Application du réseau des neurones artificiels à la prévision des débits horaires: Cas du bassin versant de l’Eure, France

Abstract

Rainfall–runoff models are frequently used in flood-risk management and flood forecasting. In this study, we present a rainfall–runoff model for the prediction of hourly rates based on the technique of artificial neural networks (ANN). This model was developed and applied on the Eure basin in northwest France with the aim of overcoming the problems due to the nonlinearity of the rainfall–runoff relationship and the inaccuracy of the data collected. The development of this model required several steps in which we were able to determine the model parameters required for understanding the level of hydrological complexity and the production of information required for forecasting. The process resulted in an ANN model that could produce effective flood forecasts, in several seconds, to a forecast horizon of 48 h. These results confirm that ANN models can play an important role in forecasting, being able to model the nonlinearity of the rainfall–runoff relationship that is encountered in certain basins