A method for independent validation of surface fluxes from atmospheric inversion: Application to CO2

Validating fluxes from an atmospheric inversion is a challenging problem because there are often no direct flux measurements at comparable spatiotemporal scales whereas there are often relevant independent observables, e.g., trace gas concentrations. In this paper, we propose a method that validates posterior fluxes by projecting the errors between posterior and prior observable model states and independent data to the spatiotemporal differences between posterior and prior fluxes with an atmospheric transport adjoint model. We prove theoretically the conditions for which observed error reductions lead to error reductions in fluxes. We apply this approach to the atmospheric CO2 inversion problem using the NASA Carbon Monitoring System Flux project with an Observing System Simulation Experiment. We show that the posterior fluxes are more accurate than the prior over the region that significantly contributes to the reduction of CO2 errors, which is consistent with the theory.