pcrecon: An R Package for Nested Principal Component Regression Climate Reconstructions

Keywords: Dendrochronology, R, Climate, Proxy Reconstruction
Session Type: Virtual Paper
Day: Friday
Session Start / End Time: 4/9/2021 03:05 PM (Pacific Time (US & Canada)) - 4/9/2021 04:20 PM (Pacific Time (US & Canada))
Room: Virtual 30

Abstract

The process of tree-ring analysis frequently relies on the use of multiple, stand-alone software programs for cross-dating, ring-width file editing, standardization, chronology development, and more. Moving between these programs, each with a very specific application and user interface, presents workflow issues for the user as well as general barriers to transparency and reproducibility. In the last 10-15 years, major efforts have been made to port many of these applications to the R statistical programming environment, most notably with the package dplR (Bunn et. al, 2007). Legacy fixed-width file formats common to dendrochronology can be read into R using functions within the dplR package, and all manner of analysis (tree-ring specific or not) performed within that single environment. A record of the data analysis process is kept via scripts, text files containing R code, thus improving reproducibility and transparency in the process. We present pcrecon, a new software package for the development of nested principal component regression (PCR) climate reconstructions from tree ring data in R. This package duplicates and extends the well-known Fortran program PcReg. The use of a standard, versatile, common, and cross-disciplinary platform for analysis is important for clear communication and reproducibility of results, as well as collaborations with subject matter experts in other fields.