Quick Search:    advanced search

GSW Home   GeoRef Home   My GSW Alerts   Contact GSW   About GSW   Journals List   Help

This Article Free via Open Access: OA OA Abstract OA Full Text Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Ott, N. Articles by Tassara, A. Search for Related Content GeoRef GeoRef Citation

GIS analyses and favorability mapping of optimized satellite data in northern Chile to improve exploration for copper mineral deposits

¹ Freie Universität Berlin, Institute for Geological Sciences, Malteserstrasse 74-100, Hs. N, D-12249 Berlin, Germany

View larger version (131K): [in this window] [in a new window]   Figure 1. Morphotectonic units of the northern Chile forearc. Inset shows the location of the study area in the context of the Central Andes. Black triangles are Holocene volcanoes. Red points with white names are world-class porphyry copper deposits of Eocene–Oligocene age.

View larger version (58K): [in this window] [in a new window]   Figure 2. Perspective view of different thematic layers of the database in the vicinity of La Escondida mining district. Upper layers represent optimized Landsat data derived from band ratioing, principal component analysis (PCA), and inverse PCA. Lower layers represent topographic data, lithology, and aeromagnetic data. Bottom layer is one of the calculated favorability maps. This study focuses on optimized Landsat data, whereas the other data sets are not used at this time.

View larger version (119K): [in this window] [in a new window]   Figure 3. The atmosphere selectively scatters shorter wavelengths of light. Atmospheric scattering produces haze, which results in low image contrast and poor brightness. To reduce these effects, the value of an intercept offset is substracted from the digital numbers of each spectral band (histogram minimum method). The left image shows uncorrected Landsat Thematic Mapper (TM) data (bands 7, 4, and 1 in red, green, and blue) of the Chilean Coastal Cordillera with apparent haze and dull colors. The right image shows corrected Landsat TM data with resultant high contrast and color saturation. Correction of atmospheric scattering is essential for further image optimization and classification.

View larger version (125K): [in this window] [in a new window]   Figure 4. Data adjustment of Landsat data by histogram matching is necessary when two or more contiguous images are to be joined to form a mosaic. Landsat data acquired at different times under different seasonal and climatic conditions will result in inhomogeneous image appearance. Note the oblique line in the left image, which represents the boundary between two different Landsat frames. The boundary is indicated by a crosshair. Rocks and other surfaces show different hues. The upper part of the image is darker in general, whereas the lower part is brighter and yellower. Thus, the differing appearance needs to be adjusted. The resulting image mosaic at the right shows homogeneous hues without any apparent boundary of Landsat frames.

View larger version (40K): [in this window] [in a new window]   Figure 5. Simplified model of hydrothermal alteration zones with porphyry copper deposits (modified from Lowell and Guilbert, 1970).

View larger version (54K): [in this window] [in a new window]   Figure 6. Extensive mining activities produce dust and other pollutants that cover the wider mining area. Spectral signatures of rocks are contaminated by dust coverage and generate anomalous colors. Synmining satellite data include such contamination and are not suitable for spectral mapping on the basis of spectral properties. Pre-mining satellite data are not affected by contamination and show authentic spectral properties. To demonstrate the influence of mining activities, a difference image is calculated for La Escondida mining district of Escondida Norte, and Zaldivar pre-mining Landsat TM and synmining Enhanced Thematic Mapper (ETM+) data for detection of changes. Black indicates no change of digital numbers between TM and ETM data. Red indicates a decrease of digital numbers by 10% (shadow and water), whereas green indicates an increase of 10% generated from dust coverage with resultant higher digital numbers. Therefore pre-mining TM data are used for image optimization.

View larger version (12K): [in this window] [in a new window]   Figure 7. Bivariate plots of data from two Landsat bands produce an elongate ellipse of points in the two-dimensional feature space because of strong correlation. Principal component analysis (PCA) begins by shifting the origin of the plot (A) to a point defined by the mean values of the two data sets (B). The axes are then rotated so one is aligned with the maximum variance in the data (C). This axis becomes the first principal component (PC), combining contributions from both bands. The second axis, perpendicular to the first, expresses the lower variance in the data and becomes the second principal component. Furthermore, because successive components are chosen to be orthogonal to all previous ones, the data are uncorrelated.

View larger version (85K): [in this window] [in a new window]   Figure 8. The PCA image is the result of the fifth (clay), fourth (lithology), and third (lithology) principal components displayed in red, green, and blue. This PCA color image reflects best the distribution of altered diorites and clay minerals. Spectral anomalies of altered rocks are highlighted in purple to red and can be recognized easily. Anomalous colors represent various rock types. Sedimentary rocks are displayed in bluish to greenish colors, felsic volcanics are displayed in pinkish to purple colors, and intrusives are displayed in deep purple and red colors. Quaternary deposits are shown in various but bright colors. In summary, spectral differences between rocks may be more apparent in PC images than in individual bands. Anomalous PC colors derived do not correspond to spectral reflectance and absorption of rocks because of the data transformation. Current open pits at La Escondida mining district are marked by symbols.

View larger version (11K): [in this window] [in a new window]   Figure 9. Bivariate plots of two bands with principal component axes (A). The first principal component has been stretched after rotation of the axes to principal component space (B). In the next step the second principal component has been stretched (C). This produces a decorrelation in the principal component space. Decorrelated data are rotated back to the original feature space (D). Stretching and back rotation of principal components are called inverse principal component analysis. The effect presented by this technique produces chromatically enhanced images.

