In today’s post, I’m going to critically examine another widely used tree ring chronology: the Icefields (Alberta) MXD RCS chronology of Luckman and Wilson (2005 pdf), used most recently in Wilson et al 2016.
I’ll show that the RCS technique used in the LW2005 MXD chronology eliminated high medieval values as a tautology of their method, not as a property of the data and that the Icefields data provides equal (or greater) justification for MXD RCS chronologies with elevated medieval values. The measure of potential difference is previewed in Figure 1 below, which compares the reported LW2005 chronology (top panel) to an MXD chronology with elevated medieval values, calculated using equally (or more plausible) RCS methodology, with several other similar variations shown in the post.
Figure 1. Top – MXD RCS chronology from Luckman and Wilson 2005 Figure 2; bottom – MXD RCS chronology calculated under alternative assumptions – see Figure 3 below.
I will use random effects statistical techniques both to analyse the data and to place prior analysis in a more formal statistical context. Because LW2005 coauthor Rob Wilson stands alone for civility in the paleoclimate world and because the present post is critical of past analysis, in some ways, I would have preferred to use another example. However, on the other hand, it is also possible that Rob will recognize that the techniques applied in the present post – techniques that are unfamiliar to dendros – can yield fresh and important insight into the data and will look at the Icefields data a little differently.
Although the article in consideration was published more than a decade ago, the analysis in today’s article was impossible until relatively recently, because coauthor Luckman withheld the relevant data for over a decade.