ABSTRACT
In the water resources systems planning and management prediction of ground water level is an important parameter. Several techniques such as stochastic models, fuzzy models, artificial neural networks and others methods can be used for this purpose. Stochastic model is one these techniques that formed based on time series. And also an Artificial Neural Networks (ANNs) is flexible computing frameworks and universal approximates that can be applied to a wide range of forecasting problems with a high degree of accuracy. Therefore in this study we used this ANNs and stochastic models for nonlinear time-series for predicting Groundwater Level (GWL) fluctuations for Shahrood Plain of Iran. For this purpose the rain, relative humidity, evaporation, temperature and groundwater table data as monthly collected at the study area and these data were used to train and validate the ANN model. The ANN model was performed by varying the network parameters to minimize the prediction error and determine the optimum network configuration. Also in this research different stochastic models are fitted to monthly data of ground water table. After performance of necessary tests, PARMA (2, 1) model with the least Akaike Information Criterion (AIC) has been selected as suitable model. The results show that the performance of the MLP/BP neural network was good in predicting the groundwater level rather than stochastic model. Therefore it can be used for proper water management studies in that area.
