Research ArticleCLIMATE CHANGE

Assessing recent warming using instrumentally homogeneous sea surface temperature records

+ See all authors and affiliations

Science Advances  04 Jan 2017:
Vol. 3, no. 1, e1601207
DOI: 10.1126/sciadv.1601207
  • Fig. 1 Comparison of the different ERSSTv3b, ERSSTv4, buoy-only, and CCI SST monthly anomalies from January 1997 to December 2015, restricting all series to common coverage.

    ERSSTv4 is shown as a broad band for visualization purposes; this band does not represent an uncertainty range. The series are aligned on the 1997–2001 period for comparison purposes. Spatial trend maps are also available in fig. S1, and a similar comparison with Argo data is shown in fig. S2.

  • Fig. 2 Twelve-month centered moving average of temperature difference series between composite and buoy-only, CCI, and ARC SST anomalies.

    Values below 0 indicate that the composite series has a cool bias relative to the IHSST record.

  • Fig. 3 Twelve-month centered moving average of temperature difference series between composite and Argo near-SST anomalies.
  • Fig. 4 Trends and 95% confidence intervals (°C per decade) in difference series for each IHSST and composite SST series, masked to common composite SST coverage.

    Each difference series represents a composite series minus an IHSST series. Confidence intervals for trends are calculated using an ARMA(1, 1) autocorrelation model. Values below 0 indicate that the composite series has a lower trend than the IHSST series over the period examined. The two trend periods examined are January 1997 to December 2015 and January 2005 to December 2015.

  • Fig. 5 Twelve-month centered moving average of temperature difference series between collocated ERSSTv4/HadSST3 ensemble realizations and IHSST anomalies.

    The left column shows the difference series with the buoy-only record. The right column shows the difference series with the CCI record. The top row shows 1000 ERSSTv4 ensemble members, with operational versions of ERSSTv3b and v4 highlighted in black (note that the ERSST ensemble runs only go through 2014). The middle row shows the 100 published HadSST3 ensemble members, with the operational version in black. The bottom row displays the 1000 expanded ensemble members, as discussed in the text.

  • Fig. 6 Twelve-month centered moving average of differences between CCI ATSR + AVHRR and ATSR-only ARC SST records during the period of overlap.

    The earlier IHSST ARC shows small differences to the newer combined version; however, the differences are minor compared to the differences relative to the composite SST records.

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/3/1/e1601207/DC1

    fig. S1. Trend maps on the 2005–2015 period for all of the composite records, and for the buoy, Argo, and CCI records.

    fig. S2. Comparison of ERSSTv3b and ERSSTv4 with three different Argo-based near-SST records, using the same spatial restrictions as in Fig. 1, but with ERSSTv4 aligned to 1997–2001 (inclusive), with all other series aligned onto ERSSTv4 using the 2005–2007 period because of the limited time span with Argo data.

    fig. S3. Trends and 95% confidence intervals (°C per decade) for the 1997–2012 period for buoy, ARC, and CCI IHSSTs and each composite SST series, masked to common composite SST coverage.

    fig. S4. Cross-validated uncertainties for the buoy record, whether with no climatology or with daily climatologies derived from the OISSTv2 daily reanalysis data.

    fig. S5. Twelve-month centered moving average of temperature difference series between composite and buoy-only, CCI, and ARC SST anomalies restricted to common coverage across all series shown (four composites, buoys, and ARC/CCI).

    fig. S6. Twelve-month centered moving average of temperature difference series between composite and Argo near-SST anomalies restricted to common coverage across all series with records from 2005 to 2015 (four composites, three Argos, buoy-only, and CCI).

    fig. S7. Trends and 95% confidence intervals (°C per decade) in difference series for each IHSST and composite SST series, masked to common coverage for all series available.

    fig. S8. Twelve-month centered moving average of temperature difference series between composite and buoy-only, CCI, and ARC SST anomalies, with the buoy and HadSST3 series kriged and all series reduced to common coverage to ensure consistent complete spatial and temporal coverage.

    fig. S9. Twelve-month centered moving average of temperature difference series between composite and Argo near-SST anomalies with the buoy and HadSST3 series kriged and all series reduced to common coverage to ensure consistent complete spatial and temporal coverage.

    fig. S10. Trends and 95% confidence intervals (°C per decade) in difference series for each IHSST and composite SST series, with the buoy and HadSST3 series kriged and all series reduced to common coverage to ensure consistent complete spatial and temporal coverage.

    fig. S11. Trend difference maps from January 1997 to December 2015 for the difference between ERSSTv4 and CCI.

    fig. S12. Differences between ERSSTv4 and CCI by latitude zone.

    fig. S13. Buoy-only and ship-only temperature anomalies from January 1997 to December 2015, with no matching of coverage.

    fig. S14. Difference between ship-only record and the three Argo series using a 12-month centered moving average.

    fig. S15. Comparison of COBE-SST and COBE-SST2 to the IHSSTs using a 12-month centered moving average.

    fig. S16. Comparison of HadSST3 and HadNMAT2 to the IHSSTs using a 12-month centered moving average.

