Author(s): I. Fraga; L. Cea; J. Puertas; M. Alvarez-Enjo; S. Salson; A. Petazzi
Keywords: No keywords
Abstract: The increasing importance of hydrological models as management and prediction tools has triggered the need of quantifying the uncertainty of their predictions. These uncertainties result from multiple factors. In this paper we present a new methodology to account for rainfall uncertainty. The methodology is based on adding an error function to the rainfall prediction in every point. This error function is determined from cross-validation analysis of the available rain gauge data. The error functions are then sampled using random fields. The proposed methodology is firstly validated using rain data from 7 rain events. Then, a fully distributed hydrological model based on the 2D shallow water equations is then used to simulate an additional rain event in a 24 km 2 catchment, taking into account the rainfall prediction uncertainties and quantifying their effect on the computed discharge at the catchment outlet.