INTEGRATION OF SATELLITE REMOTE SENSING DATA FOR GOLD PROSPECTING IN TROPICAL REGIONS

Transcription

1 INTEGRATION OF SATELLITE REMOTE SENSING DATA FOR GOLD PROSPECTING IN TROPICAL REGIONS Amin Beiranvand Pour, Mazlan Hashim Geoscience and Digital Earth Centre (Geo-DEC), Research Institute for Sustainability and Environment (RISE), Universiti Teknologi Malaysia (UTM), UTM Skudai, Johor Bahru, Malaysia KEY WORDS: Landsat Thematic Mapper (TM); Gold prospecting; geological mapping; Peninsular Malaysia ABSTRACT The area under investigation is the Bau gold mining district in the State of Sarawak, East Malaysia, on the island of Borneo. It has tropical climate with limited bedrock exposures and the other constraints imposed in the environment. Bau is a gold field with Carlin style gold deposits. Geological analyses coupled with remote sensing data were used to detect structure elements and hydrothermally altered rocks associated with gold mineralization. The Landsat Enhanced Thematic Mapper + (ETM+), the Phased Array type L-band Synthetic Aperture Radar (PALSAR) and the Hyperion data were integrated to carry out lithological-structural mapping of mineralized zones in the study area and surrounding terrains. Hydrothermally alteration mineral zones were detected along the SSW to NNE structural trend of the Tai Parit fault that corresponds to the areas of occurrence of the gold mineralization in the Bau Limestone. The results show that the known gold prospects and potentially interesting areas are recognizable by the methods used, despite limited bedrock exposure in this region and the constraints imposed by the tropical environment. The approach used in this study can be more broadly applicable to provide an opportunity for detecting potentially interesting areas of gold mineralization using the integration of ETM+, PALSAR and the Hyperion data in the tropical/sub-tropical regions. 1. INTRODUCTION In this study, the possibility of identifying hydrothermally altered rocks, faults and fractures associated with hydrothermal ore mineralization in the tropical environments is examined using the Landsat Enhanced Thematic Mapper + (ETM+), the Phased Array type L-band Synthetic Aperture Radar (PALSAR) and the Hyperion remote sensing data. Bau gold mining district in Sarawak province, eastern Malaysia, on the island of Borneo in Southeast Asia has been selected (Fig. 1). It is located between 1 25 N in longitude and E in latitude in 25 km southwest of Kuching city, Sarawak. Figure 1. Location of the study area in Southeast Asia. The Bau gold mining district is on westernmost Borneo. It is located in the state of Sarawak which is part of Malaysia. Bau is a gold field with Carlin style gold deposits. Carlin-type gold deposits are sediment-hosted disseminated gold

