Scroll To Top

Thursday, January 5, 2017

The winner is NOAA - for global sea surface temperature

Sou | 6:00 AM Go to the first of 8 comments. Add a comment

There's a new paper out that shows that, contrary to what you'll read on denier blogs, NOAA's latest version of global sea surface temperature is probably the best and most accurate around. It's the closest to observations, when you compare it to measurements from moored and floating buoys, Argo floats and radiometer-based satellite records of sea surface temperature.

Umpteen denier protests

Lamar Smith
You might remember how climate hoax conspiracy theorists, professional disinformers and other deniers protested loud and long when NOAA scientists published a paper about the revised NOAA temperature data. The US House Committee on Science, Space, and Technology, led by arch denier Lamar Smith, harassed NOAA endlessly with subpoena after subpoena. A lot of the changes to the NOAA temperature record were a result of a new version of the global sea surface temperature data set, known as Extended Reconstructed Sea Surface Temperature, or ERSST v4. The papers on that were published in February 2015 (see below). The protests only came, though, in June 2015 when there was a paper by Karl et al. That paper pushed denier buttons because it challenged the so-called "hiatus". You can read about Karl15 here, and the paper itself is here.

Deniers tried every trick in the book to demonise the paper. Some claimed (wrongly) that there was "something wrong" with adjustments to correct the bias from the shift from ship-based temperature readings to buoy readings. Others like Bob Tisdale thought (wrongly) that the global trend in sea surface temperature, using readings from buoys and ships, should exactly match the trend just using night time marine air temperature. Anthony Watts went as far as accusing the NOAA scientists of fraud, which didn't go down well with Andy Revkin of the New York Times. Other deniers didn't understand what was being discussed, they just knew that if scientists had a paper published in Science then it had to be a scientific conspiracy, and wrong.

The odd part about all this is that the rate of warming over the full NOAA record as reported in Karl15 in the new version was lower, not higher, than the uncorrected version. As well as that, the NOAA global mean surface temperature (land and sea) has a trend that is middle of the road compared to other data sets. The table below is a bit out of date - it's the linear trends per decade from 1970 to the end of 2015, as reported here.

Deniers are never satisfied - "something must be wrong" with everything science. I expect this study won't satisfy them either. I wonder how long it will take for WUWT to write about it, and what form their protests will take. (Usually they love satellites and Argo floats. I suspect they'll suspend that adoration for the duration.)

New analysis shows that the NOAA sea surface temperature record is the best of the lot

In a new paper published in Science Advances today, a team of scientists led by Zeke Hausfather analysed the different sea surface temperature records, comparing them with observations from buoys, Argo floats, and radiometer-based satellite measurements, which measure the temperature of the sea "skin" or surface. You might recognise some of the other members of the team too. They are: Kevin Cowtan, David C. Clarke, Peter Jacobs, Mark Richardson and Robert Rohde.

Argo floats. Source: JAMSTEC
What this team found was that the latest NOAA sea surface temperature record, ERSST v4 was superior to the other three records examined: the previous NOAA version ERSST v3b, HadSST3 from the UK Met Office Hadley Centre, and a record I hadn't come across before, COBE-SST. That last one is from the Japan Meteorological Agency (JMA).

Using observations from moored and floating buoys, Argo floats and satellites, the authors found that the old NOAA sea surface temperature record (ERSST version 3b) was way too cool. They also found that the sea surface temperatures reported by the Hadley Centre (UK Met Office) and Japan Meteorological Agency were also too cool, though not as off as NOAA's ERSST version 3b. They found that the new NOAA record (ERSST version 4) agreed rather well with the instrumental records.

Putting together a long-term sea surface temperature record is very complicated

You're probably wondering why the difference. Well, the different organisations use slightly different observation sets when working out global sea surface temperature changes. They also have different techniques for analysing the data and putting it all together to form a composite global record. The authors explain in a background article on Kevin Cowtan's website:
Both Hadley and NOAA (as well as the Japanese COBE-SST record) are what we call composite records. That is, they try and take data from multiple different types of instruments that are changing over time and combine them in a single long-term climate record. This poses challenges when the way measurements taken change (e.g. switching from buckets thrown over the side of ships to engine room intake valves, or more recently from ships to buoys) and requires some judgment calls of how to calibrate each set of data and correct for any biases. 

