Abstract
Increased knowledge of the present global carbon cycle is important for our ability to understand and to predict the future carbon cycle and global climate. Approximately half of the anthropogenic carbon released to the atmosphere from fossil fuel burning is stored in the ocean, although distribution and regional fluxes of the ocean sink are debated. Estimates of anthropogenic carbon (C(ant)) in the oceans remain prone to error arising from (i) a need to estimate preindustrial reference concentrations of carbon for different oceanic regions, and (ii) differing behavior of transient ocean tracers used to infer C(ant). We introduce an empirical approach to estimate C(ant) that circumvents both problems by using measurement of the decadal change of ocean carbon concentrations and the exponential nature of the atmospheric C(ant) increase. In contrast to prior approaches, the results are independent of tracer data but are shown to be qualitatively and quantitatively consistent with tracer-derived estimates. The approach reveals more C(ant) in the deep ocean than prior studies; with possible implications for future carbon uptake and deep ocean carbonate dissolution. Our results suggest that this approachs applied on the unprecedented global data archive provides a means of estimating the C(ant) for large parts of the world's ocean.