectives of the case studies described here. For coastal defence schemes, for example, wave overtopping volumes and the timing and development of defence failure can be The importance of accurate topographic data However, it is extremely important to recognise that the accuracy of predictions remains only as good as the information that has been incorporated in the flood model. Significant advances in technology in recent years have assisted in this process somewhat. The use of lidar remote surveys (elevation accuracy possible to approximately +/- 150 mm) and more efficient global positioning system (GPS) surveying techniques (elevation accuracy possible to approximately +/- 10 mm) has improved the level of accuracy of topographical information available. Where possible, these should be substantiated with as much level ground data as possible to include drainage channels, road and rail embankments, relic defences and so on. Ordnance Survey Landline and Address Point data now linked to general flood loss assessment information report categories have been used successfully in the evaluation of larger scale schemes. Smaller, or more economically marginal, schemes may require more indepth information. For example, where the economic justification is sensitive to the damages attributed to very few assets such as large industrial or commercial units, schools and offices it may be appropriate to spend time on a land-based survey of some critical threshold levels. A case for confidence The techniques presented in this paper are now well established within the authors firm, but are still the subject of ongoing research and development in order to improve their application and ease of use. In particular, ongoing research to include flood mapping and 34 C I V I L E N G I N E E R I N G

6 FLOOD RISK MODELLING: A CRISIS OF CONFIDENCE? Engineers need to work closely with economists and statisticians to deliver realistic assessments of flood risk, validated against real data economic evaluation as bolt-on modules to the Isis software suite is under further development. Such a system would allow the more efficient use of engineering resources by concentrating more on the testing of various flood defence options, rather than time spent in spreadsheet analysis and calculations. Adoption of the type of methodologies presented here has the potential to increase the confidence that may be placed in assessments of flood risk, particularly where validation data are available. As a result of the unfortunate events of 2000, there are now much better records of flood extents, depths and damage values. The industry should be encouraged to use these data to validate the tools available against deterministic (real) events. In so doing, the benefits or otherwise of applying a dynamic approach to flood risk appraisal should be evaluated. If, as the authors suggest, the improvements in understanding are significant while the increased effort (time and money) is relatively small, the wide-scale adoption of this approach should be encouraged. The availability of flood damage data also provides an opportunity to re-focus the assessment of flood risk in a practical sense. To date, much of the development of risk appraisal techniques for the evaluation of potential economic damage has not been undertaken with the (unfortunate) benefit of practical experience. Engineers need to work closely with economists and statisticians to deliver realistic assessments of flood risk, validated against real data. The methods described in this paper have been used on a number of flood defence evaluation schemes and demonstrate the efficiency of computational, dynamic techniques. Whereas the models available are the subject of constant development, they are already able to reduce uncertainty through improved hydraulic representation and, importantly, through their suitability for multiple tests of scenarios to assess sensitivity to key assumptions. The tools are appropriate for use in evaluating flood risk from regional to local level and, while accurate in areas of good calibration data, allow an increased understanding of flood risk and its sensitivity to key assumptions. The future direction of flood risk appraisal Flood risk is a function of probability and consequence. Engineers can reduce the probability of failure, although not eliminate it, and other measures are widely accepted as necessary to manage the consequences. These range from flood warning, planning and development control, and better information through the Environment Agency s education programme. Measures to increase insurance premiums or to refuse insurance cover for property in high-risk areas are already starting to receive public attention, and will do so increasingly in years to come. Practical measures, such as the inclusion of an assessment of flood risk as a standard part of property searches, should be expected to follow. However, we should be cautious in our estimation of the degree of impact that insurance initiatives and planning controls are likely to provide in addressing flood risk. In the past 20 years, for example, more than residential properties have been built on floodplains, of which were built in the period The pressure on space for development limits the impact that planning control can have on this issue, although the part it can play remains significant. There is a consensus against development in green areas and government policies drive forward development in brownfield sites through regional development plans (particularly in the developed and at-risk South East). Current targets are for an additional three million homes by 2016, with many of the potential brownfield sites located in flood risk areas. The UK flood defence industry spend remains below the 360 million budget required to maintain current defence standards. With this fact comes the bold acceptance that we will continue to encounter flooding in our river valleys and around our coasts for the foreseeable future and received wisdom suggests that this will occur at an increased frequency and severity. The paucity of resources confirms the need for more consistent and justified processes for the assessment of flood risk in order to focus expenditure in the most efficient manner. References 1. MINISTRY OF AGRICULTURE,FISHERIES AND FOOD. Flood and Coastal Defence Project Appraisal Guidance Economic Appraisal. The Stationery Office, London, 1999, Report FCDPAG3. 2. HALCROW GROUP LTD,HR WALLINGFORD AND JOHN CHATTERTON ASSOCIATES. National Appraisal of Assets at Risk from Flooding and Coastal Erosion including the Effects of Climate Change. Department for Environment, Food and Rural Affairs, London, New Civil Engineer, 24 January HAMER B.A.,WEBB D. R., DALE W. C. and HOME R. Improved methodology for predicting flood limits in coastal regions. Proceedings of 35th MAFF Conference of River and Coastal Engineers. Ministry of Agriculture Fisheries and Food, London, WEBB D. R., MOCKE R., ENEVER D. and NAYLOR S. Implementation of MAFF project appraisal guidance in complex river systems. Proceedings of 36th MAFF Conference of River and Coastal Engineers. Department for Environment, Food and Rural Affairs, London, Bibliography CARPENTER G. United Kingdom Floods, Guy Carpenter Publication, HALCROW GROUP LTD. Washlands investigation phase II. Report prepared for the Environment agency, HALCROW GROUP LTD. Happisburgh to Winterton sea defence strategy.project Appraisal Report prepared for the Environment agency, INSTITUTE OF HYDROLOGY. National assessment of river flood risk. 1995, Report 130. MIDDLESEX UNIVERSITY FLOOD HAZARD RESEARCH CENTRE. Flood loss assessment information report (FLAIR), What do you think? If you would like to comment on this paper, please up to 500 words to the editor at simon.fullalove@ice.org.uk. If you would like to write a paper up to 2000 words about your own experience in this or any related area of civil engineering, the editor will be happy to provide any help or advice you need. C I V I L E N G I N E E R I N G 35