Regional Lithological Mapping Using ASTER-TIR Data: Case Study for the Tibetan Plateau and the Surrounding Area

Abstract

The mineralogical indices the Quartz Index (QI), Carbonate Index (CI) and Mafic Index (MI) for ASTER multispectral thermal infrared (TIR) data were applied to various geological materials for regional lithological mapping on the Tibetan Plateau. Many lithological and structural features are not currently well understood in the central Tibetan Plateau, including the distribution of mafic-ultramafic rocks related to the suture zones, the quartzose and carbonate sedimentary rocks accreted to the Eurasian continent, and sulfate layers related to the Tethys and neo-Tethys geological setting. These rock types can now be mapped with the interpretation of the processed ASTER TIR images described in this paper. A methodology is described for the processing of ASTER TIR data applied to a very wide region of the Tibetan Plateau. The geometrical and radiometric performance of the processed images is discussed, and the advantages of using ortho-rectified data are shown. The challenges of using ASTER data with a small footprint in addition to selecting an appropriate subset of scenes are also examined. ASTER scenes possess a narrow swath width when compared to LANDSAT data (60 km vs. 185 km, respectively). Furthermore, the ASTER data archive is vast, consisting of approximately three million images. These details can present an added level of complexity during an image processing workflow. Finally, geological interpretations made on the maps of the indices are compared with prior geological field studies. The results from the investigations suggest that the indices perform well in the classification of quartzose rocks based on the carbonate and mafic mineral content, in addition to the granitic rocks based on the feldspar content.

1. Introduction

Spectroscopic studies of inorganic materials in the thermal infrared (TIR) region (8–12 μm) [1,2] have demonstrated strong absorption features caused by the stretching vibrations of the Si–O bonds in silicate minerals, the major components of the terrestrial surface, whereas the Si–O bonds do not cause any prominent spectral features in the visible and near-infrared (VNIR) region of the spectrum (0.4–2.5 μm) [3]. Studies carried out during the Apollo missions to the Moon in the 1960s confirmed relationships between the TIR spectral features and the geology of the lunar surface composition (e.g., References [4,5]). For example, it was observed that the emissivity minima due to Si–O stretching vibrations (i.e., reststrahlen bands) moved to longer wavelengths as the composition of the target igneous rocks changed from felsic to ultramafic [6].

The airborne and satellite remote sensing systems were developed after the results of the spectroscopic studies shown above. They demonstrated the capability of detecting rock composition, despite the radiometric, spectral and spatial resolutions being much worse when compared to a modern system such as ASTER. For example, an airborne scanner with two detector elements could distinguish quartz sandstone from nonsilicate surface materials [7]. Additionally, the infrared interferometric spectrometer onboard the Nimbus-4 satellite derived a global surface emissivity map at 9 μm, which discovered the significantly low emissivities over arid areas due to quartz, even with a quite low spatial resolution (~100 km) [8].

The airborne Thermal Infrared Multispectral Scanner called TIMS was the first modern thermal infrared remote sensing system with resolutions adequate for geological applications (e.g., References [9,10]). TIMS and similar airborne systems (e.g., Reference [11]) confirmed their capabilities of detecting the systematic shift of the reststrahlen bands in the silicate rocks and minerals.

The Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) sensor developed by the Japanese government was launched onboard Terra, the first of NASA’s Earth Observing System (EOS) series of satellites, in December 1999 [12]. It is composed of three subsystems: (1) the VNIR subsystem observing the visible to near-infrared region (three bands between 0.5 and 0.9 μm, 15 m resolution with a stereoscopic capability for the NIR band); (2) the SWIR subsystem observing the shortwave infrared region (six bands between 1.6 and 2.5 μm, 30 m resolution); and (3) the TIR subsystem observing the thermal infrared region (five bands between 8 and 12 μm, 90 m resolution, NEΔT < 0.3 K at 300 K), respectively. The spatial coverage of an ASTER image at the surface is approximately 60 × 60 km.

The authors have proposed mineralogical indices defined with the ratioing approach of the ASTER data for detecting lithology or mineralogy [13,14,15,16,17]. Especially the Quartz Index (QI), Carbonate Index (CI) and Mafic Index (MI) defined for the ASTER-TIR data are finely analyzed theoretically and practically, and the performance of the indices in extracting geological information is confirmed [14]. Many studies have applied the indices (e.g., References [13,14,15,16,17,18,19]) and other workers have proved the usefulness and the stability of the indices in the local geological case studies (e.g., References [20,21,22,23,24,25,26,27,28,29]). Despite ASTER radiance registered at the sensor data without atmospheric corrections being adopted for the calculation of the indices, the recognized scene dependency is quite weak if the atmospherically well-conditioned data are selected [17].

One of the main goals of this project is the mapping of large-scale structure and lithology. We therefore commenced research for regional-scale mapping covering a wide area with the spectral indices. Considering the relatively narrow spatial coverage of each ASTER data compared to that of LANDSAT, many scenes need to be mosaicked for mapping a wide area. Additionally, the ASTER data archive is vast and constantly expanding due to the collection of data over the past 16 years that ASTER has been in operation. This makes it difficult to locate suitable scenes when mapping a region of interest. To save time and labor, the selection of well-conditioned data, in addition to prioritizing the selected data and calculating the indices and mosaicking the imagery, was followed [30].

In this paper, we present the regional maps of the indices covering the region of the central Tibetan Plateau and the surrounding area, after summarizing the geology of the study area and introducing the methodology of calculating the indices and producing the regional maps. In total, 386 ASTER products are selected from a database of more than 10,000 scenes to produce the regional maps. The characteristics of the indices for typical geological and the other materials and the radiometric and geometrical performances of the produced maps are discussed. Finally, geological interpretations are made in association with the maps.

2. Materials and Methods

2.1. Study Area

The study area is the central part of the Tibetan Plateau and the surrounding area, bounded by latitude 28° N to 37° N and longitude 84° E to 90° E. The previous studies (e.g., [31,32]) have revealed that the geology of the Tibetan Plateau involves multiple terranes, suture zones containing ophiolites and ophiolitic mélanges, high-P metamorphic belts, magmatic arcs, and accretionary prism complexes that largely developed during the Paleozoic-Mesozoic Eras.

Knowledge of the pre-Cenozoic lithospheric architecture of Tibet is critical for providing initial conditions for models of Tibetan Plateau formation, for establishing strain markers to quantify Cenozoic deformation, and for the development of hypotheses to explain heterogeneous tectonism related to the Cenozoic Indo-Asian collision. From northeast to southwest, Tibet is comprised of the Altyn Tagh–East Kunlun–Qilian, Songpan–Ganzi, Qiangtang, Lhasa and Himalayan terranes, which are separated by east-striking suture zones of Paleozoic to Mesozoic age (Figure 1). Many researchers [31,32] proposed that the formation and evolution of the Tibetan Plateau have been intrinsically related to those of the eastern Tethys, recorded by the Longmu Co-Shuanghu ophiolite mélange zone, the Southern Qiangtang Paleozoic accretionary arc-basin system, the Bangong-Nujiang suture zone, and their associated, composite island arc-basin systems. The present-day Bangong–Shuanghu–Changning–Menglian suture system marks the final closure of the Tethyan Ocean. The Indus-Tsangbo Ocean opened as a back-arc basin in response to the southward subduction of the Tethyan Ocean lithosphere in the Middle Triassic. In addition, it closed as a result of the India-Asia collision at the end of Cretaceous, followed by the northward indention of the Indian plate that resulted in significant intra-continental deformation and plateau uplift in the Cenozoic. Cenozoic deformation mainly produced a crustal scale shear zone and décollement between the Precambrian basement and the cover sequences, regional thrust faults and folds, and strike-slip fault zones.

