Broad-Scale Assessment of Urban Flood Risk Mark G. E. Adamson 1

Transcription

1 Broad-Scale Assessment of Urban Flood Risk Mark G. E. Adamson 1 1 Office of Public Works, Trim, Co. Meath, Ireland Abstract The Directive on the assessment and management of flood risks (2007/60/EC The Floods Directive) requires Member States to identify areas of potentially significant flood risk through the Preliminary Flood Risk Assessment (PFRA). This assessment needs to consider flooding from a range of sources, and the associated risk in terms of adverse consequences to human health, the environment, cultural heritage and the economy. The implementation of the PFRA in Ireland has required the development of indicative mapping of areas prone to flooding from rivers, coasts and estuaries, groundwater and intense rainfall events, using rapid and cost-efficient methods to provide national coverage. An innovative approach has also been developed to define a quantitative, but non-monetarised Flood Risk Index (FRI) based on the probability of flooding and the susceptibility to damage and potential degree of damage or loss of a wide range of receptors, including cultural heritage assets and environmentally designated (Natura 2000) sites. The FRI provides a consistent metric to permit the comparison of indicative risk between individual assets or sites of value and a range of urban areas comprising different mixes of types of receptor. This paper will outline the processes involved in defining the potentially flood-prone areas, how the susceptibility to damage and the potential degree of damage for a range of receptors was determined, and how these factors are used to define the Flood Risk Index and meet the requirements of the PFRA. Keywords: Floods Directive, Flood Maps, Flood Risk, PFRA, Risk Assessment Introduction The EU Directive on the assessment and management of flood risks (2007/60/EC The Floods Directive) requires Member States to identify areas of potentially significant flood risk (referred to in Ireland as Areas for Further Assessment, or AFAs ) through the Preliminary Flood Risk Assessment (PFRA). This assessment needs to consider flooding from a range of sources, and the associated risk in terms of adverse consequences to human health, the environment, cultural heritage and the economy, based on available or readilyderivable information. The PFRA is an important step in the process of implementation of the Directive, as it is only for the AFAs where detailed flood maps then need to be developed, and for where objectives for the management and reduction of flood risk need to be defined, along with a prioritized set of measures aimed at achieving those objectives. At the outset of undertaking the PFRA in Ireland, flood maps with national coverage did not exist, and a national flood risk assessment had not been undertaken. Costeffective methods were hence required to address these gaps and complete the PFRA. Page 1 of 7

2 This paper outlines the processes involved in developing national indicative flood maps for a number of sources of floodwaters, and in the preliminary risk assessment then undertaken to meet the requirements of the PFRA. Due to the breadth of the topic, only brief descriptions of the mapping and risk assessment methodologies are given herein. Further detail on each of the methodologies is available in the relevant technical reports that are available from National CFRAM Website ( National Indicative Flood Mapping The Directive does not limit the scope of the analysis required in terms of sources of floodwater (with the optional exemption of floods from sewerage systems). While a range of sources were considered in Ireland as part of the PFRA, this paper will describe only the mapping related to floods from rivers (fluvial), coastal waters (including estuaries) and direct overland rainfall runoff and ponding (pluvial). Descriptions of the analysis related to other sources are available from the reports published by the OPW for the PFRA on the National CFRAM Website, as cited above. Fluvial Indicative Flood Mapping The method used to develop the fluvial indicative maps is based on the Normal- Depth method, by the following steps: - Major and Minor Nodes were established along the rivers systems at 500m and at points up and downstream of confluences, and at 100m centres, respectively, for catchments of area down to 1km 2, where the catchments were derived from DTM analysis. - Flood flows were derived for a number of annual event probabilities (up to 0.1% AEP) for each of the Major Nodes, based on the 6-parameter physical catchment descriptor equation developed to determine Q (2) in ungauged catchments in Ireland under the Irish Flood Studies Update research (unpublished at the time of writing). - Out of bank flows (Q fp(t) ) were assumed to be equal to Q (T) Q (2), based on observation and experience of Irish rivers. - Cross-sections were taken at both Major and Minor Nodes from the national IFSAR-based DTM. Cross-sections were taken to be perpendicular to the direction of flow averaged over a distance up and down stream, rather than only at the particular node, to capture valley flow direction and compensate for local meanders. - Manning s Equation was applied at each Major Node to determine a flood level for each event probability. Crosssectional parameters were derived from DTM-based cross-sections, while the floodplain roughness values were assigned based on the CORINE Land Class (EPA, 2006), and the slope calculated based on a reach extending up and downstream of the Node. - A simple rule was applied to account for backwater effects. - Levels from Major Nodes were interpolated to Minor Nodes based on an inverse distance weighted method. - A number of corrections were applied to reduce level errors arising from issues such as localised DTM error and overestimation at Nodes at confluences. - Flood extents at both Major and Minor Nodes were determined by extrapolation of the flood level for each event probability across the DTM-based crosssection to the point of intersection. Page 2 of 7

