Simulation of groundwater quality is important due to the water shortages in arid and semiarid areas and for managing water resources. Geostatistical models have developed zoning techniques that are used to spatial prediction and interpolation of groundwater parameters. Recently, the ability of hybrid intelligent models has been proven in the simulation of dynamic systems. In this research, hybrid intelligent models based on a neuro-fuzzy system integrated with fuzzy c-means data clustering (FCM) and grid partition (GP) models as well as artificial neural networks integrated with Particle Swarm Optimization algorithm were used to predict spatial distribution of chlorine (Cl), electrical conductivity (EC), and sodium absorption ratio (SAR) parameters of groundwater. The results of the hybrid models were compared with geostatistical methods including Kriging, RBF, and IDW. The latitude and longitude values of observation wells and qualitative parameters in the three states of maximum, average, and minimum were introduced as inputs and outputs to the models, respectively. In order to evaluate the models, RMSE, CC, and MAE statistical criteria were used. The results showed that in the hybrid models, NF-GP with the lowest RMSE and MAE and highest CC is the most suitable model for prediction of parameters. The RMSE, MAE, and CC values in the average state for Cl were 107.175 (mg/L), 79.804 (mg/L), and 0.924, for electrical conductivity were 518.544 (µmho/cm), 444.152 (µmho/cm), and 0.882, for sodium absorption ratio were 1.596, 1.350, and 0.582, respectively. Among the geostatistical models, Kriging was more accurate. Using the coordinates of the wells will eventually allow the NF-GP to be used for more sampling and replace the visual techniques that require time, cost, and facilities.