Figure 1. Simplified tectonic map of Tibet and surrounding region, showing the main suture zones in Tibet (modified from References [26,27,28]. Blocks and terranes: ALT-EKL-QL: Altyn Tagh–East Kunlun–Qilian terrane; BS: Baoshan terrane; HM: Himalayan terrane; IC: Indo-China block; KA: Kohistan Arc terrane; LA: Ladakh arc terrane; LC: Lincang–Sukhothai–Chanthaburi Arc terrane; LST: Lhasa terrane; NCB: North China Block; NQT: North Qiangtang terrane; QT: Qiangtang terrane; SP: South Pamir terrane; SPGZ: Songpan-Ganze terrane; SQT: South Qiangtang terrane; TC: Tengchong terrane; TSH: Tianshuihai terrane; WB: West Burma terrane; WKL: West Kunlun terrane. Suture zones: BNS: Bangong-Nujiang Suture; EKLS: East Kunlun Suture; ITS: Indus-Tsangbo Suture; JSS: Jinsha Suture; LSS: Longmu Tso–Shuanghu–Menglian–Inthanon Suture; WKLS: West Kunlun Suture. Basins: QB: Qaidam Basin; KB: Kumkol Basin. Faults: ALT: Altyn Tagh Thrust; ALTF: Altyn Tagh Fault; KKF: Karakorum Fault; LMST: Longmen Shan Thrust; MFT: Main Frontal Thrust; NQLT: North Qilian Thrust; RRF: Red River Fault; SGF: Sagaing Fault; XXF: Xianshui River–Xiaojiang Fault. The dashed red rectangle represents the study area.

Figure 1 shows the tectonic setting of the Tibetan Plateau and the surrounding area, modified from the previous studies [33,34,35]. As shown in Figure 1, this study area is composed of the blocks of the Altyn Tagh–East Kunlun–Qilian terrane, Songpan-Ganzi terrane, North Qiangtang terrane, South Qiangtang terrane, Lhasa terrane and Himalayan terrane, from north to south. The most important geological features targeted in this study include the ophiolite belts appearing at the surface along the sutures of the blocks, the quartzose and carbonite sedimentary rocks deposited as the accretionary wedges associated with the subduction of the oceanic plate beneath the continental plate, the granitic rocks intruded with the island arc geological setting, and the distributions of the sulfate layers as the geological evidence of the neo-Tethys environment. The field geological studies so far (e.g., [36,37,38,39,40,41,42,43]) have revealed the distribution of the target lithological features at some limited local areas in the Tibetan Plateau. However, these terranes and suture zones are poorly characterized in central Tibet, with a >1000-km-long and up to 500-km-wide tectonic belt, where the field geological surveys have not yet been accomplished well for most of the region because this area is remote and inaccessible (Figure 1).

The geological map at the scale of 1:1,500,000 is published [44] for the entire Tibetan Plateau region including the study area. Larger-scale geological maps are published at the limited locations, for example the geological map of the ophiolite zone at the scale of 1:200,000 in the middle segment of the Indus-Tsangbo suture zone [45].

2.2. ASTER Data

Since the launch of ASTER in December 1999, the VNIR and TIR subsystems have observed the earth for more than 16 years. This has far exceeded its original lifespan of five years. The SWIR subsystem stopped operation in 2008 due to over saturation and severe striping of the detector. Ownership of the ASTER data had been held by the Ministry of Economy, Trades and Industry (METI) of Japan and NASA, and the ASTER data were distributed by the Japan Space Systems (JSS) and the USGS Eros Data Center, respectively. It was necessary to pay a nominal fee to METI for ordering ASTER data products except for the freely distributed ASTER global digital elevation model (GDEM) products [46] until March 2016. In April 2016, a change in policy was announced [47]. As a result, all ASTER data products are available at no cost. Furthermore, the ASTER data distributor in Japan has changed from JSS to the National Institute of Advanced Industrial Science and Technology (AIST).

The data observed by the ASTER sensors (Level-0 data) are transmitted to ground stations, where the data for each scene is processed to generate the standard Level-1A data product which retains the raw data observed with the sensor. Until March 2016, the ASTER data product was generated and provided as described below. Each standard and semi-standard product of the ASTER data was generated as a Hierarchy Data Format (HDF) file. The standard Level-1B data product was generated automatically for all the Level-1A data satisfying a condition on the detected cloud coverage, as well as based on the user request, which apply the fundamental geometrical and radiometric corrections essential for most of the geoscience applications. Each pixel data in Level-1B is in the measure of radiance registered at the sensor [48,49,50,51]. The standard higher level Level-2B data products of physical parameters including the surface radiance data with the atmospheric corrections on Level-1B data (Level-2B01) [52,53] and the surface emissivity data with the temperature separation processing on the TIR data of the Level-2B01 data (Level-2B04) [54] were generated on the basis of the user request. Also, the semi-standard Level-4 data product of the DEMs was generated from the along-track NIR stereo images (band 3N and 3B) with the downward- and backward-viewing telescopes [55]. The semi-standard Level-3A data product with the ortho-rectification processing had been provided by JSS, which includes the radiance registered at the sensor data equivalent to the one of the Level-1B product and the ASTER-driven DEM [56].

After April 2016, AIST supplies the ortho-rectified data products corresponding to Level-3A in the GeoTIFF files [57], and USGS supplies the data products including the ortho-rectified data of the radiance registered at the sensor (Level-1T) in the HDF files [58] freely to the public covering all over the world, in addition to the ASTER GDEM products already provided to public as described above.

This study uses the ASTER products of radiance at the sensor (e.g., Level-1B, Level-3A). The DN of ASTER-TIR band i in the Level-1B or Level-3A data is linearly related to radiance registered at the sensor (Wm⁻²·sr⁻¹·µm⁻¹) by [49]

$${L_{sen}}^i=coe{f}^i\times (D{N}^i-1),$$

(1)

where coef¹⁰ = 0.006882, coef¹¹ = 0.006780, coef¹² = 0.006590, coef¹³ = 0.005693, and coef¹⁴ = 0.005225.

Figure 2 shows a false color composite mosaic generated with the ASTER data used in this study annotated with typical geographical information in this study area. All images of the regional maps including Figure 2 are projected using zone 43 of the Universal Transverse Mercator (UTM). As the mosaic VNIR imaging is not the purpose of this study, only an automatic color balancing of the linear stretching in the histogram region covering “average DN ± 3 × σ” was applied, except for the southern part of the study area bounded by latitude 28° N to 30° N, where the color balancing is carefully achieved manually.

Figure 2. Mosaic VNIR false color image generated with the ASTER data used in this study, assigned as blue for band 1, green for band 2 and red for band 3 covering the study area bounded by latitude 28° N to 37° N and longitude 84° E to 90° E, using a Universal Transverse Mercator (UTM) map projection. Geographical locations of some cities, mountains, etc., are shown. The yellow rectangle represents the area of Figure 7.

