A climateaudit first:- here is the first sample-by-sample àŽⳏ18 for an entire Thompson drill core – in this case KNIF2 and KNIF3 from Kilimanjaro. I had hoped that the data would be properly archived, but it was sent to me by Science and is webbed up here pending a more official archive, which will presumably take place in the future. In the mining business, you need to see complete drill cores – why should ice cores be any different?
I have been attempting to get Thompson to create proper permanent archives of data a long time (with an initial data inquiry over 2 years ago). The two data files here are very inadequate, since they are only 2 of 6 cores at Kilimanjaro and include only àŽⳏ18, but not ion and chemical measurements, and do not deal with other important published tropical sites such as Dunde, Guliya, Dasuopu, Quelccaya, Sajama and Huascaran (not to speak of relatively work at Puruogangri and Quelccaya that is rapidly becoming stale for not being published). However, it’s a small first dividend. Lonnie Thompson is one of the “wise men”, who recently wrote the Barton Committee advising them to butt out.
Here is a quick plot of the data and a few thoughts.
First, a simple that there is no annual cycle of àŽⳏ18 values observable in the Kilimanjaro cores, comparable to other tropical cores. Thompson doesn’t claim otherwise, but it wasn’t mentioned. Second, one obviously doesn’t get any sense of àŽⳏ18 going upwards off the page in the 20th Century. An obvious corollary is that the Hockey Team uses Dunde or Guliya àŽⳏ18 rather than Kilimanjaro àŽⳏ18. If àŽⳏ18 is a good temperature proxy, why isn’t it showing up? Third, the most negative àŽⳏ18 value in the entire KNIF2 core occurs at a depth of only 0.544 m (dated after 1950) – what was going on here?
Figure 1.Kilimanjaro àŽⳏ18 top-KNIF2, bottom KNIF3
Both cores are about 50 m long. KNIF3 is assigned a start date of 11700 BP, while KNIF2 is (implicitly) assigned a start date of about 3800 BP. about 50 m long. I’ve been mildly curious as to how the start date of 11700 is assigned to this relatively thin glacier, as, for example, Quelccaya is much thicker and is assigned a start date of about 470 AD.
One of the key pieces of dating information cited in the article is that a radiocarbon date of 8280 BP (1950) “determined from small quantities of organic samples from the deepest section (47.9 to 49.0 m) of NIF3 supports the presence of early Holocene ice”. The basal age of 11700 BP was assigned by comparing the general shape of the NIF3 àŽⳏ18 to the shape of the well-dated Soreq Cave àŽⳏ18 curve. No information is given as to what the organic material was – it would be interesting to know. I’ve mentioned before that if a glacier in a few centuries were to overrun a temperate site like Sheep Mountain CA, some of the highest organics that it would engulf would be the Holocene Optimum bristlecone stumps, which would date to 6000 BP or so. Wouldn’t an included organic simply be a terminus ad quem for the date of formation of the glacier. Archaeologists worry about this type of thing all the time in dating strata in a tell. I’ve plotted below all the radiocarbon dates by depth from the Thompson SI together with a rough approximation of the age model for the KNIF3 core? There are some remarkably young organic samples in the core, which much surely raise some question as to how much weight can be placed on the interpretation of the dated organics. I’ll try to consider the Soreq Cave on another occasion. For now, let’s just say that the wiggle-matching seems a little arbitrary to me and other interpretations may be possible.
Figure 2. Kilimanjaro Radiocarbon Samples from Thompson et al  SI
One aspect of the Kilimanjaro report that I find troubling is a proposed re-scaling of the KNIF2 core. Obviously I haven’t dealt with this core in detail and am not trying to do anything here other than ask questions. The re-scaling is justified on the basis of àŽⳏ18 wiggle-matching in which “events” in KNIF2 are allocated to “events” in KNIF3. Wiggle-matching seems to me to be sometimes in the eye of the beholder. For example, if you look at my plots in Figure 1 above, it seems to me that one could plausibly allocate the downspike near 40 m in KNIF2 to the downspike near 37 m in KNIF3 and that one could plausibly view the two cores as being more or less the same age – whatever that age was. Instead, Thompson wiggle-matches the downspike near 40 m of KNIF2 with the downspike near 30m in KNIF3 and compresses much KNIF2 length into a shorter length of KNIF3 – see below.
Thompson et al  Fig. 2 (A and C). The 10-year average àŽⳏ18 records from the (A) NIF2 and (C) NIF3 cores are shown for their entire lengths. (B) NIF2 depths are rescaled to the NIF3 depth scale by matching similar àŽⳏ18 features, showing that NIF2 is contemporaneous with the upper 32 m of NIF3. The àŽⳏ18 events labeled 1 to 9 are assumed to be coeval.
In his Supplementary Information, Thompson plots 50-year averages of various ions, showing the effect of the KNIF2 compression. Obviously I haven’t seen the ion data in detail, but the matches in the SI do not seem very impressive. The upspikes in KINF2 occur earlier than in KNIF3, suggesting at least the possibility that the àŽⳏ18 wiggle-matching might not be 100% on the money.
Thompson et al.  Supplementary Information Figure .
A few other oddities that caught my eye in this article.
Thompson has reported massive 20th century downwasting of the Kilimanjaro glacier – this seems unarguable. However at KNIF3, he reported that the high Cl-36 levels from a 1952 thermonuclear explosion could be observed in the core. He attempted to reconcile the downwasting with this layer still being extant by hypothesizing a large precipitation event in the early 1960s. I don’t see that this really does what he wants. One idea here: at tropical glaciers in the Himalayas, the most negative àŽⳏ18 values occur in the summer due to a monsoon rainout effect, reversing the pattern in Arctic regions. (This drives Thompson crazy since he wants to argue that higher àŽⳏ18 values in the 20th century at Dunde and Guliya are evidence of exceptional warmth. As I’ve discussed before, he does some very unimpressive correlation analysis to try to argue this. I’m not saying it’s impossible, only that his statistical arguments don’t really work.) Anyway if you look back at the negative àŽⳏ18 excursion at 0.544 m in KNIF2, maybe this reflects the massive precipitation event, with a monsoon rainout effect in the very negative àŽⳏ18 values. On earlier occasions, Thompson interprets highly negative àŽⳏ18 downspikes as signs of coldness and aridity. I’m not saying that this interpretation is incorrect, but wouldn’t you like to see an explanation of the modern values?
For core SIF2, Thompson reports a 6-meter interval of “rogue” ice that was “removed from the record because the isotopic signature was unique within the set of 5 cores”. The “rogue” ice was attributed to a “brief mass wasting or crevasse formation event”. I am sure that that’s a possible explanation, but it would be nice to see what the data looked like.
The formation of the Kilimanjaro glacier in the warm Holocene Optimum seems a little odd to me. If anyone were trying to argue that the “Holocene Optimum” was a regional hodge-podge (the current Hockey Team line on the Medieval Warm Period), then the supposed formation of the Kilimanjaro glacier in this period would undoubtedly crop up as evidence for local “coldness”. However, if there were problems with the Kilimanjaro glacier dating, then the matter might be reconciled in a different way.