Statistical models of dendrochronology

Mining the British Isles oak tree-ring data set. Part A: Rationale, data, software, and proof of concept

“…The fundamentals of Baillie's methodology (high-pass filtering; minimum series length; sliding series against each other; correlation-based goodness-of-fit statistic, and; statistical significance testing-commonly Student's t) were well established by the 1970s and persist to the present day. There have been some local developments, such as alternative statistical methods developed by the Nottingham lab (Litton and Zainodin, 1991), and there are areas of disagreement amongst practitioners (e.g. varying opinions on the minimum series length that should be considered eligible for crossdating).…”

A simple Bayesian approach to tree‐ring dating

“…Each year a tree grows a new extra ring, and the width of that ring depends on several factors; mainly the climate and growth trend. It is assumed that trees of the same species within the same geographical region receive the same climatic signal throughout a given year because they are exposed to similar climatic growth conditions, but this signal varies from year to year (Litton and Zainodin ). Good growing seasons usually provide wide rings, whereas poor ones produce narrow rings.…”

“…Over the past decades, many efforts have been made to develop statistical models for dendrochronology. The earlier statistical models, such as those suggested by Fritts () and Litton and Zainodin (), were typically based on extending linear, empirically determined relationship between tree‐ring indices and climate. This paper explores the use of the statistical model first proposed by Litton and Zainodin.…”

“…The routine use and current success of Bayesian methods in radiocarbon dating, through modern radiocarbon calibration software such as OxCal (Bronk‐Ramsey ) and BCal (Buck et al . ), means that it now seems timely to explore the Litton and Zainodin () model once more. Those unfamiliar with the Bayesian approach to radiocarbon dating might like to consult Buck et al .…”

“…The next section provides a brief literature review of the existing statistical methods that are key to understanding the process of dendrochronology, including pre‐processing of tree‐ring data, cross‐matching sequences, constructing chronologies and modelling tree‐ring growth. The third section introduces the Bayesian approach to tree‐ring dating, which involves formalization of the cross‐matching process, and implementation alongside the model proposed by Litton and Zainodin (). The fourth section includes results of the Bayesian implementation using both simulated and real data to investigate the proposed model and matching process.…”