Predicting the groundwater level of Chahardoli aquifer in Qorveh city using artificial neural network and support vector machine

Estimating water level is one of the most important and basic issues in agricultural planning, water resources management and determining the water needs of plants. In this study, the efficiency of artificial neural network models and support vector machines in estimating the groundwater level of Chahardoli aquifer in Qorveh city was investigated. To run the models, from the data of precipitation, flow and temperature and water level level in the previous month as input variables and water level in the next month as output variable on a monthly time scale during the statistical period (2006-2017) was used. The criteria of correlation coefficient, root mean square error and mean absolute error value were used to evaluate and also compare the performance of the models. For modeling with the mentioned methods, training data from 2006-2015 and model validation data from 2015-2016 were used. The results showed that both models had acceptable accuracy in estimating the water table level. So that the coefficient of determination in the calibration stage in the models of artificial neural network and support vector machine were equal to 0.74 and 0.94. Comparison of the two models showed that the support vector machine model performs better than the artificial neural network and the prediction accuracy has been decreased for one year in this model.