The new paper is detailed and, as is usual with these authors, easy to follow and full of good information. The authors discuss the work that must be done to develop a comprehensive record of sea surface temperature, particularly because the sources have changed over time. They wrote about the shift from ship measurements to buoy measurements:
Before the past two decades, a large majority of SST measurements were taken by ships, first with buckets thrown over the side and increasingly through engine room intakes (ERIs) after 1940. Since 1990, the number of SST measurements coming from buoys has increased around 25-fold, whereas the number of observations from ships has fallen by around 25% (3, 4). The observations have gone from 80% ship-based in 1990 to 80% buoy-based in 2015. Modern ship-based measurements (primarily ERI, although hull contact sensors and other devices are also used) tend to be biased warm by around 0.12°C relative to buoys, whose sensors are directly in contact with the ocean’s surface (1, 5, 6). As the number of ships actively taking measurements available in the International Comprehensive Ocean-Atmosphere Data Set (ICOADS) database (4) has fallen, a growing portion of ships are also using non-ERI systems that may introduce further changes in the combined record (1). Although buoy records are widely considered to be more accurate than ship-based measurements, their integration with ship records into longer SST series poses a number of challenges (3).

How the new study compared observations with composite data sets

The authors of the new paper looked at relatively recent observations and compared them individually with the composite data sets. That is, they compared each data set with buoy-only data, and with satellite-only data (both from 1997 to the end of 2015). Satellite data measures the temperature of the "sea skin" - or right at the surface. Moored buoys record temperature at about three metres below the surface and floating buoys at around half a metre below the surface. They also compared them with only data from Argo floats, which is only useful from 2005 to the present. Most of the measurements from Argo floats (in this exercise) are taken at about five metres below the surface, and the authors refer to those measures as "near-SST". (Ships today take in water through engine room intake valves, which are at depths of seven to 11 metres for large ships and at one to three metres for small ships.)

The scientists selected only the period after the records were dominated by ship data (buoys are about 80% of the NOAA data over recent years), and a time span when there are sufficient data from other sources to provide reliable information around the globe. In this way, they were able to do a like for like comparison without having to make all those pesky adjustments that the longer composite records require. The composite records don't just have to adjust for differences between ships and buoys, they also have to make adjustments for different designs of ships over the years. It is a very complicated exercise.

By looking at recent records using the most modern instrumentation available, the authors were able to see how the composite records compared with these up-to-date instrumental measurements.

Below is a chart from Kevin Cowtan's website that highlights the differences between the observations from buoys and satellite data, and three different composite data sets.

As you can see, NOAA's latest version of sea surface temperature, ERSSTv4 (red line) is almost indistinguishable from the buoy-only (green line) and satellite-only data (brown line). The older version, ERSSTv3b (blue line), and the Hadley Centre version, HadSST3 (purple line), are cooler than the other measures shown.

Caveats, coverage and a video

It's worth reading the paper (it's open access), because the authors explain their study in considerable detail and talk about the differences in coverage of the different data sources at different periods. They also suggest reasons for the differences between the composite data sets (NOAA, HadSST3 and JMA's COBE-SST.

The video from Peter Sinclair (below) is worth watching too. It's only five minutes long, and has Zeke Hausfather explaining the work. If you're Australian (or British), don't be too distracted by Zeke's "boo-eys" :D


References and further reading

Zeke Hausfather, Kevin Cowtan, David C. Clarke, Peter Jacobs, Mark Richardson, and Robert Rohde. "Assessing recent warming using instrumentally homogeneous sea surface temperature records." Science Advances 3. no. 1 (2017): e1601207 (open access)

Thomas R. Karl, Anthony Arguez, Boyin Huang, Jay H. Lawrimore, James R. McMahon, Matthew J. Menne, Thomas C. Peterson, Russell S. Vose, Huai-Min Zhang. "Possible artifacts of data biases in the recent global surface warming hiatus." 4 June 2015. doi:10.1126/science.aaa5632 (pdf here)

