Estimation of Quantity and Quality Parameters of Groundwater Using Numerical Models (Case Study: Mighan Desert Basin, Arak)

In this paper, salinity, total dissolved solids (TDS), groundwater level (GWL) and electrical conductivity (EC) of the Arak Plain, located in Markazi Province, Iran, were simulated using four novel artificial intelligence models including extreme learning machine (ELM), wavelet extreme learning machine (WELM), online sequential extreme learning machine (OSELM) and wavelet online sequential extreme learning machine (OSELM) as well as the MODFLOW software (MT3D model). In order to develop the hybrid artificial intelligence models, the wavelet transform was employed. First, the effective lags in estimating the quality and quantity parameters of the groundwater were identified using the autocorrelation function (ACF) and the partial autocorrelation function (PACF) analysis. After that, four different models were developed using the effective lags for each of the artificial intelligence methods. Then, the superior models in simulating the groundwater quality and quantity parameters were detected by conducting a sensitivity analysis. Subsequently, the most effective lags in estimating these parameters were introduced. In addition, the results of The MODFLOW model were compared with the artificial intelligence models, and it was concluded that the latter were more accurate. For instance, the scatter index and Nash-Sutcliffe efficiency coefficient values for TDS simulation by the superior model were 5.34E-03 and 0.991, respectively. Additionally, RMSE and MAE for estimating groundwater level using the superior model were obtained 0.078 and 0.061, respectively. Finally, uncertainty analysis for the superior models was carried out.