A Monte Carlo Investigation of the Impact of Aerosol and Albedo Uncertainty on XCO2 Retrieval Errors

Abstract:

The Orbiting Carbon Observatory-2 (OCO-2) uses an optimal estimation (OE) retrieval to provide estimates and associated posterior variances for atmospheric CO₂ concentrations from satellite radiances. The actual error distribution for the retrieved quantity of interest, X_CO2, results from the inherent variability of the atmospheric state, measurement variability, and the choices made in implementing the retrieval algorithm. In this Bayesian framework, the OCO-2 retrieval error distribution is particularly sensitive to the specification of surface albedo and aerosols, including the prior distribution for aerosol optical depth. The impact of the aerosol and albedo prior specification is studied through Monte Carlo experiments with a physically-based surrogate model. Retrieval bias and covariance depend on the underlying true distribution of the atmospheric state as well as the aerosol parameterization implemented in the retrieval.