NOAA: No pause in the global surface temperature

Anthony Watts prepared the ground on Monday, with an article, in which he alleged fraud and skulduggery at NOAA. At the time I wrote the response, I didn't know what he was referring to (and neither did anyone at WUWT). Peter Thorne was awake to it, however, and gave me time to prepare for the upcoming "frenzy of denial". He was coyly referring to a new Science paper about global surface temperature.

Anthony Watts has since written another article, breaking the embargo on this new paper, which was published in ScienceXpress today. He rationalised his behaviour by saying that "multiple sources have sent me a press release and advance copies of a paper", meaning maybe two people.

The new paper is by a team of scientists, mostly from NOAA's National Centers for Environmental Information (NCEI). The title is:

Possible artifacts of data biases in the recent global surface warming hiatus

The paper describes an analysis of global surface temperature trends, using much more data from land stations in particular. It also includes corrections, particularly for time-related bias in sea surface temperatures. The main finding is that the trend in surface temperature this century is indistinguishable from that for the second half of last century. Global warming continues at the surface.

Update: I've added more links to blog articles elsewhere about this new paper.

While scientists are cautious, deniers throw caution to the wind

While the scientists themselves are suitably cautious (as seen in the paper's title), Anthony Watts is not. He seems keen to tempt NOAA into suing him for defamation. This time Anthony doesn't accuse the NOAA of fraud, he accuses it of lying, writing it is "the most mendacious attempt yet":

Tune in here tomorrow at 2PM EDT (11AM PDT) and you’ll see why this is the most mendacious attempt yet to save their climate science from the terrible ravages of an uncooperative planet.

Anthony, who can't get his head around anomalies from a baseline, claims that "WUWT has already found the fatal weakness in the paper". Notice that he stops short of claiming that it was he, Anthony, who found a fatal weakness. He probably didn't understand the paper, if he managed to read it. From the comments Anthony passed the buck to Perennially Puzzled Bob Tisdale. So you can expect a tedious, verbose, and dense article with lots of irrelevant charts of sea surface temperature in various oceans, mixed up with El Niño, sunlight-fueled water, complaints that climate models aren't weather forecasts, and greenhouse effect denial.

I'll be writing a separate article (probably two) on the "frenzy of denial" about the paper. This one is about the paper itself.

The surface keeps getting hotter, warming at 0.116°C a decade so far this century

The new paper has been published in ScienceXpress today. What it demonstrates is that the so-called "hiatus" in surface warming didn't exist, if you didn't already know that. The authors state:

...there is no discernable (statistical or otherwise) decrease in the rate of warming between the second half of the 20th century and the first 15 years of the 21st century.

This research puts the rate of warming from 1950 to 1999 at 0.113°C decade⁻¹ and the rate of warming from 2000 to 2014 at 0.116°C decade⁻¹. The difference between the two periods is not distinguishable, statistically.

Below is Figure 1 from the paper, which shows global surface temperature trends (°C per decade) for several periods, comparing the old analysis with the new analysis. The “old” analysis refers to the analysis based on ERSST version 3b for ocean areas and GHCN–Monthly version 3 for land areas. The "new" analysis is that described in this paper - using ERSST v4 and an expanded land dataset incorporating GHCN plus ISTI data - as described further down.

The new analysis is marked by squares, and the old analysis is marked by circles. The triangles in the global panel are the new analysis with polar interpolation included. (Click to enlarge it.)

Fig. 1. Effect of new analysis on global surface temperature trends for several periods. Temperature trends are shown for data with the “new” analysis (squares) and “old” analysis (circles) for several periods of interest. Also indicated are global values calculated with the new corrections and the polar interpolation method (triangles). Consistent with IPCC (1), the error bars represent the 90% confidence intervals (CIs). The additional error associated with uncertainty of our corrections extends the 90% CI and is depicted with a horizontal dash. (A) and (B) show the base period (1951–2012) and “hiatus” period used in IPCC (1). (C) An alternate base period—the second half of the 20th century. (D) The 21st century through 2014. (E) 1998 (a strong El Niño year) through the 21st century. See Table S1 for source data. Source: Karl15

As you can see, most of the difference in the trend between the old (circles) and new (squares) analysis is in the recent sea surface temperature (SST) data, and mainly for the period since 1998. The land surface temperature data is almost unchanged - just moved up by a smidgen.

With this new analysis (squares), there is little difference between the trend this century and the trend from 1950 to 1999 in global, sea or land surface temperature trends.

