Data: 17/09/2014 à 19/09/2014
Local: São Paulo - Brazil
Cascade of Uncertainties in Flood Damage Estimations (PAP014432)
Julian Cardoso Eleutério, Sylvain Payraudeau, Robert Mosé, Anne Rozan
Flood forecasting and early warning systems
The quantification of flood risk involves several modelling steps each of which comprisesuncertainties. This work compares the impact of different sources of epistemic uncertainty in potentialflood damage estimates. We distinguish the uncertainties linked to models, methods and data, i.e. modeluncertainties; and the uncertainties correlated with the hypotheses and choices to be introduced in themodels, i.e. parametric uncertainties. In order to measure the global uncertainty of damage estimations,different data acquisition and modelling strategies were proposed for the four fundamental modules of theassessment: (1) hydrological analyses and considerations for determining discharges for different eventprobabilities; (2) the types of hydraulic model built and considerations when integrating topographical andbathymetric data; (3) the datasets and methods used to characterise the vulnerability of buildings tofloods; and (4) the damage functions used and the errors related to characterising the value of the stakes.We propagate uncertainties linked to different strategies in the assessment results (sensitivity testsrelated to each assessment module) and we measured the results variability generated. The method wasapplied to two case studies in the French part of the Rhine River basin. The results of this analysisshowed that the uncertainty of each module of the assessment depends on several factors that are highlydependent on the characteristics of the sites studied. However, the role played by flood hazard modellingwas preponderant in assessing flood risk to buildings, especially for the most frequent floods. Thisshowed that great attention must be given when modelling frequent floods for damage assessmentpurposes. The results of this study highlighted that the uncertainty linked to protection structures (dikesand dams) is a significant source of uncertainty in the damage assessment process.