An evapotranspiration model driven by remote sensing data for assessing groundwater resource in karst watershed

Highlights: • SimpKcET (Simple Crop coefficient for Evapotranspiration) reproduces the dynamics of ET of a wide range of vegetation cover • A better estimation of ET with SimpKcET of a large watershed improves simulations of its outflow discharges. • The shading of the soil by the presence of pieces of rock on its surface limits the evaporation, this process plays a major role in the estimation of ET. • Despite the simplicity of the assumptions, the estimates are below 0.5 mm.d⁻¹. Aquifer recharge may depend mainly on the difference between precipitation and evapotranspiration. Hydrological models used to estimate groundwater reserves use evapotranspiration models that are mainly determined by climate demand. In particular, mechanisms of plant transpiration are neglected, although transpiration constitutes 70% of evapotranspiration. This is problematic when considering karst watershed, which are poorly documented at the interface between soil and atmosphere where vegetation and soil properties control water flows. To fill this gap, we propose an evapotranspiration model that integrates the processes of plant transpiration and soil evaporation. The dynamics of vegetation is evaluated using the Enhanced Vegetation Indexes from the Terra and Aqua Moderate Resolution Imaging Spectroradiometers. The soil evaporation calculation account for the impact of coarse elements at soil surface. The "Simple Crop coefficient for Evapotranspiration" (SimpKcET) model is tested at flux tower sites over forest of Font-Blanche, Puechabon and the agricultural area of Avignon. The simulated daily evapotranspirations are very close to the observations (RMSE ~0.5 mm.d^-1), while the model is simple compared to other models proposed in the literature. The SimpKcET is implemented in a karst hydrological model to evaluate the impact of evapotranspiration estimation on the aquifer flow rate simulation. This approach is applied to the vast watershed of Fontaine de Vaucluse. In comparison to the water bucket model that is frequently used in karst models, SimpKcET provide ET simulations that are more in line with ET processes. A cross wavelet analysis highlighted the improvement of the simulated recharge and observed flow rate relationship brought by the consideration of evaporation and transpiration processes. The use of remote sensing data related to plant activity makes it possible to propose a parsimonious model that can be applied to all types of vegetation (agricultural, natural, mixed forest) and that can be transferred to other karst models.