Artificial Neural Networks for Improved Predictions in Flow Eestimation

Abstract

In order to assure the smooth function of a water reservoir, we have to estimate streams flow. Flow estimation is usually recognized as a proper tool for regional climatic condition description in respect to soil erosion by water. It is also a basic input to simple and widespread soil erosion prediction models. However its calculation on the base of original precipitation records is a very laborious operation and is completely impossible for many locations without a precise precipitation data. The aim of the research was to develop a simple method of flow values estimation on the base of general precipitation data. We examined the possibility of implementing artificial neural networks for flow values estimation on the base of minimum-maximum temperatures and daily precipitations. The research was conducted with the use of a database containing calculated precipitation and flow values from 3 meteorological stations in Cyprus. As a result of the study 3 radial basis function networks (RBF) of two to five hidden layer neurons and 2 multilayer perceptrons networks (MLP) with one and two hidden layers were developed. The study results suggested the possibility of neural networks technology introduction for flow values estimation on the base of daily precipitations, with a preference for MLP networks, instead of simple statistical regionalrelationships.