2 deposits, which are hosted by a variety of permeable sedimentary rocks, especially thinly bedded, silty dolomites or limestones, cut by high-angle faults. Nearly all the deposits contain felsic intrusive rocks, commonly in the form of dikes or sills. Orebodies may be confined to fault zones or may be irregularly shaped replacements in the adjoining rocks. Gold-bearing rocks underwent decalcification, silicification, and argillic alteration, and are associated closely with structurally localized replacement of carbonate rock by jasperoid. These characteristics can be detected as indicators for the initial stages of Carlin-type gold exploration using remote sensing data (Pour et al., 2014). To date, remote sensing study is not carried out in the Bau mining district to identify hydrothermally altered rocks and structure elements associated with ore mineralization for exploring potentially interesting areas of Carlin style gold mineralization. Therefore, this investigation attempted to acquire comprehensive and accurate information for exploring potentially interesting areas of gold mineralization using the integration of the ETM+, PALSAR, and Hyperion remote sensing data. The objectives of this research are: (1) To introduce an approach for detecting hydrothermally altered rocks using ETM+ and Hyperion data at regional and district scales in heavily vegetated tropical rainforest regions. (2) To identify structure elements associated with fault-controlled and stratiform gold orebodies using the PALSAR data. (3) To discover the potentially interesting areas of gold mineralization at the Bau mining district. 2. MATERIALS AND METHODS Landsat Enhanced Thematic Mapper+ (ETM+), the Phased Array type L-band Synthetic Aperture Radar (PALSAR) and the Hyperion data were used in this investigation. ETM+ image was obtained through the U.S. Geological Survey Earth Resources Observation System (EROS) Data Center (EDC). It was acquired during dry season with 5% cloud coverage on August 3, 1998 for the Bau mining district and surrounding areas. The images were pre-georeferenced to UTM zone 40 North projection using the WGS-84 datum. Full polarimetry (multi-polarization), off nadir pointing function and other functions of PALSAR improved the accuracy of analyzing geological structure, distribution of rocks, and expected to be used for the first stage of ore deposits and hydrocarbon exploration and environmental protection. Furthermore, in the tropical climate synthetic aperture radar data are particularly appropriate, because microwave signals can penetrate the persistent cloud coverage. Structural geology investigations that are searching for ore mineral deposits and hydrocarbon traps can be developed using this type of data in tropical/sub-tropical terrains. A PALSAR Fine mode Level 1.5 scene was obtained from the Earth and Remote Sensing Data Analysis Center (ERSDAC) Japan for the Bau mining district and surrounding areas. It was acquired on 14 March Level 1.5 product is the product of high resolution mode by single polarization (HH or VV), which is geo-reference and geo-coded. The data used in this study has high resolution mode with 6.25 m pixel spacing and off-nadir angle (34.3 ). A cloud-free level 1B Hyperion image was obtained through the U.S. Geological Survey Earth Resources Observation System (EROS) Data Center (EDC). Hyperion scene was also obtained during dry season on August 28, 2004 for the Bau mining district, which was pre-georeferenced to UTM zone 40 North projection using the WGS-84 datum. The Hyperion data were tested for hydrothermal alteration mapping at district scale in the tropical environment. ETM+, PALSAR and Hyperion data of the Bau were processed using the ENVI (Environment for Visualizing Images) version 5.0 software package. Spectral signatures observed in the tropical remote sensing imagery are related almost entirely to vegetation because it is dominated material on earths surface in the tropical/sub-tropical environment. So, the separation and suppression of vegetation effects is significant to test the application of optical remote sensing data for detecting mineralogy of soils and rocks in areas of tropical/sub-tropical climate. To suppress and separate the erroneous effects of vegetation in hydrothermal alteration mapping and unveiling the lithology of the tropical terrain the principal components analysis (PCA - Singh and Harrison, 1985) was implemented on specific spectral indices of ETM+ data. Vegetation index (band ratio of 4/3), clay minerals index (band ratio of 5/7), ferric iron oxide index (band ratio of 3/1), and ferrous iron oxide index (band ratio of 5/4) were used to generate PCA image components. The image eigenvectors and eigenvalues were obtained from PCA using covariance matrix on indices. PCA outputs are presented as tables of statistic factors and selected PCA images from these transformations are reproduced in figures to support the discussion. Statistic results are shown in Table 1. Table 1: Principal components analysis on band ratio indices of ETM+, selected spatial subset scene covering the Bau

