Groundwater is an important resource of water in all parts of the world, especially in arid and semiarid regions. In recent years, over extraction of groundwater resources has caused instability for groundwater resources that finally caused more drawdown in groundwater level in many aquifers of the country. However, due to shortage of water resources, efficient use and management of water resources is particularly important task in Iran. In order to effective management of water resources, we need to predict the groundwater level fluctuations exactly. In this study, the Artificial Neural Networks (A) approach is applied for predicting mean groundwater level in Najafabad Plain in central Iran. Najafabad Plain is one of the most important aquifer in Zayandeh-Rud Basin, which has faced with negative hydrologic balance and hence the quality of groundwater affected by excessive use of groundwater resources for agricultural uses. Artificial Neural Networks were used for simulation of this hydro system improved by Particle Swarm Algorithm for training of the network. For the neural network training, the weights and biases of the network are the decision variables, and the network error is the objective function that must be minimized. Comparison of the predicted results of ANN-PSO and measured mean groundwater level using 58 piezometers with monthly time steps for 20 years showed that the designed neural network is capable of predicting groundwater level fluctuations of the aquifer in different situations and can be used as a reliable tool for evaluating the water resources management scenarios in this aquifer. Three management scenarios under two climate change scenarios A2 and B1 were defined and its impacts on the plain were investigated. The first scenario aims to continue the current trend of extraction of groundwater as a base line scenario over the next 5 years, the second scenario aims to decrease pumping from the wells, and the third scenario aims to increase the surface water to irrigation channels in order to preserve the agriculture lands. For each scenario the volumes of groundwater extractions and surface water diversion were determined. Mean groundwater level changes under two climate change scenarios A2 and B1 and three management scenarios showed that for reducing drawdown of groundwater levels over the next five years, we need to reduce pumping from the Najafabad aquifer besides increasing the surface water diversion to the irrigation lands. Keywords : Water Resources Management, Simulation, Artificial Neural Network, Optimization, Particle Swarm Optimization, Management Scenarios, Najafabad Plain.
