Groundwater level fluctuation forecasting Using Artificial Neural Network in Arid and Semi-Arid Environment

In arid and semi-arid environments, groundwater plays a significant role in the ecosystem. In the last decades, groundwater levels have decreased due to the increasing demand for water, weak irrigation management and soil damage. For the effective management of groundwater, it is important to model and predict fluctuations in groundwater levels. In this study, groundwater table in Kashan plain aquifer forecasted using Artificial Neural Networks. MLP and RBF models were used to simulate the ground water table, but, because of the high number of wells studied, the samples were first organized into 5 clusters based on a Vard cluster analysis algorithm. The results indicated that, for all clusters, MLP showed good precision for predicting water depth in 37 months ahead. The correction coefficient within clusters 1, 2, 3, 4, and 5 were, respectively, 0.86, 0.88, 0.93, 0.55, and 0.79. The results showed that by change of data, education algorithm and transport function the model can be changed into the best. In 60, 20 and 20 percent of models, Delta-Bar-Delta, Momentum and Levenberg-Marquardt were best Education Algorithm, respectively. In 60, 20 and 20 percent of models hyperbolic tangent Axon, Sigmoid Axon and Linear hyperbolic tangent Axon were best transport function, respectively.