Transfer Functions for Paleoclimate Reconstructions — Theory and Methods
In this chapter, all within the KIHZ-community used methods for deriving transfer functions in order to reconstruct paleoclimatic variables are introduced. The aim is to provide the conceptual and statistical background of the methods which are used throughout this book. A short review is given on the use of probability density functions in paleoecology. We then focus on linear and non-linear regression methods, among them are weighted averaging, weighted averaging partial least squares regression, both frequently used in paleolimnology, as well as neural network regression, which is recently used in pollen-based paleoclimate reconstructions. In addition, we review some of the validation methods for transfer functions.
KeywordsEurope Covariance Shrinkage Autocorrelation Cond
Unable to display preview. Download preview PDF.