Predictive Modeling of Nitrate (NO_3^-) Contamination Risk in Shallow Groundwater Utilizing Population Density and Village Sanitation Statistics (PODES)

Abstract

Shallow groundwater in Indonesia's suburban areas faces nitrate (NO₃⁻) contamination risks from domestic wastewater due to high population density and inadequate sanitation, posing health threats like methemoglobinemia, yet traditional assessments are costly. This study develops a Geographically Weighted Regression (GWR) model using BPS secondary data Population Density and PODES-derived Sanitation Risk Index (SRI) to predict NO₃⁻ risk across Java Island suburbs, spatially joining variables with administrative boundaries, normalizing data, and validating via R², RMSE, and Moran's I. Results show the GWR model outperforms OLS (adjusted R² = 0.785 vs. 0.612; RMSE = 0.075), with PD coefficients (0.00001–0.00005) dominant in northern urbanizing zones and SRI coefficients (0.650–0.980) stronger in southern established areas; high-risk hotspots (18.5% of 215 units) cluster where PD > 4000 people/km² and SRI > 0.60. The model suggests that BPS data can serve as robust predictors for spatially variable contamination risk, enabling hotspot maps for targeted PDAM monitoring and sanitation upgrades in Indonesia.