Chapter 3 - A data-driven approach for estimating contaminants in natural water

Abstract

Studies across the world indicate that high chloride concentration in groundwater is a major issue that adversely affects the available freshwater resources. The measurement of chloride in a laboratory is both time-consuming and chemical-intensive. Few studies have reported the estimation of chloride concentration using correlation expression from a proxy, electrical conductivity, which is routinely determined in the field. However, neither independent validation nor errors associated with the estimates have been reported. We present a data-driven approach using a decision tree that provides acceptable estimates of chloride concentrations using a different proxy, the total dissolved solids. The estimated chloride concentration in a geographical area facilitates the distinction between different water sources based on their distinct signatures and understanding the underlying geochemical processes that influence the overall quality of groundwater. The influence of season (pre-or postmonsoon) as well as the degree of saturation of soil (waterlogged or water depleted), when considered in model development, yielded better accuracy in estimations.