Prévision de la sécheresse par réseaux de neurones artificiels et séries chronologiques d'indices de sécheresse

Abstract

Drought is considered among the natural disasters and extreme events that have affected the environment over the past decades. Drought prediction plays an important role in monitoring and identifying this phenomenon The main objective of this work is to develop a drought forecasting model in the Isser catchment. Using the time series of the standard precipitation index (SPI) for 6 and 12 months as a database to form and evaluate the multilayer artificial neural network model. The prediction results obtained by the artificial neural networks show the performance of this model and are more accurate for a series of SPIs of longer durations (SPI 12 months compared to SPI 6 months).