Simulations of the Impact of OCO-2 Retrieval Bias on the Determination of the Inter-hemispheric Gradient of XCO2

The Orbiting Carbon Observatory (OCO-2) was designed to make measurements of the dry air column mole fraction of atmospheric carbon dioxide (XCO2) with the precision and accuracy to infer regional scales sources and sinks of CO2. The OCO-2 retrieval algorithm to obtain XCO2 from the three near-infrared bands observed in reflected sunlight is dependent on accounting accurately for the effects of aerosols in the radiative transfer calculations. Input to the forward model are aerosol type, mean height, width, and depth (as Gaussian vertical profile) each with a prior mean and covariance as part of the optimal estimation-based retrievals. We investigated the impact of choices for aerosol type and prior, including the omission of aerosols from the retrieval, using a simulation system to control the specification and calculation of the radiances. We test the significance of the bias in retrieved XCO2 from these simulations over a representative annual cycle of 2015 and 2016 using a simple 2 box model to confirm the error in net hemispheric surface flux.