Even if you start the trend at 1998, the year of the super El Niño, there is only a small difference globally between that and the trend just for the twentieth century (or the period 1950 to 1999). There is a bigger difference for land only, between the IPCC base period 1951-2012 and the IPCC's so-called "hiatus" period 1998 to 2012. But this disappears when you extend the periods slightly (and don't start in an abnormally hot year). That is, there is little difference in the trend on land between 1950-1999 and 2000 to 2014,

What you'll no doubt also notice is that the error bars are overlapping for the period 1998-2012. In other words, the difference that this new analysis brings shows two things:

• the new analysis (squares) has an almost identical trend for the 1950-1999 as for 2000 to 2014, for all surfaces, suggesting there was not any "hiatus" in recent years, and
• the error bars for the old analysis (circles) overlap with the error bars for the new analysis (squares) for all periods for all surfaces. The old analysis has larger error bars than the new. 

Corrections reduce the rate of warming, not increase it!

There is more. Figure 2 from the paper shows what's been discussed here before. Applying the corrections to the sea surface temperature data reduces, not increases, the rate of warming over the instrumental period. This is the opposite to what deniers often claim - that all adjustments increase warming!

Fig. 2. Global (land and ocean) surface temperature anomaly time series with new analysis, old analysis, and with and without time-dependent bias corrections. (A) The new analysis (solid black) compared to the old analysis (red). (B) The new analysis (solid black) versus no corrections for time-dependent biases (cyan). Source: Karl15

The authors say that "...the long-term trend would be significantly higher (by 0.085°C a decade; Fig. 2B) without corrections for other historical biases..."

More data plus corrections to sea surface temperature

The main things this new version, version 4, of ERSST brings are more data plus corrections and fine-tuning of the sea surface temperature. The authors indicate (my paras):

Changes of particular importance include:

1. an increasing amount of ocean data from buoys, which are slightly different than data from ships;
2. an increasing amount of ship data from engine intake thermometers, which are slightly different than data from bucket sea-water temperatures; and
3. a large increase in land-station data that enables better analysis of key regions that may be warming faster or slower than the global average.

The paper goes into some detail explaining these changes and point out that none of those corrections were included in the analysis used in the recent IPCC report.


i Buoy data vs ship data

When comparing measurements from buoys and ships located at the same place, the ship measurements were systematically warmer than the buoy data. (This has been found in other studies, not just this one.) Averaged globally, the difference was 0.12°C. Because there were no buoys before 1970, the correction of +0.12°C was applied to buoy data at every grid cell. (It couldn't reasonably be applied to ship data prior to 1970 and with anomalies, it makes no difference to the trend whichever way it was done.)


ii Bucket vs engine intake thermometers

Just before World War II, there was a big change in the way that ships took observations of sea surface temperature. It used to be by buckets, and that changed to using thermometers in engine intake. However with improved metadata it became apparent that some ships even now take measurements by bucket. So ERSST v4 has made corrections for this, based on comparisons with night marine temperature. The authors state:

Of the 11 improvements in ERSST version 4 (13), the continuation of the ship correction had the largest impact on trends for the 2000-2014 time period, accounting for 0.030°C of the 0.064°C trend difference with version 3b. (The buoy offset correction contributed 0.014°C dec⁻¹ to the difference, and the additional weight given to the buoys because of their greater accuracy contributed 0.012°C dec⁻¹. ..) 

iii More land-station data (and Arctic warming)

This study uses data from the International Surface Temperature Initiative (ISTI). This databank integrates the Global Historical Climatology Network (GHCN) Daily dataset with more than 40 other historical data sources, providing more than twice as many stations in total. The authors say that this improves spatial coverage of land temperature data, including the Arctic (which is warming very quickly).

The authors used the same methods of data processing and corrections as are used for GHCN monthly version 3. That is for "quality control, time-dependent bias corrections, and other data processing steps (21) ... to address artificial shifts in the data caused by changes in station location, temperature instrumentation, observing practice, urbanization, siting conditions, etc.".

They merged the expanded land temperature dataset with the sea surface temperature (ERSST v4) to provide a new global land and sea surface temperature analysis.

The authors also refer to Cowtan and Way (2014), writing how incomplete coverage over the Arctic led to an underestimate of recent warming (since 1997), and that they address that issue in this paper, too.

A year (or two) makes a difference

The authors point out that adding the extra two years of temperature data (2013 and 2014) underscores the fact that the so-called "hiatus" was too short to say that "global warming has slowed" (it never stopped). Added to that, 1998 was an unusually warm year because of the super El Niño. Adding another couple of years increases the trend. From the paper (my emphasis):

Specifically, the central trend estimate in our new analysis for 1998–2014 is 0.020°C dec⁻¹ higher compared to 1998–2012. Likewise, global trends for 2000–2014 are 0.030°C dec⁻¹ higher than for 1998–2012. In other words, changing the start and end date by two years does in fact have a notable impact on the assessment of the rate of warming, but less compared to the impact of new time-dependent bias corrections. 

What the researchers are saying is that the impact of correcting sea surface temperature data for time-dependent bias is greater than that from adding temperature data for 2014.

