Sea surface temperatures (SSTs) and temperature measurements from terrestrial weather stations are used to detect global change and evaluate both natural and anthropogenic influences on Earth’s climate. Historical records of SSTs extend back to the mid-19th century, but the technology used to make these measurements has varied significantly over the past 150 years. Such changes in instrumentation can result in systematic errors comparable in magnitude to the climatic variations observed during the same period, so these inaccuracies must be corrected, and the uncertainties must be quantified, for scientists to isolate actual climate trends.
Despite earlier attempts to identify and correct systematic uncertainties in historical SST records, such as discrepancies between measurements made by ships versus buoys, significant differences still exist between key data sets. These differences are especially apparent for the time period around World War II, which coincided with a protracted El Niño, variations in international shipping patterns, and poorly documented changes in how SST measurements were made.
Now Kennedy et al. present a major update to the U.K. Met Office Hadley Centre SST (HadSST), a crucial climate data set that documents changes in sea surface temperatures from 1850 to 2018. In this fourth version of HadSST, the researchers revisit and improve upon previous data adjustments and quantify remaining uncertainties.
The team incorporated several innovations in its recent work, including an updated ensemble approach for estimating biases between different SST measurement methods (for example, using an insulated versus uninsulated bucket to sample water), which helped quantify the range of uncertainty of the data adjustments.
The results affect the estimation of biases throughout the SST record, especially in the post–World War II period, although they do not change the conclusion that SSTs show a clear trend of long-term global warming. Compared to the prior version of HadSST, the newly estimated bias from 1950 to 1970 is substantially lower, for example, and the correction is more consistent with other SST analyses. In the modern era, the latest version of the data set agrees well with independent satellite and Argo float observations. The new adjustments also highlight an early 1990s discrepancy, which has yet to be resolved, between the Met Office’s data set and the National Oceanic and Atmospheric Administration’s Extended Reconstructed Sea Surface Temperature version 5 (ERSSTv5) data set, another key source of SST observations.
The update to HadSST improves our understanding of historical SST measurements and their associated uncertainties and reduces systematic errors in the data. It thus provides an enhanced data set that scientists can use to track one of the most important variables for ascertaining the past and present state of Earth’s climate system—and for making predictions about its future. (Journal of Geophysical Research: Atmospheres, https://doi.org/10.1029/2018JD029867, 2019)
—Terri Cook, Freelance Writer