Drought forecasting using feed-forward recursive neural network

Drought affects natural environment of an area when it persists for a longer period. So, drought forecasting plays an important role in the planning and management of natural resources and water resource systems of a river basin. During last decade neural networks have shown great ability in modeling and forecasting nonlinear and non-stationary time series. This paper compares linear stochastic models (ARIMA/SARIMA), recursive multi-step neural network (RMSNN) and direct multi-step neural network (DMSNN) for drought forecasting. The models were applied to forecast droughts using standardized precipitation index (SPI) series as drought index in the Kansabati River Basin, which lies in the Purulia district of West Bengal, India. The results obtained from three models and their potential to forecast drought over different lead times are presented in this paper.