Contrasting perspectives on soil evaporation in soil science and land surface modelling

Soil evaporation is an essential component of the hydrological cycle. Within soil science, the fundamental mechanisms involved in soil evaporation are well-documented. However, within the realm of land surface modelling, the coarse spatial and temporal scale, as well as the computational limitations result in a simplified representation of this highly non-linear process.
Here, we evaluated the current representation of soil evaporation within the RMI evapotranspiration (ET) and surface turbulent fluxes (STF) model applied in the frame of  the EUMETSAT Satellite Applications Facility  (LSA)  on support to Land Surface Analysis (SAF) (http://lsa-saf.eumetsat.int/). This model is used to produce remote-sensing based estimates of the fluxes, using Meteosat Second Generation (MSG) observations. With 30 minutes interval, estimates of these fluxes are provided in near real time, resulting in a data record that spans over 20 years.
We highlighted the discrepancies between the simplified representation of soil evaporation and the soil physical solution. To achieve this, synthetic experiments were performed using Hydrus as a reference for comparison with the LSA SAF ET-STF model. Additionally, a comparison was made with formulations in other land surface models (Surfex, ECLand & GLEAM), the resulting texture-dependent bias was demonstrated and impact of sub-grid heterogeneity was shown. Finally, an updated formulation was tested in large-scale ET simulations and evaluated using in situ observations.
Though widely recognised as one of the fundamental processes in the hydrological cycle, the perspective on soil evaporation is very different in soil physics compared to land surface modelling. Here, we attempted to harmonize both approaches in a pragmatic manner.