3. METHODS

3.1 IMAGE PROCESSING
3.1.1 Rectification. A Landsat MSS image quarter which showed the subject features was obtained and rectified, using 14 widely spaced, visually chosen ground control points on the image, to a tiled layout of the following paper USGS 1�x 2� quadrangles: Rolla, St. Louis, Springfield, and Jefferson City. A second-order transformation matrix was chosen, with the acceptable total root mean square (RMS) error limited to a 2 pixel distance. After the discard of two ground control points, a total RMS error of 1.81443 was achieved. The nearest neighbor resampling method was then used during the transformation, since preservation of the lineament on the image was desired. Projection was in the Universal Transverse Mercator (UTM) coordinate system, using the Clarke 1866 projection spheroid. Since the Landsat mission number was unknown, a pixel resolution of 79 x 79 meters was chosen, as this is the most common. ERDAS 7.4 was used for the rectification, and for all ensuing registration and digitization. All data was then imported into ERDAS Imagine for raster manipulation and annotation overlay of the maps discussed below.

3.1.2 Enhancement. Initial enhancement for clarity was limited to simple contrast, with the brightness set to 30% and the contrast at 70%. Because one of the causes of circular features, artifact of illumination, is a product of the imaging process, and because the origin of the arcuate features on the image was not known, individual bands were inspected in order to determine the spectrographic origin of the feature. To this end, a principle component analysis (PCA) was also performed as were three unsupervised classifications using 20, 10, and 4 classes each. Band ratios using all possible combinations were made, and a color composite ratio image chosen using a modified form of Chavez�s Optimum Index Factor (Jensen, 1996): Since statistical methods of correlation would by definition consider all the data in the image, which was not desired, the most highly correlated bands for the features of interest were chosen visually. In addition, the image was put through 3 individual low pass filters of 3x3, 5x5, and 7x7 convolution kernels each. Since the first cover in the area of interest is vegetation, common indices used in plant discrimination, specifically NDVI (Normalized Difference Vegetation Index), [IR/R]1/2 (square root of the Infrared/Red wavelengths), and the Tasseled Cap transformation, were also used.

3.2 GEOLOGICAL METHODS
3.2.1 Map Analyses. To determine the location of the arcuate features, overlays of the public land survey system townships and of county boundaries were also digitized from the current Rolla, St. Louis, Jefferson City, and Springfield 1�x 2� quadrangle topographic sheets. Several existing maps were digitized for use as annotation layers. Most of these were taken from the CUSMAP study done on the Rolla 1�x 2� quadrangle, and include mylar overlays of 1) A Bouguer gravity map (Hildebrand et al., 1979, and 2) A USGS 1979 aeromagnetic map; and paper maps showing 3) Structural contours on top of the Precambrian (Kisvarsanyi, 1979), 4) The Upper Cambrian strata (Anderson, 1979), and 5) Surface geology (Pratt et al., 1992). This later map also shows sinkholes in a selected area of Phelps County, and faults for the entire quadrangle. Surface topography was not digitized: The area of interest is on the eroded surface of an almost horizontally bedded plateau so significant elevation changes were evident on the geologic map. In addition, any further possible correlation of the circular expression with terrain was determined by comparing the digitized townships on the overlay to those on the Rolla 1�x 2� quadrangle. Isopachs of the individual Cambrian formations (Thacker & Anderson, 1979a, b, c) were also visually inspected.

Fig. 3.1
3.2.2 Petrologic and Field Studies. Thin sections made from three existing drill cores and one existing rotary well drilled to basement in the feature were obtained from the DNR. Unpublished notes on file at the DNR, describing these cores and cuttings were also obtained, since the actual core samples from these four holes were not available. Two complete cores, which had been drilled to basement in the feature were available for inspection. The locations of all six of these cores were plotted as an annotation overlay [Fig. 3.1].

The two available cores, numbers 4 and 6 on fig. 3.1, were inspected macroscopically, and 4 thin sections made from their dominant rock types. Identification of the component minerals in the rocks was made by optical methods under the polarizing microscope. Plagioclase composition was determined by the Michel-L�vy method, and percent composition of minerals was determined by visual estimate. The thin sections from the remaining cores were also examined, with whole-rock descriptions taken from DNR notes.

In addition to the petrographic study, a surface reconnaissance of the area was undertaken. A Magellan GPS 2000 XL global positioning system unit was used in the field to determine the location of any unusual geologic or cultural feature which might contribute to the circular satellite image pattern.


Table of Contents | Bibliography | Next Page

Hosted by www.Geocities.ws

1