HR: 15:05h
AN: B42B-06    [PDF]
TI: Inferring Terrestrial CO$_{2}$ Fluxes From A Global-Scale Carbon Cycle Data Assimilation System (CCDAS)
AU: * Rayner, P J
EM: peter.rayner@csiro.au
AF: CSIRO Atmospheric Research, 07-121 Station Street, Aspendale, Vic 3195 Australia
AU: Scholze, M
EM: scholze@dkrz.de
AF: Max-Planck-Institute for Meteorology, Bundesstr. 55, Hamburg, 20146 Germany
AU: Knorr, W
EM: wknorr@bgc-jena.mpg.de
AF: Max-Planck-Institute for Biogeochemistry, Hans-Knoell-Str 10, Jena, 07745 Germany
AU: Kaminski, T
EM: thomas.kaminski@fastopt.de
AF: Fastopt, Schanzenstr. 36, Hamburg, 20357 Germany
AU: Giering, R
EM: ralf.giering@fastopt.de
AF: Fastopt, Schanzenstr. 36, Hamburg, 20357 Germany
AU: Widmann, H
EM: heiner.widmann@dkrz.de
AF: Max-Planck-Institute for Biogeochemistry, Hans-Knoell-Str 10, Jena, 07745 Germany
AB: Various indirect methods suggest that much of the year-year variability in the growth-rate of atmospheric CO$_{2}$ over the last two decades is attributable to variability in the net terrestrial flux. It is much less clear which processes are responsible for this terrestrial variability. In this talk we present results from a carbon cycle data assimilation system (CCDAS) in which the controlling parameters in a terrestrial carbon cycle model are inferred by nonlinear optimization based on the model's adjoint. Uncertainties in the parameters are inferred from observational and model uncertainties via the model's Hessian and then mapped forward on predicted quantities such as net CO$_{2}$ fluxes to the atmosphere via the model's Jacobian. The adjoint, Hessian, and Jacobian are generated by automatic differentiation of the model's source code. The dataset is the set of extended CO$_{2}$ concentration records from 41 observing sites. The model is able to fit the observations moderately well although it slightly overpredicts the long-term growth rate. This occurs despite the increase in terrestrial uptake through the 1990s over the 1980s. The increase, in turn, occurs despite a reduction in net primary productivity and is hence caused by a larger decrease in soil respiration. It appears that the requirement to fit both the seasonal cycle and interannual dynamics in the CO$_{2}$ record is a strong constraint on model formulation. We will demonstrate this by describing some of the problems with the description of respiration which are highlighted by this approach.
DE: 0315 Biosphere/atmosphere interactions
DE: 0322 Constituent sources and sinks
DE: 0330 Geochemical cycles
SC: Biogeosciences [B]
MN: 2003 Fall Meeting