Chronology-building using objective identification of annual signals in trace element profiles of stalagmites.

Smith, Claire and Fairchild, Ian J. and Spötl, Christoph and Frisia, Silvia and Borsato, Andrea and Moreton, Steven G. and Wynn, Peter M. (2009) Chronology-building using objective identification of annual signals in trace element profiles of stalagmites. Quaternary Geochronology, 4 (1). pp. 11-21. ISSN 1871-1014

Abstract

Determination of annual lamination provides important additional constraints to radiometric dates on speleothems, both for dating the duration of specific growth intervals and optimizing growth models. In the absence of visible laminae, however, speleothem age models are reliant upon curve fitting through discretely dated points and are therefore inherently more uncertain than annual chronologies from laminae. Given that the impact of seasonality on speleothems is expected to be strong enough to generate an annual pulse in trace element chemistry regardless of whether or not visible or fluorescent growth laminae are visible, we demonstrate the potential for deriving high-resolution stalagmite chronologies from non-laminated samples using annual chemical variations in stalagmites from two Alpine caves (Obir, Austria and Ernesto, NE Italy). Trace element data were obtained by ion microprobe analyses for H, P, Mg, Na, Sr and Ba and the annual signal was sought using spectral and wavelet analysis. An automated chemical peak-counting software tool was developed in Matlab©. It counts significant peaks using criteria of 1) minimum amplitude in relation to the local standard deviation of signal variation and 2) minimum separation between peaks. Verification of the tool using visibly-laminated samples suggests the software is a reliable and accurate method of chronology building, with hit ratios greater than 0.93 and less than 0.75% false alarm occurrences. Used in conjunction with other dating methods such as radiocarbon, U-Th and sulphur peak dating, the automated chemical laminae chronology-building approach would provide a more meaningful alternative to simple age-depth curve fitting for non-laminated samples.