Process-based Regional Flood Risk Assessment: Methodology And Applications

An increasing-trend in extreme flood events has been observed in the global scope in recent years and that makes people to rethink traditional flood control strategies profoundly. At present, risk-based flood management has been a core strategy in flood control activities in many countries around the world. Proper flood management strategies should be formulated and adopted based on the risk analysis and risk assessment. Although quite a lot of researches have been conducted in the field of flood risk assessment, most of the existing models have to suffer from some significant limitations, e.g. without consideration of risk transfers among flood defence system and other flood control works. So it is necessary to carry out more in-depth study for flood risk assessment theory and method. This thesis aims to develop a new process-based flood risk assessment model with full consideration of flood risk transfer mechanisms and also includes several applications, the main content as follows:(1) Comprehensive summary of existing researches in the field of flood risk assessment. All the existing methods are classified to three types including index system-based method, historical flood hazard-based method and simulation-based method, and the three types of methods are reviewed respectively. The results show index system-based method can only be used for rough assessments, historical flood hazard-based method cannot be used for flood risk assessments independently and simulation-based method with clear physical mechanisms has many important advantages in spite of two constraints, i.e. data availability and computational complexity.(2) Anlysis of flooding process and related uncertain factors. Analyze flooding process from the view of material flow and energy flow, indentify related uncertain factors and discuss how to deal with these factors. Monte Carlo method is used to simulate the most important uncertain factor, i.e. flood load. A united sampling method combining Latin Hypercuber algorithm with sample point sorting algorithm is used to simulate synthesis situations of main stream floods and tributary floods. A stepwise sampling method is used to consider randomicities of peak discharges and flood hydrographs.(3) A new macro-level model based on flooding process simulation for large-scale flood risk assessment. Modules simulating sub-processes of the flooding process are selcected according to general principles of model selection and a conceptual event tree of flood disaster considering both natural flooding process and artificial coutermeasures is established, flood risk transfers among flood defence system are simulated by using state transition analysis method and banching and bounding algorithm, and then a flood risk assessment model under Monte Carlo simulation framework is presented. Formulas to calculate failure probabilities of flood defence system and its subsystem are deduced with consideration of time effects of flood action and length effects of dikes, flood damage estimating model are established by using approximate GIS method for simulating flood inundation, and then fomulas to calculate regional and sub-regional flood risks are presented. The model presented in this thesis can be used for both complete probabilistic flood risk assessment and conditional flood risk assessment upon a certain flood scenario.(4) Adaptive integrated flood management mode characterized by social learning process. Expound defects of tranditional flood management mode and challenges of flood management, introduce integrated flood management principles, definite a flooding system corresponding to these management principles and then estabilish an adaptive integrated flood management mode by introducing social learning process into the integrated flood management mode. Besides the above, process-based flood management under the constraint of the existing flood defence system and some important problems in flood emergency management are also dicussed.(5) Scenario analysis method for future flood risk. Integrated flood management principles are introduced into the existing framework of scenario analysis for future flood risk and social learning process is embedded into the PSIR (Pressure-State-Impact-Response) model of flood management. A method of periodically reviewing and adjusting flood management policies according to the variation trend of future flood risk is put forward, relationships between management policies and drivers of flood risk or states of flooding system are discussed and the way to reflect drivers in the flood risk assessment is proposed. The case study shows the method presented in this thesis can give more accurate predicted flood risk in the future, and so enlightments to current management policies will be more profound.