Geochemical and isotopic studies of the Douda-Damerjogue aquifer (Republic of Djibouti): Origin of high nitrate and fluoride, spatial distribution, associated health risk assessment and prediction of water quality using machine learning

: Groundwater from the East African Rift System (EARS), for which there is limited data available, is often characterized by high levels of dissolved fluoride and nitrate, which pose inherent health risks. Within the EARS, the Douda-Damerjogue aquifer system was overexploited and subjected to anthropogenic and/or geogenic pollution with high NO3-concentrations (up to 375.4 mg L-1) and F-(up to 4.5 mg L-1). This study is the first to examine the origin and cumulative health risk assessment of groundwater with high F- and NO3- contents in rifting zones, as well as the spatial patterns and the water quality forecasting. This study use a combination of geochemical and thermodynamic tools, geospatial analysis, MixSIAR model, Machine Learning (ML) model, as well as stable isotope ratios, including δ18O(H2O), δ2H(H2O), δ15N(NO3-), and δ18O(NO3-). A ML framework was developed to forecast NO3-, Electrical Conductivity (EC), and Irrigation Water Quality Index (IWQI) in such data-scarce environments. The key geochemical processes controlling the groundwater composition in the study area were: (i) basalt weathering; (ii) ion exchange; (iii) mixing with fossil groundwater; and (v) seawater intrusion. Fluoride enrichment (> 1.5 mg L-1) in the groundwater was likely driven by the dissolution of fluoride-bearing minerals and desorption from sorbent surfaces. The combined application of the MixSIAR model, stable nitrate isotopes, and the NO3/Cl vs Cl diagram identified soil organic nitrogen, NH4-fertilizers, sewage and manure as the primary anthropogenic sources of NO3- in the groundwater. Groundwater chemistry showed that 76 % of samples exceeded the permissible limits for fluoride and nitrate, posing potential health risks, especially for teenagers and infants. The proposed ML-based framework provides a robust, scalable solution for forecasting water quality in Djibouti and other regions facing similar challenges.