Characterizing Recharge in Southern Mali Using a Combination of Modeling and Stable Isotopes

Abstract

Groundwater recharge in southern Mali is investigated using interpretive recharge models and stable isotopes to identify the dominant recharge mechanism and explore how local variations in geological materials influence the recharge characteristics. At a regional scale, the groundwater level hydrographs from across southern Mali (1998–2002) are relatively consistent, showing seasonal variations, suggesting diffuse recharge is the dominant mechanism. Groundwater samples plot within the range of the weighted mean monthly δ18O and δ2H concentrations for July-August-September rainfall, and below the weighted mean annual δ18O and δ2H concentrations for rainfall, suggesting a dominantly rainy season source of recharge. Recharge is simulated for four representative unsaturated zone environments, each with varying soil, laterite and sedimentary bedrock layers, and three ranges of water table depths, for a total of 12 combinations. The simulated recharge response starts in July, 1 month after the arrival of the rainy season, and recharge is greatly accelerated through August to its peak in September. On an annual basis, ~72% of annual rainfall occurs between July and September, and nearly 60% of simulated recharge occurs between August and October. The simulated regional average annual recharge is 519 mm/year (479–560 mm/year range among models). By comparison, recharge estimated from the observed storage anomaly hydrographs using the water table fluctuation method is 384 mm/year (189–619 mm/year) using a specific yield of 0.05, although the range could be as high as 83–772 mm/year given the uncertainty in specific yield values (0.02–0.07). The simulated recharge also agrees with the timing of regional observed storage anomalies for all observation wells, but somewhat less so for the regional GRACE storage anomaly (2002–2008), which has a slower rate of rise in storage and a faster rate of recession compared to the observed storage anomalies and simulated recharge response.