Global nitrate concentrations and exceedance probability prediction map (GeoTIFF rasters)

#1. Forecast atlas of nitrate concentrations in global groundwater. The map was developed by applying all selected integrated machine-learning models to the 15 variables that exhibited the most statistically significant predictive power. Concentration prediction maps provide precise locations and specific concentrations of groundwater nitrates, effectively capturing high-value areas observed in field measurements. However, they do not represent the absolute true values of the aquifer onsite. Aligned with the objectives of exceedance probability map sets, the focus lies in identifying regions where elevated nitrate levels may exist, especially in heterogeneous aquifers where field investigations are challenging or impossible, thereby estimating the impact on the potentially affected population. Notably, high-value areas are widespread in African regions, despite limited modeling data, extensive research supports the distribution of elevated nitrate concentrations in Africa. However, predictions of nitrate concentrations in spatially contiguous desert zones in northern Africa were not considered. This decision was made after analyzing a combination of measured data (training data), land use type, and population distribution. #2. Probability of exceeding the WHO drinking water standard (50 mg/L) for nitrates in groundwater. The map of nitrate exceedance probabilities was created by applying the final integrated machine-learning model to 45 natural and anthropogenic-dominated predictor variables.