DEVELOPMENT OF MICROGRAPHIC INFORMATION SYSTEMS (MIS) FOR PETROGRAPHIC STUDY: AN EXAMPLE FROM THE HARTFORD BASIN, CONNECTICUT

Contextual spatial analysis of thin-section images is being used to develop an MIS for the West Rock diabase sill of the western Hartford Basin. Petrographic analysis, including the determination of modal abundances, textural relationships, crystal-size distributions, rock fabrics, and mineral orientations, is a fundamental component of the current investigation. Petrographic information is traditionally gathered with a petrographic microscope in a generally qualitative manner aimed at describing minerals and textures. Acquiring this petrographic information currently involves manual interpretation of mineral relationships and manual digitization of crystal outlines. The latter produces polygons used to determine crystal size distributions and crystal orientations for crystallization models and compaction analysis. The method currently under development is implemented with eCognition by Definiens, an object-oriented image analysis program designed for remote sensing applications, and involves the segmentation of images of rock thin sections into objects available for classification as different mineral phases present in the rock sample. A fuzzy-logic, supervised classification is then applied, based on the spectral and spatial properties of the minerals found in the sample. Because polygons are semi-automatically drawn and classified about individual minerals in the program, the time-consuming task of digitizing crystal boundaries becomes unnecessary. Micrographic information for each mineral, including intrinsic crystal properties (color in plane and polarized light, size, shape), topological relationships (location in the sample, neighbor relationships, distance to other minerals), and contextual relationships (the petrographic texture), is then available for transfer to an MIS database for extensive spatial analysis in ArcGIS by ESRI. This petrographic information can be combined with larger-scale data from electron microprobe and electron microscope analysis in the MIS, making it also available for query in further spatial analysis. The ultimate goal of this work is to outfit the petrographer with a method of image analysis that will successfully quantify the petrographic texture of a thin section and make that data available for query in an MIS database.