Hydrothermal Alteration Mapping in SW Birjand, Iran, Using the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) Image Processing

MATERIALS AND METHODS

In this research, we have analyzed ASTER level 1B data for the identification of hydrothermal alteration. Processing procedures were done by ENVI 4.0 software. The investigated area is within an approximately 400 km² subset of the ASTER scene (1b-090601-322, Acquisition data 2001/06/18). VNIR and SWIR bands have been used for this study. The images have been corrected by using control points from topographic sheets. Both subscenes are jointed together to form a single image. Different image processing techniques such as false color composite images, ratio images, color composite ratio image, Principal Component Analysis (PCA) image, Intensity-Hue-Saturation (IHS) transformation, filtering, supervised classification and unsupervised classification are normally used for delineating the favorable areas for further exploration. In this study, we have used four different processing techniques for detection of porphyry Cu-Au and epithermal Au deposits in eastern Iran. These are as follow: 1-false color composite, 2-color composite ratio images, 3-Principal Component Analysis (PCA) and 4-supervised classification.

RESULTS AND DISCUSSION

Figure 2 shows reflectance spectra of alunite and three common hydrothermal clay minerals illite, kaolinite and montmorillonite. These minerals have distinctive absorption features (reflectance minima) as wavelengths within ASTER bands 5 (2.145-2.185 μm), 6 (2.185-2.225 μm) and 8 (2.295-2.365 μm). The alteration minerals have higher reflectance values within ASTER band 4 (1.600-1.700 μm) (Fig. 2).

Figure 3 shows spectra of the iron oxide minerals which have low reflectance and high reflectance within ASTER band 1 (0.52-0.60 μm) and band 2 (0.63-0.69 μm), respectively (Fig. 3). These bands provide a useful tool for detection of alteration minerals assemblage in different image processing techniques described below.

False color composite image: Several false color composite images were produced for enhancing the identification of hydrothermal alteration in the study area. Figure 4 is a false color composite image of ASTER bands 6-3-1 shown in red, green and blue, respectively. Sericitic and argillic alteration zones in A, B, C, D and E prospect areas can be distinguished by a white, bright brown and bright green colors. The propylitic zones were recognized by red color. In D area, this is clearly recognizable due to intense chloritization and epidotization. Vegetation locations were shown by yellow-green color in the study area.

Fig. 2:	Reflectance spectra of some common clay minerals

Fig. 3:	Spectral reflectance curves for jarosite, hematite and goethite

The color composite RGB:631 did not show very good results for lithologic (rock type) discrimination. The Kuh-e-Shah dacite and volcanic rocks south of it were recognized by green and blue-violate color, respectively. Microgranodiorite unit in the southeastern part of district were characterized by white color, whereas tuff unit in northwestern part of the area has a similar color. Volcanic rocks in the northern part of the study area were detected by dark brown, violet and other dark colors (Fig. 4).

Color composite ratio images: In the Ratio technique, bands with high reflectance are divided by bands with high absorption. Ratio images can emphasize and quantify spectral differences between these bands (Sabins, 1999). Therefore ratios such as band 4/band 5, band 4/band 6, band 4/band 8, band 7/band 6, band 5 + band 7/band 6, band 7/band 5 and etc. would yield very high values for altered zones comprising dominantly hydroxyl-bearing minerals.

Fig. 4:	Red-Green-Blue color composite of bands 6, 3 and 1

Fig. 5:	Red-Green-Blue color composite ratio image of 2/1, 5+7/6, 7/5

This characteristic of phyllosilicates has been used in numerous mineral exploration investigations. With the same technique, iron oxide minerals can be detected by the band 2/band 1 ratio.

Color composite ratio images are produced by combining three ratio images in red, green and blue. An advantage of the color composite ratio images is that it combines the distribution patterns of both iron minerals and hydrothermal clays. A disadvantage is that the color patterns are not as distinct as in the individual density-sliced images (Sabins, 1999).

More than 15 color composite ratio images have been produced for the target locations. One of these is presented here. In the RGB: 2/1, 5+7/6, 7/5 (Fig. 5), 2/1 ratio can detect iron oxide minerals. 5+7/6 and 7/5 ratios enhance sericite and clay minerals, respectively. The area with sericitic and argillic alteration can be recognized by white to bright violate colors. The five prospect areas can be detected by this technique. Propylitic alteration can be distinguished by green color in this area. Vegetation is detected by yellow color (Fig. 5). Discrimination of different rock unites is better than earlier technique shown in Fig. 4. In violate is Kuh-e-Shah dacite, in purple is unaltered volcanic rocks south of it, in green-white is microgranodiorite unit and in bright yellow is tuff unit of northwestern study area. Volcanic rocks in the northern part of the study area are in brown, violate and other dark colors (Fig. 5).

Principal Component Analysis (PCA): The Principal Component Analysis (PCA) is widely used for alteration mapping in metallogenic provinces (Tangestani and Moore, 2002; Jing and Panahi, 2006). This technique is a multivariate statistical technique that selects uncorrelated linear combinations (eigenvector loadings) of variables in such a way that each successively extracted linear combination, or Principal Component (PC), has a smaller variance (Singh and Harrison, 1985).

