MULTIDIMENSIONAL SCALING IN PROVENANCE STUDIES OF PLATFORM, BASIN, AND PASSIVE MARGIN SUCCESSIONS OF NORTH AMERICA

Highly efficient, low-cost grain age acquisition for detrital zircon samples has paved the way for the exponentially growing field of zircon provenance analysis. Ever-increasing datasets, while providing improved understanding of sediment dispersal through time and space, require effective means to qualitatively and quantitatively visualize and deconvolve changes in provenance dispersal. The application of multidimensional scaling (MDS) to detrital zircon data proves to be an effective visualization tool by translating pairwise dissimilarities between U/Pb ages of multiple samples into Euclidean space, with greater dissimilarities represented as greater interpoint distances. By applying MDS to a suite of synthetic age data, we to examine several key issues relating to the interpretation of zircon provenances. These include: 1) how varying methodologies (e.g. PDPs, KDEs) for representing intra-sample age distributions and the measure of their dissimilarities can affect MDS configurations; 2) how random sampling can alter resulting MDS ordination, and 3) how MDS configurations, and more importantly differentiation between sample groupings, is affected by varying degrees of overlapping, shared, and/or unique age components and MDS dimensionality. With these limitations in hand, we apply MDS to several real-world detrital zircon datasets from North America. The approach clearly identifies abrupt upsection stratigraphic variations within Paleozoic-Mesozoic deposits of the Colorado Plateau and the Big Horn Basin, as well as lateral variations in Paleogene sediment sources to the Eocene Wilcox Formation of the central Gulf Coast. The application of MDS to both synthetic and real-world examples demonstrates that the approach affords significant advantages in the geologic interpretation of detrital zircon data.