Statistical Procedures of Testing for Orbitally Derived Periodicity: With Application to Pleistocene Deep Sea Cores

Michelle A. Kominz

A fundamental problem in testing for orbitally forced periodicity of cyclic sediment packages is the assignment of time to observed sediment thicknesses. In some cases, this problem can be overcome by applying the gamma method; that is, by assuming that cycles are approximately periodic and that facies have independent, but constant accumulation rates.

The statistical accuracy of results of this procedure have been obtained by forward modeling of cyclicity generated by a time series which includes eccentricity, obliquity and precessional signals. Sedimentation rates, with a specified degree of random error, are assigned to facies and gamma results are obtained. Correlation coefficients compare the true time series and the true spectra with observed and gamma corrected time series and spectra. While gamma tuning does improve the time series, correlation coefficients are not high (0.05 to 0.4 on observed data and 0.25 to 0.5 after gamma correction). The correlation coefficients of the spectral results are stronger, increasing from about 0.5 for the observed spectra to 0.7 to 0.9 after tuning. Gamma correction consistently improves the relative sedimentation rates, but does not give exactly correct rates. Small errors in the time series result in a low correlation coefficient although the spectrum is quite accurate.

The gamma method has been applied to two pelagic sequences from Ocean Drilling Project Cores, sites 609 and 677. Gamma tuning was applied using the highest frequency cycles, which were assumed to be a response to the precessional index beat frequency at about 12 k.y. Tuning produces the predicted eccentricity, obliquity and precession peaks, as well as an indication of the precessional beat frequency peaks. Comparison with statistical results from forward modeling suggests that calculated sedimentation rates are accurate within a factor of 2 and that variation on sedimentation rates may range from 10 to 40%.

AAPG Search and Discovery Article #91020©1995 AAPG Annual Convention, Houston, Texas, May 5-8, 1995