Observations of sea surface and land–near-surface merged temperature anomalies are used to monitor climate variations and to evaluate climate simulations; therefore, it is important to make analyses of these data as accurate as possible. Analysis uncertainty occurs because of data errors and incomplete sampling over the historical period. This manuscript documents recent improvements in NOAA’s merged global surface temperature anomaly analysis, monthly, in spatial 5° grid boxes. These improvements allow better analysis of temperatures throughout the record, with the greatest improvements in the late nineteenth century and since 1985. Improvements in the late nineteenth century are due to improved tuning of the analysis methods. Beginning in 1985, improvements are due to the inclusion of bias-adjusted satellite data. The old analysis (version 2) was documented in 2005, and this improved analysis is called version 3.

An improved SST reconstruction for the 1854–1997 period is developed. Compared to the version 1 analysis, in the western tropical Pacific, the tropical Atlantic, and Indian Oceans, more variance is resolved in the new analysis. This improved analysis also uses sea ice concentrations to improve the high-latitude SST analysis and a modified historical bias correction for the 1939–41 period. In addition, the new analysis includes an improved error estimate. Analysis uncertainty is largest in the nineteenth century and during the two world wars due to sparse sampling. The near-global average SST in the new analysis is consistent with the version 1 reconstruction. The 95% confidence uncertainty for the near-global average is 0.4°C or more in the nineteenth century, near 0.2°C for the first half of the twentieth century, and 0.1°C or less after 1950. Abstract An improved SST reconstruction for the 1854–1997 period is developed. Compared to the version 1 analysis, in the western tropical Pacific, the tropical Atlantic, and Indian Oceans, more variance is resolved in the new analysis. This improved analysis also uses sea ice concentrations to improve the high-latitude SST analysis and a modified historical bias correction for the 1939–41 period. In addition, the new analysis includes an improved error estimate. Analysis uncertainty is largest in the nineteenth century and during the two world wars due to sparse sampling. The near-global average SST in the new analysis is consistent with the version 1 reconstruction. The 95% confidence uncertainty for the near-global average is 0.4°C or more in the nineteenth century, near 0.2°C for the first half of the twentieth century, and 0.1°C or less after 1950.

A monthly extended reconstruction of global SST (ERSST) is produced based on Comprehensive Ocean–Atmosphere Data Set (COADS) release 2 observations from the 1854–1997 period. Improvements come from the use of updated COADS observations with new quality control procedures and from improved reconstruction methods. In addition error estimates are computed, which include uncertainty from both sampling and analysis errors. Using this method, little global variance can be reconstructed before the 1880s because data are too sparse to resolve enough modes for that period. Error estimates indicate that except in the North Atlantic ERSST is of limited value before 1880, when the uncertainty of the near-global average is almost as large as the signal. In most regions, the uncertainty decreases through most of the period and is smallest after 1950. The large-scale variations of ERSST are broadly consistent with those associated with the Hadley Centre Global Sea Ice and Sea Surface Temperature (HadISST) reconstruction produced by the Met Office. There are differences due to both the use of different historical bias corrections as well as different data and analysis procedures, but these differences do not change the overall character of the SST variations. Procedures used here produce a smoother analysis compared to HadISST. The smoother ERSST has the advantage of filtering out more noise at the possible cost of filtering out some real variations when sampling is sparse. A rotated EOF analysis of the ERSST anomalies shows that the dominant modes of variation include ENSO and modes associated with trends. Projection of the HadISST data onto the rotated eigenvectors produces time series similar to those for ERSST, indicating that the dominant modes of variation are consistent in both.

A merged land–air–sea surface temperature reconstruction analysis is developed for monthly anomalies. The reconstruction is global and spatially complete. Reconstructed anomalies damp toward zero in regions with insufficient sampling. Error estimates account for the damping associated with sparse sampling, and also for bias uncertainty in both the land and sea observations. Averages of the reconstruction are similar to simple averages of the unanalyzed data for most of the analysis period. For the nineteenth century, when sampling is most sparse and the error estimates are largest, the differences between the averaged reconstruction and the simple averages are largest. Sampling is always sparse poleward of 60° latitude, and historic reconstructions for the polar regions should be used with caution.

Arctic sea surface temperatures (SSTs) are estimated mostly from satellite sea ice concentration (SIC) estimates. In regions with sea ice the SST is the temperature of open water or of the water under the ice. A number of different proxy SST estimates based on SIC have been developed. In recent years more Arctic quality-control buoy SSTs have become available, allowing better validation of different estimates and the development of improved proxy estimates. Here proxy SSTs from different approaches are evaluated and an improved proxy SST method is shown. The improved proxy SSTs were tested in an SST analysis, and showed reduced bias and random errors compared to the Arctic buoy SSTs. Almost all reduction in errors is in the warm melt season. In the cold season the SIC is typically high and all estimates tend to have low errors. The improved method will be incorporated into an operational SST analysis.

Because of changes in SST sampling methods in the 1940s and earlier, there are biases in the earlier period SSTs relative to the most recent 50 years. Published results from the Met Office have shown the need for historic bias correction and have developed several correction techniques. An independent bias-correction method is developed here from an analysis using nighttime marine air temperatures and SST observations from the Comprehensive Ocean–Atmosphere Data Set (COADS). Because this method is independent from methods proposed by the Met Office, the differences indicate uncertainties and similarities indicate where users may have more confidence in the bias correction. The new method gives results that are broadly consistent with the latest Met Office bias estimates. However, this bias estimate has a stronger annual cycle of bias in the Northern Hemisphere in comparison with the Met Office estimate. Both estimates have midlatitude annual cycles, with the greatest bias in the cold season, and ... Abstract Because of changes in SST sampling methods in the 1940s and earlier, there are biases in the earlier period SSTs relative to the most recent 50 years. Published results from the Met Office have shown the need for historic bias correction and have developed several correction techniques. An independent bias-correction method is developed here from an analysis using nighttime marine air temperatures and SST observations from the Comprehensive Ocean–Atmosphere Data Set (COADS). Because this method is independent from methods proposed by the Met Office, the differences indicate uncertainties and similarities indicate where users may have more confidence in the bias correction. The new method gives results that are broadly consistent with the latest Met Office bias estimates. However, this bias estimate has a stronger annual cycle of bias in the Northern Hemisphere in comparison with the Met Office estimate. Both estimates have midlatitude annual cycles, with the greatest bias in the cold season, and ...