The major features of data curves are often matched, to a first order, by bump and wiggle
matching to arrive at an offset between data sets. This poster describes a simple statistical
correlation program that has proved useful during this stage by determining the optimal
correlation between geomagnetic curves using a variety of fixed and floating windows. Its
utility is suggested by the fact that it is simple to run, yet generates meaningful
data comparisons, often when data noise precludes the obvious matching of curve
features.
Data sets can be scaled, smoothed, normalised and standardised, before all possible
correlations are carried out between selected overlapping portions of each curve. Best-fit
offset curves can then be displayed graphically. The program was used to cross-correlate
directional and palaeointensity data from Holocene lake sediments (Stanton et al., submitted)
and Holocene lava flows. Some example curve matches are shown, including some that
illustrate the potential of this technique when examining particularly sparse data
sets.
Stanton, T., Snowball, I., Zillén, L. and Wastegård, S., submitted. Detecting potential errors
in varve chronology and 14C ages using palaeosecular variation curves, lead pollution history
and statistical correlation. Quaternary Geochronology. |