2.3. Mineralogical Indices for ASTER-TIR

Based on analysis of the spectral properties for typical terrestrial rocks and minerals shown in Figure 3, the indices, QI, CI and MI were proposed [13]. After the precise analysis on the performance of the indices against the temperature and the atmospheric conditions as well as concerning the separation among the mineral types, the indices were defined [14,17] as described below.

Figure 3. Emissivity spectra for common rock types (a) carbonate rock; (b) quartzose sedimentary rock; (c) felsic rock (granite); (d) intermediate rock (diorite); (e) mafic rock (gabbro); (f) ultramafic rock (dunite); (g) gypsum (modified after Ninomiya et al. [17]). Each tick in the Y axis denotes 1.0/0.75 of emissivity except for (b) where it denotes 1.0/0.5 of emissivity.

To emphasize the specific spectral feature for quartz in the major rock-forming minerals, e.g., higher emissivity at the 8.6 μm region (corresponding to ASTER band 11) compared to the emissivity at the 8.3 and 9.1 μm regions (corresponding to ASTER band 10 and band 12, respectively) as shown in Figure 3b, QI was defined as [13]

$$QI=\frac{D_{11}\times D_{11}}{D_{10}\times D_{12}},$$

(2)

where Di represents ASTER band i. In this paper, we use radiance data without atmospheric corrections of ASTER Level-1B or Level-3A data. The DN value of each pixel in Level-1B or Level-3A data is linearly converted to radiance registered at the sensor (Wm⁻²·sr⁻¹·µm⁻¹) with Equation (1). QI is expected to be high for crystallized quartz, and on the other hand, to be low for the common sulfates, typically gypsum because the spectral features for the sulfates in ASTER bands 10 to 12 are the opposite of quartz as recognized in the spectrum of gypsum shown in Figure 3g. Not as strong but similar as the sulfates, K-feldspars and Na-feldspars exhibit the opposite spectral features of quartz. If the rocks are rich in both quartz and feldspar, which often occurs in igneous rock, typically for the granitic rocks, QI is expected to be relatively low [15,17,19,30]. The spectral emissivity property of the rock type shown in Figure 3c supports this expectation. The mixture of quartz and feldspar does not generally occur in sedimentary rocks, which suggests high QI functions well as an indicator of quartz-rich sedimentary rocks. The previous studies have confirmed that high QI is a distinct indicator for quartz-rich sedimentary rocks [14,15,16,17,18,19,21,22,23,26,29,30], and low QI is a distinct indicator primarily of sulfate minerals [15,16,17,30] and secondarily of feldspar-rich silicate rocks [15,17,19,30].

The spectral emissivity of carbonate rock which is composed of the two major carbonate-forming minerals on Earth, calcite and dolomite, is shown in Figure 3a. This indicates the absorption feature is only at the 11.3 μm region (corresponding to ASTER band 11), and CI for ASTER TIR data was defined as [13]

$$CI=\frac{D_{13}}{D_{14}}$$

(3)

CI is expected to be high for calcite and dolomite. Prior studies have confirmed that a high CI signal is a good indicator for the presence of marine sedimentary rocks that are composed of calcite and/or dolomite [14,15,16,17,18,19,21,24,25]. It should be considered that the atmospheric anomalies such as the clouds and the mists can also cause relatively high CI as discussed in Section 4.1.

The spectral emissivity for silica and silicate rocks (typically igneous rocks), as shown in Figure 3b,c,d,e,f, demonstrates that the ratio of the emissivity at band 12 or the lower wavelengths to band 13 or the higher wavelengths increases as the chemical SiO₂ content decreases (i.e., as the rock type changes from silicic, felsic to mafic). The ratio value indicates the medium value in the carbonates. To accomplish the separation of the carbonate rocks in the spectral features described above, MI was defined as [13]

$$MI=\frac{D_{12}}{D_{13}\cdot C{I}^3}=\frac{D_{12}\cdot {D_{14}}^3}{{D_{13}}^4}.$$

(4)

MI is expected to have an inverse correlation to the chemical SiO₂ content in silicate rocks. It is expected to be high for ultramafic rocks, and to move systematically lower as the rock type changes to felsic, and to be quite low for silicic rocks. Similarly, it has been confirmed that high MI is a distinct indicator of ultramafic rocks [14,15,16,17,18,25,26,27,28,29,30]. Low MI is sensitive for high SiO₂ content rocks with both crystallized and amorphous silica [14,17]. Considering QI is sensitive only for the crystallized quartz, the discrimination between crystallized and disordered quartzose rocks is possible using both QI and MI.

A theoretical analysis on the stability of the indices against the surface temperature and the atmospheric parameters indicated QI and MI are well stabilized against temperature if the atmospheric conditions are good, but CI exhibits a strong dependency on the surface temperature even in atmospherically good conditions [14]. The properties of the indices were confirmed by a practical analysis comparing the results applied to multi-temporal scenes observing a study area [14]. It then showed the effectiveness of the normalization processing of the brightness temperature at band 13 to a fixed temperature on improving the heavy dependency of CI to the surface temperature. The normalized radiance registered at the sensor at band i was defined after the Planck function as [14,17]

$$n{L^i}_{sen}={L^i}_{sen}\times \frac{\exp \{\frac{{\lambda }^{13}}{{\lambda }^i}\cdot \ln \left(\frac{c_1}{\pi \cdot {{\lambda }^{13}}^5\times {L^{13}}_{sen}}+1\right)\}-1}{\exp \left(\frac{c_2}{{\lambda }^i\times \raisebox{1ex}{$nT$}\!\left/ \!\raisebox{-1ex}{${\epsilon }_{a^{13}}$}\right.}\right)-1},$$

(5)

where the radiation constants are c₁ = 3.742 × 10⁸ (Wm⁻²·µm⁻⁴) and c₂ = 1.439 × 10⁴ (µm·W), λⁱ is the center wavelength (µm) of ASTER band i (i.e., λ¹⁰ = 8.3, λ¹¹ = 8.65, λ¹² = 9.1, λ¹³ = 10.6, λ¹⁴ = 11.3), Lⁱ_sen is the radiance registered at the sensor in band i as shown in Equation (1), ε_a¹³ is the assumed emissivity in band 13, nT is the fixed temperature (K) to be normalized. In this study, ε_a¹³ is adopted as 1.0, and nT is adopted as 300. This normalization process of the radiance is equivalent to a type of temperature separation processing without atmospheric corrections. That is, as nL_sen¹³ corresponds to the blackbody radiation of 300 K at λ¹³, when nL_senⁱ (Equation (5)) is divided with the blackbody radiation of 300 K at λⁱ, it becomes the emissivity with the constant unity value at band 13.

In this paper, the indices QI, CI and MI are calculated on the normalized radiance registered at the sensor data (Equation (5)) for the ASTER Level-1B or Level-3A data. One of the reasons for using radiance registered at the sensor data (e.g., Level-1B and Level-3A) is the higher availability compared to the emissivity data (e.g., Level-2B04). Another reason is the higher geometrical accuracy in the ortho-rectified image available for the radiance registered at the sensor data (e.g., Level-3A), which is discussed in detail in Section 4.2. The Level-2B04 emissivity data are partially applied just for the purpose of the comparison with the Level-1B or Level-3A data within a part of the study area.

