Climate Audit

Al Gore’s hockey stick is from Lonnie Thompson’s ice cores. [Update: subsequent to this, we discovered that Al Gore’s hockey stick is not “DR Thompson’s thermometer” but Mann’s hockey stick wearing a wig.] On previous occasions, we’ve talked about the Guliya ice core -Thompson’s failure to archive data; the use of three different and inconsistent versions in three different 2006 peer-reviewed publications and the questionable use of a core in the flow zone. Paul Dennis wrote to Thompson trying to clarify the Guliya situation and got nowhere.

I recently discussed Dasuopu Core 1, which could be dated back only to 1922. I was recently looking at Dasuopu for a different reason. A new version of Dasuopu information is contained in Thompson et al (Quat Int 2006) showing increased detail in the 1990s. We’ve seen how Emanuel used bin-and-pin to make a high closing value look like a strong trend. The new graphic indicates that Dasuopu actually closes on a very low value. An Emanuel bin-and-pin applied to this data would give a very different impression than the strong 20th century trend presented by Thompson. (BTW Dasuopu dO18 is regarded as being a precipitation amount proxy even in Thompson’s original publication. Of course that does not stop Thompson from adding it together with other tropical dO!8 proxies, also probably precipitation proxies, and calling it a temperature proxy and Al Gore using it as a new hockey stick – but that’s another story.

For today, I’m simply going to collate various Thompson versions. Now the differences are not as egregious as Guliya (which has been discussed recently) but the differences are intriguing.

Thompson et al (Science 2000)
First is the version from the original publication of Core 3 in Science (2000). This shows the top 154.6 m of a 166.7 m core converted to an age scale and then presented as 10-year averages. There were 6903 samples for dO!8 (Thompson et al 2000) so there is no reason why complete information can’t be archived rather than relying on decadal averages.

The core was drilled in 1997 and the last value is presumably the 1990-1997 average. This glacier has a high accumulation and thinning rate – the interval at 149 m is dated to 1440 and at 42.2 m interval to 1962 (through tritium values). Core 1 (discussed elsewhere) was in a flow zone and the 160 m interval was only at 1922. No data was archived at the time, but the version appears to be identical to the version used in Climatic Change 2003, which was archived at WDCP in 2004 as a result of my repeated requests to Climatic Change.

Dasuopu Version 1. Thompson et al (Science 2000).

Thompson et al (Clim Chg 2003)
The next version shown here is from Thompson et al (Clim Chg 2003) which also covers the period 1000-1997 using decadal values. The data at ftp://ftp.ncdc.noaa.gov/pub/data/paleo/icecore/trop/dasuopu/dasuopu-d18o.txt reconciles to this figure. In this case, a 3-decade smooth is added to the decadal results. Thompson et al Clim Chg Figure 7 is the first presentation of the Thompson hockey stick (and this appears to be Gore’s version). It calculates “z-scores ” (standardizes) 3 Himalayan and 3 Andean cores, averages them and compares the result to MBH. Dasuopu is a very strong contributor to HS-ness in this figure.

Dasuopu Version 2. Thompson et al Clim Chg 2003 Figure 5.

Thompson et al (PNAS 2006)
The next version is from Thompson et al PNAS 2006 where Thompson presented 7 tropical ice cores for the period 1600-1995. This version is in 5 year intervals and new data was archived at the PNAS SI. Once again an average was taken after standardizing – this time, however, curiously only for the 1600-1995 period (rather than a year 1000 start as in Thompson et al Clim Chg 2003). Note that the mean used here for Dasuopu is a little different than the mean used in the Clim Chg figure.

Figure 3. Thompson 2006. PNAS Figure 5C.

Now the PNAS data converted to 10-year averages matches the Clim Chg data quite closely, but there is an interesting difference in presentation as shown in my collation of below. The values match almost exactly for all periods except the closing interval. The PNAS uptick (-16.66 for the interval 1990-1994) is noticeably higher than the closing Clim Chg value (-17.34 for the interval 1990-97.) This suggests that 1995-1997 values must be lower (-18.5 is the forced value). Of course, the Dasuopu information is not complete over the 1995-1999 period and that’s a “reason” for not including the most up-to-date Dasuopu information in the PNAS average for 1995-1999, but you’ll understand if I have a strong impression that such “reasons” from the Team always seem to be opportunistic. (Also look at Quelccaya for a similar situation).

Figure 4. Comparison of PNAS 2006 data converted to decadal averages and the Clim Chg 2003 data.

Thompson et al (Quat Int 2006)

There are three different images of Dasuopu in Thompson et al (Quat Int 2006), each shedding some light on the matter.

