Extracting the annual signal from Greenland ice-core chemistry and isotopic records

Abstract

Stratigraphic dating of ice cores by identification and counting of annual cycles in, for example, chemical measurements requires skill and experience. The work presented here investigates a method of data enhancement which is a first step towards an automated and more objective method of annual-layer counting. The method of dynamical decorrelation is briefly introduced and is applied to data from Site D and North GRIP in central Greenland. With this method the measured data series are decomposed into a number of independent source series, one of which exhibits a more pronounced annual variation than the input data themselves. The annual variation is more regular in that (1) some double and triple peaks in the measured series are replaced by single peaks in the extracted signal, and (2) the resulting annual peaks have a much more uniform height. A simple method of determining the number of annual peaks in a series is set up. Using this method, it is shown that it is easier to determine the number of annual peaks in the series produced by dynamical decorrelation than in the original data series. Dynamical decorrelation may thus be used to improve data series prior to dating.