HR: 14:25h
AN: A13G-03    [Abstracts]
TI: Fine Spatial Resolution Global CO2 Flux Estimates for 1997 to 2001 Obtained Using Remote-Sensing Derived Environmental Data Within a Geostatistical Inverse Model
AU: * Gourdji, S M
EM: sgourdji@umich.edu
AF: Department of Civil and Environmental Engineering, 2340 G.G. Brown Building, University of Michigan, Ann Arbor, MI 48109-2125 United States
AU: Mueller, K
EM: kimlm@umich.edu
AF: Department of Civil and Environmental Engineering, 2340 G.G. Brown Building, University of Michigan, Ann Arbor, MI 48109-2125 United States
AU: Humphriss, C
EM: chumphri@umich.edu
AF: Department of Civil and Environmental Engineering, 2340 G.G. Brown Building, University of Michigan, Ann Arbor, MI 48109-2125 United States
AU: Michalak, A M
EM: amichala@umich.edu
AF: Department of Civil and Environmental Engineering, 2340 G.G. Brown Building, University of Michigan, Ann Arbor, MI 48109-2125 United States
AU: Michalak, A M
EM: amichala@umich.edu
AF: Department of Atmospheric, Oceanic and Space Sciences, 2455 Hayward Street, University of Michigan, Ann Arbor, MI 48109-2143 United States
AB: This study applied the recently developed geostatistical approach to atmospheric inverse modeling (Michalak et al, 2004) to estimate carbon fluxes on a monthly gridscale from 1997 to 2001, focusing on the effect of using remote-sensing derived auxiliary environmental data to help constrain flux estimates. The results of this study represent the first application of the geostatistical approach to inverse modeling using atmospheric data from the NOAA cooperative air sampling network, and the first implementation of the approach with auxiliary environmental data. The geostatistical approach is related to the Bayesian approach in that it uses observational CO₂ concentration data and an atmospheric transport model to update prior information about carbon fluxes. In the Bayesian approach, prior fluxes are estimated from process-based models over land and ocean and fossil fuel inventories for anthropogenic emissions. In the geostatistical approach, prior information is represented in the form of a spatial and/or temporal covariance in the deviations of fluxes from mean behavior. Both the parameters of the covariance model and of the mean behavior are inferred from the observational data. The model of the mean can be improved by the addition of deterministic drift parameters known to affect carbon flux. This setup offers the opportunity to directly incorporate environmental data derived from remote sensing instruments, without assuming a priori the magnitude or statistical significance of the correlation of these data with the inferred carbon fluxes. In this study, global estimates of carbon fluxes were obtained at a 3.75° x 5° resolution, along with their estimated uncertainties. The inversion was run with both a constant model of the mean (one constant over land and another over ocean), and a variable model of the mean incorporating auxiliary environmental data. Among the variables shown to be statistically significant were several remote-sensing datasets, including Leaf Area Index, Fraction of Photosynthetically Available Radiation (fPAR), land cover and Sea Surface Temperature. The results of the inversions were compared to results from previous carbon dioxide inversion studies and to flux distributions obtained from biospheric models. Preliminary results show greater fine-scale variability for the variable model of the mean compared to the constant model. In addition, both geostatistical inversions show less seasonal extremes than the output from biospheric models. The method developed for incorporating auxiliary environmental data within a geostatistical inverse modeling framework is discussed in an accompanying presentation,``Using remote sensing data to help constrain fluxes of CO₂ in a geostatistical inverse modeling framework'' by K. Mueller et al.
DE: 0312 Air/sea constituent fluxes (3339, 4504)
DE: 0315 Biosphere/atmosphere interactions (0426, 1610)
DE: 0428 Carbon cycling (4806)
DE: 0480 Remote sensing
DE: 1615 Biogeochemical cycles, processes, and modeling (0412, 0414, 0793, 4805, 4912)
SC: Atmospheric Sciences [A]
MN: Fall Meeting 2005