Geochemistry applied to the watershed survey: hydrograph separation, erosion and soil dynamics. A case study: the basin of the Niger River, Africa

Abstract A periodic sampling (85 samples) collected fortnightly along 3 annual hydrological cycles (1990–1993) in the Niger River basin, at the outlet of Bamako (Mali), allowed the calibration of an hydrochemical model based on the hydrograph separation. As a first step, 5 reservoirs are identified: Rr the rapid runoff, Rs the superficial runoff, Rh the hypodermic or differred runoff, Ns the superficial ground water, and Np the deep ground water, also called base flow. In each reservoir, the physico-chemical composition of water is supposed to be constant with time. Along the hydrograph, or total discharge curves, and during 3 hydrological cycles, the relative proportions of each of the reservoirs fluctuate continuously. The methodology, processing step by step, is first calibrated, by using concentrations of specific tracers: dissolved Na + , HCO 3 − , and suspended sediments (TSS). Then, the proportions of Rr, Rs, Rh, Ns, and Np, contributing to the total flow measured at each instant of sampling, are calculated. As a second step, pH and the concentrations of aqueous species (K + , Ca 2+ , Mg 2+ , Cl − , SO 4 2− , DOC, SiO 2 ), are in turn calculated by regression analysis in each reservoir. Finally a test of validity of the method is presented as a very close correlation between measured and predicted fluctuating concentrations of each of the elements. From that satisfactory stage, the geochemical budget of weathering is conducted, with particular attention to SiO 2 release and CO 2 consumption. Parent-rock mineralogical compositions contributing to dissolved species in the total discharge as well as in the individual flow components have been reconstituted. Rates of both chemical erosion (soil profile formation by weathering), and mechanical erosion (soil profile and landscape denudation) have been evaluated. The methodology, successfully applied to the Niger basin, is proposed as a strategic tool for studying the watershed dynamics for any time and space scales. The model revealed its ability to extrapolate and predict the geochemical or the environmental behaviour of such basins, naturally submitted to large secular or annual, regular or even catastrophic climatic oscillations.