Thompson et al 2008, writing in Nature, assure their readers,
the data before ~1940 and after the mid-1960s are not expected to require further corrections for changes from uninsulated bucket to engine room intake measurements
Is there a shred of evidence to support this assertion? There is convincing evidence otherwise – evidence already reported here. While Thompson et al do confirm some Climate Audit observations, on essential points, their analysis is actually a step backwards from my 2007 posts.
The hypothesis of the original Hadley Center Windowed Marine De-trending program was that there was an approximate 0.3 deg C inhomogeneity between engine inlet SST measurements and canvas bucket measurements and that there was a drop-dead changeover on December 1941, a switch which continued in place to the present day. In earlier posts, I showed that there was strong documentary evidence against this assumption and hypothesized that there was a return to “business as usual” after the war. That’s not the only relevant information on the transition, as I’ll show below.
Thompson et al 2008 agree that there was a return to “business as usual” after the war, citing related but somewhat different evidence than presented here: they observed that wartime measurements were predominantly U.S., which they say were engine inlet, while U.K. measurements come back into play after the war, using a ~0.3 deg C estimate. They observe:
The Met Office Hadley Centre is currently assessing the adjustments required to compensate for the step in 1945 and subsequent changes in the SST observing network. The adjustments immediately after 1945 are expected to be as large as those made to the pre-war data (~0.3 deg C; Fig. 4).
This was also the conclusion in the prior Climate Audit post and is fair enough as a first estimate. They go on to say:
smaller adjustments are likely to be required in SSTs through at least the mid-1960s, by which time the observing fleet was relatively diverse and less susceptible to changes in the data supply from a single country of origin …
the data before ~1940 and after the mid-1960s are not expected to require further corrections for changes from uninsulated bucket to engine room intake measurements.
They’ve worded their comment on the early bucket adjustments carefully, as there’s lots of hair on these early adjustments and these adjustments need to be minutely scrutinized. But on the post-1960s period, they have completely lost their bearings and are, so to speak, lost at sea.
Thompson et al 2008 cited Kent et al 2007, an important discussion of metadata, but they completely failed to discuss or cite the most relevant graphic in Kent et al – a graphic previously reproduced at Climate Audit on a number of occasions – and reproduced one more time below. This graphic, based on a very comprehensive examination of metadata, showed the distribution of measurement type from 1970 to 2006.
Figure 2f from Kent et al 2007.
In 1970, as I observed last year, about 90% (this is a visual estimate from the graphic) of SST measurements, for which the type is known, were done by buckets. Because the proportion with metadata is a very large sample, it’s plausible to use this 90% estimate for the entire population, including the unknown population.
Between 1970 and 2006, the proportion of bucket and engine inlet measurements is more or less reversed, with about 90% of SST measurements in the 2000s being engine inlet or hull sensor, the latter by the way, being a further addition to the witches’ brew that the Nature boys didn’t mention at all. The starting point of all this was that there is about a ~0.3 deg C bias between engine inlet and buckets.
However, Thompson et al 2008 completely failed to grasp the significance of this graphic. The changeover to engine inlet measurements, previously attributed to a drop-dead date in 1941, actually took place AFTER 1970 (providing, of course, for a one-off WW2 adjustment ending in 1945). If the same ~0.3 deg C consistently used by Hadley Center is applied after 1970, as this information shows, this comes off the post-1970 SST trend (and has to be allocated much earlier, as proposed last year at Climate Audit, ) refuting the claims of Thompson et al that no substantial changes are required to the post-1965 record, a point that should be obvious to anyone thinking for 5 minutes about the problem.
Thompson et al 2008 observe that the 0.3 deg adjustment looms relatively large in 20th century terms. They observe:
thus the amplitude of the drop is roughly 40% as large as the 0.75 deg C rise in [global temperature] from 1900 to 2006,
If, as outlined here, this 0.3 deg C adjustment has to come off the post-1970 record, as implied by the information at hand, it is a very large proportion of the post-1970 temperature increase, which is much reduced and allocated earlier in the century. Because the effect is so large relative to observed changes, the knock-on impact for attribution and modeling will not be small – whatever way it goes.
One hopes that this will also lead to an end to CRU secrecy on their source code, algorithms and data versions.
[UPDATE (May 30):
A reader has contacted me to say that buckets in the 1970s were predominantly insulated buckets not uninsulated buckets and that the differential between insulated buckets and engine inlets is less than between uninsulated buckets and engine inlets (say 0.1 deg C, versus 0.25-0.3 deg C). So there may be a couple of things going on in bucket world - a change from buckets to engine inlets and a change from uninsulated buckets to insulated buckets. The latter possibility was not clearly articulated in Thompson et al, or for that matter in the predecessor articles, but may nonetheless be a real effect. IF such transition were complete by the 1970s, then this would contain adjustments in the 1980s to ones resulting from differences between insulated buckets and engine inlets, which would be less than between uninsulated buckets and engine inlets. I'll take a look at this. I'm going to look for discussion of the transition from uninsulated buckets now said to have been in use after WW2 to insulated buckets. This episode definitely confirms my very first point on these bucket adjustments: whenever the adjustments are as as large the effect being measured, then there needs to be a replicable description and careful assessment of all aspects of the adjustment process.]
Folland, C. K., D. E. Parker, and F. E. Kates. 1984. Worldwide marine temperature fluctuations 1856–1981. Nature 310, no. 5979: 670-673.
Kent, E. C., S. D. Woodruff, and D. I. Berry. 2007. Metadata from WMO Publication No. 47 and an Assessment of Voluntary Observing Ship Observation Heights in ICOADS. Journal of Atmospheric and Oceanic Technology 24, no. 2: 214-234.
Parker, D. E., C. K. Folland, and M. Jackson. 1995. Marine surface temperature: Observed variations and data requirements. Climatic Change 31, no. 2: 559-600.