Role of infiltration on land–atmosphere feedbacks in Central Europe: WRF-Hydro simulations evaluated with cosmic-ray neutron soil moisture

The skill of climate models partly relies on their ability to represent land–atmosphere feedbacks in a realistic manner, through the coupling with a land surface model. However, these models often suffer from insufficient or erroneous information on soil hydraulic parameters. In this study, the land–atmosphere model WRF-Hydro driven with ERA5 reanalysis is employed to reproduce the regional climate over Central Europe with a horizontal resolution of 4 km, for the period 2017-2020 during which cosmic-ray neutron sensor (CRNS) soil moisture is available at three Terrestrial Environmental Observatories. The soil hydraulic parameter datasets referred to as SoilGrids and EU-SoilHydroGrids, together with Campbell and van Genuchten–Mualem retention curve equations, are used to assess the role of infiltration on modeled land–atmosphere feedbacks. After calibration of the percolation parameter to better capture observed discharge amounts in the observatories, it is found that WRF-Hydro with Campbell and SoilGrids gives the lowest mean temperature and mean precipitation differences compared to the E-OBS product from European Climate Assessment & Dataset, by reducing soil moisture in the rootzone, increasing temperature, and decreasing precipitation through a positive soil moisture–precipitation feedback process. WRF-Hydro with van Genuchten–Mualem and EU-SoilHydroGrids best reproduces CRNS soil moisture daily variations, despite enhanced positive biases that generate a larger proportion of convective precipitation favored over wet soils and spurious discharge peaks. The question remains open whether an infiltration modeling option that better captures CRNS soil moisture dynamics can also lead to a clear improvement of the simulated climate.