Résumé
Gravity data with a few microgal precision were proved to be able to constrain the specific yield of various kinds of aquifer in West Africa from annual fluctuations of both the gravimetric and piezometric signals (Pfeffer et al., Geophys. J. Int., 2011; Hector et al., Geophys. J. Int., 2013). However some recent papers reported a disappointing potential of gravity measurements during a pumping experiment in a sandy aquifer (Blainey et al., WRR, 2007; Herckenrath et al., WRR, 2012) and their poor ability in constraining the transmissity and specific yield of the aquifer, which are the parameters to which pumping tests give access. Fresh basement rocks present generally a null porosity and the structure of basement aquifers is given by the weathering profile. In tropical climate, this profile consists of a few tens meter thick saprolite layer, with noticeable porosity but low permeability overlying the weathering front. This weathering front includes in many instances a fractured medium and presents a high permeability with variable porosity. It is hardly sampled in coring experiments. We present some numerical simulation results on the ability of gravity to constrain the transmissivity of this medium. Due to poroelasticity of clay minerals in the saprolite, soil subsidence is expected to occur during pumping with a significant gravity effect. Gravity measurements have therefore to be completed with leveling data at a millimetric precision. We present first the results of numerical modeling of the gravity and subsidence for a theoretical horizontally stratified basement aquifer, and show that gravity and leveling are able to provide independently the poroelasticity coefficient and a single transmissivity coefficient for the bottom of the aquifer, if the properties of the upper saprolites are known. We will discuss then the general case, where the aquifer presents a vertical fracture where the weathering profile thickens.