NO SCALE TOO SMALL: APPLYING REMOTE SENSING AND GEOGRAPHIC INFORMATION SYSTEMS (GIS) BASED SOFTWARE FOR QUANTIFYING MINERAL DISTRIBUTION PATTERNS ON PHOTOMICROGRAPHS

Spatial variations of mineral distribution patterns can provide important information about different geological processes, such as fabric development due to deformation, strain softening, etc. This presentation describes an approach to quantify mineral distribution patterns on microscopic as well as on hand specimen scales on suitable digital images using remote sensing and GIS based software, which are commonly used to analyze and quantify landscape patterns containing a mosaic of elements (patches) belonging to different classes. Digital images best suited for these types of analyses are those on which different mineral phases are associated with specific shades of color, and thus can be grouped into different classes containing individual mineral aggregates or patches. These software can be used to quantify the areal extent of different mineral phases within an image, the shape and size of individual mineral aggregates, their nearest neighbor distances, as well as their spatial distribution characteristics, such as dispersion or connectivity of each patch in relation to other patches within the same class.

For this work, digital photomicrographs captured under plane polarized light were used where the mineral phases could be broadly classified into a “dark” class where aggregates of hornblende grains made up more than 95% of the patches, and a “light” class including aggregates of quartz, feldspar and other colorless minerals. The images were classified using ERDAS Imagine 9.3® software, commonly used for analyzing remotely sensed images. The classified photomicrographs were analyzed using Fragstats 3.3®, which is a public domain spatial pattern analysis software program for quantifying landscape structures. Patch level metrics describing the shape, size and distribution characteristics of individual patches, and class level metrics including fragmentation, connectivity and dispersion of patches belonging to each class were calculated. A patch level attribute table was created for each analyzed image using ArcMap 9.3® software. Metrics thus obtained from photomicrographs from different parts of the same rock sample were compared to determine variations in mineral spatial patterns.