XCO2 and Surface Pressure retrievals from the OCO-2 spectra using a Neural Network approach

Abstract:

The OCO-2 instrument measures high-resolution spectra of the sun radiance reflected at the Earth surface or scattered in the atmosphere. These spectra are used to estimate the column integrated CO2 dry air mole fraction (XCO2) and the surface pressure. The official retrieval algorithm (NASA’s Atmospheric CO2 Observations from Space retrievals - ACOS) uses a full physics algorithm and has been extensively evaluated. We have proposed an alternative approach based on an artificial neural network (NN) technique. For the training and evaluation, we have used as a reference estimate the surface pressures from a numerical weather model and the XCO2 derived from an atmospheric transport simulation constrained by surface air-sample measurements of CO2. The evaluation of the surface pressure and XCO2 estimates derived from the NN approach, against both model values and TCCON estimates indicate a precision that is similar to that of the operational algorithm.

The neural network estimates show less spatial variability than those of the full-physic algorithm. We analyse the potential of the NN product to identify plumes from high-emitting sites and compare the spatial features from the two retrieval algorithm. We conclude on the suitability of the NN approach for the processing of XCO2 imagers such as the forthcoming CO2M mission.