In the African Sahel, agro-hydro-ecosystems are very sensitive to climate variability and land management. In turn, soil moisture, vegetation and surface fluxes provide substantial feedback on atmospheric convection and rainfall generation. It is very important to understand and correctly model the dynamics of the soil-plant-atmosphere continuum in this region, in response to contrasted meteorological and surface conditions. The aim of this study is to generate a local-scale process-based model of water and energy transfers at the soil-land-atmosphere interface for millet crop and fallow savannah, the two main land cover types in the cultivated Sahel. This study is based on a comprehensive 5-year dataset (2005-2009) from the Wankama catchment of the AMMA-CATCH observatory in south-west Niger. It includes atmospheric forcing, seasonal course of vegetation phenology, soil properties and model validation variables (net radiation, turbulent fluxes, soil heat/water profiles), for the two fields. The study area is typical of the central Sahel conditions, with 400-600 mm annual rainfall concentrated in the 4-5 month wet season. Soils are mainly sandy and prone to surface crusting, leading to a strong vertical contrast in hydrodynamic properties. The SiSPAT physically-based model is used for this study. It solves the mass and heat transfer system of equations in the soil with a vapor phase, coupled with a two-component (bare soil and vegetation) water and energy budget at the surface-atmosphere interface. The study investigates whether such a model can be accurately calibrated and validated for the two sites using realistic-parameter values. The model is calibrated over a 2-year period (2005-2006) and validated over the other three years for the two sites. The variations in water and energy variables over the five contrasted years and between land covers will be highlighted. Multi-year, field-based estimations of land surface water and energy budgets will be produced. The model could eventually be applied to longer periods to account for the marked variability of the Sahel climate. This modeling study is part of the ALMIP-2 (AMMA Land Model Intercomparison – Phase 2) project, for evaluation and comparison of land surface models at local and meso scales over the AMMA-CATCH north-south eco-climate gradient of West Africa.