Predicting Groundwater Chlorine Concentration in Dezful Aquifer Using the Panel Data Model

Groundwater resources are of great importance in arid and semi-arid regions due to their ease of access and low extraction costs. Compared to studies conducted on the quantity of groundwater resources, less research has been devoted to groundwater qulity. The present study was thus designed and implemented to forecast groundwater chlorine variations in Dazful Plain in Khuzistan Province, Iran. " Panel data" is a regression model that considers variables of different units over time. In this study, it was exploitedfor the simultaneous prediction of groundwater quality in different wells. For this purpose, meteorological parameters such as rain and ET0 as well as the quality parameters including EC, sodium, calcium, and magnesium were collected in ten wells in the study area on a seasonal basis over a period of 8 years. In the next step, the data thus collected were subjected to different "panel data" regression models including Common Effects, Fixed Effects, and Random Effects. The results showed that the Random Effects Regression Model was best suited for predicting groundwater quality. Moreover, performance indicators (R2= 0.96, RMSE= 2.445) revealed the effectiveness of this method.