ANALYZING THE METHODS USED TO DETERMINE STRATAL ORDER IN THE ROCK RECORD WITH COMPUTER GENERATED STRATIGRAPHIC LAYERS

Sedimentary rock layers are often deposited in a manner suggestive of cyclic changes in sea level due to Milankovitch forcing. A number of methods for testing the presence or absence of cyclicity in the resultant rock record have been developed. This project used cyclically produced, computer generated stratigraphic layers and the resulting simulated stratigraphic records were then tested. By modifying the simulation parameters, these methods could be analyzed to determine level of test success under varying circumstances. Stratigraphic data and rock layer information were simulated along a shallow, subsiding marine shelf using PHIL Basin Analysis software (PetroDynamics Inc.). Orbital forcing was simulated using one to six overlapping frequencies of varying period and water level change.

A number of methods used to test the rock record for stratal order were analyzed. Markov chain analysis and autocorrelation were used to demonstrate predictability in the lithofacies and to find repetition in the stratigraphic column that demonstrates a cyclic pattern as similar rock types are produced at certain water depths. One method measures thickness frequencies and compares them to a modeled Poisson (random) distribution. It has been suggested that the rock record exhibits a purely stochastic distribution due to the high correlation between rock layer thickness frequencies and the modeled random frequency. When testing thickness frequencies, the simulated records with controlled orbital forcing were compared against the theoretical random distribution of layers predicted by a Poisson distribution. Pearson's product-moment correlation was used to measure the quality of fit. For all simulations, the results indicate that cyclic processes produce patterns that are indistinguishable from random processes using this method. Changing environment has little effect on producing frequencies of that are dissimilar from the theoretical random distribution. The effect of water depth on the carbonate production curve tends to mask patterns in the rock record, and cyclicity is more apparent when production is depth invariant. Further methods need to be developed to test for cyclicity in the rock record.