The Polar Urals temperature reconstruction (Briffa et al, 1995) has been a mainstay of multiproxy studies. More data was collected at this site in 1998 (russ176), but in the two new studies (Osborn and Briffa, 2006; D’Arrigo et al., 2006), they relate their site selection to the Polar Urals, but substitute the Yamal RCS series (oddly, D’Arrigo et al. use the Polar Urals data for the spline standardization (STD)). What accounts for the Yamal substitution?
Here’s a figure comparing the Yamal version (as archived at Briffa’s website) – the plot of which looks a little different than the plot in D’Arrigo et al. 2006 for reasons that I don’t yet understand – and an RCS chronology that I calculated for the combined russ021w and russ176w datasets.
There is no explanation in either text for the Yamal substitution. Rob Wilson has sent me an explanation offline, which he can post up if he wants, but doesn’t make any sense to me. However, one rather unsavory explanation for the Yamal substitution jumps out merely by looking at the series. When you’re dealing with small datasets – e.g. the 6 series average shown in DWJ Figure 7 or robustness to 2-3 series being excluded, these little substitutions have a knock-on effect on both averages and exceedances.
Top – Yamal; bottom – RCS for Polar Urals – combined russ021 and russ176. Both scaled.