Evaluating groundwater level fluctuation by support vector regression and neuro-fuzzy methods: a comparative study

Quantitative groundwater modeling is essential in water resources management. In this article, the abilities of two different data-driven methods, support vector regression (SVR) and an adaptive neuro-fuzzy inference system (ANFIS), were investigated in estimating monthly groundwater level fluctuation in the Kashan plain, Isfahan province, Iran, by using the inputs of stream flow, evaporation, spring discharge, aquifer discharge and rainfall. Polynomial and radial basis function (RBF) was used as the kernel function of the SVR. Root mean squared error (RMSE) and correlation coefficient (R) statistics were used for evaluation of the applied models. The results indicated that the ANFIS model, having an RMSE of 3.6 m and R of 0.985, performed better than the optimal SVR_rbf model (RMSE = 13 m and R = 821) in the test period. Among the SVR methods, the SVR_rbf model was found to be better than the SVR_poly model.