New insights in the capability of climate models to simulate the impact of LUC based on temperature decomposition of paired site observations

In this study, we present a new methodology for evaluating the biogeophysical impact of land use change (LUC) in regional climate models. For this, we use observational data from paired eddy covariance flux towers in Europe, representing a LUC from forest to open land (deforestation). Two model simulations with the regional climate model COSMO-CLM² (The Consortium for Small-Scale Modelling model in climate mode COSMO-CLM coupled to the Community Land Model CLM) are performed which differ only in prescribed land use for site pair locations. The model is evaluated by comparing the observed and simulated difference in surface temperature (T_s) between open land and forests. Next, we identify the biogeophysical mechanisms responsible for T_s differences by applying a decomposition method to both observations and model simulations. This allows us to determine which LUC-related mechanisms were well represented in COSMO-CLM², and which were not. Results from observations show that deforestation leads to a significant cooling at night, which is severely underestimated by COSMO-CLM². It appears that the model is missing one crucial impact of deforestation on the nighttime surface energy budget: a reduction in downwelling longwave radiation. Results are better for daytime, as the model is able to simulate the increase in albedo and associated surface cooling following deforestation reasonably well. Also well simulated, albeit underestimated slightly, is the decrease in sensible heat flux caused by reduced surface roughness. Overall, these results stress the importance of differentiating between daytime and nighttime climate when discussing the effect of LUC on climate. Finally, we believe that they provide new insights supporting a wider application of the methodology (to other regional climate models). Key Points A new methodology for evaluating LUC's impact in climate models is presented This methodology is applied to a state of the art regional climate model The main model biases are identified.