A strategy for incorporating satellite retrievals in a regional inverse model to infer carbon flux variability over North America

Feedbacks between the climate system and the carbon cycle represent a key source of uncertainty in model projections of Earth's climate, in part due to the inability to measure large scale biosphere-atmospheric carbon fluxes. In situ measurements of CO₂mole fraction from surface flasks, towers and aircraft are used in inverse models to infer fluxes, but extensive measurement networks remain sparse, with limited or no coverage over large parts of the planet. Satellite retrievals of total column CO₂, such as those from NASA's Orbiting Carbon Observatory-2 (OCO-2), have the potential to provide unprecedented global information about CO₂spatiotemporal variability. However, these data need to be highly precise and unbiased for use in distinguishing concentration changes emanating from surface fluxes from those due to synoptic variability in the atmosphere. Systematic errors in the satellite retrievals of CO₂have been identified, and while bias correction algorithms are applied globally, inconsistencies persist at regional and smaller scales that may complicate or confound flux estimation. To evaluate the satellite retrievals and assess potential biases, we compare OCO-2 and in situ data-constrained CO₂columnsover North America. Data-constrainedCO₂columns are derived from simulated atmospheric profiles of CO_2­, estimated using surface fluxes optimized with in situ data from the NOAA's Global Greenhouse Gas Reference Network and from Environment and Climate Change Canada, using the CarbonTracker-Lagrange inverse model. Systematic errors in atmospheric transport are independently evaluated. Differences between simulated and satellite retrievals allow for identification of biases in the satellite retrievals but also in distinguishing real flux signals not detected by in situ data. North America is a useful testbed for evaluating retrieval bias-correction and inverse modeling strategies due to the relatively dense in situ measurement network. This work represents a significant step towards assimilating OCO-2 retrievals alongside in-situ data to fully leverage their complementarity and provide robust estimates of surface flux variability with rigorously quantified uncertainties.