Atmospheric transport inverse modeling using CO2 observations has been a valuable tool for estimating regional carbon budgets, but CO2 only responds to the small residual between photosynthesis and respiration. Net fluxes are notoriously noisy in regional ecosystems, but gross fluxes are much more coherent in time. Using observations to correct biases in model estimates of gross fluxes provides a stronger constraint because the longer decorrelation time scales increase the ratio of observations to unknowns.
We have tested an optimization framework for correcting model biases in gross fluxes at a number of flux tower sites and found that the new method is much more robust to uncertainty in the observations than a control in which net fluxes were estimated. Unsurprisingly, the new method requires an additional constraint to distinguish between errors in photosynthesis and respiration. We test the new method using solar induced fluorescence (SIF), and find that it is able to accurately estimate gross fluxes at the sites.
We outline a global implementation of the new algorithm using tracer transport and global observations of both CO2 and SIF from the Orbiting Carbon Observatory (OCO-2). The global method can also be extended to use observations of carbonyl sulfide (COS). Resulting global maps of imbalances in respiration and photosynthesis are interpreted in terms of model errors related to sink mechanisms such as CO2 fertilization, regrowth from past disturbance, changes in nutrient cycling, and responses to climate change.