GLOBAL PREDICTION MODEL OF GEOGENIC FLUORIDE IN GROUNDWATER

Millions of people worldwide are negatively affected by the chronic consumption of elevated concentrations of fluoride in groundwater. Due to health effects including dental mottling and skeletal fluorosis, the World Health Organization maintains a maximum guideline of 1.5 mg/L in drinking water. However, groundwater quality is not regularly tested in many areas, and it is therefore often unknown if the water in a given well or spring contains hazardous concentrations of fluoride. To help gauge the scope of the problem and identify priority areas for groundwater quality testing, we have used machine learning to create a global fluoride hazard map. Over 400,000 data points of fluoride in groundwater (10% greater than1.5 mg/L) from 77 countries were used along with 12 predictor variables out of an initial set of 62 variables relating to geology, soil, climate and topography. The model performs very well, as evidenced by an average AUC of 0.90. An additional model was created that also incorporates physicochemical parameters measured in situ. Although this model performed even better (e.g. AUC of 0.95), it could not be used to create a map due to the utilization of non-spatially continuous variables. Both the spatially continuous and in-situ predictor variables confirm globally that arid conditions promote the dissolution of fluoride in groundwater. Hotspots identified by the groundwater fluoride hazard map include parts of central Australia, western North America, eastern Brazil and many areas of Africa and Asia. This fluoride hazard model was used to estimate the global at-risk human population at about 180 million people, most of whom live in Asia and Africa.