Using Machine Learning for Cell Identification in Ring-Porous Oak Species
Author: Adam Johnston
Faculty Supervisor: Alexander Stine
Department: School of the Environment
Measurements of tree ring growth have played an important role in our reconstruction of the climate in the last 2000 years. Traditionally, climate reconstructions have been derived from measuring the width of annual growth rings in gymnosperm species. The geometry of vessel cells in angiosperm species also encodes information about the environmental conditions the tree grew in. My research draws from measurements of vessel cells in oak species to expand our record of climate in space and time. In my work I am developing a machine learning framework that extracts climate signals from high resolution images of ring-porous oaks. This framework identifies and outlines vessel cells to build a proxy for past climate. The goal of my research is to build a framework for future climatic reconstructions using ring-porous angiosperm species. My work has a special focus on oak species from California and Ireland.