Prediction models for groundwater quality parameters using a multiple linear regression (MLR): a case study of Kermanshah, Iran

Groundwater is one of the major sources of exploitation in arid and semiarid regions. Spatial and temporal quality distribution is an important factor in groundwater management. Thus for protecting groundwater quality, data production on spatial and temporal distribution is essential. The present study has applied multiple linear regression (MLR) techniques to predict the fitness of groundwater quality in Kermanshah province, west of Iran. The parameters examined were Total dissolved solids (TDS), Total hardness (TH), Sodium adsorption ratio (SAR). the quality variables were modelled by MLR. Finally, the performance of the models was assessed using the coefficient of determination (R-2). The relationship between parameters by MLR showed that TDS and water quality parameters in semi-deep wells and aquifers had a strong positive correlation (r = 0.94, r = 0.98) and there was a strong positive significant correlation between SAR and water quality parameters in deep wells and aquifers (r = 0.98, r = 0.99). Also, TH and water quality parameters in all water sources had a strong positive cor-relation (r = 1). The MLR model could serve as an alternative and cost-effective tool for groundwater quality prediction where there is limitation in laboratory facilities, trained expertise or time. Consequently, the usefulness of these linear regression equations in predicting the groundwater quality is an approach, which can be applied in any other locations.

Automated summary

This study applied multiple linear regression (MLR) techniques to predict groundwater quality in Kermanshah province, Iran. It found strong positive correlations between total dissolved solids (TDS), total hardness (TH), sodium adsorption ratio (SAR) and groundwater quality. The MLR model can serve as a cost-effective tool for predicting groundwater quality.