A parametric sensitivity analysis for prioritizing regolith knowledge needs for modeling water transfers in the West African critical zone

Abstract

Abstract Hard rock aquifers (HRAs) in West Africa (WA) are located within a thick regolith layer. The representation of thick tropical regolith in integrated hydrological models lacks consensus on aquifer geometries and parameter ranges. Our main objective was to determine the knowledge requirements on saturated hydraulic conductivity (Ks) to model the critical zone (CZ) of HRAs in WA. A parametric sensitivity analysis with a focus on the representation of the Ks heterogeneity of the regolith was conducted with a critical zone model (Parflow‐CLM [Community Land Model]) of the Upper Ouémé catchment in Benin (14,000 km2) at a 1‐ × 1‐km2 resolution. The impact of parameter changes in the near subsurface (0.3‐to‐5‐m depth) and in the deeper regolith aquifer (24‐ and 48‐m maximum depth) was assessed in five modeling experiments. Streamflow was largely dependent on Ks and on clay distribution in the near subsurface and less on the properties of the deeper subsurface. Groundwater table depths and amplitudes were controlled by vegetation and topography as observed on instrumented hillslopes and for Ks within the literature range. Experiments with higher Ks suggested a Ks threshold where dynamics become less determined by one‐dimensional vertical and more determined by lateral processes. Such heterogeneity impacts from smaller scales need to be accounted for when hydrological models are upscaled to larger domains (1‐ × 1‐km2 resolution or coarser). Our findings highlight the need for a new conceptual approach to represent clay distribution in order to develop catchment‐scale CZ models of HRAs in WA that capture the observed processes.