A PROBABILISTIC OPTIMISATION METHODOLOGY FOR FLOOD RISK MANAGEMENT IN THE SCHELDT AND LYS BASIN

摘要

In response to the European Flood Directive a flood risk management plan (FRMP) is being prepared for the Ghent canals and the Belgian part of the Upper Scheldt and the Leie/Lys basins (ca. 10.000 km~2 drainage area in total) using a full probabilistic risk approach. According to the Directive FRMP's should focus on preparedness, prevention and protection, considering climate change (CC) and land use change (LUC) impacts. In this study a methodology is presented which allows for selecting an optimal combination of flood risk mitigation measures according to economic and societal criteria, i.e. the net present value (NPV) and people at risk (P@R). Risk analysis requires considering the entire frequency domain of flood consequences. Hence statistical extrapolation is necessary. This is done by generating 2500 synthetic extreme events for the different catchments, taking into account joint occurrences by means of an extreme value copula. After simulation with the hydrodynamic flood model, empirical frequency distributions are drawn for flood depths in the floodplains. As a validation, these distributions are plotted against historical events, displaying good similarity. At this point, flood risk maps could be drawn, while CC and LUC are taken into account by future projections for the year 2050, through perturbed rainfall and evaporation series and urban development scenario's. Protection measures include different river channel widening configurations, levee increase, dike displacement, local protection walls and pump stations. Prevention measures include resilience and retreat measures for buildings, and preparedness measures include an early flood warning system. In order to detect the optimal scenario, any possible combination of protection, prevention and preparedness measures is considered. This yields over 1.000 possible FRMP's, being evaluated by means of the hydraulic model and a GIS loss model. Each scenario produces a benefit, defined as a decrease of monetary and societal risk. At the same time construction, operational and maintenance costs for each combination are assessed. Considering a lifetime of 40 years, NPV and P@R are calculated for any individual scenario. Eventually a bivariate Pareto front is obtained, allowing to select an optimised FRMP according to economic and societal criteria.