3 gold mining district. Input band ratio indices Band ratio 4/3 Band ratio 5/7 Band ratio 3/1 Band ratio 5/4 Eigenvalues (% ) PC PC PC PC The second moment co-occurrence texture filter was applied to the Lee-filter resultant PALSAR image for extracting faults and lineaments. It is based on the co-occurrence matrix, including mean, variance, homogeneity, contrast, dissimilarity, entropy, second moment and correlation. The co-occurrence texture filter uses gray-tone spatial dependence matrix to calculate texture values. This is a matrix of relative frequencies with which pixel values occurs in two neighboring processing windows separated by a specified distance and direction. It shows the number of occurrences of the relationship between a pixel and its specified neighbor (Park and Chen, 2001). We used the pixels in the 3*3 base window and the pixels in a 3*3 window that was shifted by 1 pixel for implementing the co-occurrence matrix. High gray scale 64 was setting to reduce the shade and highlight the dark pixels of fracture zones. Linear spectral unmixing was applied on VNIR and SWIR bands of Hyperion for mapping iron oxide/hydroxide minerals and clay mineral assemblages associated with gold mineralization at district scale. Linear Spectral Unmixing (LSU) method is used to determine the relative abundance of materials that are depicted in multispectral or hyperspectral imagery based on the materials' spectral characteristics (Shimabukuro and Smith, 1991). The reflectance at each pixel of the image is assumed to be a linear combination of the reflectance of each material (or end-member) present within the pixel. This technique also known as sub-pixel sampling, or spectral mixture analysis, is a widely used procedure to determine the proportion of constituent materials within a pixel based on the materials spectral characteristics. AIG-developed hyperspectral analysis processing methods (Kruse et al., 2003) were used to extract end-member spectra from Hyperion subsets for applying linear spectral unmixing technique in this study. Linear Spectral Unmixing (LSU) was applied to a selected spatial subset scene covering the Bau gold mining district for district scale mineral mapping purposes. Two spectral subsets of Hyperion data were analyzed separately to detect iron oxide/hydroxide minerals and hydroxyl-bearing (clay) alteration mineral assemblages. The first subset (VNIR) covering 90 bands (excluded overlapping and inactive bands) between 0.4 to 1.3 μm is used for highlighting iron oxide/hydroxide minerals, and second subset (SWIR) of 40 bands between 2.00 to 2.40 μm (185 to 225) is processed for detecting hydroxyl-bearing (clay) alteration mineral. Vegetation end-member spectrum has been removed from the output to suppress the erroneous effects of vegetation in hydrothermal alteration mapping. In fact, the spectral unmixing bands dominated by erroneous effects of vegetation were not used for the alteration mapping analysis. 3. RESULTS AND DISCUSSION After analyzing the results of principal component analysis transformation for specific spectral indices of ETM+ data, considering magnitude and sign of the eigenvector loadings and percentage of eigenvalues for selected spatial subset scene covering the Bau gold mining district, it is realized that the first principal component (PC1) accounts percent of total eigenvalue, which is higher value among the PCA images in the scene (Table 1). A PCA image with higher eigenvalue contains most of the spectral information in the scene. All of the eigenvector loadings for the PC1 are positive (Table 1), thus the differentiation between materials (vegetation and alteration minerals) using the specific spectral indices in the PC1 image is unattainable. Vegetation index 4/3 band ratio is very sensitive to vegetation amount and relatively insensitive to lithologic variation. Clay mineral index 5/7 band ratio generally varies with the abundance of hydroxyl-bearing minerals. However, the clay mineral index can also be affected with vegetation amount. Eigenvector loadings for the PC2 indicate that PC2 image has highest value of vegetation variations, which is statistically dominant in the image due to the high positive contribution from the vegetation index (0.730) and negative contribution for the specific hydrothermal mineral indices (Table 1).

4 PC3 image manifests desired information related to Al (OH)-bearing (clay) minerals with very squat disturbance of vegetation due to very low negative (contribution) eigenvector loadings for the vegetation index (-0.097) and very strong positive eigenvector loadings for the clay mineral index (0.870). Eigenvector loadings for ferric iron oxide index (3/1) and ferrous iron oxide index (5/4) are (-0.329) and (-0.352), respectively. Hence, the contribution of iron oxide minerals in the extracted spectral information from the PC3 image is very low. Figure 5 shows a regional view of unveiling the lithology in the background of PC3 image for the study area. Geological structures (Bau anticline), bedrock lithology, hydrothermally rocks (as bright pixels), alluvial patterns and rivers can be easily distinguished in the PC3 image (Fig. 2). Figure 2. A regional view of unveiling the lithology in the background of PC3 image. Figure 3 indicates selected spatial subset scene covering the Bau gold mining district and surrounding terrains, hydrothermally altered rocks are especially marked in the scene by red color. Most of the detected altered zones are well-known prospects that pointed by arrow in the Figure 3, and some identified potentially interesting areas for gold mineralization can also be considered for more investigations in the study area. However, the identified new potentially interesting areas have very small rock exposures in a regional scale of lithological mapping. Jugan, Bukit Sarin, Sirenggok, Jumbusan and Kapor prospects are marked by arrows in the Figure 3 to illustrate the location of the well-known prospects in the Bau area. Figure 3. Selected spatial subset scene covering the Bau gold mining district and surrounding terrains, hydrothermally altered rocks (clay minerals) are indicated as red color.

