Assessing groundwater vulnerability to salinity using modified drastic model and machine learning

Abstract

Groundwater vulnerability to salinity assessment serves as a measure of potential groundwater salinity pollution in a target area. The study focused on modified DRASTIC model to assess groundwater vulnerability to salinity, with machine learning (random forest) to successfully predict groundwater saline sensitive zones. However, research to predict groundwater salinity risk using machine learning was limited. The aims of the study, to assess inherent sensitivity of groundwater to pollution, determine spatial variations in vulnerability due to additional factor of land use, to develop ground water salinity map based on saline concentration and to predict potential ground water saline zones using machine learning. The methods employed on the objectives, were the DRASTIC, DRASTIC-LU incorporating the impact of anthropogenic influences, validation of the DRASTIC-LU with groundwater quality data (EC as an indicator of salinity levels) and implementation in machine learning model with the threshold value of EC as the predictor class value. Groundwater salinity was increasingly recognized as a significant environmental hazard, particularly in areas highly influenced by the impact of anthropogenic activities such as mining, agricultural practices, and different land use changes. Random forest classifier models were developed, using depth to water (D), recharge (R), aquifer media (A), soil media (S), topography (T), vadose zone (I) and hydraulic conductivity (C), on the basis of EC class. To assess the performance of groundwater salinity risk model, training and validation datasets were evaluated using the ROC curve and gained accuracy of 94%. The results revealed that approximately 20% of the study area fell within the high salinity potential, and around 50% exhibited moderate salinity risk while 30% very low to salinity risk concentration. Random forest algorithm could serve as efficient technique for evaluating and managing groundwater resources. The findings underscored relatively poor groundwater quality in the Busire, with excessive aquifer exploitation by the agricultural sector and infiltration of urban sewage and industrial waste identified as the primary causes of groundwater salinity. The implications of these findings were crucial for devising strategies and implementing preventive measures to mitigate water resource vulnerability and associated health risks in Busitema Subcounty.