Journal cover Journal topic
Biogeosciences An interactive open-access journal of the European Geosciences Union
Biogeosciences, 10, 7793-7815, 2013
© Author(s) 2013. This work is distributed
under the Creative Commons Attribution 3.0 License.
Research article
29 Nov 2013
A neural network-based estimate of the seasonal to inter-annual variability of the Atlantic Ocean carbon sink
P. Landschützer1, N. Gruber2, D. C. E. Bakker1, U. Schuster1,*, S. Nakaoka3, M. R. Payne2,4, T. P. Sasse5, and J. Zeng3 1School of Environmental Sciences, University of East Anglia, Norwich Research Park, Norwich, UK
2Institute of Biogeochemistry and Pollutant Dynamics, ETH Zürich, Zürich, Switzerland
3Center for Global Environmental Research, National Institute for Environmental Studies, Tsukuba, Ibaraki, Japan
4Section for Oceanography and Climate, National Institute for Aquatic Resources, Technical University of Denmark, Charlottenlund, Denmark
5Climate Change Research Centre, University of New South Wales, Sydney, Australia
*now at: College of Life and Environmental Sciences, Hatherly Laboratories, University of Exeter, Exeter, UK
Abstract. The Atlantic Ocean is one of the most important sinks for atmospheric carbon dioxide (CO2), but this sink has been shown to vary substantially in time. Here we use surface ocean CO2 observations to estimate this sink and the temporal variability from 1998 through 2007 in the Atlantic Ocean. We benefit from (i) a continuous improvement of the observations, i.e. the Surface Ocean CO2 Atlas (SOCAT) v1.5 database and (ii) a newly developed technique to interpolate the observations in space and time. In particular, we use a two-step neural network approach to reconstruct basin-wide monthly maps of the sea surface partial pressure of CO2 (pCO2) at a resolution of 1° × 1°. From those, we compute the air–sea CO2 flux maps using a standard gas exchange parameterization and high-resolution wind speeds. The neural networks fit the observed pCO2 data with a root mean square error (RMSE) of about 10 μatm and with almost no bias. A check against independent time-series data and new data from SOCAT v2 reveals a larger RMSE of 22.8 μatm for the entire Atlantic Ocean, which decreases to 16.3 μatm for data south of 40° N. We estimate a decadal mean uptake flux of −0.45 ± 0.15 Pg C yr−1 for the Atlantic between 44° S and 79° N, representing the sum of a strong uptake north of 18° N (−0.39 ± 0.10 Pg C yr−1), outgassing in the tropics (18° S–18° N, 0.11 ± 0.07 Pg C yr−1), and uptake in the subtropical/temperate South Atlantic south of 18° S (−0.16 ± 0.06 Pg C yr−1), consistent with recent studies. The strongest seasonal variability of the CO2 flux occurs in the temperature-driven subtropical North Atlantic, with uptake in winter and outgassing in summer. The seasonal cycle is antiphased in the subpolar latitudes relative to the subtropics largely as a result of the biologically driven winter-to-summer drawdown of CO2. Over the 10 yr analysis period (1998 through 2007), sea surface pCO2 increased faster than that of the atmosphere in large areas poleward of 40° N, while in other regions of the North Atlantic the sea surface pCO2 increased at a slower rate, resulting in a barely changing Atlantic carbon sink north of the Equator (−0.01 ± 0.02 Pg C yr−1 decade−1). Surface ocean pCO2 increased at a slower rate relative to atmospheric CO2 over most of the Atlantic south of the Equator, leading to a substantial trend toward a stronger CO2 sink for the entire South Atlantic (−0.14 ± 0.02 Pg C yr−1 decade−1). In contrast to the 10 yr trends, the Atlantic Ocean carbon sink varies relatively little on inter-annual timescales (±0.04 Pg C yr−1; 1 σ).

Citation: Landschützer, P., Gruber, N., Bakker, D. C. E., Schuster, U., Nakaoka, S., Payne, M. R., Sasse, T. P., and Zeng, J.: A neural network-based estimate of the seasonal to inter-annual variability of the Atlantic Ocean carbon sink, Biogeosciences, 10, 7793-7815, doi:10.5194/bg-10-7793-2013, 2013.
Publications Copernicus