Quantifying the natural components of CO2 surface flux is key to accurately representing Earth’s carbon cycle. Existing methods struggle to separate the natural components because data on atmospheric CO2 concentrations are a function of the net flux. However, the advent of solar-induced fluorescence (SIF) satellite data offers a promising new mechanism for isolating natural fluxes by anchoring gross primary production (GPP), one of the principal natural components. Here, we extend the WOMBAT v2.0 (Wollongong Methodology for Bayesian Assimilation of Trace-gases, version 2.0) statistical flux-inversion framework with a model of spatio-temporal dependence between SIF and GPP. The resulting Bayesian model, WOMBAT v2.S, gives posterior estimates of natural fluxes over the globe during a recent six-year period, which we compare to those from WOMBAT v2.0 and to estimates from a non-inversion method. In a simulation experiment, the inclusion of SIF yields increased accuracy of the flux estimates and better characterisation of the uncertainties. Using data on CO2 and SIF from NASA's Orbiting Carbon Observatory-2 satellite and other sources, we observe that connecting GPP to SIF has little effect on the net flux but leads to larger positive- and negative-component flux estimates than those typically seen in models of the biosphere.
