Estimation of the Spatial Distribution of the Groundwater Quality Using the Combined Method of Geostatistics_ Artificial Neural Networks (Case study: the Miandoab aquifer))

Drought crisis, although traditionally limited to central provinces, desert, and hot and dry places, but recent research has shown that in recent years, the plains of the Lake Urmia, such as the Miandoab plain, have been affected by drought, and has undergone a sharp decline in groundwater levels and subsequently reduced quality. n this research, the rate of variation in quality parameters of groundwater such as TH, TDS, EC, pH, SAR, which was collected by the regional water company of West Azarbaijan in the years 2002 and 2011, has been investigated. The statistical data of 2002 were mapped by statistical and Kriging method and stored in a regular grid of 31 * 26 in GIS software. This data is stored as a text file and used in the simulation of the artificial neural network. The results showed that the MLP model with M6 structure has a correlation coefficient of 0.92 and a mean square error of 0.562, and it can simulate groundwater quality in Mianodab Plain. Also, in predicting values of the absorption sodium ratio between 2003 and 2011, the correlation coefficient showed 0.69 satisfactory results. Finally, with sensitivity analysis, respectively, chlorine, acidity, and phosphate have the greatest effect on simulation and prediction of sodium adsorption rates.

Estimation of the Spatial Distribution of the Groundwater Quality Using the Combined Method of Geostatistics_ Artificial Neural Networks (Case study: the Miandoab aquifer)) Keywords:

Prediction , Geostatistics , Artificial neural network , Sodium Absorption Ratio , Miandoab