Modelling for decision making in water risk scenario

The Oxford Dictionary defines a model "A simplified mathematical description of a system or process, used to assist calculations and predictions. Just from the mere definition, the importance of a model to manage and analyze a process is underlined. The scenario modelling has a vital importance in all of the branches of the decision analysis. In fact, the cognitive process that leads to reach a decision, that is the basis of the decision making, needs a deep analysis and knowledge of the "real world". This knowledge is achieved by means of studying the physics of the process and the empirical observations. For that, the decision makers needs to analyze different hypothesis and scenarios, that are not often available in the empirical observation. Furthermore to test and apply decision models, different scenario-developers are needed, to give an input for the model itself. For that, starting from the physical knowledge, the modelling discipline has been developed. When different scenarios are needed, the main aim of the modeller is to find a solution to represent the real world in a consistent way. The representation must also be a simple, non strongly data hungry and easily reproducible. In this thesis, different modelling techniques have been analyzed. In the first part a weather model has been developed and applied, involving knowledge in statistics, probability theory, hydrology, and empirical observations. This part has been developed in order to give back a weather generator to test a rainfall-runoff model. Starting with the literature analysis, a new concept of weather generator has been developed, in order to combine the Markov Chain approach with the Circulation Patterns data. The second part analyzes the development of an hydraulic flood spreading model to give back inundation scenarios for the risk analysis in one very "hydrologically" interesting area, the Thames estuary. This part analyzes the process to assess this kind of model, starting from the choice of the scenarios to be represented, the boundary conditions, the field work and the assessment of the model itself. Furthermore, and interesting historical analysis has been developed in order to reconstruct the 1953 flood scenario. In this part, the modelling technique has been explored and presented, and the main original contribution concerns the application to an interesting pilot site for the present scenario and to an historical case study. In the third part, has been analyzed a field that is not often developed from the modellers, that is, the flood evacuation modelling. The analysis of the problem from different point of views and the review of some of the existing models have been firstly developed. Furthermore, the testing and the application of a model in a deeply densely populated area has been performed. In this thesis, an important contribution in the testing and the updating of this model has been reached, while for the first time, an evacuation model has been performed in the London estuary pilot sites. In the forth part, different decision techniques has been developed and tested to analyze the results of the modelling. This techniques are developed beneath the work of this thesis, and they are also tested thanks to the availability of the evacuation model dataset. Finally, a small evacuation model has been developed and applied to the case study, in order to show an alternative way to apply the optimization techniques for the evacuation problem.