Une méthode prédictive non biaisée et géoréférencée d'estimation des dommages résidentiels d'inondation

Flood risk management for residences requires an economical analysis involving the mean annual damage by floods, taking into account the whole range of probability of floods and the cost of projected fluvial enhancements and measures, taking also into account the residual level of risk. Efficient methods are therefore necessary to estimate these risk values. The proposed approach is of a "distributed" type; it involves numerical modeling for estimating "residential submersion depth", a variable, which explains most of the direct damages to residences. The method relies on an individualized georeferenced definition of each residence. Measured submersion data and the compensations obtained from the huge Saguenay flood in 1996 (Ville de Laterrière) were used to build empirical laws based on submersion. Four distinct relationships were established: residences with or without a basement and valued below or above $50,000 each were assigned a specific relationship. With these relationships, direct residential damages in Laterrière were assessed by using only simulation results at the georeferenced position of sector residences as input. It was then possible to evaluate the total amount of direct home damages in Laterrière.Key words: flood, risk, damage rating curves, Saguenay floods, numerical modeling, predictive model, georeference, geographic information system.