Return to search

Discriminant Analysis of XRF Data from Sandstones of Like Facies and Appearance: A Method for Identifying a Regional Unconformity, Paleotopography,and Diagenetic Histories

The placement of an unconformable surface within a stratal succession affects the interpreted thickness of units and sequences in contact with that surface. Unit thickness influences the interpretation of basin subsidence, paleotopography, diagenesis, and depositional style. Accurate placement of an unconformity results in true formational thicknesses for formations associated with that unconformity. True thicknesses aid in producing more precise surface to subsurface correlations, isopach maps, and paleogeographic maps. An unconformity may be difficult to identify in the stratal succession due to similar rocks above and below the unconformity and the presence of multiple candidate surfaces. Using statistical discriminant analysis of XRF data, formations bounding an unconformity can be discriminated by elemental composition which results in delineation of the associated unconformity. This discrimination is possible even for rocks that do not have significant differences in provenance if they have experienced distinct diagenetic histories. Elemental differences can be explained by quantity and type of cement. Three discriminant models were created. These models were tested with samples from three formations of similar facies, appearance, and provenance that are all associated with the same regional unconformity. All data, regardless of location, facies, or tectonic feature were used to create the first model. This model achieved moderate success by correctly classifying 80% of known samples. In a second model, data were grouped by facies trends. Separating the data by facies resulted in 94% of known samples being correctly classified. This model was most useful for delineation of an unconformity and discrimination of formations. A third model based solely on location or local tectonic feature produced the best results statistically. 96% of known samples were classified correctly. This third model does not compare locations to each other, thus making it less robust. This last model contributes by adding detail to interpretations made with the facies trend model.

Identiferoai:union.ndltd.org:BGMYU2/oai:scholarsarchive.byu.edu:etd-4372
Date29 September 2012
CreatorsPhillips, Stephen Paul
PublisherBYU ScholarsArchive
Source SetsBrigham Young University
Detected LanguageEnglish
Typetext
Formatapplication/pdf
SourceTheses and Dissertations
Rightshttp://lib.byu.edu/about/copyright/

Page generated in 0.0026 seconds