    fig. S17. Trends in differences for ERSSTv4 records versus IHSST records, with common coverage from 1997 (buoys and CCI only as dashed lines) and common coverage from 2005 (buoys, CCI, and Argos as solid lines).

    fig. S18. Differences between Argo series at minimum reported depth, and differences within each Argo series as minimum reported, 20- and 50-m depths.

    fig. S19. Comparison of buoy records composed of all buoys (drifting + moored) and only drifting buoys.

    fig. S20. Comparison of drifting buoy–based IHSST records for different quality control and homogenization choices.

    fig. S21. Twelve-month centered moving average of differences between IHSST series from January 1997 to December 2015 when reduced to common coverage for each separate pairing.

    fig. S22. Trend difference maps during 2005–2015 for the composite records versus Buoy, CCI, and Argo (H2008).

    fig. S23. Trends in differences for composite versus buoy (solid lines) and CCI (dashed lines) IHSST records with common coverage.

    fig. S24. Number of observations over time by instrument type in the ICOADS (version 2.5) database.

    fig. S25. Similar to fig. S24, but showing the percentage of ICOADS observations in each year from each instrument type.

  • Supplementary Materials

    This PDF file includes:

    • fig. S1. Trend maps on the 2005–2015 period for all of the composite records, and for the buoy, Argo, and CCI records.
    • fig. S2. Comparison of ERSSTv3b and ERSSTv4 with three different Argo-based near-SST records, using the same spatial restrictions as in Fig. 1, but with ERSSTv4 aligned to 1997–2001 (inclusive), with all other series aligned onto ERSSTv4 using the 2005–2007 period because of the limited time span with Argo data.
    • fig. S3. Trends and 95% confidence intervals (°C per decade) for the 1997–2012 period for buoy, ARC, and CCI IHSSTs and each composite SST series, masked to common composite SST coverage.
    • fig. S4. Cross-validated uncertainties for the buoy record, whether with no climatology or with daily climatologies derived from the OISSTv2 daily reanalysis data.
    • fig. S5. Twelve-month centered moving average of temperature difference series between composite and buoy-only, CCI, and ARC SST anomalies restricted to common coverage across all series shown (four composites, buoys, and ARC/CCI).
    • fig. S6. Twelve-month centered moving average of temperature difference series between composite and Argo near-SST anomalies restricted to common coverage across all series with records from 2005 to 2015 (four composites, three Argos, buoy-only, and CCI).
    • fig. S7. Trends and 95% confidence intervals (°C per decade) in difference series for each IHSST and composite SST series, masked to common coverage for all series available.
    • fig. S8. Twelve-month centered moving average of temperature difference series between composite and buoy-only, CCI, and ARC SST anomalies, with the buoy and HadSST3 series kriged and all series reduced to common coverage to ensure consistent complete spatial and temporal coverage.
    • fig. S9. Twelve-month centered moving average of temperature difference series between composite and Argo near-SST anomalies with the buoy and HadSST3 series kriged and all series reduced to common coverage to ensure consistent complete spatial and temporal coverage.
    • fig. S10. Trends and 95% confidence intervals (°C per decade) in difference series for each IHSST and composite SST series, with the buoy and HadSST3 series kriged and all series reduced to common coverage to ensure consistent complete spatial and temporal coverage.
    • fig. S11. Trend difference maps from January 1997 to December 2015 for the difference between ERSSTv4 and CCI.
    • fig. S12. Differences between ERSSTv4 and CCI by latitude zone.
    • fig. S13. Buoy-only and ship-only temperature anomalies from January 1997 to December 2015, with no matching of coverage.
    • fig. S14. Difference between ship-only record and the three Argo series using a 12-month centered moving average.
    • fig. S15. Comparison of COBE-SST and COBE-SST2 to the IHSSTs using a 12-month centered moving average.
    • fig. S16. Comparison of HadSST3 and HadNMAT2 to the IHSSTs using a 12-month centered moving average.
    • fig. S17. Trends in differences for ERSSTv4 records versus IHSST records, with common coverage from 1997 (buoys and CCI only as dashed lines) and common coverage from 2005 (buoys, CCI, and Argos as solid lines).
    • fig. S18. Differences between Argo series at minimum reported depth, and differences within each Argo series as minimum reported, 20- and 50-m depths.
    • fig. S19. Comparison of buoy records composed of all buoys (drifting + moored) and only drifting buoys.
    • fig. S20. Comparison of drifting buoy–based IHSST records for different quality control and homogenization choices.
    • fig. S21. Twelve-month centered moving average of differences between IHSST series from January 1997 to December 2015 when reduced to common coverage for each separate pairing.
    • fig. S22. Trend difference maps during 2005–2015 for the composite records versus Buoy, CCI, and Argo (H2008).
    • fig. S23. Trends in differences for composite versus buoy (solid lines) and CCI (dashed lines) IHSST records with common coverage.
    • fig. S24. Number of observations over time by instrument type in the ICOADS (version 2.5) database.
    • fig. S25. Similar to fig. S24, but showing the percentage of ICOADS observations in each year from each instrument type.

    Download PDF

    Files in this Data Supplement: