Comparison of model land skin temperature with remotely sensed estimates and assessment of surface‐atmosphere coupling

Abstract The coupling between land surface and the atmosphere is a key feature in Earth System Modeling for exploiting the predictability of slowly evolving geophysical variables (e.g., soil moisture or vegetation state), and for correctly representing rapid variations within the diurnal cycle, particularly relevant in data assimilation applications. In this study, land surface temperature (LST) estimated from Meteosat Second Generation (MSG) is used to assess the European Centre for Medium‐Range Weather Forecasts (ECMWF) skin temperature, which can be interpreted as a radiative temperature of the model surface. It is shown that the ECMWF model tends to slightly overestimate skin temperature during nighttime and underestimate daytime values. Such underestimation of daily amplitudes is particularly pronounced in (semiarid) arid regions, suggesting a misrepresentation of surface energy fluxes in those areas. The LST estimated from MSG is used to evaluate the impact of changes in some of the ECMWF model surface parameters. The introduction of more realistic model vegetation is shown to have a positive but limited impact on skin temperature: long integration leads to an equilibrium state where changes in the latent heat flux and soil moisture availability compensate each other. Revised surface roughness lengths for heat and momentum, however, lead to overall positive impact on daytime skin temperature, mostly due to a reduction of sensible heat flux. This is particularly relevant in nonvegetated areas, unaffected by model vegetation. The reduction of skin conductivity, a parameter which controls the heat transfer to ground by diffusion, is shown to further improve the model skin temperature.