Modeling the Groundwater Level of Reshtkhar Plain Using Artificial Neural Network Method and Election and Imperialist Competituve Algorithms

Evaluating the groundwater level in arid and semi-arid regions of the country requires accurate prediction and efficiency of its fluctuations. The use of modern methods, including evolutionary algorithms, artificial neural networks and fuzzy methods, is very useful for prediction the groundwater level and generating artificial water surface data due to its high efficiency. In this research, by using Election and Imperialist Competitive Algorithms, artificial neural network, monthly data for 9 years as well as groundwater level of 10 wells, predicted the 7-year the groundwater level of Reshtkhar plain in khorasan-Razavi. In order to train the models, the statistic data was provided on 10 observation wells with a 9-year (2002-2014), which 70% of the data was introduced as training data to the model and 30% of the data was used as a test for calibration of the model. The results of the Election Algorithm predicted Reshtkhar groundwater level for the year 1400, between 14 to 16.5 meters in diffirent areas of the plain. Based on the calculations and the results obtained from the statistical parameters, the Election algorithm was RMSE, R2 and NSE, 0.029, 0.90 and 0.73 respectively, compared with the two methods of artificial neural network and Imperialist Competitive Algorithm has a significant ability to predicte the groundwater level.