|
Techniques: Chronology and Correlations |
1 Department of Geology & Geophysics, University of Edinburgh, West Mains Road, Edinburgh EH9 3JW, UK
2 Department of Mathematics, Monash University, Victoria, Australia
An automated core correlation procedure has been developed in order to match sequences, such as core logs, based on a mathematical sequence-slotting approach. The programme can be applied to univariate, multivariate or compositional data. In addition to finding the optimal slotting, the programme also quantifies the quality of match. It can also be used to detect sections of the sequences which are tightly constrained, by the match, or only fit together more loosely. An important aspect of any multivariate analysis is scaling. This topic is addressed through a simplex optimization method.
We have used the sequence-slotting method to analyse high quality oxygen-isotope data from ODP Site 658 on the continental slope, west of Cap Blanc, on the northwest African coast. Excellent agreement is found with the global oxygen-18 reference curve and this allows the sediments to be dated. By and large the resulting depth-age profile is very similar to that derived in previous studies. However, small changes in the age estimates between the original stratigraphy and our sequence-slotting lead to major differences in the pattern of sediment accumulation rate.
An hiatus spanning 24 500 SPECMAP years some 50 000 years ago has been proposed in the Site 658 sediments. However, as we find no compelling mathematical evidence for the hiatus, Site 658 may contain an even better stratigraphic record of palaeoclimatic change than previously thought.
