Evaluation of Gene Expression Model in Spatial Prediction of Groundwater Salinity and Its Comparison with Geostatistical Models (Case study: Mashhad plain)

Groundwater is an important source of exploitation in arid and semi-arid regions. For this reason, in order to maintain groundwater quality and its optimal management, it is important to know their spatial and temporal distribution and their monitoring and zoning should be considered as an important principle in the country's water resources planning. The aim of this study was to zoning the electrical conductivity of groundwater in the Mashhad plain aquifer using 5 methods of distance inverse geostatistics (IDW), local estimator (GPI), general estimator (LPI), kriging and cokriging and also evaluating the gene expression programming model in predicting this The parameter is using spatial data. For the present study, data from 122 observation wells in the aquifer area of Mashhad plain were used. To compare the methods used, two squares evaluation criteria were the mean squared error (RMSE) and the mean absolute error (MAE). The semi-variable plot in GS + showed that the electrical conductivity data fit best in the spherical model. The results of the present study showed that among the mentioned methods, the inverse distance method (IDW) with error rate RMSE=0.3 mos/cm and MAE=0.16 mos/cm and then the model of gene expression programming with RMSE=275.54 mos/cm MAE=223.15 mos/cm with the highest accuracy and local estimator method (GPI) with RMSE=996.11 mos/cm and MAE=755.56 mos/cm, had the least accuracy in this field.