Improving water and carbon flux simulations through multi-source earth observation data assimilation

Accurate representation of terrestrial carbon–water dynamics remains a challenge in Earth system modelling, particularly under extreme hydro-climatic conditions such as droughts and heatwaves. While data assimilation (DA) of satellite-based brightness temperature (BT) and soil moisture (SM) retrievals improves near-surface moisture estimates, its impact on evapotranspiration (ET) and carbon fluxes is limited. Recent studies demonstrate that the joint assimilation of multiple Earth observation streams, such as vegetation indices, can improve estimates of both hydrological and biogeochemical state variables.

In this study, we developed a DA framework coupled to the Encore Community Land Model (eCLM), a fork of the Community Land Model version 5.0, with some extensions. The framework is applied over the EURO-CORDEX domain at 0.11-degree resolution. Assimilation is performed using the Ensemble Smoother with Multiple Data Assimilation (ES-MDA) with 64 ensemble members, allowing for the joint updating of key land surface parameters governing both soil hydrology and vegetation physiology. Assimilation observations include satellite-derived SM retrievals from the Soil Moisture Active Passive (SMAP) mission, ET from the Integrated Carbon Observation System (ICOS) eddy covariance (ET) flux towers, and leaf area index (LAI) from the Moderate Resolution Imaging Spectroradiometer (MODIS). Experiments are performed for the period 2018-2020, which include several recent European hydro-climatic extremes. Model performance is evaluated against in situ eddy covariance (EC) flux tower measurements of SM, ET, and net ecosystem exchange (NEE) across multiple European sites. We demonstrate that joint assimilation enhances the model’s ability to reproduce observed water–carbon fluxes and improves representation of land surface responses under recent extreme drought conditions.