View larger version (78K): [in this window] [in a new window]   Figure 10. Image is a color composite of the highly decorrelated bands 5, 3, and 1 in red, green, and blue. This inverse PCA color composite reflects best the distribution of altered intrusive rocks and clay minerals. Spectral anomalies of altered intrusive rocks are displayed in yellow and reddish colors. Sedimentary rocks are displayed in various colors, ranging from pink to blue and green. Felsic volcanic rocks are displayed in deep blue to blue green. Alluvial deposits show characteristic pink and purple colors but vary with chemical composition of bedrock. Because of more or less original hues of rocks, the image is more interpretable than the principal components with resulting anomalous hues. Current open pits at La Escondida mining district are marked by symbols.

View larger version (49K): [in this window] [in a new window]   Figure 11. Spectral reflectance curves of iron minerals superimposed with Landsat TM bands from the visible and near infrared region. In the visible blue region (Landsat band 1) iron minerals show low reflectance owing to strong absorption (Fe-O charge transfer), whereas in the visible red and near infrared regions (Landsat bands 3 and 4) there is high reflectance owing to strong reflection. Calculation of band ratios highlights the occurrence of iron minerals in rocks (modified from Drury, 2001).

View larger version (86K): [in this window] [in a new window]   Figure 12. This image shows a color ratio image (band ratios 5/7, 3/1, and 4/3 in red, green, and blue) derived from Landsat TM data. Digital enhancement and information extraction allow discrimination of altered intrusive rocks from unaltered rocks. Altered intrusive rocks are highlighted in yellow and red colors. Sedimentary rocks are displayed in blue colors. Felsic volcanic rocks show brown to purple colors together with green. Alluvial deposits show pink to purple and blue colors, depending on the bedrock composition. Current open pits at La Escondida mining district are marked by symbols.

View larger version (47K): [in this window] [in a new window]   Figure 13. Clay-band ratio image (5/7) derived from Landsat TM data. The 5/7 band ratio has bright signatures for altered rocks, because the lower reflectance values of band 7 are in the denominator, which results in higher ratio values. Rocks with high clay-mineral content can be clearly identified from spectral anomalies with high pixel values displayed in white. Color variations of ratio images express more geologic information than conventional color images. Current open pits at La Escondida mining district are marked by symbols. Note the high spatial correlation of spectral anomalies with current open pits.

View larger version (15K): [in this window] [in a new window]   Figure 14. Spectral reflectance curves and properties of rock samples taken from the open pit at Escondida Norte. Reflectance curve is used as a reference for supervised spectral mapping of dioritic complexes. Note low reflectance in Landsat TM bands 1 and 7, whereas there is high reflectance in bands 3 and 5. Band 6 is part of the thermal infrared and is not suitable for spectral mapping.

View larger version (44K): [in this window] [in a new window]   Figure 15. This image shows a spectral map of Landsat TM data wherein the data are trained and classified with spectral properties of dioritic rocks. Rocks of dioritic composition can be clearly identified from spectral anomalies with high pixel values displayed in white. Note the high spatial correlation of spectral anomalies with current open pits.

View larger version (79K): [in this window] [in a new window]   Figure 16. Classification result derived from calculated PC image (see Fig. 8). Target areas of altered rocks in the classified PC image correspond to reddish to purple colors in the PC image. Clustering of target areas is well defined in the central part of the image, whereas in the southern and northern parts, clustering is thinned out. Number of classified target-area pixels derived from PCA is 3.0% of total pixel number.

View larger version (76K): [in this window] [in a new window]   Figure 17. Classification result derived from inverse PC image (see Fig. 10). Target areas in the classified inverse PC image correspond to yellow colors in the inverse PCA image. Clustering of target areas is well defined in the central, southern, and northern parts of the image. In other areas, the target clustering is thinned out. Number of target-area pixels derived from inverse PCA is 3.2% of total pixel number.

View larger version (78K): [in this window] [in a new window]   Figure 18. Classification result derived from clay-band ratio image (see Fig. 13). Target areas in the classified clay-band ratio image correspond to light colors that represent high digital numbers in the clay-band ratio. Clustering of target areas is well defined in the central and southern parts of the image, whereas in the northern parts, clustering is thinned out. Number of target-area pixels derived from clay-band ratioing is 3.8% of total pixel number.

View larger version (78K): [in this window] [in a new window]   Figure 19. Classification result derived from spectral mapping (see Fig. 15). Target areas in the classified spectral-mapping image correspond to light colors in the spectral map. Clustering of target areas is well defined all over the image, but in small areas. Number of target-area pixels from spectral mapping is 0.4% of total pixel number.

View larger version (83K): [in this window] [in a new window]   Figure 20. Favorability map from the training site at La Escondida mining district. Calculated favorability map shows the spatial distribution of altered rocks. Note the high spatial correlation of host rocks with the current open pit at La Escondida (location 1). In the pre-mining Landsat data, there is evidence for altered magmatic rocks for the current pit at Escondida Norte (location 2), because it is calculated to be a major target area and verifies the results of this favorability mapping. Current open pit at Zaldivar (location 3) is not calculated as a major target area. A reason for this might be that Zaldivar (1) is a supergene copper deposit, formed by transported aqueous solutions in addition to precipitation by groundwater and is therefore not correlative with altered rocks near the surface, or (2) is under postmineralization rock cover and cannot be detected by satellite sensors.

View larger version (77K): [in this window] [in a new window]   Figure 21. Favorability map from the training site at Quebrada Blanca mining district. The calculated favorability map shows the spatial distribution of altered rocks. Note the high spatial correlation of host rocks with the current open pit at Quebrada Blanca (location 1).