Two PCA techniques have been used for ASTER images processing: (1) standard PCA on VNIR + SWIR bands and (2) selective PCA on six (SWIR bands), four or three bands.

In this study, the general statistics and principal component eigenvectors and eigenvalues for VNIR + SWIR bands, SWIR bands, four and three bands have been calculated in different ways. More than 15 different alteration maps of this area have been produced by these PCA techniques and their color compositions. The best images have been selected and are described here.

The statistics and principal component transformation (eigenvectors and eigenvalues) are described in Table 2A-C, using VNIR + SWIR ASTER bands as input bands (bands 1 to 9). It is observed the PC1 does not contain spectral features relevant in this analysis as it is a combination of all bands. This component contains 97.59% of the variance of the nine bands (Table 2C). The PC1 gives information mainly on albedo and topography. The PC3 has higher loadings of band 4 (-0.80).

Table 2A:	General statistics for 9 bands (VNIR+SWIR)

Table 2B:	Correlation matrix for 9 bands (VNIR+SWIR)

Table 2C:	Principal components for 9 bands (VNIR+SWIR)

Fig. 6:	The hydroxyl (-PC3) image shows the altered areas in bright pixels

This PC should detect the hydroxyl minerals in dark pixels because of negative contributions of band 4 and positive contributions of band 5 (0.35) or 6 (0.39). The PC8 has a higher negative loading of band 2 (-0.61). This PC should represent iron oxides in dark pixels because of negative contributions of band 2 and positive contributions of band 1 (Table 2C). The PC3 image clearly detects the location of alteration zones in dark pixels. In order to show the areas with hydroxyl minerals in bright pixels (Fig. 6) an inverse of this PC is obtained by using Eq. 1.

The selected principal component analysis on 6 and 3 bands has also been applied in study area. The statistics and PC eigenvectors and eigenvalues have been calculated in Table 3A-C, using bands 4, 5, 6, 7, 8 and 9 (SWIR) as input bands. The PC2 has higher positive loadings of band 4 (0.81). This PC detects the hydroxyl minerals in bright pixels because of positive contributions of band 4. The PC3 has a higher negative loading of band 4 (-0.34). This PC should represent hydroxyl minerals in dark pixels because of negative contributions of band 4 (Table 3C). In order to show the areas with hydroxyl minerals in bright pixels an inverse of this PC is obtained by using Eq. 2.

The statistics and PC eigenvectors and eigenvalues have been calculated in Table 4A-C, 5A-C and 6A-C using bands 4, 6 and 7, 4, 5 and 7 and 2, 5 and 8 as input bands, respectively.

Table 4C shows that clay minerals and sericite should be enhanced in PC2 because of higher loadings of band 4 (0.74). This PC should detect these minerals with dark pixel due to positive contributions of band 4 and negative contributions of bands 6 and 7. In order to show the areas with hydroxyl minerals in bright pixels an inverse of this PC is obtained by using Eq. 3.

Table 5C shows that alunite should be detected in PC2 because of higher loadings of band 4 (0.74). This PC should detect these minerals with dark pixel due to positive contributions of band 4 and negative contributions of bands 5 and 7. In order to show the areas with hydroxyl minerals in bright pixels an inverse of this PC is obtained by using Eq. 4.

Table 6C shows that propylitic zone should be detected in PC2 because of higher loadings of band 2 (0.78). This PC should detect this zone with dark pixel due to positive contributions of band 2 and negative contributions of bands 5 (-0.62) and 8 (-0.04). In order to show the areas with hydroxyl minerals in bright pixels an inverse of this PC is obtained by using Eq. 5.

Table 3A:	General statistics for 6 bands (SWIR)

Table 3B:	Correlation matrix for 6 bands (SWIR)

Table 3C:	Principal components for 6 bands (SWIR)

Table 4A:	General statistics for 3 bands

Table 4B:	Correlation matrix for 3 bands

Table 4C:	Principal components for 3 bands

Table 5A:	General statistics for 3 bands

Table 5B:	Correlation matrix for 3 bands

Table 5C:	Principal components for 3 bands

Color composition PC images are also useful technique for hydrothermal alteration mapping (Ranjbar et al., 2004; Roustaei et al., 2006).

Fig. 7:	Red-Green-Blue color composite image of PC2 (467), PC2 (457) and PC3 (258)

Table 6A:	General statistics for 3 bands

Table 6B:	Correlation matrix for 3 bands

Table 6C:	Principal components for 3 bands

This technique is produced by combining three PC images in red, green and blue. Color composition of PC2 (467), PC2 (457) and PC3 (258) in red-green-blue is shown in Fig. 7. Sericitic and argillic alteration have been characterized by yellow-white color. Discrimination of these zones is not possible in this method. In southern part of (B) prospect area can also be distinguished by yellow pixels differ from other areas. Propylitic zone is in brown-green color which can recognize in D area. Discrimination of different rock units is not good in this image. All of rocks are in blue or minor green color (Fig. 7).

