There’s a new study by Hegerl et al. in this week’s Nature, which, among other things, describes the performance of something called the CH-blend, a secret blend of 12 proxies – a secret somewhere up there with the Caramilk secret. As I mentioned previously, I requested the identity of the sites in the blend as part of the IPCC review process and was refused and even threatened with expulsion as a reviewer if I made any further attempts to obtain the data from Hegerl et al. At the time, I was told that such considerations were the province of the journal and that it was not the job of an IPCC reviewer to carry out any additional due diligence on data.
I brought this refusal up at the NAS Panel discussed here. Hegerl explained to the NAS Panel the paper was under review by Nature, implying that showing the site locations to an IPCC reviewer would be a breach of Nature embargo. While I objected, I got the impression that, with the mention of a Nature embargo, a hush fell over the NAS panel and everyone immediately "understood".
So naturally when Hegerl et al. was finally published in Nature this week, I immediately went to see if the Caramilk secret was finally revealed.
Three other paleo reconstructions are mentioned (Mann and Jones , Briffa et al  and Esper et al  as re-calibrated in Cook et al , but most of the discussion is about the CH-blend, about which comments such as the following are made:
For CH-blend, our estimate of climate sensitivity fully accounts for the uncertainty in the amplitude of the record.
Results for the CH-blend (short) reconstruction, for which we have the most reliable uncertainty estimate, yield a 5-95% range for sensitivity of 1.4K to 6.1K and a median sensitivity of 2.6K over the preinstrumental period 1505-1850 (Figure 3a).
The resulting 5-95% ranges for CH-blend (short) shrink to 1.6K to 4.6K, those for all proxy data combined to 1.5K to 6.2K. This result reduces the probability from 36% to 15% or less that climate sensitivity exceeds the upper limit of the IPCC range of 4.5K.
So what is the CH-blend? Well, it is said to be:
our own new decadal reconstruction termed “CH-blend” of annual average 30-90°N temperature  (Figure 1). A version of CH-blend using 12 records extends from 1505 to 1960; and a reconstruction based on 9 sites (“CH-blend long”) is used from 1270. Both reconstructions use a relatively small number of well spaced sites (often based on multiple records, including some regional reconstructions).
The acknowledgements also said that "TC provided the reconstruction of past forcing and developed the CH-blend reconstruction" – that must be the Tom fellow that Hegerl talked about so much in her NAS panel presentation.
However, they didn’t actually mention what the sites were. OK, it’s not unreasonable to put lists in Supplementary Information and Hegerl et al provided a Supplementary Information here . The SI webpage tantalizingly shows a table described as follows:
This file contains our new reconstruction and its 2.5% and 5% uncertainty ranges. It contains data necessary to reproduce our result.
Aha, the Caramilk secret finally. But, no. The Table only provided a digital version of the CH-blend from 1251 to 1960. To 1960 ?!? Why only 1960? Was there a Divergence Problem?
So what are the 12 sites? Well, they are nowhere listed in the Nature article. Instead, they are attributed to: Hegerl, G. C. et al. Detection of human influence on a new, validated 1500 yr temperature reconstruction. J. Climate, submitted. (2006).
A couple of points:
(1) Journal of Climate has no rules putting an embargo on identifying data. So this excuse of a Nature embargo has been total nonsense from beginning to end.
(2) IPCC refused data to me on the basis that that was the job of journal reviewers. Here we have an example of a paper which I have been following through the process and there is not a speck of evidence that the site details were either provided to or considered by Nature reviewers.
Just for fun, I’m going to try to guess what the Hegerl et al sites are, from the limited information available and from the other guiding Hockey Team principle -maximum non-independence of data. The article says that they used 12 series from 1505 on and 9 series from 1270 on (although the SI starts in 1251). Why didn’t they use the series from 1251 on? So here are my guesses together with reasons:
(1) the van Engeln composite starts in 1251 – the same year as the CH-blend starts in the SI. The van Engeln composite is used in Jones and Mann 2004 and Osborn and Briffa 2006. So it’s my top pick as being in the CH-blend.
(2) my next pick is the Yang China composite. If they used a van Engeln series in Europe, there’s a good chance that they used the Yang composite – it has the added attraction to the Hockey Team of including the Dunde and Guliya ice cores, which impart a HS to it. It was used in Mann and Jones 2003, Moberg and Osborn and Briffa 2006. So it’s my #2 pick.
(3) if they used these two composites, they will also probably use a Greenland ice core. Probably the Fisher dO18 series from West Greenland – which is used over and over – MBH, Jones et al 1998, etc. I strongly doubt that they would a Greenland borehole as these have high MWP values. Crowley and Lowery 2000 used GISP and the Tom fellow might have returned to this series. But on balance, I’ll guess the Fisher composite.
With only 12 series, I think that there will be 9 tree ring series, so I’ll try to guess from these.
(4) The information shows that 3 series are added in between 1270 and 1505. Osborn and Briffa have 3 tree ring series joining in during that period: Quebec, Tirol and Mangazeja. I might vary these guesses as I think about it a little more.
(5) So we have 6 more series to guess. I’ll guess that they will be old faithful’s in Osborn and Briffa: the Luckman-Wilson Alberta series, Tornetrask, Yamal, Taimyr, Jacoby’s Mongolia and yes even the PC1 from Mann and Jones .
But whatever the answer the Caramilk secret is still locked in the safe at the Nicholas School of Environmental Studies.
BTW, I’m going to immediately send in a Materials Complaint to Nature about the failure to provide the data for the CH-blend. They agreed that Moberg had to provide his data and even required Moberg to issue a Corrigendum a month ago related to data access. So there’s little doubt in my mind that they will require Hegerl et al to disclose the sites and the site data. But it’s ridiculous to have to do this all over again. It takes about 15 seconds to see the problem. You’d think that they would do this ahead of time.
Gabriele C. Hegerl , Thomas J. Crowley, William T. Hyde and David J. Frame, 2006. constraints on climate sensitivity from temperature reconstructions of the last millenium