Coupling predicted model of arsenic in groundwater with endemic arsenism occurrence in Shanxi Province, Northern China

► Variables from topography, soil property, hydrology, gravity and remote sensing information can be used to build the model. ► Area more than 3000 km2 was found to have high risk of arsenic contamination. ► About 800,000 people may be at risk of high arsenic exposure. ► Arsenicosis occurrence rate can be partly predicted using predictive probability of As concentration above 50 μg L−1.