5 The linear spectral unmixing technique was applied to the selected spatial subset scene of the Hyperion covering the Bau gold mining district for detailed hydrothermal alteration mapping at district scale. The end-member spectra were extracted using the AIG-developed hyperspectral analysis processing methods, and then compared with USGS spectral library as reference spectra. Considering the shape and position of absorption feature, the minerals are characterized as hematite-limonite-goethite for first subset. The extracted signatures for second subset suggest the existence of hydroxyl minerals such as kaolinite-sericite. Figure 4 shows image map of the selected spatial subset scene for first subset (VNIR), showing spectrally predominant iron oxide minerals as colored pixels that are overlaid on the gray-scale image background of Hyperion band 36. The abundance of iron oxide minerals is represented as green (high abundance) and purple pixels (low abundance) in the Bau subset scene (Fig. 4). Figure 4. Image map of VNIR bands of Hyperion shows the abundance of iron oxide minerals in Bau gold mining district. Gold prospects and potentially interesting areas are pointed by arrow. Figure 5 shows image map of the selected spatial subset scene for second subset (SWIR). Kaolinite-sericite abundance is illustrated as red (high abundance) and yellow (low abundance) pixels that are overlaid on the gray-scale image background of Hyperion band 205. Detected pixels are associated with fault and fracture structures in the Bau scene. Most of the gold prospects, including Kapor, Taiton, Umbut, Bukit Young, Tai Parit, Bukit Sarin, Jugan, Bekajang, Jambusan are located in high abundance clay minerals area that pointed by arrows in the scene (Fig.5). The Co-occurrence texture filter identified the edges of the surrounding pixels, highlighting the fault and lineament structures in the PALSAR image. Hence, geological structures are obviously detected in the study area at both regional and district scales. Figure 6 shows the geological stuctures, including Bau anticline, Bau anticline axis, Tai Parit fault and several linear and curvature features in the Bau and surrounding areas at a regional scale. Bau anticline (NE-SW trending) and its axis are located in the south and southwestern part of the PALSAR image (Fig. 6). Tai Parit fault (NNE-trending) and several linear and curvature features (WNW-to NW-trending) dissected the Bau limestone, which are identifiable in the north western part of the scene (Fig. 6).

6 Figure 5. Image map of SWIR bands of Hyperion shows the abundance of clay minerals in Bau gold mining district. Gold prospects and potentially interesting areas are pointed by arrow. Figure 14. A regional view of the geological stuctures in the Bau and surrounding areas. 4. CONCLUSIONS The results presented in this study demonstrate the importance and advantages of the combined use and integration of ETM+, PALSAR and the Hyperion remote sensing data in detecting potentially interesting areas of Carlin style gold mineralization in tropical/sub-tropical regions. Structurally controlled gold mineralization indicators, including iron oxides, clay minerals and faults and fractures in the Bau gold mining district have been detected using the remote sensing satellite data and the approach used in this investigation.

7 ACKNOWDEGMENT We are thankful to the Universiti Teknologi Malaysia (UTM) for providing the facilities for this investigation. We would also like to express our great appreciation to Research Institute for Sustainability and Environment (RISE) of Universiti Teknologi Malaysia (UTM) for supporting research invironments. REFERENCES Kruse, F.A., Bordman, J.W. and Huntington, J.F., Comparison of airborne hyperspectral data and EO-1 Hyperion for mineral mapping. IEEE Transactions of Geosciences and Remote Sensing, 41(6), Park, B., Chen, Y.R., Co-occurrence matrix texture features of multi-spectral images on poultry carcasses. Journal of Agriculture Engineering Research, 78 (2), Pour, B. A. and Hashim, M., 2011a. Application of advanced spaceborne thermal emission and reflection radiometer (ASTER) data in geological mapping. International Journal of the Physical Sciences, 6 (33), pp Pour BA, & Hashim, M., 2011b. The Earth Observing-1 (EO-1) satellite data for geological mapping, southeastern segment of the Central Iranian Volcanic Belt, Iran. International Journal of the Physical Sciences, 6(33), Pour, B. A. and Hashim, M., ASTER, ALI and Hyperion sensors data for lithological mapping and ore mineral exploration. Springerplus, 3(130), pp Pour, B. A., Hashim, M. and Marghany. M., Exploration of gold mineralization in a tropical region using Earth Observing-1 (EO1) and JERS-1 SAR data: a case study from Bau gold field, Sarawak, Malaysia. Arabaian Journal of Geosciences 7(6), pp Shimabukuro, Y.E., and Smith, J.A., The least-squares mixing models to generate fraction images derived from remote sensing multispectral data. IEEE Transactions of Geosciences and Remote Sensing, 29, Singh, A. and Harrison, A., Standardized principal components. International Journal of Remote Sensing 6, pp