3 - Extents were linearly connected between nodes to provide a flood map for each event probability. The indicative flood extent maps produced through this process, implemented by Compass Informatics Ltd (Compass, 2011) on behalf of the OPW, were then validated against maps produced using detailed hydraulic modeling under the Lee Catchment Flood Risk Assessment and Management (CFRAM) Study. It was found that there was a slight overestimation of the mean flood extent topwidth from the indicative flood maps, as can be seen in Figure 1 below. The percentages of cross-sections within certain error residual bands are set out in Table 1 below. Frequency Count of Compared Cross-Sections Figure 1: Error Residuals for Q100 Indicative Fluvial Mapping Error Residuals (% No. of Cross-Sections) <+ / - 20m 22% < + / - 50m 42% < + / - 100m 62% < + / - 200m 81% Table 1. Error Residuals for the Fluvial Indicative Mapping Residual Difference in Flood Extent Widths (m) Errors were noted as being most significant in flat areas with wide, gently sloping floodplains, where areas of clear gross error (typically identifiable by long lateral spikes or wedges extending out from the floodplain) could be seen in some locations. These were removed from the mapping before use. The method also does not take account of flood defences (in order to identify the natural flood risk) or local channel improvements or constraints. The results indicate that while not locally accurate, the indicative mapping was broadly reasonable, and hence suitable for use for the PFRA, which is only a preliminary, screening exercise to determine areas where the risk may be potentially significant. Coastal Indicative Flood Mapping Consultants RPS (RPS, ), in conjunction with the OPW, undertook the Irish Coastal Protection Strategy Study (ICPSS) to develop maps indicating coastal and estuarine areas prone to flooding from the sea. The predicted flood extents are based on: - Analysis of historic recorded sea levels - Numerical modelling (MIKE-21) and statistical analysis of combined tide levels and storm surges to estimate extreme water levels along the national coastline for defined probabilities - Calculation of the extent of the predictive flooding, by comparing calculated extreme tide and surge waters levels along the coast with ground level based on DTM data (LiDAR data where available, else IFSAR data). Page 3 of 7

4 The maps have been produced at a strategic level and have certain limitations and potential sources of local error, notably: - The flood extents are determined by horizontal projection in-land of the extreme sea levels, and do not provide for the inland propagation and recession of the tidal flood cycle, providing for a potential over-estimation of the flood extent in large coastal floodplain areas. - Flood defences have not been included in the preparation of the maps, given the scale of the exercise and to allow for mapping of natural flood risk. - The methods (and maps) do not take account of (or represent flooding from) wave action or overtopping (which will be assessed only at the detailed floodmapping stage). Indicative Pluvial Flood Mapping Pluvial flooding appears to be an increasing problem in Ireland, as evidenced by a number of such events leading to significant damage, and indeed loss of life, in the Dublin area in recent years. A project was undertaken by H.R. Wallingford Ltd (HR Wallingford, 2010) on behalf of the OPW to provide national indicative mapping of potential pluvial flooding. The process for the development of these maps included: - Use of geo-referenced 2km gridded depth-duration-frequency rainfall data, developed by Met. Eireann under the Flood Studies Update research, for a number of annual event probabilities. - The base rainfall data was processed by applying areal reductions factors, storm profiles and the application of losses for: depression / storage (5mm) and infiltration ( mm/hr depending on soil type) in rural areas, or built storm-water drainage capacity (15mm/hr) in urban areas - The application of the net rainfall profiles to the H.R. Wallingford Rapid Flood Spreading Model (RFSM), which is a quasi-2d model using simple routing to allow for rapid computation over large areas. The models were built using a high resolution 10m grid derived from DTM data, and were developed for basins throughout the country to determine flood extents, depths and velocities for a range of events. - Final flood extent maps excluded areas where the depth was less than 50mm, on the basis of limited damage (flooding likely to limited to roads in urban areas). The value selected for urban losses was based on a comparison of mapping using the above process against flood extents derived from more detailed Infoworks CS modeling undertaken as part of the Greater Dublin Strategic Drainage Study. Figure 2: Sample Indicative Pluvial Flood Depth Map Page 4 of 7

