Modeling the Groundwater System Response to Varaiations of the Consumption and Surface Discharge

Investigating groundwater level variations is very important for sustainable management and planning of water resources. In the present study, groundwater system response to the consumptions and surface water discharge were utilized as input variations of nonlinear regression and gene expression programming models to estimate groundwater table. The used data consisted of monthly consumption amounts, surface discharge rates and groundwater levels of Bam Normashir plain (Kerman province) during the period of 2002 to 2011. The cross-correlation analysis indicated that 12 lagged monthly surface discharge as well as the current monthly consumption values had the highest impact on groundwater level fluctuations. The global relationship between these three variables was obtained using the two equations produced by ordinary nonlinear regression and gene expression programming (GEP) model. For estimating the groundwater level fluctuations, the consumption magnitudes and surface discharge values were predicted using artificial neural network (ANN) and Thomas-Firing methods, respectively. Then, the values were putted in the regression-based equations to predict the groundwater level. The obtained result revealed that in the case of using observational data, the performance of GEP was slightly better than regression models (RMSE and MAE values were 0.793, 0.636 meter respectively), while by use of the data produced by employing Thomas-Firing and ANN, the regression models gave a promising result with RMSE and MAE values of 1.437 and 1.118 meter, respectively.