Application of neural networks to the problem of forecasting the flow of the River Nile

This paper applies multilayer neural networks to the problem of forecasting the flow of the River Nile in Egypt. Estimating the flow of the River Nile can have significant economic impact, since it can help in managing scarce irrigation water. The second goal of the paper is utilize the time series as a benchmark to compare between different neural network forecasting methods. We compare between four different methods for input and output preprocessing, including a novel method proposed here based on the discrete Fourier series. We also consider the problem of forecasting several steps ahead. We compare three methods for the multistep ahead prediction problem.