There is no "hiatus"

In concluding, the authors say that:

In summary, newly corrected and updated global surface temperature data from NOAA’s NCEI do not support the notion of a global warming “hiatus.” 

They make the following points, my words though, not theirs:

• There isn't any discernible decrease between the rate of warming from 1950 to 1999 (0.113°C decade⁻¹) and the rate of warming from 2000 to 2014 (0.116°C decade⁻¹)
• Even if you cherry pick a start date of abnormally high temperature, the super El Niño in 1998, the rate of warming from 1998 to 2014 is 0.106°C decade⁻¹.
• Better coverage of the Arctic would probably add to the trend, not decrease it.

About ERSST v4

The new ERSST v4 dataset might still be in beta, I'm not sure that it's been made publicly available yet. A scientist from NOAA told me that it is a considerable improvement on ERSST v3b, and that "These improvements result from using the latest and more observational data, and more reasonable internal parameters based on validation tests."

I was also told what would be obvious to most people, that "the robustness of ERSSTv4 need to be further tested or confirmed by the users and scientists in the scientific community. Only the users and scientists have the final conclusion on the robustness".

There is a paper listed below (Huang15) that discusses the changes in ERSST v4 in considerable detail, and compares ERSST v4 with HadSST v3. Both ERSST and HadSST have corrected for some of the same things (like the shift away from buckets), but not in the same manner. As written in Karl15 in relation to the "bucket" bias, in ERSST v4, the correction was applied to the buoy SSTs at every grid cell, whereas:

[Notably, IPCC (1) used a global analysis from the UK Met Office that found the same average ship-buoy difference globally, although the corrections in that analysis were constrained by differences observed within each ocean basin (18).]

The different treatments would go some way to explaining the differences in this chart from an article at SkepticalScience.com in January this year. It's a comparison of HadSST v3 with ERSST v4 over map cells where both include values.

Figure 4: Comparison of HadSST3 and ERSSTv4 over map cells where both include values.
Source: SkepticalScience.com Credit: Kevin Cowtan

After contacting the NOAA authors, I got in touch with Kevin Cowtan, who wrote the SkS article, and authored the coverage bias paper (see references below). He told me (alluding to the spike in sea surface temperatures around 1940) that: "Most of the change in the new NOAA record comes from the sea surface temperatures. I initially couldn't explain the 21st century warming in ERSSTv4, but having talked to the authors I now think the results are at least plausible. I think more work is required on the period around world war 2 and prior to 1900 though. It's a challenging problem."

From what I've seen of Kevin Cowtan's work he is meticulous to a fault, maybe as concerned to get things right as are the NOAA scientists. So I take his comment as a thumbs up for ERSST 4, or as "thumbs up" as you're likely to get for any dataset :D

Zeke Hausfather supported what the NOAA scientists said about ERSST, and commented: "The major changes involve extending bucket adjustments past 1941, and adjust for the recent widespread transition from ship engine intake measurements to buoys. Our research has shown that both of these factors will bias the ocean temperature record if not addressed, and the lack of corrections in the old ERSST v3 temperature series has long been seen as a problem."

References and further reading

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." http://www.sciencemag.org/content/early/recent 4 June 2015. doi:10.1126/science.aaa5632 (subs req'd)

The Recent Global Surface Warming Hiatus - NOAA press release (with an unfortunate "seepage" title)

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: http://dx.doi.org/10.1175/JCLI-D-14-00006.1 (subs req'd)

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: http://dx.doi.org/10.1175/2011BAMS3124.1 (open access)

Kevin Cowtan, Robert G. Way. "Coverage bias in the HadCRUT4 temperature series and its impact on recent temperature trends." Q. J. Roy. Met. Soc. 140, 1935–1944 (2014). doi:10.1002/qj.2297 (open access)


Related websites

• International Surface Temperature Initiative website
• Extended Reconstructed Sea Surface Temperature (ERSST) website (currently v3b)
• Cowtan and Way: Coverage bias in the HadCRUT4 temperature record website

From HotWhopper and elsewhere

• Anthony Watts alleges fraud by the NOAA - June 2015
• On Seeps and SCAMS Part I: Lessons for Climate Scientists - (on the myth of the "hiatus") - May 2015 - 
• Slowdown Skeptic  - by Tamino - April 2015
• Recent global warming trends: significant or paused or what? - by Stefan Rahmstorf at RealClimate.org
• NOAA temperature record updates and the ‘hiatus’ - new article at realclimate.org by Gavin Schmidt, about this NOAA paper (GISTemp perspective)
• New Research On Global Warming Hiatus - Greg Laden on this NOAA paper
• On the existence of the hiatus - Doug McNeall (UK perspective)
• New research suggests global warming is accelerating - John Abraham in the UK Guardian
• Puzzling global warming 'pause' was illusion - Jeff Tollefson at Nature