2.4. Regional Mapping

The main goal of geological remote sensing requires the mapping of regions containing large-scale structure and lithology and this necessitates the mosaicking of large numbers of satellite scenes. The coverage of one ASTER scene corresponds to approximately 60 km × 60 km on the ground. This is much smaller compared to LANDSAT-TM, which has ground coverage of approximately 185 km × 170 km. Therefore, the use of ASTER data for geological remote sensing requires the downloading and mosaicking of large numbers of scenes which can become unwieldy. On the other hand, the ASTER data archive is vast and has been actively collecting data for the past 16 years. The procedures to achieve the effective regional lithological mapping with the ASTER mineralogical indices are described below.

Firstly, the browse images of the ASTER data covering the required areas are checked before downloading. In this study, we had obtained ASTER Level-1B products from the JSS site which had closed because of the change in the ASTER data distribution policy which had been announced on April 2016 as described in Section 2.2. An alternative method for obtaining ASTER data is to check the browse images and submit an order at the sites of the new distributors (AIST and USGS) [53,54].

The next step is to calculate the indices (i.e., QI, CI and MI) for the respective data, and carefully check the images of the VNIR data and the indices manually. The CI index images should be checked carefully as they can be heavily affected by the atmospheric water content and as a result should be omitted as useless data. During the course of the checking, the priority ranking of each ASTER data is determined, and the mosaic images of the indices covering the area of interest are produced as the regional maps.

In the case of this study, we produce the mosaic images mapped for every 1° × 1° region of the Earth surface (hereinafter, we call the region as “tile”), and then joined them together to produce the regional maps covering the entire study area with 54 tiles in total.

More than 10,000 ASTER scenes in archive covering the study area were checked with the browse images, and greater than 5000 Level-1B images were downloaded. The best 380 Level-1B products were selected to produce the regional maps covering the entire study area.

For the purpose of heightening the value of the geometrical positioning accuracy in the regional maps, each of the Level-1B products constructing the mosaic image is replaced with the Level-3A product of the same ASTER observation. However, the regional maps with the replaced Level-3A products do not cover a part of the area because the coverage of the Level-3A product is a little bit narrower than that of the Level-1B product to ensure the geometrical accuracy. The original Level-1B products are again utilized for the area not covered with the Level-3A products. Finally, the regional maps are consisting of the 380 Level-3A products together with the six Level-1B products. In total, the 386 ASTER image products are utilized for composing the regional maps as summarized in Appendix Table A1.

3. Results

Figure 4 shows the mosaic color composite image of QI, CI and MI with a linear contrast stretch applied of 0.97–1.055 (QI; Red), 1.005–1.055 (CI; Green), and 0.79–0.95 (MI; Blue) with the ASTER Level-3A products. Two small areas are not covered due to reasons explained in an earlier section. For those areas not covered by the Level-3A products as indicated in Figure 4 with blue polygons, the Level-1B products are alternatively utilized. Annotations have been added using points, lines and rectangles, as shown in Figure 4, indicating the locations which are discussed in the next chapter. Grayscale images with the linear contrast stretches for QI (0.95–1.1), CI (1.005–1.055), and MI (0.75–0.98) were produced using the same ASTER dataset as in Figure 4, as shown in Figure 5. The granule ID, observation date and location at the scene center of the ASTER data products consisting of the regional maps (Figure 4 and Figure 5) are summarized in Table A1, which will be useful in reproducing the results shown here.

Figure 4. Mosaic color composite image of QI, CI and MI with a linear contrast stretch of 0.97–1.055 (QI; Red), 1.005–1.055 (CI; Green) and 0.79–0.95 (MI; Blue), with ASTER Level-3A products. The areas bounded by the blue polygons were mapped with Level-1B products when Level-3A products were not available. The annotations indicate the locations discussed in the text. The extent of the study area and the map projection are described in the caption in Figure 2.

Figure 5. Grayscale image of the indices (a) QI with a linear contrast stretch of the histogram region 0.95 to 1.1; (b) CI with a linear contrast stretch of the histogram region 1.005 to 1.055; and (c) MI with a linear contrast stretch of the histogram region 0.75 to 0.98, produced using the same ASTER dataset of Figure 4 covering the study area. The extent of the study area and the map projection are described in the caption in Figure 2.

To assist readers in locating details of the regional maps, the un-annotated, higher-resolution version of the VNIR false color image (Figure 2), the color composite image of the indices (Figure 4), and the grayscale images of QI (Figure 5a), CI (Figure 5b) and MI (Figure 5c) are presented in the Appendix Figure A1, Figure A2, Figure A3, Figure A4 and Figure A5. For the interactive display with the geographical information and DEM that will achieve further convenience to the readers, the color composite image of the indices (Figure 4) is converted to a Google Earth compatible KMZ file as presented in the Supplementary Figure S1.

4. Discussion

4.1. Properties of Indices for Geological and Other Materials

Many kinds of geological and the other materials are characterized with the indices. To avoid erroneous attention to the colors appearing in the composite image of the indices (Figure 4) for the targets without intensive features in each index (QI, CI or MI; Figure 5), the combined use of the grayscale and the color index images (Figure 4 and Figure 5) is essential. Also, to distinguish the geological materials from the other materials (e.g., cloud, vegetation, snow, ice and water) in the images of indices (Figure 4 and Figure 5), it is important to make interpretations referring to the VNIR false color image (Figure 2). In the discussions below, the location of the target is indicated with the annotations in Figure 4. The colors indicated in the discussion are of the composite index image (Figure 4).

High QI (>1.05) is a distinct indicator of quartz-rich/feldspar-poor rocks, usually quartzose sedimentary rocks [14,15,16,17], which are widely distributed in the study area as the whitish pixels in the QI image (Figure 5a). Looking at the details, there are several types in the high-QI rocks. The high-QI rock bodies along line “A” show relatively low CI (<1.02), whereas most of the others in the more northern part and the ones along line “B” show intermediate CI (>1.02) (Figure 5a,b). The former look pure reddish and the latter look yellowish-red (Figure 4). The differences of CI in the high-QI rocks can be reflecting the content of the carbonates in the quartzose rocks.

Taking notice of the relationship between QI and MI, since MI is designed inversely correlating to the SiO₂ content in the silicate rocks, quartz with 100% SiO₂ content indicates low MI. Specifically, high-QI (>1.05) rocks simultaneously have low MI (<0.80) (Figure 5a,c) in essence, as expected with the spectral property of quartz (Figure 3a). However, high-QI (>1.05) rocks together with intermediate MI (>0.82) are often observed in and around the distribution area of the high-QI/low-CI rocks described above, especially along line “A”. These rocks are expected to be high in quartz content, coexisting with silicate minerals except feldspars. They are usually distributed around metamorphic domes in the north Himalayan metamorphic belt [38] as discussed later. This type of rock looks pure red to purplish-red (Figure 4).

In addition, there are some bodies of intermediate QI (1.02 to 1.04) together with low MI (<0.80), for example the ones along line “C” and around points “D”, “E” and “F” (Figure 5a,c). Most of the bodies are in the fluvial deposits, as recognized with the VNIR image (Figure 2), and can be non-crystallized, silica-rich deposits in which the spectral features of quartz in the ASTER band 10 to band 12 region (Figure 3b) are lost with the weathering processes. These materials look mixed dark-reddish (Figure 4). Also, the mixture of quartzose and sulfatic deposits can cause the characteristics of intermediate QI with low MI, which is usual in the northern part of the study area as the sulfate layers are often interbedded within the quartzose layers, as described below.

