Application of Clustering Regression Time Series in Dendrochronology

• Activity Number: 284
• Type: Topic Contributed
• Date/Time: Tuesday, August 5, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistical Computing
• Author(s): Semhar Michael*+ and Volodymyr Melnykov
• Companies: University of Alabama and University of Alabama
• Keywords: finite mixture model ; regression time series model ; dendrochronology ; tree rings ; Kalman filter

Abstract

Dendrochronology is a field of science that studies sequences of growth rings in trees. Clustering of growth rings has different applications in variety of fields such as archaeology and climatology. Past attempts in grouping tree rings based on traditional methods have met with difficulties resulting in groups with unclear interpretations. Hence, researchers often relied upon subjective assessment based on the resemblance of the tree ring sequences. In view of this, we develop two techniques to cluster tree ring sequences via finite mixtures of regression ARMA models. One approach is based on the conditional maximum likelihood estimation and the other relies on the maximum likelihood estimation using the Kalman filter. The advantages and shortcomings of both methods are discussed. Simulation studies and the analysis of tree ring data shows promising result.