Supervised classification: Multispectral classification is a computer routine for information extraction that assigns pixels into classes based on similar spectral properties. In a supervised multispectral classification, the operator specifies the classes that will be used (Sabins, 1999). This method has been used for hydrothermal alteration mapping in porphyry copper deposit (e.g., Infiernillo, Argentina, Tommaso and Rubinstein, 2007).

In this research, we have used Spectral Angle Mapper (SAM) method for supervised classification. Quantitative spectral mapping techniques, such as the Spectral Angle Mapper (SAM) method (Kruse et al., 1993), operate by comparing image spectra with reference spectra. It was developed for hyper spectral images, but can be used in lower resolution systems (Rowan and Mars, 2003). Hyper spectral and ASTER images have up to 255 and 14 bands, respectively, therefore hyper spectral is more accuracy to mapped the minerals.

The image spectra were compared with the USGS Digital Spectral Library (Minerals) (Clark et al., 1993) loaded in ENVI 4.0 software in the study area. Three minerals spectral representative of propylitic zone include chlorite, epidote and zoisite and five minerals spectral representative of sericitic and argillic alteration consist of quartz, sericite, kaolinite, montmorillonite and pyrophyllite were selected.

Fig. 8:	Chlorite, epidote and zoisite spectral library plots from USGS library as representative of propylitic zone

Fig. 9:	Kaolinite, montmorillonite, pyrophyllite, sericite and quartz spectral library plots from USGS library as representative of sericitic and argillic zones

At the same method, alunite, jarosite, gypsum, hematite, goethite and limonite spectrals can be also selected as secondary minerals zone representative. Fig. 8 to 10 show selected minerals spectral library plots.

This technique shows very good results for identification of type of alteration, its developing, mineral assemblages and locating of each mineral. Figure 11 shows the SAM’s result of propylitic zone overlain on band 1 include of chlorite in green pixels, epidote in red and zoisite in blue. Propylitic alteration is recognized in A, B, C and D prospect areas associated with different amount. High chlorite, epidote and zoisite represented in D area due to intense propylitic alteration. Some alteration minerals distinguished in drainages in southern part of D area as result of altered rocks erosion (Fig. 11).

Figure 12 shows the SAM^, s result of sericitic and argillic zones overlain on band 1 include of quartz in red pixels, sericite in yellow, kaolinite in blue, montmorillonite in bright green and pyrophyllite in green. Five altered areas can be clearly distinguished by this method. Quartz and sericite are the most common minerals in A, B, C and D prospect areas.

Fig. 10:	Alunite, jarosite, gypsum, hematite, goethite and limonite spectral library plots from USGS library as representative of secondary minerals zone

Fig. 11:	Results of SAM classification method for propylitic zone, overlain on band 1

Kaolinite is also common clay mineral in this region. E area is recognized by argillic alteration. South of B prospect area (Kuh-e-Shah) can also been clearly enhanced by argillic zone. Location of clay minerals, quartz and sericite sometimes coincides with together due to approximately spectral similarity (Fig. 9). Microgranodiorite unit, in southeastern part of area, is also detected by argillic alteration that had not been clearly recognized in earlier techniques (Fig. 12).

Figure 13 shows the SAM’s result of secondary minerals zone overlain on band 1 include of alunie in green pixels, jarosite in bright green, gypsum in brown, hematite in red, limonite in yellow and goethite in blue. Hematite and jarosite are the most common gossan zone minerals in A, B, C and D prospect areas. Alunite can clearly detect in southern part B prospect area. Some alunite has also been recognized the east of B area (Fig. 13).

Fig. 12:	Results of SAM classification method for sericitic and argillic zone, overlain on band 1

Fig. 13:	Results of SAM classification method for secondary minerals zone, overlain on band 1

The obtained results from ASTER SAM classification method have been confirmed by on the ground field geology such as alteration and mineralization mapping in the district.

CONCLUSION

Most of mineral resources in eastern Iran have remained unknown and virgin due to semi-arid to arid type of climate, presence of developed desert, lack of suitable access roads and finally imperfect information. The Lut block has a great potential for different types of mineralization, particularly porphyry copper and epithermal gold deposits, as result of its suitable subduction zone tectonic setting resulting in the presence of calc-alkaline subvolcanic rocks. Five porphyry and epithermal prospect areas in this area are the best documents for this claim.

ASTER images processing is a successfully low-cost and short-time technique for delineating favorable areas for further exploration. This technique works very well in eastern Iran due to low precipitation causing very minor soil development and the small amount vegetation and therefore good rock exposure.

In this study, different image processing methods were used. SAM classification is the best method for identification of type of alteration and different alteration mineral assemblages. The SAM method have also detected argillic and alunitization zones in the south part of the (B) area and argillic alteration in microgranodiorite unit in southeastern area that had not been clearly recognized in earlier techniques.

ACKNOWLEDGMENT

This research project was sponsored by Ferdowsi University of Mashhad, Iran with grant No. {P/85- 2008-03-20).