As CA readers know, Phil Jones keeps his CRU data secret. Embarrassingly, the UK Met Office relies on this secret data and says that it is unable to provide this supporting data for the most relied upon temperature data set in the world. Their statements in response to FOI requests as to what they actually hold seem contradictory, but most recently they state that they do not hold original data, but only the “value added version” provided to them by Phil Jones. Whether they are entitled to keep the “value added version” secret is something that their FOI officer is presently considering.
Recently, Anthony Watts discovered that the Honolulu Observatory data, which NOAA and NASA lost track of in the 1980s, continued to the present day.
Anthony observed the substantial difference between trends at Honolulu airport and at more rural sites.
When I’ve done previous benchmarking of GISS data, I’ve usually tried to use relatively isolated stations so that the effect of data inclusions and exclusions could be simplified. Since Hawaii is relatively isolated, it seemed like it would be an interesting exercise to look at the Hawaiian gridcell, to get a preview of whether the “discovery” of a long data set might have an impact at the gridded level.
As so often, when one goes down a climate science rabbit hole, wonderland awaits.
First, here’s a graphic showing an interesting contrast between the CRUTEM gridded data and the NOAA/GHCN gridded data for the Hawaii gridcell (157.5W, 22.5N). In one of his FOI obstructions, Phil Jones argued that CRUTEM data was already available at GHCN. But as you see, the CRUTEM gridded version for the Hawaii gridcell is remarkably different from the NOAA GHCN version.
To illustrate the stunning difference between the two series, here’s a plot of the difference. CRU has increased relative to GHCN by approximately 2 deg C during the 20th century. ([snark] I guess that’s what the Met Office means by the “value added” product/ [/snark])
Reconciling these sorts of differences is quite time consuming and one would think that this would be the sort of thing that the temperature index providers would spend time doing. But unfortunately they seem to use funding for temperature indices to support other activities that they enjoy more and spend negligible time on QC – Gavin Schmidt justified the lack of QC on the basis that is unreasonable to expect NASA to be able to keep track of such things as they only allocate about 0.25 man-years annually to quality control.
I took a quick look at features of the underlying data to get some ideas. However, this sort of reconciliation is detailed and NOAA/NASA should either do it themselves or contract someone to do it for them.
A while ago, after a number of FOI requests, we managed to get a list of Phil Jones’ stations (online here.) Honolulu Observatory is not on the list of CRU stations within 5 degrees of Honolulu.
country name id lat long alt
2733 HAWAII LIHUE WSO A 911650 22.0 -159.4 31
2734 USA KANEOHE BAY/MCAS 911761 21.5 -157.8 3
2735 USA BARBERS POINT/NAS 911782 21.3 -158.1 15
2736 HAWAII HONOLULU WSFO A 911820 21.3 -157.9 2
2737 HAWAII MOLOKAI A 911860 21.2 -157.1 137
2738 HAWAII LAHAINA 911897 20.9 -156.7 14
2739 USA HI*KAHULUI WSO (PUU 911900 20.9 -156.4 20
2740 HAWAII LANAI CITY 911905 20.8 -156.9 494
2748 HAWAII HILO WSO A 912850 19.7 -155.1 8
2749 HAWAII HILO HAWAII 912857 19.7 -155.1 12
All or virtually all of the Jones sites are airports. Six of 10 sites have name evidence of being an airport and are classified by GISS/GHCN as airports: Lihue WSO, Kaneohe Bay MCAS, Honolulu WSFO, Molokai A(irport), Kahului WSO and Hilo WSO. Barbers Point NAS (Naval Air Station) and Lanai City (see NWS here) seem to be airports, but are misclassified by GISS. (Gavin, are you on the job? Or did Harry exhaust you?) Hilo 912857 might be a version of Hilo WSO 912850, but the ID doesn’t tie directly to anything at GHCN. That leaves Lahaina as the only candidate for a non-airport site. Honolulu Observatory was not on the CRU list.
Many of these sites are lost to NASA/NOAA. To the extent that Jones statement that he uses GHCN is true, this means that they are also lost to CRU. Early ends occur, for example, in GISS data for Molokai Airport (ends 1981), Lanai City (ends 1970) and Lahaina (ends 1980). Again, these sites do not appear to be really “lost”. Lanai City weather is online here and Lahaina here.
GISS sites in Hawaii classified by them as R-sites are Lihue (current), Molokai (ends 1981), Lanai (ends 1970), Lahaina (ends 1980), Hana (ends 1980), Kualapu (ends 1954), Puunene (1950s only) and Barking Sands – a military facility ?? – (ends 2003 and is erratically available). Honolulu Observatory is classed U.
As far as I can tell, the only current CRU sites in their Hawaii gridcell are Honolulu Airport, Lihue Airport and Hilo Airport.
But why is there such a difference between Phil Jones’ “value added” version and the NOAA GHCN version? Dunno.
It’s a secret.
Update: A further observation on the construction of these series. Both the CRU and GHCN versions prior to 1905 are based entirely on the series the current avatar of which is at Honolulu airport. GHCN applies an adjustment, presumably to reflect UHI (though I don’t for sure what they do.) It appears that Jones does no adjustment given the similarity between portions of his gridded series and unadjusted Honolulu airport.