Assessment of the SiSPAT SVAT model for irrigation estimation in south-east France

In this study, we assess the interest of using a Soil – Vegetation – Atmosphere – Transfer model, the SiSPAT (Simple Soil Vegetation Atmosphere Transfer) model, which solves the surface energy balance, for the evaluation of theoretical crop water requirements in south-east France. First the relevance of the model results, when parameterized using information extracted from a soil data base and pedotransfer functions for the estimation of soil hydraulic properties, and when vegetation characteristics are prescribed using available data bases is assessed. We use long term time series of soil water content profiles for this purpose. The results show that evapotranspiration, as simulated by SiSPAT is sensitive to the soil parameter specification leading to large uncertainties in the model results. Then, we present two methods implemented in SiSPAT to compute irrigation requirements. The first option mimics the soil water balance model principles by estimating the irrigation from the available soil water capacity filling. The second option relies on the model physics and estimates the difference between actual transpiration and the value corresponding to a minimal stomatal resistance, i.e. without water stress. Aspersion and drip irrigation can be simulated. Nine crop are chosen for the model evaluation. A comparison with two other water balance models is performed. The three models are consistent with determination coefficient between the simulated annual irrigation generally larger than 0.4. However, differences of the interannual irrigation needs, larger than several 100 mm, are sometimes found, especially for drip irrigation. This work provides a quantification of expected uncertainties when using water balance models or physically-based models for irrigation needs estimation.