Inverse modeling of the global CO cycle 1. Inversion of CO mixing ratios

A three-dimensional modeling study on atmospheric carbon monoxide is presen ted, based on the TM2 model. A Bayesian inverse technique is applied to opt imize the agreement between model and observational data, including a prior i source information as regularization term. Using the National Oceanic and Atmospheric Administration Climate Monitoring and Diagnostics Laboratory d ata set for CO mixing ratios at 31 globally distributed sites, a posteriori CO budgets can be derived, which allow the model to reproduce the observat ions at most sites within two standard deviations of monthly mean values. U se of different spatiotemporal emission distributions for terpenes (Global Emissions Inventory Activity, similar to 80% of emissions in the tropics; H ough [1991], similar to 70% of emissions in the extratropical Northern Hemi sphere) showed a large impact on calculated a posteriori source strengths a nd on the modeled partitioning among individual CO sources. In order to rep roduce the interhemispheric gradient of observed CO mixing ratios, a ratio between total sources in the Northern Hemisphere and those in the Southern Hemisphere of similar to 1.8 is required. While it is obvious that this asy mmetry is mainly due to CO emissions from technological sources, the invers ion results suggest that either(1) the global technological CO source stren gth is higher (similar to 800 Tg CO/yr) than present inventory based estima tes or (2) CO from terpenes or vegetation (or additional sources with domin ant emissions in the Northern Hemisphere) have a significant impact on the northern hemispheric mixing ratios. Further sensitivity studies showed that a posteriori results slightly depend on biomass burning seasonality (shift ed by 1 month), but they are virtually identical for the two different OH f ields (CH4-nonmethanehydrocarbons chemistry vs. CH4-only chemistry). Invers ion results, however, were sensitive to model wind fields used (based on me teorological observations of 1987 and 1986, respectively), mainly due to st ations near source regions. Use of a reduced set of stations resulted in vi rtually identical a posteriori source strengths for both model wind fields. The analysis is extended in the companion paper which considers the additi onal information on the CO budget provided by measurements of the stable is otope ratios (C-13/C-12, O-18/O-16).