Home > Geospatial Application Papers > Geology > Mineral & Mining


Using Principal techniques on ETM+ 2002 for arid and semi-arid environment Central Iran


3. Materials and Methodology
Methodology in brief is given in Figure 2. Geological map and digital topography maps are used for ground truth measurement and accuracy purposes. Landsat 7 satellite, sensor-ETM+ with spatial resolution is 28.5 meter in bands-1,2,3,4,5,6 and band 7 were used for digital analyses to extract the thematic information from the digital image. This allows to discriminate the additional information about the mineralogy and its associated features of the ground. With the use of the ETM+ sensor data and more sophisticated computer software, more insight is gained into the minerals and hence the types of rocks present. Ground-truth data is used to identify interesting targets (such as an existing igneous body) and the computer is instructed to search for regions that appear similar in all bands. The role of OIF contains high values of OIF and gives more spectral information of the object. In OIF method, seven bands of ETM+ data are used (Table 2), spectral range (8-14 micrometers) and gives rock and minerals ( Lillisand and Leild, 1994) information content. OIF is given as in equation 1. Field observations are taken in four steps as follows:

  1. Before applying remote sensing techniques
  2. During applying OIF method
  3. During applying Crosta method
  4. After three above steps to study the accuracy separately as well as combination of OIF and Crosta methods.

Fig. 2. Flowchart represents brief methodology adopted


3.1. Digital analysis
The analyses are carried out on satellite images for classification, enhancement of normalize differential vegetation index (NDVI), band ratioing and Principal Component. These enhancement techniques are differentiated thematic information and their associated features. Software packages have been used for digital analyses are namely ER-Mapper version 6.1 and ENVI version 4. Satellite image of ETM+ sensor is geometrically corrected using ground control points (GCPs) of digital topography map at scale 1:25000 applying ER-Mapper software. 27 FCCs are prepared to select the best images for differentiating the rock types and minerals. The OIF formula (Eq 1) is applied for different composite spectral (Table.1).

Table 1 OIF values for different composite spectral (FCC)


Principal component transformation (eigenvector and eigenvalues) are applied to six ETM+ bands (1,2,3,4,5 and 7) in Table 2. Crosta method (Wilson, 1992) is based on one of the statistical techniques. In this study Crosta method and OIF bands combination are compared to derive the optimum results.

Table 2 Homougeneous bands matrix


3.1.1. Supervised classification
Supervised classification procedure is performed using ENVI 4 software package. The training sets are identified on the image for extraction of thematic information. Six classes are chosen based on ground truth data evidences. The supervise classification is carried out using maximum probability algorithm (Wilson, 1992) and different FCC images are produced. With respect to Table 1, seven FCC images in different band combination of visible, Infra-red and Thermal infra-red with maximum OIF have been used for classification purposes. Post classification confusion matrix using ground truth images and region of interest (ROIs) are carried out for accuracy measure (Table 3). On the basis of OIF method, best band combination FCC images are selected for enhancing thematic information content and classifications.

Table 3 Accuracy of classification in different spectral composite


4. Data analysis and discussion
The use of ETM+ bands (Fig 3) for interpreting igneous and minerals which are providing more accurate delineation of the thematic area. Amphiboles and clay minerals are enhanced on digital images. To evaluate the enhancement capability of the ETM+ sensor, few test sites in west Esfahan city have been identified and studied as training window (areas) for image classification techniques.


Fig. 3. Showing FCC of 7,4&1 ETM image


Clay minerals are more enhanced using band ration 5/7. PC3 has high positive loading in band 7 and negative loading in band 1 respectively. This PC enhances amphibole minerals and amphibolite rocks. The bright pixel on the image (Fig 4) is showing amphibole minerals and amphibolite rocks. The spectral reflectance curve diagram of amphibole (Fig 5) indicates that the amphibole minerals reflects in band 7 and absorb in band 1. Thus the amphibole minerals are showing in bright pixels on the image. For amphibole the digital image (Fig 4) is prepared using eigenvector loading of PC3 (Eq 2) as follows:

{ ( PC3= 0.48(B1)-0.39(B2)-0.32(B3)-0.22(B4)-0.48(B5)+0.49(B7)}………Eq.2


Fig. 4. Represents amphibole minerals using band ratio (7/3) image in PC3



Fig. 5. Represents behavior of spectral reflectance in the study area


Band ratios were also applied for the study which shows that the band ratio: (7/1) and (7/3) on the image is also enhanced the amphibole minerals. In PC3 the granite body is seen in dark color on the image while orthoclase and quartz minerals visible band and Infra-red band (IR) have homogeneous and uniform reflection.

The combination of Band Vs Principal component contains 86.8 percentage of variance of six bands.
PCA analysis shows that PC1 does not contain spectral feature relevant to enhancing the igneous body and amphibole minerals as it is combination of all bands and contain 86.8 % of the variance of six bands. This PC provides information mainly on albedo and topography (Ranjbar and Honarmand, 2004). Vegetation cover is enhanced in PC2 (Fig 6) as this has higher loading of band 4 (0.001). Thus, in PC2 vegetation cover display in light color with moderate tone on the image due to maximum negative values (Table 4).


Fig. 6. Showing NDVI (PC2)


Page 2 of 3
| Previous | Next |