More Curiosities about NOAA’s New “Pause Busting” Sea Surface Temperature Dataset

UPDATE 2: KNMI added the HadNMAT2 data to their Climate Explorer, so we no longer have to rely on my replication of data from a graph. See the update before the closing.
# # #
UPDATE: I was just informed by the UKMO via email and by blogger Andrew on this thread that the UKMO has provided links to the HadNMAT2 data on their webpage here. For those of you with programming skills, have at it. Thank you, Met Office.

# # #

The Night Marine Air Temperature dataset HadNMAT2 from the UKMO is used for bias adjustments of the new NOAA ERSST.v4 “pause-buster” sea surface temperature data over nearly its full term, from 1875 to 2010. But the UKMO HadNMAT2 data are not available online so that the public can easily verify the NOAA ERSST.v4 results. That’s small fish compared to an even bigger problem for NOAA. A preliminary investigation of the UKMO dataset suggests that the HadNMAT2 data do not support NOAA’s claims of no slowdown in global surface warming. In other words, the HadNMAT2 data have a much lower warming rate than the new NOAA “pause buster” ERSST.v4 data since 1998.

INTRODUCTION

In the post NOAA/NCDC’s new ‘pause-buster’ paper: a laughable attempt to create warming by adjusting past data, we discussed the new paper about NOAA’s latest revisions to their global surface temperature dataset. That paper was Karl et al (2015) Possible artifacts of data biases in the recent global surface warming hiatus. The changes to the sea surface temperature component of the new NOAA/NCEI global land+ocean surface temperature dataset were the biggest contributors to claims that the new data show no hiatus or slowdown in global warming since 1998. NOAA’s new extended reconstructed sea surface temperature dataset is called ERSST.v4. Those adjustments and the fact that the oceans cover about 70% of Earth’s surface made the new sea surface temperature data the governing factor in NOAA’s new claims that there was no slowdown in surface warming.

One of the oddities of the new NOAA sea surface temperature dataset shown in our earlier post was that the warming rate of the sea surface temperature portion of NOAA’s new data was an outlier since 1998…that it had a much higher warming rate than all other sea surface temperature datasets during the recent slowdown in global surface warming. In other words, there were no other sea surface temperature datasets that supported the high warming trend of the new NOAA data. See Figures 6 through 9 from the earlier post about Karl et al.

What I did not compare to NOAA’s new sea surface temperature dataset in that post was the temperature dataset that served as the reference for bias adjustments over the full term of the data, and that reference dataset was the HadNMAT2 Night Marine Air Temperature data from the UK Met Office. My preliminary investigation reveals that the reference HadNMAT2 data also do not support the excessive warming rate since 1998 of NOAA’s new sea surface temperature data. And that’s a bad sign…a really bad sign.

BACKGROUND

The two papers that present NOAA’s new Extended Reconstructed Sea Surface Temperature dataset ERSST.v4 are (both are paywalled):

• Huang et al. (2014) Extended Reconstructed Sea Surface Temperature version 4 (ERSST.v4), Part I. Upgrades and Intercomparisons, and
• Liu et al. (2014) Extended Reconstructed Sea Surface Temperature version 4 (ERSST.v4): Part II. Parametric and Structural Uncertainty Estimations.

We discussed the new NOAA ERSST.v4 data in a few posts last year, including Has NOAA Once Again Tried to Adjust Data to Match Climate Models?

Like their earlier ERSST.v3b dataset, the new ERSST.v4 were bias adjusted by a marine air temperature dataset. But NOAA added a new feature this time through. For the bias adjustments in the earlier ERSST.v3b dataset, NOAA used a night marine air temperature dataset that was part of the ICOADS Release 2.4 (R2.4) to adjust the sea surface temperature for the period of 1875 to 1941. On the other hand, two of the features of the new NOAA ERSST.v4 data was that (1) NOAA used a newer and improved night marine air temperature dataset HadNMAT2 from the UKMO and (2) NOAA extended its use through to 2010. In other words, the HadNMAT2 data were used for bias adjustments for nearly the full term of the new ERSST.v4 data from NOAA. See my Table 1, which is Table 1 from Huang et al. (2014). I’ve highlighted the relevant portion.

Table 1

As they write in Huang et al. (2014) (my boldface):

Firstly, ERSST.v3b does not provide SST bias adjustment after 1941 whereas subsequent analyses (e.g. Thompson et al. 2008) have highlighted potential post-1941 data issues and some newer datasets have addressed these issues (Kennedy et al. 2011; Hirahara et al. 2014). The latest release of Hadley NMAT version 2 (HadNMAT2) from 1856 to 2010 (Kent et al. 2013) provided better quality controlled NMAT, which includes adjustments for increased ship deck height, removal of artifacts, and increased spatial coverage due to added records. These NMAT data are better suited to identifying SST biases in ERSST, and therefore the bias adjustments in ERSST version 4 (ERSST.v4) have been estimated throughout the period of record instead of exclusively to account for pre-1941 biases as in v3b.

