Research ArticleOCEANOGRAPHY

Predicting the variable ocean carbon sink

See allHide authors and affiliations

Science Advances  17 Apr 2019:
Vol. 5, no. 4, eaav6471
DOI: 10.1126/sciadv.aav6471
  • Fig. 1 Temporal evolution and predictive skill of the global CO2 flux into the ocean.

    (A) Time series of anomalous CO2 flux into the ocean from SOM-FFN data–based estimates (15) and MPI-ESM simulations with respect to the climatological mean. Gray bars show SOM-FFN data. Assimilation, initialized simulation at a lead time of 2 years, and uninitialized simulations are shown in black, red, and blue, respectively. Here, the anomalies refer to conditions in which the respective climatological means are removed, and the positive values indicate anomalous uptake of CO2 by the ocean. The numbers show the correlation coefficients and root mean square error (in parentheses) against SOM-FFN data. (B) Correlation skill of the global CO2 flux into the ocean against SOM-FFN data–based estimates. The blue dot and red dots show the uninitialized correlation and initialized correlation at different lead years, respectively. The blue dashed line extends the uninitialized correlation for easy comparison. The vertical bars provide 90% confidence intervals, and the numbers above the bars show the P values based on a bootstrap approach (27). (C) The same as (B), but for correlation skill against assimilation.

  • Fig. 2 Predictive skill of the ΔpCO2 (pCO2sea − pCO2atm) at a lead time of 2 years.

    We show skill against SOM-FFN data–based estimates (left panels: A, C, and E) and against assimilation (right panels: B, D, and F). First row: Correlations with initialized simulations. Second row: Correlations with uninitialized simulations. Third row: Difference of the correlations between initialized and uninitialized simulations. The values are computed over the years from 1982 to 2013 with the MurCSS tool for central evaluation of predictive skill (28). Crosses denote significant skill at the 95% confidence level based on a bootstrap approach (27).

  • Fig. 3 Predictive skill of ocean surface pCO2 thermal and nonthermal components at a lead time of 2 years.

    Columns 1 (A, D, and G) and 2 (B, E, and H) show the correlations between the initialized simulations and SOM-FFN data–based estimates for the thermal and nonthermal components, respectively. Column 3 (C, F, and I) shows the nonthermal part correlation between the initialized simulations and the assimilation. First row (A–C): Correlations with initialized simulations. Second row (D–F): Correlations with uninitialized simulations. Third row (G–I): Difference of the correlations between the initialized and uninitialized simulations. The values are computed over the years from 1982 to 2013 with the MurCSS tool for central evaluation of predictive skill (28). Crosses denote significant skill at the 95% confidence level based on a bootstrap approach (27). The thermal component correlation is based on pCO2sea, and the nonthermal component correlation is based on ΔpCO2 (pCO2sea − pCO2atm) (14).

  • Fig. 4 Lead years with improved skill.

    Left maps: The spatial pattern of the lead years with improved skill for ΔpCO2 (A and B) and its thermal (C and D) and nonthermal (E and F) components, i.e., when correlations of the initialized simulations are larger than 0 and are larger than the uninitialized simulations, against SOM-FFN data–based estimates (left) and the assimilation (right). Crosses indicate that the improved skill is statistically significant at the 95% confidence level for the first 5 lead years. Right bar charts: Zonal mean of lead years with improved predictive skill of ΔpCO2 (gray bars) and its thermal and nonthermal components (red and blue dotted lines, respectively) against SOM-FFN estimates (G) and assimilation (H).

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/5/4/eaav6471/DC1

    Fig. S1. Time series of the CO2 flux into the ocean from MPI-ESM-HR simulations.

    Fig. S2. Decadal trends of CO2 flux into the ocean.

    Fig. S3. The same as Fig. 1, but for detrended global CO2 uptake by the ocean.

    Fig. S4. Predictive skill of the air-sea CO2 flux at a lead time of 2 years.

    Fig. S5. Monthly ocean surface pCO2 (gray bars) and its thermal (red) and nonthermal (blue) components.

    Fig. S6. Decadal trends of ΔpCO2 (pCO2sea − pCO2atm) and its thermal and nonthermal components from 1992 to 2001.

    Fig. S7. The same as fig. S6, but for 2002–2011.

    Fig. S8. Predictive skill of pCO2 thermal and nonthermal components at a lead time of 2 years against assimilation.

    Table S1. Summary of the simulations used in this study based on the MPI-ESM-HR decadal prediction system.

  • Supplementary Materials

    This PDF file includes:

    • Fig. S1. Time series of the CO2 flux into the ocean from MPI-ESM-HR simulations.
    • Fig. S2. Decadal trends of CO2 flux into the ocean.
    • Fig. S3. The same as Fig. 1, but for detrended global CO2 uptake by the ocean.
    • Fig. S4. Predictive skill of the air-sea CO2 flux at a lead time of 2 years.
    • Fig. S5. Monthly ocean surface pCO2 (gray bars) and its thermal (red) and nonthermal (blue) components.
    • Fig. S6. Decadal trends of ΔpCO2 (pCO2sea − pCO2atm) and its thermal and nonthermal components from 1992 to 2001.
    • Fig. S7. The same as fig. S6, but for 2002–2011.
    • Fig. S8. Predictive skill of pCO2 thermal and nonthermal components at a lead time of 2 years against assimilation.
    • Table S1. Summary of the simulations used in this study based on the MPI-ESM-HR decadal prediction system.

    Download PDF

    Files in this Data Supplement:

Navigate This Article