5 As with the mapping for other sources described herein, the mapping is, by necessity, indicative. It does not take account of drainage paths through barriers captured within the DTM (e.g., culverts through large embankments), nor of the local urban storm water drainage system capacity, which varies significant both within and between urban areas. Flood Risk Analysis For the PFRA, it is necessary to consider flood risk to a variety of assets that could suffer harm or damage in the event of a flood (referred to as receptors ), including risk to people and social infrastructure as well as to economic, cultural and environmental assets. The degree of risk then needs to be compared between individual receptors and communities across the country, which would include differing degrees of risk to various mixes of types of receptors in different communities. It is common practice to measure flood risk in economic terms (i.e., annual average damages and long-term damages measured in Euro). However, these assessments typically only evaluate flood risk to a limited range of assets, such as damage to the structure and contents of houses and other properties, for which measuring economic damages can be done relatively readily, based on analysis of records of financial losses collected over a long period of time or detailed property inspection. These assessments generally do not quantify or measure the financial risk for the other types of receptor referred to above, for which monetarising damages is particularly uncertain, and for which there is a relative deficit of research. It was hence decided to develop a new, indicative system (Mott Macdonald, ) to measure flood risk in terms of a Flood Risk Index (FRI), taking into account both the probability and potential consequences of flooding. This is a quantitative but nonmonetarised metric that allows the risk to different types of receptors (e.g., home, business, monument, utility asset, etc.) to be expressed consistently and numerically, which in turn allows risk to the different types of receptor to be compared, and also for different mixes of risk to be compared between different communities. Under this method, the FRI is calculated based on the vulnerability category of a given receptor type, and the probability of flooding of that receptor. Vulnerability Assessment To apply the adopted risk assessment system, it was necessary to develop a method for vulnerability categorization, to describe the potential consequences of flooding that would be consistent across different receptor types. The receptor types were determined from under a number of categories and tiers making use of recent work commissioned by the OPW in this area (undertaken by Halcrow and the Flood Hazard Research Centre, Middlesex University Report not published). Approximately 30 receptor types were defined, although many sub-types were defined and categorized for some broad types, such as national monuments. The vulnerability categorization for each receptor type or sub-type was then determined based on: - The importance of the receptor - The potential damage to the receptor in the event of a flood. Page 5 of 7