First here is their Figure 4, showing a “birdseye” view of the core over a very long scale. Both Dasuopu and Puruogangri are represented as relatively “young” at their base, with no part of the core attributed to the last Ice Age. The oldest part of the Dasuopu core is shown as having a substantial “warming” – something that was not reported in any of the earlier articles (and not discussed in this article.) A similar “warming” at the base of the Puruogangri core is shown in this graphic as well, with both synchronized to a corresponding “warming” in the Guliya core (which is held to have a much longer history).

Thompson et al (QI 2006) Figure 4 excerpt

This uptick is an interesting feature that deserves some discussion (not provided by Thompson) and is perhaps relevant to the issue of when Dasuopu started – which is actually why I started this post. Elsewhere Thompson presented evidence from Quelccaya (radiocarbon dated peat moss) to argue that the glacier was receding. Rodbell’s data on proglacial lakes at Quelccaya suggested to me that the Quelccaya glacier did not exist in the Holocene Optimum. The graphs here on Dasuopu and Puruograngri here suggest that these glaciers started forming during the Holocene Optimum, which doesn’t intuitively make a lot of sense. As at Kilimanjaro, I have no confidence in Thompson’s dating for the early portion of the core – and regard his dates as only an upper limit.

Figure 5, which is denominated in meters rather than years, sheds some further light on this. This graph provides the first information on record below 154.6 m (dated 1000AD), the earliest reported portion in the previous articles. On the far right of this figure, you can see a “warm” uptick, which is shown a little more clearly in their Figure 4 shown above. Thompson observes that there is an increase in methane levels at the base of the core (see bottom panel).

Thompson et al (QI 2006) Figure 5

A question about methane levels in Core 2: Core 2 was said in Thompson et al 2000 to have reached bedrock at 149 m., but this graphic shows methane values below 150 m attributed to Core 2? Where do they come from? Is it possible that Thompson has converted everything to dates and then plotted Core 2 on a Core 3 depth scale (a practice that crept in his Kilimanjaro publication) – but this is Thomspon, so who the hell can tell what anything means. You will notice that all the methane values are rather high, even though no Holocene pre-industrial methane values at ice cores in Greenland and Antarctic rise above 800 ppm. What are these high methane values at the start of the core? Thompson notes that the core does not contain the very low methane values characteristic of the LGM, but does not explain the extraordinarily high methane values in the base of the core – these levels are not matched in the core other than in the 20th century. The methane values are really a curiosity. Recall this comment in Davis et al 2005 about Dasuopu Core 1:

“The time series extends back only to 1922 A.D. at 160 m, and below 74 m depth (corresponding to 1945 A.D.) the stratigraphy is severely compromised by ice flow.”

Is it possible that Cores 2 and 3 are compromised as well?

The most recent (left) portion of this graphic is also interesting. The core was drilled in 1997 and it is known from other evidence that the 42.2 meter mark is dated 35 years earlier in 1962. So the 42 different one-meter averages shown here will be a somewhat blurred representation of annual information for the corresponding 35 years from 1962-1997. Now look at the most recent value of this series (on the left): here the most recent value is very “cold” (-21), one of the “coldest” values in the recent record, while the PNAS version closes at a very “warm” value above -16.

A rather similar (but slightly truncated) version of Figure 5 occurs in Thompson et al Figure 4 as shown below. In this case, the “warming” in the earliest portion of the graphic is truncated.

Thompson et al QI 2006 Figure 4 third panel.

Yang et al (QSR 2007)
This year, there is a new images from Dasuopu from the Thompson group in Yang et al (QSR 2007). Yang et al Figure 2 (shown below) shows results from 1000 to present. This version appears to be identical to the Clim Chg version (which I’ve confirmed by plotting the Clim Chg digital version, which is available, in the style of Yang et al and observed a match. Imagine that – the same version twice. Unprecedennnted.

References:
Thompson et al Science 2000
Thompson et al Climatic Change 2003
Thompson, L.G., E. Mosley-Thompson, M. E. Davis, T. A. Mashiotta, K. A. Henderson, P.-N. Lin, and Y. Tandong. 2006. Ice core evidence for asynchronous glaciation on the Tibetan Plateau. Quaternary International, 154/155, 3-10 http://www-bprc.mps.ohio-state.edu/Icecore/Abstracts/Thompson%20et%20al%20Quat%20International%202006.pdf
Yang, Bao, Achim Braeuning, Tandong Yao, Mary E. Davis, QSR 2007, Correlation between the oxygen isotope record from Dasuopu ice core and the Asian Southwest Monsoon during the last millennium.