Assessing the impact of satellite, aircraft, and surface observations on CO2flux estimation using an ensemble-based 4-D data assimilation system

Abstract

[1] The potential impacts of various types of CO2 concentration data obtained from surface, satellite (by the GOSAT project), and aircraft (by the CONTRAIL project) measurements on the estimation of surface CO2 fluxes have been investigated using an ensemble-based data assimilation approach. A four-dimensional ensemble Kalman filter with a 3 day assimilation window was used for analyzing surface fluxes of CO2 at every model grid point (horizontal resolution of 2.8°). Observation system simulation experiments have demonstrated a way to make efficient use of various observations and have shown that conventional surface network data contribute to large flux error reductions in the continental areas of the northern extratropics, while GOSAT XCO2 and CONTRAIL profile data provide strong additional constraints. The GOSAT data show a large error reduction over North and South America, South Africa, and temperate and boreal Asia, but the correction in tropical fluxes is lower than expected because of the poor data coverage caused by cloud abstraction. The CONTRAIL data provide large error reductions over Europe and tropical and temperate Asia. The assimilation of the upper tropospheric data gathered by CONTRAIL results in distinct error reductions over Siberia. By combining the information obtained from all the data sets, the global flux estimation is significantly improved. Meanwhile, many sources of error in the observations and the transport model strongly decrease the usefulness of each observation, and this can become a limiting factor in real data assimilation; for example, realistic systematic errors in the GOSAT data can reduce their usefulness by a factor of 2.