Low QI (<0.98) is a distinct indicator for the common sulfate minerals (typically gypsum) [15,16,17] represented as the blackish pixels in Figure 5a. These rocks and deposits, for example the ones around the points “G”, “H” and “I”, look deep greenish (Figure 4). We can find the bodies denoting the deep greenish color widely in the northern (>33° N) part of the study area, often interbedded within the high-QI quartzose layers, which can be the sulfate layers accumulated in the neo-Tethys geological setting. Not as strong as sulfates, feldspar-rich rocks are typically a part of the granitic rocks and also show low QI [15,17,19]. The bodies around line “J” and point “K” in the Gandise Mountains (annotated in Figure 2), as well as the bodies around points “L”, “M” and “N”, are the examples of the rock type with a dark bluish-green color (Figure 4).

Several types of intrusive rocks can be discriminated by the combined analysis of the indices with the unique paternal texture information reflecting the geomorphological features of dissection and shape (Figure 4 and Figure 5). One type is the feldspar-rich granitic rock described above. Gabbroic or basaltic rocks are also detectable, with the deep bluish color (Figure 4) reflecting the characteristics of relatively high MI (≈0.90) and relatively low QI and CI (Figure 5) [15,17], typically the rock body pointed out as “$”. Another type is granitic rock related to the north Himalayan metamorphic domes [21,36,37,38]. The domes are the granitic bodies intruding to the accretionary wedges of the south Tibetan detachment system [36], which looks like a dark pinkish-purple color with a unique texture and shape (Figure 4) along line “O”. Figure 4 suggests that the metamorphic domes are closely associated with the high-QI (>1.05)/low-CI (<1.02) rocks and the high-QI (>1.05)/intermediate-MI (0.82 to 0.86) rocks along line “A” discussed earlier in this section. According to the previous study [36], the high-QI (>1.05)/low-CI (<1.02) rocks can be linked to low-grade metamorphosed sediments and the high-QI (>1.05)/intermediate-MI (0.82 to 0.86) ones to higher-grade ones. A more detailed analysis of the indices related to the metamorphic domes with reference to the published articles is presented later in Section 4.3.

High CI (>1.05) is a distinct indicator of the common carbonate minerals, calcite and dolomite [14,15,16,17]. The high-CI rocks are widely distributed in the study area with the NWW trending, where the ones around lines “P”, “Q”, “R”, “S”, “T” and “U” are especially prominent (Figure 5b). They look light greenish in color (Figure 4). It should be noted the atmospheric water bodies (typically clouds and mist) also indicate high CI and look light greenish, similar to the carbonate minerals [14,15,16,17], as described later.

According to the definition of MI (Equation (4)), high MI (>0.92) is a distinct indicator of the ultramafic rocks with a quite low SiO₂ content (<40%) [14,15,16,17]. High-MI bodies are distributed widely in the study area, prominently in the southern part (Figure 5c). They look light purplish in color (Figure 4). The ophiolitic ultramafic rocks represent ancient remnants of the oceanic lithosphere and underlying upper mantle, which have appeared in the surface suture zones during the course of tectonic activities, especially in the subduction of the oceanic plate beneath the continental plate. The most prominent high-MI rocks appear along line “V” (Figure 4 and Figure 5c), which is related to the Indus-Tsangbo Suture (Figure 1). The high-MI rocks around lines “W” and “X” are known as related to the Bangong-Nujiang Suture (Figure 1). The ultramafic exposures related to the other sutures in this study area, i.e., the Longmu Tso–Shuanghu–Menglian–Inthanon, the Jinsha and the East Kunlun suture zones (Figure 1), are not well studied. There are the other high-MI bodies, for example the ones around lines “Y”, “Z” and “#”. Additionally, the ones around lines “%” and “&” are located in the northern and the southern part of the Lhasa terrane (Figure 1), respectively. The ultramafic rocks also indicate low CI (<1.02) (Figure 5b), which is another prominent feature in the indices other than high MI (>0.92). Further analysis on the ultramafic rocks with reference to the published articles is shown later in Section 4.3.

Some of the high-MI bodies, for example in and around the rectangle regions “@” and “*”, are in the lacustrine deposits. The possible origin is the eroded mafic-ultramafic materials from the ophiolite carried downstream, overflowing from the lake [18]. Another possible origin of the high-MI signal is the existence of the chloride materials according to the evaporation processes in the deposits, as a part of the chloride materials is known to indicate high MI [17]. Further studies are necessary to confirm, however, that chlorides can be discriminated with mafic-ultramafic minerals using the feature of relatively high measures in all three indices [17].

Considering the characteristics appearing in the indices discussed above, a thematic map is produced by coloring the pixels, satisfying given conditions on the indices, as shown in Figure 6. As described above, high QI has been confirmed to be a distinct indicator for high-quartz/low-feldspar sedimentary rocks [14,15,16,17,18,19,21,22,23,26,29,30], but this study presents the first attempt to classify rock bodies within the high QI. The high-QI (>1.05) bodies are divided into three classes with the conditions on CI and MI. The first class is for low MI (<0.80)/medium CI (>1.02), the second class is for low MI (<0.80)/low CI (<1.02) and the third class is for relatively high MI (>0.82). A further discussion on the discriminations in the high-QI bodies related to the metamorphic domes is made in Section 4.3.

Figure 6. Detected possible rock types colored for each pixel with the given conditions (see legend) on the indices mapped on the grayscale image of MI (0.75–0.9).

Figure 6 shows the thematic mapping for eight classes of possible rock types, i.e., high quartz/low silicate with some carbonates, high quartz/low silicate with minor carbonates, high quartz/low feldspar with some mafic minerals, carbonates, sulfates, sulfate- or feldspar-rich, ultramafic and mafic-ultramafic. This map suggests good detection of each rock type with the given criteria on the indices.

The criterion to detect ultramafic rocks adopted in Figure 6 is MI > 0.92, which has been confirmed as a distinct one [14,15,16,17]. Since there are not so many high-MI bodies in the northern half of the study area, the relatively low criterion, MI > 0.905, is adopted for the mafic-ultramafic rocks to detect more candidates of this target. That is why many pixels are erroneously detected as colored blue in Figure 6, especially in the southern part of the study area.

As shown in the VNIR false color image (Figure 2), a lot of lakes are distributed in this study area. The CI image (Figure 5b) indicates the surface water bodies of the lakes have various values of CI, which perhaps reflects the status (solid or liquid), the temperature, and the composition of the water and so on. Further studies are necessary for clarifying the reason for taking a wide range of CI for the surface water bodies.

The CI is very sensitive to atmospheric water, as described in Section 2.4. The bodies with a high concentration of the atmospheric water indicate high CI, similar to the carbonate minerals. We can detect the clouds with the VNIR image (Figure 2). However, the bodies concentrated with water vapor are not detectable with VNIR. Therefore, it is important to compare the multi-temporal images of CI. If the features are stably appearing through the multi-temporal images, they are reflecting the surface materials. Otherwise, the features are caused by the atmospheric water concentrations. In this study, we tried to find as much cloud-free ASTER data as possible. However, there are still several small fractions of clouds in the maps, for example around the points (28.1° N, 85.5° E), (34.9° N, 85.0° E) and (37.0° N, 89.75° E), annotated as “Clouds” in Figure 4. The clouds, as well as a part of the snow, ice and lakes, are detected as high-CI pixels in Figure 6.

