The Polar Urals tree ring site is another staple of multiproxy studies, being used in Bradley and Jones , Hughes and Diaz, Overpeck et al. , Jones et al. , MBH98, MBH99, Crowley and Lowery , Esper et al. , Bradley, Hughes and Diaz  and Rutherford et al.  (which recycles MBH98). In fact, it seems to have been used in every study.
The Polar Urals site is considered in a series of studies, starting with Graybill and Shiyatov, continuing with Briffa et al [Nature, 1995] and Briffa et al. [Nature, 1996].
In an earlier posting, I pointed out the striking change in altitude of samples at this site, with 13th century samples averaging from an altitude of about 310 m. (NATO 1996, Figure 4, top panel), while modern samples all come from between 200 and 250 m. Briffa says that the elevations of the subfossil cores are known precisely, but not the modern cores (p. 35). I have requested elevations of the subfossil cores without any success. At a typical lapse rate of 6.5 deg C/km altitude, a difference of 60-110 m between the altitudes of modern samples and 13th century samples is pretty significant: 0.39-0.72 deg C., say 0.55 deg. C and something that would seem to require adjustment. NATO Figure 4 shows different altitudes than a similar graphic from Shiyatov, shown as Figure 1 below, which shows a maximum difference in treeline of 70 m. I presume that the "mean elevation" of NATO Figure 4 and the "altitude of the upper treeline" of the Shiyatov figure measure slightly different things, but the amount of displacement in the two concepts seems pretty similar.
So how is the problem of changing altitude handled by Briffa (and Jones)?
“We examined the evidence for a time-dependent elevation bias in the reconstruction by regressing mean MXD against mean sample elevation for different age classes of trees … These results (shown in [their] Table 2) indicate no significant elevational influence on mean density, at least over the range of elevation involved in these calculations”
Here is their Table 2:
TABLE 2. Correlation (Number of Samples) between Mean MXD and Elevation by Age Class, Polar Urals.
|Age Class||Subfossil Only||All Samples|
|1-50||0.09 (37)||-0.07 (62)|
|51-100||0.05 (49)||-0.12 (90)|
|101-150||0.10 (40)||-0.05 (71)|
|151-200||-0.20 (17)||-0.07 (39)|
|201-250||0.10 (8)||0.22 (18)|
|251-300||-0.25 ( 3)||-0.08 ( 9)|
The original caption stated:
“The density data in each age class are averaged over different time periods so that effects of climate variability should hopefully largely cancel out. Two sets of correlations are shown: one based only on the subfossil series and the other including the living tree material whose precise elevations are not known and have been set here to a constant elevation of 250 m. None of the correlations is significant indicating that there is little evidence for an elevation influence on ring density and hence little age-dependent bias in the temperature reconstruction arising out of the differences in sample heights shown in Figure 4.”
But think for a moment about what is actually being said here: if MXD is measuring temperature and there is no elevation effect, then the temperatures being estimated in the Briffa reconstruction are for varying elevations. So if you want a temperature reconstruction for a constant elevation, you need to adjust for the varying elevations i.e. increase the 13th century results by about 0.55 deg C. If the effect went the other way, how long do you think it would have taken the Hockey Team to do that adjustment?
When I get a digital version of the changing altitudes, I’ll do an adjusted version of the Polar Urals data set and update this page. There are only 3 proxies in Jones et al.  in the first part of the warm 11th century: Tornetrask, Polar Urals and GISP2 — so you can probably guess where I’m headed with these details.
There’s another interesting issue about data quality in the 11th century, which I’ll get to sometime this week.