Propagation of Hydraulic Modelling Uncertainty on Damage Estimates (PAP014433)

Código

PAP014433

Autores

Julian Cardoso Eleutério, Anne Rozan, Robert Mosé

Tema

Flood forecasting and early warning systems

Resumo

This work focuses on the impact of strategies used to model and map flood hazard indamage estimations. We consider different strategies to model the flood phenomenon, as a function of:(1) the type of hydraulic model used ? 1D, hybrid 1D/2D or 2D software, i.e. model selection; and (2) thesimplifications made by the modeller when representing the topography and bathymetry of the riversystem, i.e. parametric choices. With this purpose, different hazard models and damage estimations wererealised in the town of Fislis, in eastern France. Hec-RAS, Mike 21 and Mike Flood hydraulic models wereused in order to measure the selection of model effects. We considered different scale of analysis (levelof details) when constructing the different modelling scenarios ? density of cross-sections and hydraulicstructures (1D models), and the digital elevation model cell resolution (2D models). Thirty-two modelswere used to simulate floods with different return-periods. The 2D models tended to overestimate theflood surfaces and the 1D models tended to overestimate floodwater depths. The results of uncertaintypropagation tests on damage estimations revealed that the choice of the scale of analysis was the mainlyuncertainty influencing aspect of the evaluation. These parametric choices were responsible for 75% ofglobal uncertainty, against 18% for model selection. Furthermore, we notice that the increase of theprecision of hazard modelling has different impacts on flood maps and damage estimations, according tothe type of model. For 1D models, the more detailed the models are (higher density of cross-sections),the higher damage estimation results are. For 2D models, the more the models are precise (small gridcells)the lower damage estimates are. The results of damage estimations are strongly influenced byhydraulic modelling choices, therefore the production of flood maps for this purpose should be deeplyanalysed.