6 The two criteria are necessary given that some receptors may be of great importance but unlikely to suffer any significant damage during a flood (e.g., a megalithic stone monument), whereas others may be of lesser value, but prone to total destruction in the event of a flood (e.g., a regional records office or museum with volumes of paperbased records). Each receptor type or subtype was assessed against the criteria to determine its vulnerability categorization. Of particular note as part of this exercise was the categorization for national monuments, which in Ireland are all classified as national monuments and are afforded the same degree of protection, but which have significantly variable importance from an archaeological perspective and equally variable susceptibility to flood damage. A specific study (Mott Macdonald, ) was undertaken to determine the vulnerability of hundreds of different sub-types of monument, for which the importance took into account factors such as period, rarity and diversity, while the susceptibility to flood damage took account of age, material and location in or adjacent to water bodies. The vulnerability for each receptor type or sub-type was defined based on this assessment against the two criteria by one of five categories; ranging from Critical down to Low. This categorization system is by its very nature subjective in particular the assignment of the value or importance category in that it is not based on directly measurable damage metrics, such as cost. However, the categorization was subject to consultation with relevant authorities, and is intended to represent a societal value that could be assigned to a given receptor, and to the societal loss in the event of that receptor being flooded. Determination of FRI Score The numerical FRI score for a specific receptor was determined on a matrix-based process according to the vulnerability categorization (potential consequences) and the probability of flooding, taken using the indicative mapping described above that was produced for three flood event probabilities. This process was undertaken for each individual receptor within the defined receptor groups, making use of a GIS-based tool. Some individual receptors might be deemed to represent potentially significant risk in their own right, such as nationally critical infrastructural assets. These would however be relatively limited in number, as it would not be practicably possible to define each home and business, for example, individually as an AFA. A process was hence developed, based on adjoining 0.25km 2 grid squares of risk to define at-risk areas, within which the FRI score for each individual receptor was accumulated to determine a community FRI score. This was run nationally to provide community FRI scores for all communities throughout the country, ranging from major urban centres such as Dublin down to small villages and dispersed receptors in rural areas. Determination of AFAs The predictive flood risk analysis described above, and the resultant FRI score for all individual receptors and communities, was one of key approaches used in the PFRA in Ireland, which was taken into consideration along with the outcomes of historical flood risk analysis and consultation to determine probable and possible AFAs in Ireland. Page 6 of 7

7 The AFAs were subject to both formal public consultation and professional on-site assessment and review to derive the final list of 300 AFAs that have been formally designated in Ireland. Conclusion There are scientifically advanced methods of robustly determining flood hazard, and also for determining the potentially resultant risk for some categories of receptor. Such methods can be applied locally where the degree of risk merits such detail, and associated cost. These methods are generally not however appropriate, due to cost and / or data demands, for broad-scale indicative assessments such as the PFRA, which is intended only to be a high-level screening exercise to provide an indication of where the greatest degree of risk might potentially exist nationally, without attempting to accurately determine the specific risk in each location. To deliver the PFRA in Ireland, the OPW, with the assistance of its consultants, has developed and implemented a highly costeffective process, based on predictive risk assessment, aimed at meeting the specific aims of this element of the Directive. The predictive assessment has, in conjunction with a review of historical flood risk and thorough consultation and professional review, lead to the identification of areas of potentially significant flood risk (AFAs) that can be considered robust within the context of the PFRA, given the information available at the start of the process. It was not intended that this process would accurately identify the flood risk distribution within urban areas, but it does indicate nationally those urban areas where the flood risk, from a range of sources, is a matter of particular concern. These areas are now subject to a process of detailed analysis through the ongoing National CFRAM Programme, under which detailed modeling and risk analysis will be applied to determine the prioritised set of flood risk management policies, strategies and measures required to provide for sustainable and cost-effective flood risk management in Ireland into the future. References Compass Informatics 2011, Preliminary Flood Risk Assessment: Fluvial Flood Hazard Mapping Normal-Depth Method, July 2011 HR Wallingford National Pluvial Screening project for Ireland, EX 6335, Oct 2010 Mott Macdonald, , Preliminary Flood Risk Assessment, The Predictive Assessment of Potentially Significant Flood Risk, Mott MacDonald, June 2011 Mott Macdonald, , Preliminary Flood Risk Assessment, Methodology for Classifying the Vulnerability of National Monuments from Flooding in The Republic of Ireland, Mott MacDonald, July 2011 RPS , Irish Coastal Protection Strategy Study, Technical Reports IBE 0104, 0071 & 0388 Acknowledgements The OPW would like to thank Mott MacDonald Ltd., Compass Informatics Ltd., H. R. Wallingford Ltd., National Parks and Wildlife Service and Bird Watch Ireland for their work and assistance with the PFRA in Ireland, including that described herein. Page 7 of 7