Spatial prediction of geogenic manganese in Southeast Asian groundwater

Manganese (Mn) is a highly abundant element in the Earth’s crust that can become enriched in groundwater through reductive dissolution of Mn-containing minerals. Although an essential trace element for humans, manganese poses a health threat mainly in the form of neurotoxicity when consumed in high quantities. Mn has thus far received little attention, particularly in comparison to the other geogenic contaminants of arsenic and fluoride. The World Health Organisation (WHO) recently lowered the guideline from 400 µg/l to 80 µg/l, but it is not known in how many regions of the world the new level is exceeded, let alone how many people are exposed.

We have therefore collected thousands of manganese measurements from Southeast Asia and used machine learning (ML) modelling to investigate the factors related to enrichment in groundwater. By using spatially continuous predictors, we are able to apply the ML model to produce prediction maps of manganese in groundwater. This upscaling of existing manganese measurements provides insights into managanese is nearby areas based on comparable values of the associated predictor variables. The maps are then compared with existing ones of arsenic and iron, which likewise get released in groundwater along the same sequence of biologically mediated redox reactions. By creating prediction maps of both the old and new WHO drinking water guidelines of 400 µg/l and 80 µg/l, respectively, the extent of the increase in areas and populations potentially exposed to hazardous Mn concentrations can be visualized and better appreciated. This serves to raise awareness of the impact that the reduction of the guideline value may pose.