Kriging-interpolated mapping and predictive modeling of groundwater F- and NO3- contamination with chemometric and health risk assessments in Ghana's Birimian Province.

Abstract

Increasing global reports of fluoride (F^-) and nitrate (NO₃^-) contamination in groundwater highlight the urgency of identifying pollution hotspots to safeguard public health. This study investigates groundwater quality in two agricultural regions of Ghana's Birimian province, filling a vital research gap. This study utilized a diverse set of tools, including physicochemical analyses, violin plot visualizations, the Pollution Index of Groundwater (PIG), the Water Pollution Index (WPI), health risk assessments, Pearson's correlation analysis, and artificial neural network modeling. These approaches evaluated the key factors affecting groundwater quality, identified contamination sources and hotspots, and assessed associated human health risks. Results revealed predominantly alkaline groundwater (pH 7-9), with F^- ranging from 0.0 to 1.5 mg/L and NO₃^- exceeding 500 mg/L in some areas. The PIG and WPI rated 81.94-94.44% of samples suitable for consumption, with mean scores of 0.54 and 0.51, respectively, highlighting NO₃^-, pH, and K⁺ as primary quality influencers. Violin plots showed multimodal distributions in TDS, NO₃^-, Ca²⁺, and Mg²⁺, suggesting complex hydrogeochemical dynamics. Health risk assessments indicated oral exposure risks ranging from low to very high, with NO₃^- posing a sixfold greater threat than F^-. Spatial analysis tied F^- contamination in central and southern areas to geological formations, while higher NO₃^- in the northern part aligned with agricultural activities. Correlation analysis and neural network modeling confirmed the geogenic origin of F^- whereas the mixed sources of NO₃^- strongly tied to anthropogenic inputs. These insights urge targeted remediation and offer a scalable framework for global groundwater challenges.