This dissertation presents the results of a detailed physical and hydrogeophysical study of two soil-filled sinkholes mantled by ancient New River fluvial terrace deposits. Research was performed at the Virginia Tech Kentland Experimental Farms in Whitethorne, Virginia, USA between fall 2003 and spring 2007, and focused on characterizing infiltration, deep drainage, and recharge through soil-filled sinkholes. Using hydrogeophysical methods, the spatial and temporal distribution of soil moisture was modeled and potential recharge was quantified in two soil-filled sinkholes. Access-tube time domain reflectometry (TDR) was used to derive one-dimensional (1-D) soil moisture profiles. During access-tube installation, 470 soil samples were obtained from depths between 0.3 and to 9.0 m and characterized both physically and chemically. Using these data, a TDR calibration method was developed. Physio-chemical, TDR moisture, and 1-D electrical resistivity tomography (ERT) data were used to derive a numerically optimized form of Archie's Law which was used to convert ERT measurements into volumetric soil moisture. These results led to development of 2-D ERT-derived distributions of soil moisture in three transects across the two sinkholes in two terraces. Potential recharge was quantified using time-series ERT data with comparison to modeled cumulative potential evapotranspiration (PET) and cumulative precipitation between May 17 and October 9, 2006. The patterns of ERT-derived potential recharge values compared well with those expected from PET and precipitation data. Over the monitoring period from late spring to early fall during this study, results showed that a period of intense rain followed by a 31-day period of consistent rain, in which the rate of precipitation was equal to or exceeded PET, were the only periods in which significant amounts of potential recharge occurred (from 19 to 31% of cumulative precipitation during the study). Spatial distributions of ERT-derived moisture clearly revealed that significant amounts of infiltration occurred on sinkhole flanks and bottoms. Runoff during periods of intense rain flowed to the topographically lowest point in the sinkholes where it infiltrated and resulted in localized zones of enhanced infiltration and potential recharge to the water table. / Ph. D.

La forte croissance démographique (~3%.an-1) observée au cours des dernières décennies en Afrique subsaharienne a engendré la conversion de grandes étendues de savane arborée en cultures pluviales ou irriguées et la diminution de la durée des jachères. Afin de déterminer l'impact de tels changements d'usage du sol sur la recharge des nappes phréatiques, la zone non saturée a été investiguée pendant trois ans (2009-2011) pour deux régions du Niger (sud-ouest et sud-est) aux conditions climatiques et géomorphologiques différentes.Une approche expérimentale in-situ a tout d'abord permis de caractériser qualitativement les propriétés de la zone non saturée et les flux hydriques associés (profondeur 0-10 m) pour les principaux usages du sol : savane naturelle ; jachère ; culture pluviale de mil ; et culture irriguée de poivron. Des profils granulométriques et de résistivité électrique apparente ont été réalisés et des suivis temporels de la teneur en eau et du potentiel matriciel du sol ont été mis en place. A partir des résultats et de la synthèse des données ainsi recueillies, deux analyses détaillées ont été proposées.Afin de quantifier le différentiel de recharge diffuse entre une jachère à Guiera senegalensis et une culture pluviale de mil (Pennisetum sp.) au sud-ouest Niger, une modélisation à base physique via le code Hydrus-1D a été appliquée. Dans un premier temps, une inversion basée sur la méthode GLUE a permis d'établir les densités de probabilité pour les paramètres hydrodynamiques de la zone non saturée ; dans un second temps, des simulations pluriannuelles (2 × 100 ans) du drainage profond (0-10 m) lors d'une transition jachère-mil ont été réalisées. Il a ainsi été montré que l'expansion des surfaces cultivées en mil pourrait se traduire par une augmentation du drainage profond, de 20 à 25 mm.an-1 après un délai de 35 à 60 ans.Afin d'étudier les conséquences des mises en culture pluviales et irriguées sur le potentiel de salinisation des sols et des eaux souterraines au sud-est Niger, les concentrations en ions majeurs dissous dans l'eau des pores de la zone non saturée ont été mesurées, puis comparées à la composition géochimique des apports d'eau associés (pluie, eau d'irrigation). Il a été ainsi mis en évidence que la mise en culture pluviale a un effet négligeable sur la qualité des eaux interstitielles et souterraines tandis que l'irrigation est associée à un enrichissement de l'eau de la zone non saturée en solutés qui pourrait induire, à terme, une salinisation de la nappe par lessivage des sols devenus salins/sodiques. / In semiarid sub-Saharan Africa, the rapid population growth (~3%.yr-1) during the past few decades has resulted in land clearing and large-scale conversion from savannah and fallow to rainfed or irrigation crop fields. The traditional duration of fallow has been also shortened. In order to estimate the impact of these land use changes on groundwater recharge, the vadose zone was investigated during three years (2009-2011) for two regions located in Niger (south-west and south-east).A qualitative analysis was first carried out for identifying vadose zone properties and for characterizing the corresponding water fluxes (0-10 m depth) for the main land use types: natural savannah; fallow; rainfed millet crop; and irrigated sweet pepper crop. Grain size and electrical resistivity profiles were established and soil water content and matric potential were monitored. Based on the results of this analysis, two more detailed investigations were performed.In southwestern Niger, in order to estimate changes in diffuse recharge from a fallow with Guiera senegalensis to a rainfed millet crop (Pennisetum sp.), a physically-based modeling with Hydrus-1D code was completed. Probability density functions were first built for the soil hydraulic parameters based on the GLUE approach; then, deep drainage (0-10 m depth) was simulated for a 2 × 100 year time-period including a fallow-millet conversion. It was shown that the increase in millet crop areas could result in an increase in deep drainage from 20 to 25 mm.yr-1 after a delay of 35 to 60 years.In southeastern Niger, in order to assess the impact of rainfed and irrigated cropping development on soil and groundwater salinization, major ion concentrations in pore water of the vadose zone were measured and compared with the geochemical composition of water inputs (rainfall, irrigation). It was shown that rainfed cropping does not affect soil water and groundwater quality whereas irrigation results in an increase of solutes concentrations in soil pore water, which could lead to a groundwater salinization at mid-term through soil leaching.