Application of machine learning in delineating groundwater contamination at present times and in climate change scenarios

Abstract

Rapidly deteriorating groundwater quality due to geogenic and anthropogenic causes has impacted the lives of millions of groundwater-dependent population globally. The situation necessitates the requirement for monitoring of groundwater resources. This brief review summarizes the use of machine learning (ML) to predict groundwater contaminants across the world. Proper functioning of these model relies heavily on the quality of the data and the selection of features. Among the various ML models reviewed, tree based model, specifically random forest (RF), have provided more accurate predictions and are extensively used. Comprehending the factors that dictate the concentration of these contaminants in groundwater is imperative in developing robust prediction models. Furthermore it is essential to evaluate the limitations and uncertainties of these ML models in harnessing their true potential. The present study provide valuable insights that can be utilized to strategize and implement mitigation approaches to protect groundwater reserves from pollutants.