Forecasting Hydrologic Time Series Using Artificial Neural Networks

Forecasting a hydrologic time series has been one of the most complicated tasks owing to the wide range of data, the uncertainties in the parameters influencing the time series and also due to the non availability of adequate data. Recently Artificial Neural Networks (ANN) have become quite popular in time series forecasting in various fields. This paper demonstrates the use of ANN to forecast the monthly river flows. Two widely used networks namely the feed forward network and the recurrent neural network are chosen. The feed forward network is trained using the conventional back-propagation algorithm and the recurrent neural network is trained using the method of ordered partial derivatives. The selection of architecture and the training procedure for both the networks is presented. The selected ANN models were used to train and forecast the monthly flows of a river with a catchment area of 5,189 sq. km. upto the gauging site. The trained networks are used both for single step ahead and multiple step ahead forecasting. A comparative study for both the networks indicates that the recurrent neural network performed better than the feed forward network. In addition, the size of the architecture and the training time required were less for the recurrent neural network. The recurrent neural network gave better results for both single step ahead and multiple step ahead forecasting. Hence Recurrent neural networks are recommended as a tool for river flow forecasting.