MACHINE LEARNING-BASED DEPTH-CLASSIFIED MODELING OF GROUNDWATER ARSENIC IN THE GANGES DELTA
In this study, we employ a random forest machine-learning algorithm to model the depth-wise distribution of arsenic contamination within the delta aquifers. By probabilistically mapping the occurrence of arsenic concentrations exceeding the WHO guideline of 10 μg/L, we delineate arsenic hazards across three depth zones: shallow (<70 meters below MSL), intermediate (70–150 meters below MSL), and deep (>150 meters below MSL).
Our findings show that while arsenic concentrations generally decrease with depth, elevated levels persist in the northern and western margins of the delta across all three zones. In contrast, southern regions, particularly those near the active delta mouth, exhibit significantly lower contamination risks.
These results underscore the urgent need for targeted investigations to determine the spatial extent of arsenic contamination in deep aquifers before prioritizing costly deep-well installations. To support effective resource allocation and mitigation strategies, we advocate for local-scale arsenic risk modeling integrated with assessments of vulnerable populations to assist policymakers and stakeholders in addressing this ongoing public health crisis.