There are many snow or ice bodies that do not disappear through the years, as shown in the VNIR image (Figure 2), although other snow bodies usually easily disappear because of the high evaporation rate at the study area which is located in the world’s highest-altitude region. These deep snow or ice bodies usually look whitish-green (Figure 4) with relatively high CI. Many glacier bodies are observed in this study area, especially in the Himalaya Mountains, as shown in the VNIR image (Figure 2). It is not possible to get any features in the optical remote sensing data for the materials completely covered with snow or glaciers. However, we can retrieve meaningful signals around the bodies. For example, the lithological features around the peaks of Annapurna II, Manaslu, Ganesh I and Ngadi Chuli in the Himalaya Mountains (annotated as “Pk1”, “Pk2”, “Pk3” and “Pk4” in Figure 4, respectively) can be drawn from the maps of the indices (Figure 4 and Figure 5). Despite the area around the peaks being covered with snow or ice, as observed in the VNIR image (Figure 2), the indices suggest the major layers around Mt. Annapurna II are carbonate (light-greenish bodies) with minor layers of quartz (reddish bodies). On the other hand, Figure 4 suggests many kinds of layers distributed around the Manaslu Mountains. The peak region appears to be carbonate rocks (light-greenish bodies), and the regions around the glacier valleys in the southwestern part of the peak appear to be feldspar-rich granitic rocks (dark-greenish bodies). The lithological features in the indices around the Ngadi Chuli peak (“Pk4”), located south of the Mt. Manaslu, are similar to the ones around the Ganesh I peak (“Pk3”). It suggests the alternation of quartzose and carbonate layers expected in the areas. Further studies are necessary for more solid lithological interpretation with the indices in the Himalaya Mountain region.

The VNIR image (Figure 2) indicates that the southwestern part of the study area in the southern slopes of the Himalaya Mountains in Nepal is heavily covered with vegetation. Most of the signals in the indices are obscured by vegetation; however, we can find some signals in the indices (Figure 4 and Figure 5) along the valleys of the rivers in the area, which are possibly geologically meaningful.

The characteristics of the indices for the geological and other materials in the study area discussed above are summarized in Table 1.

Table 1. Characteristics of the indices for geological and other materials.

Expected Material	Typical Geology in the Study Area	Features in Indices (Figure 5)	Color in Index Image (Figure 4)	Detectability with VNIR (Figure 2)	Generality
Quartz rich/Feldspars poor	Quartzose sedimentary rocks	QI > 1.05	Pure to yellowish red	Not clear	Confirmed [14,15,16,17,18,19,21,22,23,26,29,30]
Quartz rich/Silicates poor with some carbonates	Quartzose sedimentary layers related to Tethys	QI > 1.05, MI < 0.8 and CI > 1.02	Yellowish red	Not clear	Not yet (this study is the first attempt)
Quartz rich/Silicates poor with minor carbonates	Low-grade metamorphic sediments around intrusions	QI > 1.05, M < 0.8 and CI < 1.02	Pure red	Not clear	Not yet (this study is the first attempt)
Quartz rich/Feldspars poor with some mafic minerals	Higher-grade metamorphic sediments around intrusions	QI > 1.05 and MI > 0.82	Red to purplish red	Not clear	Not yet (this study is the first attempt)
Carbonates	Marine sedimentary layers related to Tethys	CI > 1.05	light green	Not clear	Confirmed [14,15,16,17,18,19,21,24,25]
Sulfates	Evaporates related to neo-Tethys	QI < 0.98	Deep green	Not clear	Confirmed [15,16,17,30]
Ultramafic	Ophiolites related to sutures	MI > 0.92	Light purple	Not clear	Confirmed [14,15,16,17,18,25,26,27,28,29,30]
Feldspar-rich granitic rocks	Granitic rocks related to island arc (e.g., Gandise batholith)	Relatively low QI/Image pattern	Dark bluish green	Not clear	Partially confirmed [15,17,19,30]
Feldspar-poor granitic rocks	Granitic rocks related to metamorphic domes	Image pattern	Dark pinkish purple	Not clear	Not yet (this study partially attempts)
Mafic rocks	Basalt and gabbro	Relatively high MI (≈0.90)	Deep blue	Not clear	Partially confirmed [15,17]
Atmospheric water	-	High CI	Light green	Partially (for clouds)	-
Snow and ice	-	Relatively high for all the indices	Whitish green	Yes	-
Vegetation	-	Relatively high MI (≈0.89)	Greenish blue	Yes	-
Lake water	-	Relatively high MI (≈0.89), Various values in CI	Various blue to light green colors	Yes	-

4.2. Geometrical and Radiometric Performance

Since the ASTER observes with a pointing angle up to ±8.55° (for SWIR and TIR subsystems; ±24° for the VNIR subsystem) in the plane vertical to the moving direction of the satellite [44], the geometrical error caused by the parallax effect in Level-1B and the series of Level-2B data can be one-sixth of the elevation at the target. As the study area covers the world’s highest region of the Tibetan Plateau and the Himalaya Mountains, the parallax effect is quite significant. On the other hand, the ortho-rectified Level-3A data are corrected for geometrical errors. Comparing the images covering a part of the study area at just north of the Everest peak (Figure 7), the positional mis-registration in the Level-1B and the Level-2B04 images is indicated as shown below. For sub-area “A” indicated in Figure 7, the single rock unit shown in the Level-3A image (pointed out with an arrow in Figure 7c) is observed as double (pointed out with two arrows in Figure 7a,b, respectively). Likewise for sub-area “B”, the lithological boundary between the dark bluish-green unit and the dark pinkish-purple unit shown across the yellow line in Figure 7c is overlapped and a part of the area is missed in Figure 7a,b as indicated with the yellow arrows on the line. These suggest the mis-registration by the parallax effect occurred for Level-1B and Level-2B products, which are corrected for the Level-3A product.

Figure 7. Mosaic color composite image of QI, CI and MI covering the region bounded by the latitude 28° N to 29° N and the longitude 86° E to 87° E with (a) ASTER Level-1B products; (b) ASTER Level-2B04 products and (c) ASTER LEVEL-3A products. The coverage in the study area is indicated as a yellow rectangle in Figure 2. A linear contrast stretching of 0.97–1.055 (QI; Red), 1.005–1.055 (CI; Green) and 0.79–0.95 (MI; Blue) is applied on (a,c), and 0.975–1.085 (QI; Red), 1.09–1.145 (CI; Green) and 0.76–0.96 (MI; Blue) is applied on (b). The sub-areas “A” and “B” are enlarged for pointing out the parallax effects. For “A”, the rock unit shown in (c) indicated with an arrow is observed as double in (a,b) as indicated with two arrows. For “B”, the lithological boundary shown across the yellow line in (c) is overlapped and a part of the area is missed along the boundary in (a,b) as indicated with the yellow arrows on the line. These suggest the geometrical mis-registration by the parallax effect occurred for Level-1B and Level-2B products, which are corrected for the Level-3A product. The location of Mt. Cho Oyu is indicated in (c).