Huang, Boyin, Viva F. Banzon, Eric Freeman, Jay Lawrimore, Wei Liu, Thomas C. Peterson, Thomas M. Smith, Peter W. Thorne, Scott D. Woodruff, and Huai-Min Zhang. "Extended Reconstructed Sea Surface Temperature version 4 (ERSST. v4), Part I. Upgrades and Intercomparisons." Journal of Climate 2014 (2014). doi: (pdf here)

Liu, Wei, Boyin Huang, Peter W. Thorne, Viva F. Banzon, Huai-Min Zhang, Eric Freeman, Jay Lawrimore, Thomas C. Peterson, Thomas M. Smith, and Scott D. Woodruff. "Extended reconstructed sea surface temperature version 4 (ERSST. v4): part II. Parametric and structural uncertainty estimations." Journal of Climate 28, no. 3 (2015): 931-951. doi: (pdf here)

Peter W. Thorne, Kate M. Willett, Rob J. Allan, Stephan Bojinski, John R. Christy, Nigel Fox, Simon Gilbert, Ian Jolliffe, John J. Kennedy, Elizabeth Kent, Albert Klein Tank, Jay Lawrimore, David E. Parker, Nick Rayner, Adrian Simmons, Lianchun Song, Peter A. Stott, and Blair Trewin, 2011: "Guiding the Creation of A Comprehensive Surface Temperature Resource for Twenty-First-Century Climate Science." Bull. Amer. Meteor. Soc., 92, ES40–ES47. doi: (open access)

From the HotWhopper archives


  1. Wait for it:

  2. Will Judith Curry's gold standard dataset for global ocean surface temperatures change from HadSST3?
    Around June 4, 2015, she made a statement to “an (international) journalist asking for comments” on Karl 2015. “I personally see no reason to the use the NOAA ERSST dataset, I do not see any evidence that the NOAA group has done anywhere near as careful a job as the UK group in processing the ocean temperatures.
    Will JC dismiss this new paper by recycling her 2015 blog comment, “Colour me unconvinced.”?
    When and if a response to Hausfather 2017 comes from JC, colour me uninterested.

    1. That's a long long way from the dumbest thing she's said. I think most climate scientists view HadSST3 as the gold standard. What this paper indicates is that ERSSTv4 deals best with the ship-buoy transition bias over the past decade or so. However, that's not a bias which affects the rest of the record. I think most still regard HadSST3 as being more reliable for historical variations and trends.

    2. What the best dataset is depends on the study. But ERSST and HadSST are at least equals.

      Error model, period around WWII and probably regional data better in HadSST. Global average (coverage) and recent data (this study) better in ERSST.

    3. That it depends on the application and that Curry is not even willing to use the data makes her comment so ... interesting.

    4. I've yet to see any evidence that Judith understands temperature datasets. Could be why Berkeley Earth ditched her. She gets other people to prepare her presentations for her, remember. I don't think she's ever produced a temperature chart, and I doubt she's read let alone understood a paper on the subject. That's based on her blog articles.

  3. Hotwhopper: "a record I hadn't come across before, COBE-SST. That last one is from the Japan Meteorological Agency (JMA)."

    For the marine sciences it was very important when "COBE collection" came online, when the Japanese data was digitized, they had a lot of historical data from the Pacific and from whaling ships in regions where we did not have much data.

    The JMA also has a global temperature record, but the land side is simply GHCN. Their focus is the marine data. That makes it less interesting for mitigation sceptics who tend to bite at the first thing they come across and ignore that what is scientifically important. The Murdoch press proably has no Japanese newspapers, I have never heard anything about Japanese mitigation sceptics. That also helps.

    1. Thanks Victor. Appreciate you adding info about it.

      I know (and use) JMA data for various things, but hadn't come across that particular data set before. Or if I had, I wasn't aware of its name. (I've probably inadvertently used part of it, or a derivative, when I've plotted the JMA PDO index data.)


Instead of commenting as "Anonymous", please comment using "Name/URL" and your name, initials or pseudonym or whatever. You can leave the "URL" box blank. This isn't mandatory. You can also sign in using your Google ID, Wordpress ID etc as indicated. NOTE: Some Wordpress users are having trouble signing in. If that's you, try signing in using Name/URL or OpenID. Details here.

Click here to read the HotWhopper comment policy.