Building and testing age models for radiocarbon dates in Lateglacial and Early Holocene sediments

Abstract

The growing importance of understanding past abrupt climate variability at a regional and global scale has led to the realisation that independent chronologies of past environmental change need to be compared between various archives. This has in turn led to attempts at significant improvements in the required precision at which records can be dated. Radiocarbon dating is still the most prominent method for dating organic material from terrestrial and marine archives, and as such many of the recent developments in improving precision have been aimed at this technique. These include: (1) selection of the most suitable datable fractions within a record, (2) the development of better calibration curves, and (3) more precise age modelling techniques. While much attention has been focussed on the first two items, testing the possibilities of the relatively new age modelling approaches has not received much attention. Here, we test the potential for methods designed to significantly improve precision in radiocarbon-based age models, wiggle match dating and various forms of Bayesian analyses. We demonstrate that while all of the methods can perform very well, in some scenarios, caution must be taken when applying them. It appears that an integrated approach is required in real life dating situations where more than one model is applied, with strict error calculation, and with the integration of radiocarbon data with sedimentological analyses of site formation processes.

Introduction

There is significant scientific interest in understanding the nature and timing of regional and global responses to abrupt climate change during the transition from the last glacial to the Holocene, due to the numerous abrupt climatic oscillations that are now known to occur during this period as indicated by the Greenland ice core records (e.g. Alley et al., 1993). These are seen as essential natural archives for understanding climate forcing mechanisms and the underlying Earth system response (Walker et al., 2003). In order to understand these processes, it is necessary to compare different archives at sufficiently high temporal resolution to examine the timing of response between them. The abrupt nature of these climatic events, however, and the possibilities of asynchronous responses in different environments, along with regional climate gradients, requires significant improvements in the precision at which many environmental records can be dated (Lowe et al., 2001). In particular, there have been recent attempts to build age models for radiocarbon-dated sediment sequences with levels of precision normally associated with annual chronologies, such as the Greenland ice cores and dendrochronology. These include visual wiggle match dating of terrestrial sediment records, regression equations applied to dated sequences after calibration, and the application of various Bayesian statistical models to the calibration of dates (e.g. Blaauw et al., 2004; Blockley et al., 2004). All of these methods are, in fact, models of the underlying true sedimentation rate, which are based on various assumptions about the sedimentary archive, the radiocarbon dates, and the radiocarbon calibration curve itself. Some of these models are ad hoc, such as the visual wiggle match adopted by some authors, while others are based on classical or Bayesian statistical frameworks. Such models raise questions of genuine importance for many researchers involved in understanding the nature and timing of climate change in the Lateglacial and Early Holocene, regarding the reliability of the assumptions made in developing these models and their ability to accurately predict the depositional history of a given site. These questions can be examined to some extent by simulating different dating scenarios (Telford et al., 2004). Here, we compare the performance of three of the most common approaches, in four different scenarios that we believe to be representative of the range of sedimentary regimes and dating difficulties. One assumption that cannot be tested in this way, however, is whether the available radiocarbon calibration curve is a completely accurate representation of the terrestrial/atmospheric radiocarbon activity of the whole period through the Lateglacial and into the Holocene. The presently recommended calibration curve is IntCal04 (Reimer et al., 2004), and is based on dendrochronologically dated wood back to 12 460 cal BP. Thus, for this time-range it represents the natural atmospheric/terrestrial ¹⁴C content. Beyond the dendrochronological part, in the earliest part of the Lateglacial record, the IntCal04 curve is marine-based.