A comparative approach on TIR and VNIR-SWIR datasets of ASTER instrument for lithological mapping in Neyriz ophiolite zone, SW IranA. Zagros folded zone: 1- Jahrum Formation, mainly composed of Biosparite alveolina limestone and marly limestone; 2- Tarbur Formation, which includes the massive limestone with some marly limestone interbeds. B. Tertiary flysch zone: 1- EOf: Flysch type sediments, sandstone, shale, calcareous sandstone with abundant large olistoliths of orbitolina limestone, and alveolina limestone with minor ophiolitic olistolith; 2- EOof: Flysch type exposures with abundant olistoliths from ophiolites (serpentinite, pillow lava, and radiolarite) in the matrix of EOf. C. Sanandaj-Sirjan zone: 1- K1: Early Cretaceous orbitolina limestone D. Ophiolite zone: 1- d.hz: Foliated dunite-harzburgite (serpentinized), minor gabbro-pyroxenite dykes; 2- d: Foliated dunite (serpentinized); l.hz: Alternation of foliated lehrzolite-harzburgite with minor dunite (serpentinized); 3- wb: Alternation of layered webesterite clinopyrexenite, wherlite and dunite; 4- gb: Layered foliated gabbro norite and minor troctolite and clinopyroxenite; 5- Kcmu ¬: Tectonite mélange of pillow lava, rediolarite, pelagic limestone; 6- Kbu : Basaltic pillow lavas and minor diabasic dyke; 7- Kdu : Diabasic sheeted dyke; 8- Kgbu : Isotropic gabbros; 9- Pskz : Massive marble and skarn. E. Zone of Pichakan radiolarite: 1- JKr: Turbidite, alternation of radiolarite and red siliceous shale, limestone with minor spilitic pillow lava; 2- R1: Megalodon limestone turbidite; 3- Rexj : Zone of mixed limestone turbidites, radiolarian chert and serpentinite diapir. 3. SPECTRAL FEATURES Reflectance spectra have been used for many years to obtain compositional information of the Earth surface. Spectral reflectance in visible and near-infrared offers a rapid and inexpensive technique for determining the mineralogy of samples and obtaining information on chemical composition. Minerals commonly display intense absorption features due to electronic processes in transition metals such as Fe, and to molecular vibrational processes in hydroxyl- and carbonate-bearing minerals (Vincent, 1997). Peridotites and ultramafic rocks as the main components of ophiolites contain more than 40 percent olivine, orthopyroxene, clinopyroxene, and small amounts of hornblende, biotite, garnet and spinel. Reflectance spectra of these minerals are well known (Hunt and Salisbury, 1970; Hunt, et al., 1973; Clark, et al., 1990; Grove et al., 1992). Several studies have been conducted to determine reflectance spectra of ultramafic rocks (for example, Hunt et al., 1974). Olivine reflectance spectra in the 0.4 µm to 2.5 µm wavelength region show a broad asymmetric absorption feature near 1 µm (Burns, 1970) resulting from electronic transitions in Fe2+ Cations. Orthopyroxenes exhibit two main absorption features near 1 µm and 2 µm as a result crystal field transitions in ferrous iron (Burns, 1970) whose positions shift to longer wavelengths with increasing iron content (Adams, 1974; Cloutis et al., 1990). Clinopyroxenes, such as augite and diopside, display two broad absorption bands at 0.77 µm and 1.1 µm as a result of ferrous iron and a rapid fall-off of reflectance toward the blue (Adams, 1974; Hunt and Salisbury, 1970). Rock chip and hand samples were collected from the Neyriz ophiolite zone and were analyzed spectrally in order to identify the spectral features of the main rock types. Reflectance spectra of the collected samples (Figure 1) were examined in the Geosense mineral exploration company, the Netherlands, using an Analytical Spectral Devices (ASD) spectrometer, which records a reflectance spectrum across an overall spectral range of 350-2500 nm with three separate spectrometers with a 10nm individual band width. VNIR and SWIR reflectance spectra of rock samples were convolved and resampled to calibrated ASTER band positions (Figure 2). Reflectance spectra of some rock samples were selected from the spectral library of united states geological survey (USGS), and were implemented in spectral analyses.  Figure 1: Spectral curves of four main rock types, exposed at the Neyriz ophiolite zone.  