How accurate is the estimation of anthropogenic carbon in the ocean?An evaluation of the DC* methodKatsumi Matsumoto1Geological Survey of Japan, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba, Ibaraki,JapanLamont-Doherty Earth Observatory, Columbia University, Palisades, New York, USANicolas GruberInstitute of Geophysics and Planetary Physics (IGPP) and Department of Atmospheric and Oceanic Sciences, University ofCalifornia, Los Angeles, California, USAReceived 26 October 2004; revised 6 April 2005; accepted 23 May 2005; published 24 August 2005.[1] The DC* method of Gruber et al. (1996) is widely used to estimate the distribution ofanthropogenic carbon in the ocean; however, as yet, no thorough assessment of itsaccuracy has been made. Here we provide a critical re-assessment of the method anddetermine its accuracy by applying it to synthetic data from a global oceanbiogeochemistry model, for which we know the ‘‘true’’ anthropogenic CO2distribution.Our results indicate that the DC* method tends to overestimate anthropogenic carbon inrelatively young waters but underestimate it in older waters. Main sources of these biasesare (1) the time evolution of the air-sea CO2disequilibrium, which is not properlyaccounted for in the DC* method, (2) a pCFC ventilation age bias that arises from mixing,and (3) errors in identifying the different end-member water types. We largely support thefindings of Hall et al. (2004), who have also identified the first two bias sources. Anextrapolation of the errors that we quantified on a number of representative isopycnals tothe global ocean suggests a positive bias of about 7% in the DC*-derived globalanthropogenic CO2inventory. The magnitude of this bias is within the previouslyestimated 20% uncertainty of the method, but regional biases can be larger. Finally, wepropose two improvements to the DC* method in order to account for the evolution of air-sea CO2disequilibrium and the ventilation age mixing bias.Citation: Matsumoto, K., and N. Gruber (2005), How accurate is the estimation of anthropogenic carbon in the ocean? An evaluationof the DC* method, Global Biogeochem. Cycles, 19, GB3014, doi:10.1029/2004GB002397.1. Introduction[2] Understanding the fate of anthropogenic CO2is one ofthe most import ant goals in global carbon cycle research.Two developments during the last decades have maderelatively reliable, data-based estimations of anthropogeniccarbon in the ocean possible. One is the generation of alarge number of high quality inorganic carbon data from theWorld Ocean Circulation Experiment (WOCE) and the JointGlobal Ocean Flux Study (JGOFS) [Key et al., 2004]. Theother is the formulation of the DC* method, which is anempirical procedure to separate the small anthropogenicsignal from the much larger, inorganic CO2background inthe ocean [Gruber et al., 1996 ]. The DC* method has beenapplied to all three major oceanic basins [Gruber, 1998; Leeet al., 2003; Sabine et al., 2002, 1999]. In a summary of theWOCE/JGOFS data analyses, Sabine et al. [2004b] estimatea global oceanic inventory of anthropogenic carbon for anominal year of 1994 of 118 ± 19 Pg-C (Pg = 1015grams),which represents about 50% of the fossil fuel CO2emittedbetween 1800 and 1994. They inferred from this, togetherwith the well-known accumulation of CO2in the atmo-sphere and total emissions from fossil fuel burning andcement production, that the terrestrial biosphere was a netsource of CO2to the atmosphere by 39 ± 28 Pg-C over thisperiod.[3] The validity of this impo rtant conclusion hingescritically on the reliability of the DC*-derived anthropogenicCO2estimates. The uncertainty reported by Sabine et al.[2004b] accounts for random and mapping errors (i.e.,errors in extrapolating sparsely covered data to a regulargrid) but not systematic biases in the method. Like theearlier methods proposed by Brewer [1978] and Chen andMillero [1979], the DC* method starts from the measure-ment of dissolved inorganic carbon (DIC) and separatesfrom it the relatively small anthropogenic signal by remov-ing the much larger biological and CO2solubility compo-nents. The separation requires DIC data as well asmeasurements of temperature, salinity, and various nutrientsGLOBAL BIOGEOCHEMICAL CYCLES, VOL. 19, GB3014, doi:10.1029/2004GB002397, 20051Now at Department of Geology and Geophysics, University ofMinnesota, Minneapolis, Minnesota, USA.Copyright 2005 by the American Geophysical Union.0886-6236/05/2004GB002397$12.00GB3014 1of17and gases including chlorofluorocarbons (CFC). Themethod involves a number of steps and assumptions, someof which are difficult to verify or only approximations. Forexample, a central assumption is that the air-sea CO2disequilibri um (i.e., the difference between in situ DICand surface water DIC that would be in equilibrium withatmospheric pCO2) has remained constant since preindus-trial times. This assumption is not strictly correct, because itis the change over time of the CO2disequilibrium thatdrives the uptake of anthropogenic CO2by the ocean.[4] The DC* method has been a subject of much scrutiny.For example, Hall et al. [2004] examined the reliability ofchlorofluorocarbon (CFC) concentrations as means to de-termine the water parcel ventilation age, which is needed bythe method to estimate the air-sea CO2disequilibrium. Theyconclude that because the CFC-derived ventilation agestend to be biased young, the amount of anthropogenicCO2in the ocean is overestimated. Other st udies havecompared the anthropogenic CO2reconstructions from theDC* method with those from other methods as theywere applied to the same sets of hydrographic and carbondata [Coatanoan et al., 2001; Sabine and Feely, 2001;Wanninkhof et al., 1999]. They found substantial differencesin the anthropogenic CO2estimates but were unable toexplain the discrepancies and determine with confidencewhich methods were superior. More recently, Matsumoto etal. [2004] showed that the same ocean carbon cycle modelsthat can adequately simulate the d istribution of natu ralradiocarbon and CFC inventories are unable to reproducethe inventory of anthropogenic carbon as estimated by theDC* method. Since CFCs and natural radiocarbon are eitherdirectly measured or derived from measurements with littleambiguity, this suggests that either the models are makingfundamental errors simulating these transient tracers or thatthe estimated anthropogenic CO2is biased.[5] In this work, we critically