MODAL MINERALOGICAL ANALYSIS OF ROCKS IN THIN SECTION USING STATISTICAL ANALYSIS OF X-RAY MAPS IN ARCGIS

We have conducted modal analysis of thin sections of the Luna Vista sill using classification routines in ArcGIS. The technique is similar to other classification schemes (e.g. a supervised multispectral classification of Landsat data), but uses X-ray maps generated with an Energy Dispersive X-ray detector instead of spectral “bands” of electromagnetic radiation. We are using a specific adaptation of this ArcMap capability created by Ortolano and others (2014).

The Luna Vista Sill (LVS) is located in the Christmas Mountains in southern Brewster County. The LVS has a peralkaline quartz trachyte composition. The mineralogy of the rock includes alkali feldspar, arfvedsonite, eckermannite, ferro-eckermannite, aegirine augite, apatite with rare earth elements (REEs), zircon, ilmenite, hematite, calcite, and quartz. X-ray maps of thin sections of the LVS have been created for the elements Si, Ti, Al, Fe, Mg, Mn, Ca, Na, and K using an Energy Dispersive detector on a JEOL 6010LA SEM at the Sul Ross State University Analytical Laboratory. The X-ray maps are raster images in tiff format that include an 8-bit (0-255) signal per pixel for each element. The technique includes the initial application of a "Low Pass" or a "Focal Median" filter to reduce potential instrument noise, followed by a Principle Component Analysis (PCA). This technique exploits the natural variation in the multiband data by creating a new multivariate space and a new multiband raster with the same number of principle components as were in the original X-ray map set. This is followed by an unsupervised or a supervised classification to determine the mineral "classes" that exist in the thin section. The results can be used to estimate the modal mineralogical abundance in the rocks.