Modelling of the monthly and daily behaviour of the runoff of the Xallas river using Box–Jenkins and neural networks methods

This paper presents a study of the hydrological behaviour of the Xallas river basin in the northwest of Spain, based on modelling the performance of the runoff produced by the river at different temporal scales. For monthly mean runoff as well as mean rainfall forecasting, Box–Jenkins models have been used. For short-term daily flow predictions, two statistical techniques were tested and compared: the classic statistical Box–Jenkins models and artificial neural networks (ANNs). The performance of the ANN was an improvement on the Box–Jenkins results. The neural networks capability of modelling a complex rainfall-runoff relationship has been observed. Although the neural network's performance was not satisfactory for detecting some peak flows, the results were most promising.