Prediction of the Dead Sea water level using neural networks

Abstract

The Dead Sea (DS) basin plays a major role for regional economic development (industry, tourism and agriculture) in Jordan. Different studies stated that the water level of the DS is dropping an average of 3 feet per year. Accordingly there is a need to provide accurate and reliable estimates for the water level to help the researchers and geologists of the DS to make different kind of studies giving results. This achieved by a applying three Artificial Neural Networks (ANN) algorithms for the meteorological data recorded from different stations and resources inside and outside of Jordan, The models are trained and tested by BackPropagation (BP), Levenberg-Marquardt (L-M), and Generalized Regression Neural Networks (GRNN) and the results of models are verified with untrained data. The results from the different algorithms are compared with each other. The criteria of performance evaluation are calculated in order to evaluate and compare the performances of models. Finally, we can say that the proposed GRNN model provides best significant performance results comparing with other NN models using Mean Square Error (MSE).