2003GL018970-text.txt Matsumoto et al, Evaluation of ocean carbon cycle models with data-based metrics, Geophysical Research Letters, 2003. (A) Observational Uncertainty The uncertainty associated with the data based inventories is difficult to estimate. The objective mapping procedure yields an uncertainty for each grid cell; however, these individual errors are very highly correlated both horizontally and vertically. Normal propagation of these individual errors would, we believe, result in a significantly exaggerated inventory error estimate. Based on experience with the procedure and circumstantial evidence, we believe the inventory error due to mapping and integration to be about 15%. For example, using the same data and same mapping routines we were able to change an inventory estimate for the Pacific Ocean by ~10% by changing minor details of the procedure. Excluding various subsets of the data had a similar impact. In the Atlantic Ocean the CFC-11 inventory varied by 15% when using different data sets (one was a subset of the other) and radically different procedures. The error estimate incorporates noise and lack of fit with the data, but does not include any systematic bias that might be in the data values. Additionally, the error estimates do not account for the fact that WOCE survey cruises for each ocean basin were conducted over a period of a few years. A mid-year chosen for the Indian survey is 1995 and for the Pacific survey is 1993. Therefore, model results from simulation years 1995 and 1993 are used for the Indian and Pacific analyses respectively. Because CFC-11 concentration is measured accurately and precisely, we assume that the mapping error (15%) represents the bulk of the CFC-11 inventory estimate uncertainty. For anthropogenic carbon, there are additional uncertainties associated with the Delta-C* method [Gruber et al., 1996]. These uncertainties are estimated to be about 10% for the Indian and Pacific basins [Sabine et al., 2002; Sabine et al., 1999]. Therefore, we assume that the combined uncertainty for anthropogenic carbon inventory is about 25%. (B) Water Mass Boundaries For deep ocean natural Delta-14C data-model comparisons (Figures 2 and 3), we needed to define water mass boundaries. We used for the North Atlantic Deep Water (NADW): Equator-60degN, 1000-3500m; North Pacific Deep Water (NPDW): Equator-60degN, 1500-5000m; and Circumpolar Deep Water (CDW): 90degS-45degS, 1500-5000m. These boundaries have been applied to both observation and models, except NADW in models. In models, a smaller NADW depth range of 1500-2500m is used, because most models produce too shallow a NADW [Doney et al., submitted]. For data-model comparisons of CDW natural Delta-14C and Indo-Pacific inventories of CFC-11 and anthropogenic carbon (Figure 3), we needed to define the Indian and Pacific basins. For models, we used the standard OCMIP-2 basin masks, which divide the Indian and Pacific basins south of Australia (i.e., south of 40degS) along 150degE. North of Australia, the two basins are separated approximately along Indonesia archipelago. The OCMIP-2 masks in the Southern Ocean has the eastern boundary of the Pacific sector at 70degW and the western boundary of the Indian sector at 20degE. For the particular observational data sets we used, these sector boundaries are not the same: the Indian sector extends from 20degE to 120degE and the Pacific from 130degE to 70degW. Therefore, while the combined Indo-Pacific sector for observations and models are both defined between 20degE to 70degW (270deg in longitude), observations are missing 10deg in longitude (between 120degE and 130degE). To account for this difference, model inventories of the combined Indo-Pacific sector for the Southern Ocean (south of 40degS) is normalized by 0.963 (260/270). This is a reasonable adjustment to render data-model comparison meaningful, given the largely spatially homogeneous nature of the circumpolar properties in the models References Doney, S.C., K. Lindsay, K. Caldeira, J.-M. Campin, H. Drange, J.-C. Dutay, M. Follows, Y. Gao, A. Gnanadesikan, N. Gruber, A. Ishida, F. Joos, G. Madec, E. Maier-Reimer, J.C. Marshall, R.J. Matear, P. Monfray, R. Najjar, J.C. Orr, G.-K. Plattner, J. Sarmiento, R. Schlitzer, R. Slater, P.S. Swathi, I.J. Totterdell, M.-F. Weirig, Y. Yamanaka, and A. Yool, Evaluating global ocean carbon models: The importance of realistic physics, Global Biogeochemical Cycles, submitted. Gruber, N., J.L. Sarmiento, and T.F. Stocker, An improved method for detecting anthropogenic CO2 in the oceans, Global Biogeochemical Cycles, 10 (4), 809-837, 1996. Sabine, C.L., R.A. Feeley, R.M. Key, J.L. Bullister, F.J. Millero, K. Lee, T.-H. Peng, B. Tilbrook, T. Ono, and C.S. Wong, Distribution of anthropogenic CO2 in the Pacific Ocean, Global Biogeochemical Cycles, 16 (4), 10.1029/2001GB001639, 2002. Sabine, C.L., R.M. Key, K.M. Johnson, F.J. Millero, A. Poisson, J.L. Sarmiento, D.W.R. Wallace, and C.D. Winn, Anthropogenic CO2 inventory of the Indian Ocean, Global Biogeochem. Cycles, 13 (1), 179-198, 1999.