HR: 09:15h
AN: MR21A-06    [Abstracts]
TI: Multiscale Analysis of Patterns Based on Random Pattern Subsets
AU: * Gerik, A
EM: axel.gerik@mytum.de
AF: Tectonics and Material Fabrics Section, Technische Universitaet Muenchen, Arcisstr. 21, Munich, 80333, Germany
AU: Suteanu, C
EM: cristian.suteanu@smu.ca
AF: Geography Department and Environmental Studies Program, Saint Mary's University, 923 Robie St., Halifax, NS B3H 3C3, Canada
AB: Natural geological patterns such as the distribution of minerals or fractures within a rock can provide important information e.g. the genesis of the rock or its deformation. In order to develop techniques that allow for the analysis of such patterns, artificial patterns, such as mathematical fractals, serve as input data for testing purposes. While the analyses of artificial patterns can often rely on complete patterns or datasets, most geoscientific applications are limited to one or multiple available subsets of a pattern. The use of natural patterns is further complicated as many ways of obtaining them involve a thresholding process. The parameters of this thresholding process are likely to be reflected in the data point density. Depending on the internal structure of the pattern, both of these limitations can have an influence on the analyses' results. To evaluate the sensitivity of different natural and artificial pattern, we suggest a multiscale approach which uses different sizes of random subsets. The required quantitative measurements are obtained by applying different fractal geometry-based analytical methods. Preliminary results indicate an interdependence between the obtained results and the size of the subsets. Moreover, the parameters of this relationship can be used to characterize and differentiate the patterns.
DE: 3625 Petrography, microstructures, and textures
DE: 4430 Complex systems
DE: 4460 Pattern formation
SC: Mineral and Rock Physics [MR]
MN: 2007 Fall Meeting