Auxiliary Material Submission for Paper 2006GB002751 Inverse Estimates of the Oceanic Sources and Sinks of Natural CO2 and the Implied Oceanic Carbon Transport S. E. Mikaloff Fletcher Department of Atmospheric and Oceanic Sciences, University of California, Los Angeles, California, USA. Now at Atmospheric and Oceanic Sciences, Princeton University, Princeton, New Jersey, USA. N. Gruber Department of Atmospheric and Oceanic Sciences, University of California, Los Angeles, California, USA. Now at Environmental Physics, ETH Zurich, Zurich, Switzerland. A. R. Jacobson Atmospheric and Oceanic Sciences, Princeton University, Princeton, New Jersey, USA. Now at Global Monitoring Division, NOAA, Boulder, Colorado, USA. M. Gloor Institutes for Earth and Biosphere, Energy, and Environment School of Geography, University of Leeds, Leeds, UK. S. C. Doney Marine Chemistry and Geochemistry, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts, USA. S. Dutkiewicz and M. Follows Department of Earth, Atmosphere, and Planetary Sciences, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA. M. Gerber and F. Joos Climate and Environmental Physics, Physics Institute, University of Bern, Bern, Switzerland. K. Lindsay Climate and Global Dynamics Division, National Center for Atmospheric Research, Boulder, Colorado, USA. D. Menemenlis Estimating the Circulation and Climate of the Ocean, Jet Propulsion Laboratory, Pasadena, California, USA. A. Mouchet Astrophysics and Geophysics Institute, University of Liege, Liege, Belgium. S. A. Mueller Climate and Environmental Physics, Physics Institute, University of Bern, Bern, Switzerland. J. L. Sarmiento Program in Atmospheric and Oceanic Sciences, Princeton University, Princeton, New Jersey, USA. Introduction This folder contains the following materials: _____________________________________________ ---------------------------- Figures ---------------------------- 2006gb002751_fs01.eps: The locations of the observations used in this study. Each point corresponds to a station, which usually has 20 to 30 observations throughout the water column. The colors indicate the different time periods of the observations: prior to 1985 (purple), 1985-1990 (blue), 1990-1995 (green), as 1995 (red). 2006gb002751-fs02.eps: Comparison between the inverse estimates of the air-sea flux of natural CO2 from Gloor et al. [2003] and this study (PgC/yr). The inverse estimates for this study have been interpolated to the regions of Gloor et al. [2003]. and follow the naming conventions of that study. The values and error bars for this study are the weighted means and weighted standard deviations among the 10 OGCMs. Positive (negative) values indicate outgassing (uptake). 2006gb002751fs03.eps: Covariance between regional flux estimates for the 23 region aggregation (10^(-5) (PgC/yr)^2) 2006gb002751-fs04.eps: Meridional pattern of the air-sea flux (top) and meridional transport (bottom) of natural carbon. Shown are the zonally integrated fluxes and integrated transport for the 10 models that participated in this study. Positive (negative) values of transport indicate northward (southward) transport and positive (negative) values of flux indicate outgassing (uptake). 2006gb002751-fs05.eps: The natural carbon air-sea flux (top) and transport (bottom) integrated with longitude and depth over the Atlantic (left) and Pacific (right) ocean basins. Positive (negative) values of transport indicate northward (southward) transport and positive (negative) values of flux indicate outgassing (uptake). 2006gb002751-fs06.eps: The average of the residuals between the observed gas exchange tracer and the inverse gas exchange tracer estimates (umol/kg) for the 10 participating OGCMs plotted verses their model skill score. The residuals are defined as the observations minus the inverse estimates. 2006gb002751-fs07.eps: Histogram of the residuals between the observed gas exchange tracer and the inverse gas exchange tracer estimates divided by the observed gas exchange tracer concentrations (dimensionless). The gas exchange tracer observations pass through zero, which can cause artificially large ratios. In order to address this, observations are excluded from this plot if the magnitude of the gas exchange tracer is less than half of the uncertainty associated with this tracer from the propagation of random errors. This plot is based on results from the PRINCE-2 model, but is generally representative of all 10 participating OGCMs. The residuals are defined as the observations minus the inverse estimates. 2006gb002751-fs08.eps: Comparison between the zonally averaged residuals between the observed and modeled gas exchange tracer in umol/kg (shaded area) and the spatial footprint of the three basis functions in the Southern Ocean (10^(4)mol/m3, lines): the Polar Southern Ocean south of 58S (top), the South Sub-polar Atlantic between 44S and 58S (middle), and the South Sub-Polar Indian and Pacific between 44S and 58S (bottom). These results are from the PRINCE-2 model but are generally representative of all of the contributing OGCMs. Residuals are defined as the observations minus the inverse estimates. 2006gb002751-fs09.eps: Meridional section of the zonal mean of the temperature (degrees C, top) and salinity (bottom) residuals. Shown is the difference between observations (Conkright et al., 2002) and model simulations from the PRINCE-2 OGCM. Gray areas represent bathymetry or areas outside the model grid. The residuals of this model are typical of those exhibited by the other models that participated in this study (see also Doney et al., 2004). 2006gb002751-fs10.eps: Sensitivity of the inverse estimates of the natural CO2 flux (PgC/yr) to errors in the anthropogenic carbon estimates used in the calculation of the gas exchange tracer. The different estimates represent a series of different scenarios for biases in the anthropogenic carbon estimates from Mikaloff Fletcher et al. [2006], described briefly in section 4.2 of the manuscript. All estimates use the PRINCE-2 OGCM and are aggregated to 11 regions for clarity. ---------------------------- Tables ---------------------------- 2006gb002751-ts01.tex: Evaluation of model skill based on comparisons between results from natural radiocarbon simulations and observationally based estimates of natural radiocarbon. The normalized standard deviation (Norm. STD) is defined as the standard deviation of the modeled field divided by the corresponding standard deviation of the observed field. The skill score is based on the formulation of Taylor, 2001. The model simulations were undertaken as part of OCMIP-2. The observations are from the GLODAP dataset (Key et al., 2004). 2006gb002751-ts02.tex: Comparison between the natural CO2 flux estimates from the ocean inversion, the OCMIP-2 forward simulations, and the CCSM coupled with the ecosystem and biogeochemistry model described in (Moore et al., 2004). The OCMIP-2 results show only include those models that participated in both the ocean inversion and OCMIP-2 (NCAR, PRINCE-LL, PRINCE-HH, PRINCE-LHS, PRINCE-2, PRINCE-2a, and UL), and the mean and standard deviation have been weighted using the same radiocarbon skill score weighting as the ocean inversion. The reported uncertainties are the weighted standard deviations. Positive (negative) values indicate flux out of (into) the ocean.