Application of a Self-Organizing Map for Clustering the Groundwater Quality in Kerman Province and Assessment its Suitability for Drinking and Irrigation Purposes

Evaluation of groundwater hydro chemical characteristics is necessary for planning and water resources management in terms of quality. In the present study, a self-organizing map (SOM) clustering technique was used to recognize the homogeneous clusters of hydro chemical parameters in water resources (including well, spring and qanat) of Kerman province; then, the quality classification of groundwater samples was investigated for drinking and irrigation uses by employing SOM clusters. Patterns of water quality parameters were visualized by SOM planes, and similar patterns were observed for those parameters that were correlated with each other, indicating a same source. Based on the SOM results, the 729-groundwater samples in the study area were grouped into 4 clusters, such that the clusters 1, 2, 3, and 4 contained 73%, 6.2%, 6.7%, and 14.1% of groundwater samples, respectively. The increase order of electrical conductivity parameter in the clusters was as 1, 4, 3 and 2. The results of water quality index based on the entropy weighting (EWQI) showed that all of the samples with excellent and good quality (36.3% of samples) for drinking belonged to the cluster 1. According to the Wilcox diagram, 435-groundwater samples (81.7%) in the cluster 1 had the permitted quality for irrigation activities, and the other 285-groundwater samples were placed in all four clusters, indicating the unsuitable quality for irrigation. The Piper diagram also revealed that the dominant hydro chemical faces of cluster 1 were Na-Cl, Mixed Ca-Mg-Cl and Ca-HCO3, whereas the clusters 2, 3, and 4 had the Na-Cl face. This study, therefore, shows that the SOM approach can be successfully used to classify and characterize the groundwater in terms of hydrochemistry and water quality for drinking and irrigation purposes on a provincial scale.