Bridging Water─Carbon Flux Uncertainties in eCLM: Constraining Ecosystem Dynamics using Ensemble Smoother

The water and carbon cycles are tightly coupled via plant processes such as photosynthesis and transpiration, which are sensitive to changing climate conditions. Land surface models (LSMs) like the European Community Land Model (eCLM), based on CLM5.0, simulate these coupled dynamics but often face challenges due to uncertain parameterizations, particularly for soil and vegetation processes. In its biogeochemical configuration, eCLM includes hundreds of parameters related to stomatal regulation, root distribution, phenology, and soil respiration, all of which introduce substantial model uncertainty. To address this challenge, we applied an Ensemble Smoother with Multiple Data Assimilation (ES-MDA) framework to incorporate in-situ soil moisture (SM), evapotranspiration (ET), and net ecosystem exchange (NEE) observations. By constraining the model’s uncertain parameters, we aim to improve predictions of coupled water and carbon fluxes across 14 European measurement sites primarily from the Integrated Carbon Observation System (ICOS) network. We explored multiple scenarios—ranging from hydrology-only assimilation to carbon-focused and joint assimilation strategies—along with varied temporal assimilation windows and parameter subsets. Results indicate improvements in near-surface SM, as well as in simulated ET and carbon fluxes such as NEE and gross primary production (GPP), suggesting that ES-MDA can enhance eCLM representation of coupled water–carbon dynamics. However, the model still struggles to represent winter soil moisture at high-latitude sites, underscoring the complexity of soil parameterizations under colder conditions. We plan to refine assimilation settings, including inflation and step-size adjustments, to avoid unrealistic parameter updates and ensure robust, physically consistent solutions. This study highlights the potential of integrated data assimilation to reduce uncertainties in large, complex LSMs and advance our understanding of terrestrial ecosystem processes.