Inverse modeling of carbon monoxide surface emissions using Climate Monitoring and Diagnostics Laboratory network observations



[1] A three-dimensional (3-D) inverse modeling scheme is used to constrain the direct surface emissions of carbon monoxide CO. A priori estimates of CO emissions are taken from various inventories and are included in the IMAGES model to compute the distribution of CO. The modeled CO mixing ratios are compared with observations at 39 CMDL stations, averaged over the years 1990–1996. The interannual variability of CO sources is therefore ignored. We show that the method used (time-dependent synthesis inversion) is able to adjust the surface fluxes on a monthly basis in order to improve the agreement between the observed and the modeled CO mixing ratios at the stations. The Earth surface is divided into regions. The spatial distribution of CO sources is fixed inside each of these regions. The inversion scheme optimizes the intensities of the emissions fluxes for the following processes: technological activities, forest and savanna burning, agricultural waste burning and fuelwood use, soil/vegetation emissions, and oceanic emissions. The inversion significantly reduces the uncertainties on the surface sources over Europe, North America and Asia. The most striking result is the increase (almost by a factor of 2) of CO flux from Asia in all a posteriori scenarios. The uncertainties on the Southern Hemisphere emissions remain large after the inversion, because the current observational surface network is too sparse at these latitudes. The inversion, moreover, shifts the peak in biomass burning emissions in the Southern Hemisphere by one month. This temporal shift ensures a better match of the observed and modeled CO seasonal cycle at the Ascension Island station. We also attempted to optimize the annual and global productions of CO due to methane and NMHC. With the current set of data, the scheme was not able to differentiate between these two sources, and hence only the total chemical production of CO can be optimized.