Harmonizing Repeated Multi-Platform Optical and LiDAR Observations for Continental-Scale Terrestrial Carbon Cycle Data Assimilation.

Accurate monitoring of terrestrial carbon has attracted considerable attention due to the impacts of rapid climate change, disturbance, and land use on carbon storage and the growth of new global carbon markets and reporting requirements. While uncertainties in C estimates remain nontrivial, crucial opportunities exist to reduce uncertainty by rigorously harmonizing models and monitoring through data assimilation (DA). LiDAR has received considerable attention for its potential to inform carbon monitoring, and while there have been a number of papers on how to initialize models using LiDAR data, it has yet to be integrated into DA systems. In this study, we integrate GEDI LiDAR into an established DA system that already assimilates optical observations of aboveground biomass (AGB) and leaf area index (LAI), microwave soil moisture (SM), and ground-based soil organic carbon (SOC) to constrain C cycling processes across the entire North American continent. Notably, this is the first time both optical and LiDAR AGB products have been iteratively harmonized within a DA system. Results show that by assimilating both optical and LiDAR observations, we reduced the uncertainty in woody biomass by 63.7% relative to unconstrained model simulations and 18% relative to our previous DA system, and achieved an average RMSE reduction of 38.3 kg/m2 across five AGB products. Importantly, LiDAR reduces DA biases in high AGB regions where optical AGB estimates were exhibiting saturation. Analysis further identified the leading factors driving spatial error patterns within the DA system to be data quality, cloud coverage, and elevation. Overall, by bridging the spatial and temporal limitations of individual satellite products, this approach offers a practical, generalized workflow for high-resolution terrestrial carbon MRV in complex, dynamic ecosystems.