Comparison of two approaches to quantify anthropogenic CO2 in the ocean: Results from the northern Indian Ocean
|Author(s)||Coatanoan Christine1, Goyet C1, Gruber N2, Sabine Cl2, Warner M3|
|Affiliation(s)||1 : Woods Hole Oceanog Inst, Marine Chem & Geochem Dept, Woods Hole, MA 02543 USA.
2 : Princeton Univ, Atmospher & Ocean Sci Program, Princeton, NJ USA.
3 : Univ Washington, Sch Oceanog, Seattle, WA 98195 USA.
|Source||Global Biogeochemical Cycles (0886-6236) (Amer Geophysical Union), 2001-03 , Vol. 15 , N. 1 , P. 11-25|
|WOS© Times Cited||24|
|Abstract||This study compares two recent estimates of anthropogenic CO2 in the northern Indian Ocean along the World Ocean Circulation Experiment cruise I1 [Goyet et al., 1999; Sabine et al., 1999]. These two studies employed two different approaches to separate the anthropogenic CO2 signal from the large natural background variability. Sabine et al.  used the DeltaC* approach first described by Gruber et al. , whereas Goyet et al.  used an optimum multiparameter mixing analysis referred to as the MIX approach. Both approaches make use of similar assumptions in order to remove variations due to remineralization of organic matter and the dissolution of calcium carbonates (biological pumps). However, the two approaches use very different hypotheses in order to account for variations due to physical processes including mixing and the CO2 solubility pump. Consequently, substantial differences exist in the upper thermocline approximately between 200 and 600 m. Anthropogenic CO2 concentrations estimated using the DeltaC* approach average 12 +/- 4 mu mol kg(-1) higher in this depth range than concentrations estimated using the MIX approach, Below similar to 800 m, the MIX approach estimates slightly higher anthropogenic CO2 concentrations and a deeper vertical penetration. Despite this compensatory effect, water column inventories estimated in the 0-3000 m depth range by the DeltaC* approach are generally similar to 20% higher than those estimated by the MIX approach, with this difference being statistically significant beyond the 0.001 level. We examine possible causes for these differences and identify a number of critical additional measurements that will make it possible to discriminate better between the two approaches.|