As mentioned yesterday, the Law Dome series has been used from time to time in IPCC multiproxy studies, with the most remarkable use occurring, needless to say, in Mann et al 2008. As noted yesterday, despite Law Dome being very high resolution (indeed, as far as I know, the highest resolution available ice core) and the DSS core being finished in 1993 and (shallow) updates in 1997 and 1999, there hasn’t yet been a formal technical publication.
To give a rough idea of Law Dome resolution, its layer thickness between AD1000 and AD1500 averages 0.45 m, as compared to 0.034 m at Vostok, 0.07 m at EPICA Dome C, 0.1 m at Siple Dome, 0.028 m at Dunde and 0.18 m at the NGRIP site, which is regarded as very high resolution. This is a very high accumulation and very high resolution site. There were some technical difficulties with the original DSS core in the 19th century and the upper portion of the “stack” presently reported relies on the DSS97 and DSS99 cores, crossdated to the long DSS core.
The original sampling of the DSS core was at approximately 10 samples per year back to AD1304 with 0.5 meter samples in earlier portions of the core. The layer thickness (see Morgan and van Ommen 1997 Figure 6) in the period from 1304 to 1000 declined from about 0.25 m to 0.18 m. So the resolution of the original 0.5 meter sampling would be about 2-3 years. Not the desired annual to which they have been working, but still very high resolution compared to most available proxies.
The seasonally resolved O18 series from 1304-1987 (using only the somewhat problematic DSS core) was published in Morgan and van Ommen 1997 (J Glaciology). The underlying data was interpolated to monthly values; a copy was sent to Phil Jones in 1997 and was used in the SH composite of Jones et al 1998. This data was not archived. I obtained a copy of it from Phil Jones in 2004.
Phil Jones also sent this data to Canadian climate scientist David Fisher, who used it in Fisher (2002), about which more below.
The accompanying graphic in Morgan and van Ommen 1997 showed only a VERY smoothed version of the data. I’ve overlaid this graphic with a less smoothed version of the digital data – this clearly shows that the Jones et al 1998 data is the same data used to produce the graphic in Morgan and van Ommen 1997. The gap in the 1840s is visible in the overprinted data, but has been smoothed over in the graphic from the original article. Notice also that the modern portion of this version is elevated relative to the later version plotted yesterday. (More on this below).
In the early 2000s, van Ommen and associates created a “stack” of shallow cores to overcome the core problems in the upper part of the original DSS core and to update the series. (See for example Palmer et al 2001 (JGR)). By 2003, they had calculated a 2100 year series extracted at 4-year intervals. In the upper part, this composited from sub-annual data; I’m not sure why they wouldn’t have retained the annual information in the archive.) This data was sent to Phil Jones in 2003 and used in Mann and Jones 2003 (not illustrated) and later in Jones and Mann 2004 (where it was shown.)
I requested this data from van Ommen in late 2003 and followed up in Feb 2004. Van Ommen put me off, saying that the data would not be available until published ( mentioning a delay of a “couple of months”). Van Ommen passed my request on to Jones (1076336623.txt), who alerted Mann to this disturbance in the universe, leading to a vituperative recommendation from Mann that I not be given access to data. I did not get a copy of the data until over two years later, when a further inquiry to van Ommen was finally successful.
The new version extended the record back to the start of the first millennium, showing very elevated values in the 10th century. The series obviously does not accord with the expected increase of d18O values with 20th century warming. Because O18 series are so fundamental to paleoclimate, this discrepancy obviously needs to be explained. Since the series has not yet been formally published, these differences have not been addressed in academic literature. (My own take is not that this series is neccessarily “right” as a temperature history, but that it is not right to reject it, while retaining other ice core O18 series without an explanation that rises above the uninformative “regional variations” or “regional circulation”, especially when series also possibly subject to “regional circulation” are not rejected if the error goes the “right” way.)
Mann et al 2008
Fisher 2002 (Holocene) carried out a principal component analysis on a wide variety of proxies for the 210 year period from 1761-1970 when the proxies were all available. Fisher created an archive of these proxies for the period 1761-1970, one of which was Law Dome (LAWA210.ANT). This was obviously only a fraction of the available data but was the portion used in his analysis.
Even though Mann had a current and much longer version of the Law Dome O18 series that he/Jones had obtained from van Ommen, Mann et al 2008 substituted the truncated version used in the Fisher principal components analysis. The difference in the two versions is shown below.
It would take a while to calculate the effect of Mann’s use of an obsolete and truncated version of the Law Dome series on his SH reconstruction, but no one should assume that it didn’t and doesn’t “matter”. My guess is that this decision had a material impact.