Figure 2: Spectral curves of four main rock types, resampled to the VNIR-SWIR bands of ASTER. The ASTER thermal data could be used in silicate mineral enhancements. The radiation emitted from a surface in the thermal infrared wavelength is a function of both the surface temperature and emissivity. The emissivity relates to the composition of the surface and is often used for surface constituent mapping. A research work with the aim of evaluating ASTER TIR data at Mountain Pass, California (Rowan and Mars, 2003) have showed that quartzose rocks, carbonate rocks, granitic rocks, and intermediate to mafic rocks could be distinguished through analysis of the spectral emissivity data. 4. ASTER DATA ANALYSIS The ASTER level 1B data acquired on 15 December 2002, is used for the study area. A subset corresponding to the ophiolite zone and its surroundings was derived and Internal Average Relative Reflectance (IARR) calibration was carried out on the data to normalize images to a scene average spectrum. This is particularly effective for reducing hyperspectral data to "relative reflectance" in an area where no ground measurements exist and little is known about the scene. It works best for arid areas with no vegetation. An average spectrum is calculated from the entire scene and is used as the reference spectrum, which is then divided into the spectrum at each pixel of the image. The IARR calibrated dataset was used in principal components analysis approach as well as spectral angle mapping and matched filtering algorithms. The results were compared to the output images from TIR data processing, and validated by the field and map evidences. 4.1. Principal components analysis (PCA) Principal components analysis is an image enhancement technique for displaying the maximum spectral contrast from n spectral bands. This technique reduces the redundancy contained within the data by creating a new series of images in which the axes of the new coordinate systems point in the direction of decreasing variance. The resulting components are often more interpretable than the original images. The first principal component is a vector that is in the direction of the maximum variance of pixels in the scene. It accounts for more of the spectral variance in the data than any other principal component. The nth component contains all of the remaining variance and separates the most spectrally unique pixels from the rest of the pixels in the scene. All n principal components account for 100% of the variance in the data (Vincent, 1997). ASTER bands are highly correlated within and between the VNIR (bands 1-3), and SWIR (bands 4-9), which mainly comes from the spectral similarity of most geological materials within and between specific wavelength ranges. The standard and selective PCA approaches are examined on both VNIR-SWIR and TIR datasets. Performing the standard approach on six SWIR dataset revealed that PC1 is an albedo image; outcrops of marbles were enhanced in PC2 as bright pixels; PC3 showed basic alluvium in bright pixels, while marbles and gabbroic exposures in dark pixels; so this PC image could not be considered as a unique image for enhancing a specific rock unit. PCs 4, 5, and 6 are predominantly noisy images and do not discriminate a rock unit. A color composite of PCs1, 2, and 3 (Figure 3) enhances marbles in yellow, gabbros in cyan, pillow lavas in dark, and radiolarites as well as basic alluvium and peridotites in blue pixels. Selective principal components analysis has already been used on Landsat TM and ASTER data for enhancement of altered areas around specific mineral deposits or districts stained with iron-oxides (Chavez and Kwarteng, 1989; Tangestani and Moore, 2000, 2001; Tangestani et al., 2005). The selective principal components analysis approach applied in this paper consists of calculating the PC images of different subsets of ASTER bands, selected based on the spectral features of rock units in different regions of ASTER instrument. Eigenvector statistics in each PCA would identify the PC image in which the spectral information of mineral is loaded. This information usually represents, in quantitative terms, a very small fraction of the total information content of the original bands, but it is expected that the loaded information relates to the spectral signature of desired mineral. A PC image with moderate to high eigenvector loadings for diagnostic reflective and absorptive bands of a mineral or mineral group with opposite signs, enhances the same mineral (Crosta and Moore, 1989). A selective PCA approach, implemented on bands 1, 4, 6, and 8, enhances marbles in PC2 image in bright pixels. The same approach was performed on bands 10, 12, and 13 in TIR dataset and a color composite image (Figure 4) was generated using the PC images as RGB colors, respectively. Marbles were enhanced in blue pixels at the central parts, north and west of the study area. Radiolarites were enhanced in red, gabbros in green, and pillow lavas in magenta. The basic alluvium with yellow color were separated from peridotites, which were enhanced in green color. The Orbitolina limestones were also enhanced in dark green pixels. Selective PCA using thermal bands discriminates the most various rock types at the study area in comparison to other approaches.  Figure 3: A color composite image produced from PC1, PC2, and PC3 images, output results of PCA analysis on SWIR dataset of ASTER.  Figure 4: A color composite image produced from PC1, PC2, and PC3 images, output results of PCA analysis on TIR dataset of ASTER. |