Unsupervised learning techniques in groundwater quality assessment of Mewat region, Haryana, India

Abstract

Unsupervised learning techniques such as principal component analysis (PCA), cluster analysis (CA) were applied to the groundwater data of Mewat region (Haryana, India) collected in the pre monsoon season to identify the geochemical processes controlling groundwater chemistry. Thirteen physicochemical parameters were analyzed and were found to be above the permissible limits. The order of cation and anion concentration were found to be Na+ > Mg2+ > Ca2+ > K+ and Cl− > $${\text{NO}}_{3}^{ - }$$ > $${\text{SO}}_{4}^{2 - }$$ > $${\text{HCO}}_{3}^{ - }$$ > $${\text{CO}}_{3}^{2 - }$$ . The dominance of Na+ and Cl− in groundwater chemistry showed the salinity factor in the groundwater. PCA applied to the data set reduced the dimensionality to four significant factors accounting 76.66% of the total variance in the data set. The first factor can be assigned to alkalinity which originates due to the dissolution of geological minerals into the groundwater, second factor is assigned to salinity (due to salt water intrusion) and hardness which is caused by weathering of sedimentary rocks and calcium bearing minerals and other factors originate as a result of industrial wastes, domestic wastes and wastes from agricultural activities. CA classified 30 sampling sites into three clusters with relatively low salinity region, high salinity and very high salinity regions based on similar water quality characteristics.