Spatial Decomposition of Eddy Covariance Fluxes: The FLUGS Framework

Understanding how heterogeneous landscapes contribute to ecosystem-exchange fluxes remains a significant challenge for utilizing eddy covariance (EC) measurements. This work presents FLUGS, a novel framework that infers land-cover-specific ecosystem-exchange fluxes provided the EC time series of aggregated fluxes and the land cover map of the ecosystem. Using a multitask machine learning approach based on Kernel Ridge Regression combined with high-resolution flux footprints, FLUGS learns the environmental response functions (ERFs) from EC data for each land cover class simultaneously. FLUGS integrates a state-of-the-art atmospheric transport model that provides accurate, computationally efficient footprint estimates with a statistically rigorous, convex optimization method for submeso-scale flux inversion. The approach is versatile, robust to multicollinearity and yields smooth and interpretable ERFs with a unique global optimum. We evaluate FLUGS using virtual and physical experiments derived from synthetic data and observational data from an EC site in Northeast Siberia. Across all tests, the framework accurately reconstructs land-cover-specific sensible heat and ${\mathrm{CO}}_2$ fluxes and captures systematic differences between land cover classes that reflect local hydrological and ecological conditions. FLUGS also provides statistical performance metrics and cross-tower consistency diagnostics that quantify uncertainty in heterogeneous landscapes. By offering a fast, transparent workflow for spatially decomposing ecosystem fluxes, FLUGS opens new opportunities to attribute EC fluxes to ecological processes, benchmark land-surface models and improve our understanding of land-atmosphere interaction.