Evaluation des modèles de calcul des flux polluants des rejets urbains par temps de pluie

The use of urban stormwater quality models necessitates the estimation of their outputs' uncertainty level. The results of the application of a Monte Carlo Markov Chain method based on the Bayesian theory for the calibration and uncertainty analysis of a storm water quality model commonly used in available software are presented in this paper. The tested model estimates the accumulation, erosion and transport of pollutants on surfaces and in sewers using a hydrologic/hydrodynamic scheme. The model was calibrated for 4 different initial conditions of in-sewer deposits. Calibration results showed a large variability in the model's outputs in function of the initial conditions and demonstrated that the tested model's predictive capacity is very low.