Evaluating CLM5-FATES performance with different complexities of vegetation dynamics across European forested sites

Land surface models (LSMs), when coupled with vegetation dynamic models, serve as useful tools to understand how ecosystem changes impact carbon and water cycling and interact with the climate. However, simplified vegetation parameterization schemes within LSMs makes it challenging to capture the full dynamics of vegetation processes, leading to significant uncertainties in the simulated ecosystem variables. To assess these uncertainties stemming from different model complexities in vegetation representations, we conducted model simulations employing different vegetation parameterization schemes: (1) using static vegetation distribution and prescribed leaf area index (LAI), (2) simulating full carbon cycle with static vegetation, and (3) explicitly simulating fully dynamic carbon and vegetation distribution.

Here, we use a vegetation demographic model, the Functionally Assembled Terrestrial Simulator (FATES) coupled with the Community Land Model (CLM5) to evaluate the representation of vegetation dynamics and related surface fluxes across multiple forested sites in Europe selected from the Integrated Carbon Observation System (ICOS) station network. Comparison with observations showed that the CLM5-FATES model, with the full vegetation dynamics implementation, exhibited better model performance in simulating gross primary production (GPP) than the runs with prescribed leaf-area climatology. However, the model showed an underestimation of LAI with low interannual variations compared to satellite-based MODIS data, particularly for sites with evergreen forests. Additionally, the model's performance in simulating hydrological fluxes (such as soil moisture (SM) and evapotranspiration (ET)) remained consistent across all sites, irrespective of model complexity. Future work will explore uncertainties in simulated vegetation structure and distributions and parameter optimization to improve model performance in simulating forest growth and composition.