WHITHER CYCLOSTRATIGRAPHY - TESTING THE GAMMA-METHOD ON UPPER PLEISTOCENE DEEP-SEA SEDIMENTS, NORTH-ATLANTIC DEEP-SEA DRILLING PROJECT SITE-609

Gamma analysis, a procedure for tuning cyclic sediments in which gamma is defined as I/sedimentation rate, was applied to the upper Pleistoc ene gray scale record of Deep Sea Drilling Project site 609 from the N orth Atlantic. The spectrum of the gamma-tuned time series produces ec centricity, obliquity, and precession index signals when the primary p eriodicity is tuned to the precessional beat frequency. However, the t imescales are significantly too long, suggesting that some of the prim ary cycles used in the gamma tuning procedure have higher frequency th an the precessional beat. By filtering the SPECMAP-calibrated time ser ies at 10-12 kyr the spurious cycles are removed from the tuning proce dure. In this case, the resulting gamma-corrected spectra look much mo re like that of the SPECMAP-tuned time series. However, the time serie s remains slightly too long. Forward modeling is consistent with the r esults from modeling the core. That is, the gamma method is quite robu st at revealing orbital periodicity in an orbitally controlled record for which sedimentation rates are facies dependent. However, the metho d does not produce an accurate time series. Criteria for complete reje ction of gamma results are clarified as either (1) least squares resul ts include negative gamma values or (2) the resulting spectrum does no t show an orbital signature. Variation on sedimentation rates greater than +/- 5% result in incorrect time series. However, variations on se dimentation rates as high as +/- 15% generally reproduce strong orbita l spectra, and occasionally, the method can also generate an orbital s pectrum with up to +/- 25% noise. The method is not as robust in predi cting gamma values or sedimentation rates. Although it generally disti nguishes facies with relatively high sedimentation rates from facies w ith relatively low sedimentation rates, the values predicted may be in correct by a factor of 2. Thus while gamma tuning of the time series i s an improvement over observed thicknesses, the resulting timescale is not correct. The implications of these results for cyclostratigraphy are significant. The generation of an orbital spectrum from a time ser ies that is demonstrably incorrect casts doubt on cyclicity as a datin g tool.