Oddly, by using the HadNMAT2 data for bias adjustment, NOAA did not address the post-1941 problem presented in their referenced Thompson et al. (2008), which was the discontinuity in 1945. NOAA seems to have exaggerated that problem in their new ERSST.v4 data. See the earlier post here, under the heading of YOU MAY BE WONDERING…

Considering that the HadNMAT2 data played a key role in the creation of the ERSST.v4 data, it would be logical to compare the new NOAA ERSST.v4 data to the UKMO HadNMAT2 data to see how well those two datasets agree…especially during the global surface warming slowdown period from 1998 to present, which was one of the highlighted periods in Karl et al. (2015) that served as the bases for the odd claims made in that paper.

PROBLEM ONE (HAS BEEN RESOLVED BY THE UK MET OFFICE)

The HadNMAT2 data are not yet available to the public online, even though the paper that supports it was published in February 2013. And that paper is Kent et al. (2013) Global analysis of night marine air temperature and its uncertainty since 1880: The HadNMAT2 data set. I emailed the UKMO to determine if the HadNMAT2 data were available somewhere online, and I was advised that they were still trying to decide on the format in which to publish it, the length of time for the decision being unusual for them.

It’s been 2 years. I suspect the UKMO has hesitated in publishing the HadNMAT2 data, because they contradict their HADSST3 data during the period from the early-1940s to the mid-1970s. Recall that in HADSST3 the UKMO eliminated the Thompson et al (2008) “discontinuity” around 1945 and made numerous other adjustments during the 1950s, 60s and 70s. Those changes produced a slight cooling trend from the mid-1940s to the mid-1970s. The HadNMAT2 data have not received those corrections.

The fact that the HadNMAT2 data are not available online, of course, presents problems for the three NOAA papers relating to the ERSST.v4 data: Huang et al. (2015), Liu et al. (2015) and Karl et al. (2015). The results of those papers cannot be verified by the public, because one of the key reference datasets for those three papers has not been published. I wonder if the editors of those publications know that sad fact. I suspect they might be informed in the not-too-distant future.

PROBLEM TWO

The global UKMO HadNMAT2 data are presented in Figure 18 of Kent et al. (2013), alongside their HADSST3 data. See my Figure 1. The global HadNMAT2 data are shown as the black curve. The data in the graph run from 1880 to 2010 (with what appears to be a slight downturn in 2011 from an incomplete year of data).

Figure 1

(A larger version of the graph is here.)

While there is software available that will extract data from graphs, I prefer to use the x-y coordinates of MS Paint to replicate the data. My results for the HadNMAT2 data for the period of 1998 to 2010 are shown in Figure 2, compared to the global ERSST.v4 data used by Karl et al. (2015) for their claims of “no hiatus”.

Figure 2

Of course, Karl et al. presented their trends for the period of 1998 to 2012 and from 2000 to 2014. But I know of no source of the HadNMAT2 data for the years of 2011 to 2014. So this is preliminary comparison.

These preliminary results strongly suggest that even the (HadNMAT2) data used as the reference for bias adjustments in the new NOAA sea surface temperature dataset (ERSST.v4) do not support that ERSST.v4 data…or the claims in Karl et al (2015) of no slowdown in global surface warming.

Not too curiously, as far as I can tell, NOAA failed to present those differences in their papers. While NOAA may now try to justify those differences in the short-term warming rates of ERSST.v4 and HadNMAT2, they look very awkward, and any excuses NOAA gives now will simply be viewed as that…excuses.

UPDATE 2

Yesterday the UKMO published their HadNMAT2 data, and today KNMI added it to their Climate Explorer. (Scroll down to “Air Temperature”.) Thank you very much to the hard-working individuals at KNMI and UKMO who took time from their weekends.

The monthly HadNMAT2 data run from 1880 to 2010.

Figure 2 above was based on global data. It is biased by the differences in how the two datasets handle sea ice. So I’ve updated the graph in Figure 3 for the period of 1998 to 2010, using the monthly data for the latitudes of 60S-60N, the non-polar oceans.

Figure 3

NOAA’s new “pause busting” sea surface temperature data still doubles the warming rate of the HadNMAT2 data that served as the bias-adjustment reference for the new NOAA data. And as a note, the HadNMAT2 trend for 1998 to 2010 of 0.045 deg C/Decade is in line with the UKMO HADSST3 data at 0.047 deg C/Decade (not shown).

NOAA, you got some ‘splainin’ to do.

CLOSING

As noted earlier, the preliminary investigation that shows the HadNMAT2 data do not support the claims of no hiatus is a bad sign for the results of Karl et al. (2015), a very bad sign.

I suspect that the editors of the journals that published the three ERSST.v4-based papers (Huang et al. (2015), Liu et al. (2015) and Karl et al. (2015)) will soon be informed of this problem as well.

When the HadNMAT2 data are finally published online by the UKMO, that reference data for the NOAA ERSST.v4 data will very likely put NOAA and the publishers of the Huang et al. (2015), Liu et al. (2015) and Karl et al. (2015) papers in very awkward positions. Time will tell.