carbon dioxide: 2. Regional results
[1] We report here the results from a coupled ocean-atmosphere inversion, in which atmospheric CO2 gradients and transport simulations are combined with observations of ocean interior carbon concentrations and ocean transport simulations to provide a jointly constrained estimate of air-sea and air-l...
Main Authors: | , , , , |
---|---|
Other Authors: | |
Format: | Text |
Language: | English |
Subjects: | |
Online Access: | http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.144.4004 http://www.gfdl.noaa.gov/reference/bibliography/2007/arj0701.pdf |
Summary: | [1] We report here the results from a coupled ocean-atmosphere inversion, in which atmospheric CO2 gradients and transport simulations are combined with observations of ocean interior carbon concentrations and ocean transport simulations to provide a jointly constrained estimate of air-sea and air-land carbon fluxes. While atmospheric data have little impact on regional air-sea flux estimates, the inclusion of ocean data drives a substantial change in terrestrial flux estimates. Our results indicate that the tropical and southern land regions together are a large source of carbon, with a 77 % probability that their aggregate source size exceeds 1 PgC yr 1. This value is of similar magnitude to estimates of fluxes in the tropics due to land-use change alone, making the existence of a large tropical CO2 fertilization sink unlikely. This terrestrial result is strongly driven by oceanic inversion results that differ from flux estimates based on DpCO2 climatologies, including a relatively small Southern Ocean sink (south of 44°S) and a relatively large sink in the southern temperate latitudes (44°S–18°S). These conclusions are based on a formal error analysis of the results, which includes uncertainties due to observational error transport and other modeling errors, and biogeochemical assumptions. A suite of sensitivity tests shows that these results are generally robust, but they remain subject to potential sources of unquantified error stemming from the use of large inversion regions and transport biases common to the suite of available transport models. |
---|