Multi-model assessment of uncertainties in continental groundwater vulnerability and pollution risk mapping in Africa

Groundwater is among the largest freshwater storages on Earth and is a vital source of water for domestic, industrial, and irrigation purposes worldwide. In Africa, domestic water supply in both urban and rural areas largely depends on groundwater, often abstracted from shallow aquifers. Although groundwater is commonly perceived as a clean and safe water source, increasing anthropogenic pressures threaten its quality, potentially leading to negative health, social, and economic outcomes. Despite its importance, groundwater quality remains poorly monitored across much of the continent. Consequently, groundwater vulnerability and pollution risk assessments frequently rely on index-based approaches such as DRASTIC, which integrates hydrogeological factors including depth to the water table, net recharge, aquifer media, soil, topography, vadose zone, and hydraulic conductivity. At continental scales, these assessments depend heavily on global datasets and large-scale model outputs, introducing substantial uncertainty that is rarely quantified.

Here, we present the first pan-African multi-model intercomparison of groundwater vulnerability and pollution risk based on the DRASTIC framework. We analyze an ensemble of 12 groundwater vulnerability maps, generated by combining three depth-to-water-table datasets and four groundwater recharge models, and 48 groundwater pollution risk maps that additionally incorporate four gridded population datasets. Model disagreement is systematically quantified using Fleiss’ extent of agreement, enabling the identification of dominant sources and spatial patterns of uncertainty across Africa.

Our results reveal widespread disagreement among groundwater pollution risk maps across the continent, highlighting the strong sensitivity of continental-scale assessments to key hydrogeological and anthropogenic inputs. Uncertainty in population datasets drives major disagreement hotspots in the Sahel and parts of Central and East Africa, whereas differences among depth-to-water-table datasets dominate uncertainty across arid regions such as the Sahara and Kalahari deserts. Uncertainty in groundwater recharge estimates further contributes to model divergence in several humid and semi-arid regions across the continent. Using the ensemble, we explore compromise mapping approaches that synthesize model outputs to produce more informative and robust groundwater vulnerability and pollution risk maps.

Overall, our findings demonstrate that large-scale groundwater vulnerability and risk maps should be interpreted as uncertainty-informed products rather than deterministic outputs. Explicitly quantifying and communicating uncertainty is essential for improving confidence, transparency, and the responsible use of groundwater vulnerability assessments in data-scarce regions of Africa.