Direct observation of increasing CO2 in the Weddell Gyre along the Prime Meridian during 1973-2008

The World Ocean takes up a large portion of the anthropogenic CO2 emitted into the atmosphere. Determining the resulting increase in dissolved inorganic carbon (C-T, expressed in mu mol kg(-1)) is challenging, particularly in the sub-surface and deep Southern Ocean where the time rate of change of C-T (in mu mol kg(-1) decade(-1)) is commonly expected to be low. We present a determination of this time trend of C-T in a dataset of measurements that spans 35 years comprising 10 cruises in the 1973-2008 period along the 0 degrees-meridian in the Weddell Gyre. The inclusion of many cruises aims to generate results that are more robust than may be obtained by taking the difference between only one pair of cruises, each of which may suffer from errors in accuracy. To further improve consistency between cruises, data were adjusted in order to obtain time-invariant values of C-T (and other relevant parameters) over the 35 years in the least ventilated local water body, this comprising the deeper Warm Deep Water (WDW) and upper Weddell Sea Deep Water (WSDW). It is assumed that this normalization procedure will allow trends in C-T in the more intensely ventilated water masses to be more clearly observed. Time trends were determined directly in measurements of C-T, and alternatively in back-calculated values of preformed C-T (C-T(0); i.e., the C-T of the water at the time that it lost contact with the atmosphere). The determined time trends may be attributed to a combination of natural variability (in hydrography or biogeochemistry) and increased uptake of anthropogenic CO2 from the atmosphere. In order to separate these natural and anthropogenic components, an analysis of the residuals of a multivariate linear regression (MLR), involving the complete time series of all 10 cruises, was additionally performed. This approach is referred to as the Time Series Residuals (TSR) approach. Using the direct method, the time trends of C-T in the WSDW are quite small and non-significant at +0.176 +/- 0.321 mu mol kg(-1) decade(-1). On the other hand, the measured concentration of C-T in the Weddell Sea Bottom Water (WSBW) is shown to rise slowly but significantly over the period from 1973 to 2008 at a rate of +1.151 +/- 0.563 mu mol kg(-1) decade(-1). The spatial distribution of these determined increases of C-T in the deep Weddell Gyre closely resembles that of the increase of the anthropogenic tracer CFC-12, this strong similarity supporting a mostly anthropogenic cause for the increasing trend of C-T. Time trends in back-calculated values of C-T(0) appear to be obscured due to uncertainties in the measurements of O-2. Finally, the shallow waters ( < 200 m depth) do not allow for interpretation since these are strongly affected by seasonality. Due to the small time trend signal in the WSBW, the TSR approach does not allow for unambiguous attribution of the observed trend in C-T in the WSBW. The residuals of the TSR method do exhibit a time trend (considered representative of the time trend of C-ant) of +0.445 +/- 0.405 mu mol kg(-1) decade(-1) (i.e., only 38% of the direct observed time trend in C-T) thus only partly supporting the attribution of the measured time trend of C-T to uptake of anthropogenic CO2. Another TSR-derived result suggests that there is no significant time trend of biogeochemical changes. A time trend in hydrography of mixing between two deep water masses does exist, as evidenced by a slight positive time trend in the temperature of the WSBW, but is inadequate to explain the time trend of C-T. After all, the time trend in measured C-T is most straightforwardly ascribed entirely to uptake of C-ant, and assuming an exponentially growing history of storage, the observed increase of C-T in the WSBW suggests that a total amount of C-ant of 6 +/- 3 mu mol kg(-1) has accumulated in this water mass between the onset of the Industrial Revolution and 1995. Extrapolating the determined time trend, the rate of storage of C-ant in the deep Weddell Gyre ( > 3000 m, west of 20 degrees E) is calculated to be about 12 +/- 6 TgC yr(-1) over the 1973-2008 period. This rate of storage is likely somewhat lower than the rate of export of C-ant from the surface water into the deep Weddell Gyre, this due to continuous loss of C-ant with WSDW flowing out of the Weddell Gyre into the deep basins of the other oceans as AABW. (C) 2011 Elsevier Ltd. All rights reserved.