The mosaic index image of Level-1B data (Figure 7a) and the one of Level-2B04 (Figure 7b) are almost identical in their details, which suggests the radiometric performance of the mosaic index map with the radiance registered at the sensor data (i.e., Level-1B or Level-3A) is not inferior to the one with emissivity (i.e., Level-2B04). The boundary of the scenes composing the mosaic map is not significant in any case of Figure 7a,b,c.

Considering the geometrical and radiometric performances on the ASTER data and the indices discussed above, using the ortho-rectified radiance registered at the sensor data (e.g., Level-3A) is obviously advantageous in the mosaic lithological mapping of this study.

4.3. Geological Interpretation of the Results

The characteristics found in the indices (i.e., QI, CI and MI) are discussed in detail in Section 4.1, and summarized in Table 1. Also, a thematic map related to the target rock types is produced with based on the discussion, as shown in Figure 6. In this section, we show more detailed mapping for the high-MI mafic-ultramafic rock bodies and the high-MI lacustrine deposits within the entire study area, and the metamorphic domes [21,36,37] located in the north Himalayan metamorphic belt [38]. These detailed analyses are made by considering the result maps (i.e., Figure 2 and Figure 4, Figure 5 and Figure 6) and the discussion in Section 4.1, as shown in Figure 8. A part of these mapping results of the target rocks is compared with the field investigations published so far.

Figure 8. Distribution map for high-MI mafic-ultramafic rock bodies, high-MI lacustrine deposits and metamorphic domes cored with granite, interpreted with the mosaic images of the indices (Figure 4 and Figure 5), VNIR (Figure 2) and the thematic map with the indices (Figure 6) superimposed on the grayscale image of QI. The red rectangles represent the coverage of Figure 9 and Figure 10, respectively. The information of the previous articles with local geological investigations is embedded as the arrowed lines and texts.

The clearly detectable high-MI ultramafic rocks related to the Indus-Tsangbo Suture (ITS; along line “V” in Figure 4), and the ones of the Yongzhu-Guomangco section and the Dong Tso section related to the Bangong-Nujiang Suture (BNS; along lines “W” and “X”, respectively, in Figure 4), are mapped according to the criterion of MI > 0.92 accomplished in Figure 6. The other high-MI bodies are recognized with the criterion MI > 0.905. Considering the spatial extent of the detected bodies being checked with the result maps, especially Figure 4 and Figure 5c, it is determined whether it is target or not. Considering the VNIR image (Figure 2), the detected high-MI bodies are discriminated between mafic-ultramafic rocks and lacustrine deposits, as shown in Figure 8. To assist readers in locating details of the regional maps, the un-annotated, higher-resolution version of Figure 6 and Figure 8 are presented in the Appendix Figure A6 and Figure A7. For the interactive display with the geographical information and DEM that will achieve further convenience to the readers, the color composite image of the indices (Figure 7c) without embedded text and images is converted to a Google Earth compatible KMZ file as presented in the Supplementary Figure S2.

The distribution map of the ultramafic rocks given in this study using the ASTER-TIR data (Figure 8) agrees well with the published maps [39,40,41,44]. Ninomiya et al. [15,17] suggested the high correlation of the detected ultramafic rocks using ASTER-TIR data with the ones described in the published ophiolite map along the middle Yarlung Zangbo river [44] for the local area south of Xigaze (indicated in Figure 8), but this study suggests the agreement within the entire region of the map. Likewise, the good agreement in the mapping results for the ultramafic rocks of the Yongzhu-Guomangco section [39,40], the Dong Tso section [41] and the Lagkor Tso section [41,42] is suggested.

The advantage of mapping with remote sensing over mapping with a field survey is the heterogenetic observability at the constant interval regardless of the accessibility. On the other hand, whereas the field survey is limited to the area along the accessible routes, it observes many kinds of matters, for example slants and dips of the stratigraphy to reconstruct geology in three dimensions, fossils and radioactive isotope contents in hand samples to determine age, and so on; this is not possible to measure with remote sensing. In that sense, the lithological mapping with remote sensing complements the field geological survey for the materials detectable with the remote sensors in the condition where the rocks appear at the surface over vegetation, snow, and glaciers and so on. Concerning this case study with ASTER-TIR, it can support improving the geological mapping of the materials listed in Table 1, especially of quartz-rich/feldspar-poor rocks, carbonate rocks, ultramafic rocks and sulphatic rocks, which are confirmed to be detected distinctly with the indices as described in Section 4.1 and Table 1.

Figure 9 shows the comparison of the mapped ultramafic-mafic rocks of this study (Figure 9a) with the published geological map at a scale of 1:250,000 [40] (Figure 9b). The ASTER image shows the ultramafic-mafic rocks distributed along the NW-striking fault zones and provides a more detailed spatial distribution compared with the conventional geological mapping. Additionally, after the specific features for the granitic rocks discussed in Section 4.1 and summarized in Table 1, the granitic intrusive bodies in the east of Guomangco Lake are mapped in Figure 9a, which denotes a dark greenish-blue color with a unique texture and shape in the color index image (Figure 4). The color index image (the background image of Figure 9a cut and enlarged from Figure 4) also indicates the distribution of quartzose and carbonate sediments, which appear yellowish-red and light green, respectively. These mapping results suggest that ASTER-TIR index images (Figure 4 and Figure 5) can provide an efficient approach to map the rock types in the remote and inaccessible Tibetan Plateau with an average elevation of over 4500 m.

Figure 9. Comparison of (a) the Yongzhu-Guomangco ophiolite map (the polygons colored purple; enlarged from Figure 8) and the map of the granitic intrusive bodies (the polygons colored reddish purple) derived in this study using ASTER TIR data superimposed on the color composite image of the indices (Figure 4) with (b) the published geological map based on the field investigations (e.g., [39,40]). The coverage of this figure is shown as the red rectangle in Figure 8.

Figure 8 also maps the rock bodies related to the north Himalayan metamorphic domes [21,36,37,38] intruded during the Cenozoic Era to the accretionany wedges of the south Tibetan detachment system, located in the E-W direction in the Himalaya terrane at the southern part of the Indus-Tsangbo Suture. According to the compiled map by Burrell et al. [36] with the published 1:1,500,000 scale geological map [44], granite and granitic gneiss at the core of the domes are accompanied with high-grade metamorphosed gneissose sediments around the core and low-grade metamorphosed sediments around the gneiss in the eastern half of the region (in longitude 87° E and further eastern). On the other hand, it is not so well accompanied by those high-grade and low-grade metamorphosed rocks in the western half of the region (in longitude 87° E and further western).

As discussed in Section 4.1, the characteristics of the indices for the types of granitic rocks are not so distinct at the individual pixels; however, it can be mapped considered together with the unique shape and paternal textures recognized in the images of the indices (Figure 4 and Figure 5). As summarized in Table 1, granitic rocks with relatively high feldspar content appear as dark bluish-green color in the index image (Figure 4). On the other hand, the dominant granitic rocks distributed related to the domes look dark pinkish-purple (Figure 4). This color is probably reflecting less feldspar content. The relatively feldspar-rich granitic rocks of the dark bluish-green color (Figure 4) are also widely distributed in this region related to the domes.

