Assessing the performance of pan-European CLM5 simulations in capturing long-term multivariate trends in land surface varaibles.

Understanding long-term trends in Essential Climate Variables (ECVs) is important for predicting future climate impacts. This study investigates long-term trends in multiple land and climate variables, including evapotranspiration (ET), surface soil moisture (SM), snow cover (SC), snow water equivalent (SWE), total water storage (TWS) and streamflow along with variables influencing vegetation productivity, such as vapor pressure deficit (VPD), gross primary production (GPP) and plant available water, to provide a comprehensive assessment of their changes and interactions. Using a combination of observational datasets, remote sensing products, and reanalysis data, we evaluate the performance of Community Land Model, version 5.0 (CLM5) in capturing these trends over the European continent during the past 33 years (1990 - 2022). Additionally, we present a multi-model ensemble of CLM5 simulations with different configurations (Prescribed vs. prognostic vegetation) and different model resolution (0.0275^o vs. 0.11^o) to assess uncertainties in capturing trends arising from varying model complexities. All model configurations are driven by the ERA5 reanalysis dataset and share consistent datasets for the static input datasets such as topography, land cover and soil properties. 

Our preliminary analysis shows that the CLM5 model captures the interannual variability in the hydrologic states and fluxes reasonably well for ET, SWE, SC and TWS, but overestimates surface SM to satellite-derived datasets. Model performance in capturing trends varies across variables: while decreasing trend direction in snowpack variables (SC and SWE) and TWS align with remote sensing observations, surface SM trends show opposite directions. We further explore whether these discrepancies arise from trends in climatic drivers (e.g., temperature and precipitation) or differences in model configurations.This study highlights the importance of a multivariate approach to trend analysis in improving our understanding of the recent states and changes in land surface variables.