A python package implementing Direct Reconstruction Technique (DIRECT) for dendroclimatological studies

Abstract

The DIrect REConstruction Technique (DIRECT) is a dendroclimatological method that constructs a climatic response surface to account for changes of climatic response with tree age. This surface is then used as a transfer function for climatic reconstructions. Unlike widely-used standardization methods such as the traditional curve-fitting approach, the regional curve standardization, and their signal-free modifications that perform detrending explicitly, DIRECT accounts for age-size related trend in tree-ring measurements by construction of the response surface dependent on two variables: tree-ring parameter (width, blue intensity etc.) and cambial age. The method is capable of taking into account nonlinear climatic response of trees and differing response of younger and older trees. Here we describe an application of the newly developed open-source Python package that implements DIRECT (https://github.com/Gr1Lo/direct) to one real and one theoretical dataset. The package consists of functions for reading the initial data, constructing a response surface, and for reconstructing climate variables via this surface. The functions for visual assessment of the initial data and for the estimation and selection of parameters for constructing the response surface are also presented. Also here we provide a comparison of the DIRECT method with traditional standardization-reconstruction routines.