Comparing the colors of the targets (i.e., the rocks associated with the domes) in the image of the indices (Figure 4) with the previous studies [21,36,37,38] and the geological map [44], we assume that the dark pinkish-purple bodies correspond to the cored and the high-grade metamorphosed granitic rocks. Likewise, we assume the correspondences of the dark bluish-green bodies to the feldspar-rich granitic rocks, the pinkish-red bodies to the high-grade metamorphosed sediments and the pure reddish bodies to the low-grade metamorphosed sediments, respectively. The mapped granitic bodies in Figure 8 are only for the ones with a dark pinkish-purple color, because they are the most prominent target characterizing the metamorphic domes. The similar types of bodies in the Himalaya Mountains and the Gandise Mountains are omitted from the mapping (Figure 8). However, the body at the southern part of Mt. Loinbo Kangri (indicated in Figure 2), which was not mapped previously [36,37,38,45] for being recognized as a part of the Gandise granitic rock group, is included in this map (Figure 8) because it is located just south of the Gandise Mountains within the Tethyan quartzose and carbonate sediments related to the south Tibetan detachment system, as suggested in Figure 4. As this study is the first case to attempt detecting these kinds of rock types with these indices, the generality of which has not yet been confirmed (Table 1), the mapping result of the metamorphic domes shown in Figure 8 presents a preliminary case.

The Kangmar dome and the western portion of the Malashan dome, the locations of which are indicated in Figure 8, are the best-investigated ones from the previous studies [37,38]. They are dominated by the dark bluish-green (Figure 4) bodies assumed here as the feldspar-rich granitic rocks. On the other hand, most of the other domes are dominated with the dark pinkish-purple bodies which are assumed here as the relatively feldspar-poor granites and high-grade metamorphosed ones. As the target of the mapping in this study (Figure 8) is the latter, very small bodies are mapped for the Kangmar and the western Malashan units compared to the published maps [36,45].

Several domes are newly mapped in Figure 8 in addition to the known ones mapped so far [36,45], typically the one just west of the Kangmar dome (Figure 8). Further advancements are necessary in the field for the validity of the new mapping.

According to Figure 4 and Figure 6, most of the domes are accompanied by the Tethyan quartzose sediments of the pinkish-red bodies (Figure 4), assumed here as high-grade metamorphosed sediments, and reddish bodies (Figure 4), assumed here as low-grade metamorphosed sediments. Figure 4 suggests that the association is more obvious for the domes in the eastern half of the region compared to the ones in the western half, as indicated by Burrell et al. [36]. Probably, the reason for the difference is because the domes in the eastern half of the region are intruded into the accretionary wedges dominated with quartzose sediments; on the other hand, the ones in the western half are intruded into the ones dominated with carbonate sediments as inferred from the index map (Figure 4).

Figure 10 shows the comparison of the lithological mapping using the color index image (Figure 4) with the geological summary map investigated by Aoya et al. [37] around the western Malashan dome located in the western shore region of Paiku Lake. This figure indicates good agreement between our result and Aoya’s result. The spatial extents for some of the geological bodies are more closely mapped in this study, as expected with the strong points of remote sensing described earlier in this section. On the other hand, it is difficult to distinguish between two-mica granites and leucogranites with the approach of this study, which is inferred as the difference of the chemical or mineral composition in the two types of granite is not as significantly detectable with ASTER-TIR. Some high-QI bodies detected as quartzose sediments in the map of this study (Figure 10a), for example the one just west of the Paiku leucogranite body, are classified as calc schist in Aoya’s map (Figure 10b). The units mapped as pelitic schist in Aoya’s map correspond well to the high-QI/relatively low-CI bodies with a pure reddish color in the map of this study (Figure 10a), assumed here as low-grade metamorphosed quartzose sediments.

Figure 10. (a) Lithological mapping with the mosaic color composite image of QI, CI and MI around the Malashan domes. The coverage of this image is indicated in Figure 8. The interpreted rock types with this image are A: Quartzose sedimentary layers related to the Tethys, colored yellowish-red; B: Low-grade metamorphic quartzose Tethyan sediments related to the intrusion, colored pure red; C: High-grade metamorphic quartzose Tethyan sediments around metamorphic domes, colored pinkish-red; D: Relatively feldspar-poor granitic rocks related to the domes, colored dark pinkish-purple; E: Feldspar-rich granitic rocks around the domes, colored dark bluish-green; F: Carbonate layers related to the Tethys, colored light green; G: Sulfate deposits, colored deep green; H: Feldspar-rich granitic rocks at the glacier valleys in the north slope of the Himalaya Mountains; (b) Geological map around the western Malashan metamorphic dome area investigated by Aoya et al. [37] embedded in the color composite image of the indices.

Apart from the geological units mapped by Aoya et al. [37], there are many lithological features that are mappable with the color index image (Figure 10a) where alphabetic symbols are annotated in the figure to indicate the targets. As described in Section 4.1 and summarized in Table 1, quartz-rich/feldspar-poor layers, carbonate layers and sulfate deposits are distinctively detected as “A” to “C”, “F” and “G”, respectively, according to the colors based on the features in the indices. Feldspar-rich granitic rocks detected at “E” and “H” are discriminated with another type of granitic rock detected around “D”, which are mapped as the metamorphic dome in Figure 8, as the color for the former is dark bluish-green whereas that for the latter is dark pinkish-purple. This is the first study of the latter granitic rocks with reference to the indices for ASTER-TIR data. Also, this study attempts classifying the high-QI layers into three classes, i.e., relatively high-CI/low-MI layers assumed here as non-metamorphosed quartzose sediments, relatively low-CI/low-MI layers assumed here as low-grade metamorphic quartzose sediments, and relatively high-MI layers assumed here as high-grade metamorphic quartzose sediments. In Figure 10a, these rock types are distributed at “A”, “B” and “C”, respectively. The materials detected with the features in the indices given in this study for the first time are determined under assumptions, which should be confirmed with the field investigations.

As a summary, although much of the information needed for geological mapping is acquired only with the field investigations, the complementary use of this approach will improve the quality of mapping in detecting the exact spatial extent of the target materials of this study.

5. Conclusions

ASTER TIR mineralogical indices, i.e., QI, CI, and MI, were defined on the basis of the spectral properties of typical geological materials and applied to a large central region of the Tibetan Plateau. The advantages of using the ortho-rectified radiance registered at the sensor data (e.g., Level-3A products) were discussed in addition to sensor geometrical and radiometric performances. The maps were interpreted to accomplish more detailed mapping of ultramafic rock bodies and north Himalayan metamorphic granite bodies. This study sought to differentiate granitic rock types according to the abundance of feldspars, as well as the classification of the quartzose sedimentary layers according to the content of carbonates and mafic minerals. Results indicate that lithological mapping using TIR spectral indices was successful in detecting quartzose sedimentary, carbonate, and sulphatic layers, in addition to ultramafic bodies, which had been confirmed from previous remote sensing investigations. The results from the detailed mapping were compared to prior field-based geological mapping studies and it was found that they also agreed well. The advantage of using remote sensing to detect the spatial extent of the target materials, i.e., its superiority over conventional geological mapping, has been well established. However, the geological materials detected using these described techniques need to be confirmed through a vigorous campaign of ground-based field investigations. A further survey is planned where regional lithological mapping will be applied to the entire Tibetan Plateau, as well as to areas in northwestern China, such as the Tarim Basin and the surrounding area.