Yes. Three of those sites seem to be in the vicinity of the visitor centre at the Athabasca glacier. Are there not 8000 yr. old wood fragments being uncovered as that glacier is receeding? Perhaps with a little cognitive glueing they might make a good temp proxy.

The Vermillion series is unique in that it has no significant correlations with climate records from the nearest station. …The inconsistent correlations observed in Fig. 5a may reflect either varying responses of the trees to the same macroclimatic variables or similar responses of the trees to different microclimates (Fritts 1991).

LOL. Unique?? Are there ANY tree ring series that are correlated with the nearest station? It looks more and more like any correlation they get between growth and temperature is just luck. Even I understand spurious correlation.

Variance adjustment reduces low-freq variability of the past (sparser data). If we had some kind of verification tool, this would be detected right away. In addition, we have a divergence problem in 1980-present reconstructions. Funny situation, reconstructions are reliable given that there were no low-frequency variations in the past temperatures.

In many threads, Kalman smoothers etc. have been suggested for reconstruction purpose. But here’s the big problem: Kalman theory (and Wiener and other related) requires a dynamic model for the underlying stochastic process. If we choose random walk we can directly (prior to observations) say that recent temperature rises are completely natural. If we choose a white noise model, we’ll get opposite conclusion. Then we can choose anything between those two models and get different conclusions. This is a dead-end, we can’t use reconstructions for system identification (to choose the underlying stochastic model), because reconstructions depend on that model. Sorry again, too much thinking aloud I guess