Regional biospheric carbon fluxes as inferred from atmospheric CO2 measurements

Land ecosystems are currently absorbing similar to 30% of fossil CO2 emissi ons. However, the role of land ecosystems as sources or sinks of carbon in response to human perturbation is not well understood. One key issue is to better diagnose the flux of carbon exchanged between continents and the atm osphere at the regional level. We have analyzed the constraints that today' s unevenly distributed global network of atmospheric CO2 measurements put o n one's ability to understand and pinpoint biospheric CO2 fluxes. We constr ucted an inverse model using atmospheric CO2 observations and atmospheric t ransport to infer the mean spatial distribution of CO2 terrestrial fluxes. CO2 fluxes can be inferred over large regions of the globe, such as contine nts or large ocean gyres. The target period for the inversion procedure is 1985-1995. The inversion produces a global ocean uptake of 1.5 +/- 0.5 Pg C /yr and a global land sink of 1.3 +/- 1.5 Pg C/yr (1 Pg = 10(15) g = 10(9) metric tons = 1 Gt). There is a net terrestrial carbon uptake at northern m id-latitudes (2.1 +/- 1.3 Pg C/yr) and a net release in the tropics (1.1 +/ - 1 Pg C/yr). In the Southern Hemisphere, at least over South America and A frica, our results indicate that tropical deforestation either has been ove restimated or is currently offset by other sinks. However, the sparse spati al coverage of atmospheric observations around the Equator does not allow u s to partition the inferred fluxes between South America and Africa separat ely. In the Northern Hemisphere, where more stations are available, we obta in an uptake of 0.5 +/- 0.6 Pg C/yr over North America, 0.3 +/- 0.8 Pg C/yr over Europe, and 1.3 +/- 0.8 Pg C/yr over Siberia. We analyze uncertaintie s in these estimates in the light of the atmospheric measurements and the t ransport model that we used.