Hydrochemical investigation and prediction of groundwater quality in a tropical semi-arid region of southern India using machine learning

Monitoring and predicting groundwater quality is essential for managing water resources, protecting public health, and mitigating environmental impacts. This study presents a comprehensive hydrogeochemical investigation aimed at understanding the general hydrochemistry, identifying the extent of saltwater intrusion and prediction of groundwater quality in the semi-arid coastal aquifers of Tuticorin, Tamil Nadu, India. Groundwater samples were collected during both pre- and post-monsoon seasons to capture seasonal variations and groundwater quality was evaluated using the entropy weighted water quality index (EWQI) and predicted through the Random Forest (RF) machine learning technique. The findings revealed that total dissolved solids (TDS) exceeded WHO limits in 85% of samples during the pre-monsoon season and 61% during the post-monsoon season, indicating significant groundwater quality issues. Hydrogeochemical facies analysis identified Na-Cl as the dominant water type across all seasons, with a higher prevalence in coastal alluvium regions, suggesting a strong lithological influence and ongoing saline water intrusion. The EWQI coupled RF method provided high predictive accuracy, with R² values of 0.955 and 0.975 and RMSE values of 6.1 and 5.5 for the pre- and post-monsoon periods, respectively. In addition, results obtained from the RF-EWQI model indicated that ∼11.24% of the study area falls within the extremely poor water quality category. This zone is primarily associated with fluvial, fluvial-marine, and aeolian formations. In terms of spatial distribution, the RF-EWQI values for both seasons exhibit a parallel trend with the seawater mixing index (SMI), suggesting that the poor groundwater quality is primarily linked to the coastal alluvium aquifer. This underscores the significant impact of saline water intrusion on groundwater quality, particularly in the coastal alluvium aquifer. This integrated approach presented here offers valuable insights for improving groundwater quality assessment and management.