Application of high-resolution remote sensing technology for the iron ore deposits of the West Kunlun Mountains in China

This study focuses on the iron ore of Taxkorgan and Heiqia in the West Kunlun mountains as a case study, for the application of WorldView−2 and IKONOS remote sensing images as major data sources in the fabrication of a standard image map and in the adoption of image enhancement methods to extract information on the ore-controlling factors and mineralization, to interpret remote sensing for the mineral resources in these areas. ASTER, WorldView−2, and IKONOS data were applied for the extraction of alteration anomaly information. With an appropriate amount of field sampling and verification tests, this was used to establish a remote sensing geology prospecting model, that would provide the basis for future remote sensing of metalloge nic belts in West Kunlun in the hope of discovering similar minerals. Survey results showed four additional iron ore mineralization belts could be delineated in the Taxkorgan area. A comparative analysis conducted for part of the field confirmation and the known mineral deposits indicated good reliability. In Heiqia, a siderite-haematite mineralization zone was observed with copperlead-zinc formation, 60-km in length and 200–500 m wide, which includes several mineralized bodies. The ore bodies, appear as stratoid, lenticular, or podiform morphologies and were located in the transition site from clastic to carbonate rocks of the D segment in the Wenquangou Group. The ore bodies generally occur within 40°–50° strike and 68°–81° dip, in accordance to the strata. The length of the single body varies from several hundred metres to more than 9500 m. Its exposed thickness on the surface ranges from 2–50 m, and the general thickness was approximately 15 m. The surface ore minerals were mainly haematite and limonite, with a small amount of siderite. Therefore, high-resolution remote sensing technology is suitable for iron ore geological and mineral remote sensing surveying. It is advantageous in both high-ground resolution of optical characteristics and a certain spectral recognition capability, and is effective not only for information extraction from a large area, but also for recognition of local mineralization outcrops. Therefore, high-resolution remote sensing technology is valuable for popularization. an effective, prospective means of acquiring the geological anoma ly information related closely to ore-bearing strata, mineralized alteration zones, contact metamorphic zones, and tectonic zones (SHANG, 2009; WANG et al., 2011; CHEN et al., 2012; JIN et al., 2014; ZHANG et al., 2015; YANG & ZHAO, 2015; HOWARI et al., 2019; FAN et al., 2021). Located at the junction of the Palaeo–Asian and Tethyan tectonic domains, West Kunlun is an important part of the Qin–Qi– Kun tectonic belt of China, and also serves as an important area for studying the evolution of the Tethyan Ocean (PAN,1989, 1994, 1999; JIANG & ZHU, 1992; CHENG, 1994; JIANG & ZHU, 1992, 2002; YANG, 1994; PAN et al., 1995; PAN et al., 1996; CUI et al., 2006; WU et al., 2008; WANG et al., 2013; LI et al, 2011. LI, 2015; MENG et al., 2019). West Kunlun is characterized by strata exposed from Palaeoproterozoic to Mesozoic age (WANG H et al., 2016; ZHAO et al., 2010), by strong folds and faults (WU et al., 2008), and by the experience of multistage and various types of tectonic events (LU et al., 2003; YANG et al., 2004; XU et al., 2004; LI et al., 2008), or geological events in different periods that are superimposed on each other, such as multistage magmatism and multistage metamorphism with complex magArticle history: Manuscript received May 09, 2019 Revised manuscript accepted February 03, 2021 Available online February 28, 2021


INTRODUCTION
Remote sensing is an emerging technology discipline that began with the successful launch of the Multispectral Scanner (MSS) on the United States' Landsat (LANDSAT) in 1972 (WANG et al., 2011;ALIJAGIĆ & ŠAJN, 2020). Since then, coupled with the continuous development of space remote sensing and satellite sensor technology in recent years, a variety of high-resolution and improved remote sensing data has emerged both locally as well as globally. For instance, Worldview-2 (SHANG, 2009) and Quickbird have both reached metre-level resolution, while Worldview-3, launched in 2014, has achieved 0.3-m resolution and a broader spectral range. Moreover, China's launch of highresolution remote sensing satellites, such as GF_1, GF_2, Resource 2, Resource 3, Tiantu 1, and Tiantu 2, has pushed the technology and its applications to new heights, including use in the detection of mineral deposits ZHANG et al., 2015;YANG & ZHAO, 2015). Geological deposits contain different mineral and chemical compositions to their surrounding rocks, and these differences are often reflected in remote sensing images in the form of spectral anomaly information. In this regard, a series of remote sensing digital image processes becomes an effective, prospective means of acquiring the geological anoma ly information related closely to ore-bearing strata, mineralized alteration zones, contact metamorphic zones, and tectonic zones (SHANG, 2009;WANG et al., 2011;CHEN et al., 2012;JIN et al., 2014;ZHANG et al., 2015;YANG & ZHAO, 2015;HOWARI et al., 2019;FAN et al., 2021).
Located at the junction of the Palaeo-Asian and Tethyan tectonic domains, West Kunlun is an important part of the Qin-Qi-Kun tectonic belt of China, and also serves as an important area for studying the evolution of the Tethyan Ocean (PAN,1989(PAN, , 1994(PAN, , 1999JIANG & ZHU, 1992;CHENG, 1994;JIANG & ZHU, 1992, 2002YANG, 1994;PAN et al., 1995;PAN et al., 1996;CUI et al., 2006;WU et al., 2008;WANG et al., 2013;LI et al, 2011. LI, 2015MENG et al., 2019). West Kunlun is characterized by strata exposed from Palaeoproterozoic to Mesozoic age (WANG H et al., 2016;ZHAO et al., 2010), by strong folds and faults (WU et al., 2008), and by the experience of multistage and various types of tectonic events (LU et al., 2003;YANG et al., 2004;XU et al., 2004;LI et al., 2008), or geological events in different periods that are superimposed on each other, such as multistage magmatism and multistage metamorphism with complex mag-JELENKOVIĆ, WANG et al., 2018). West Kunlun's iron polymetallic ore belt includes features of a large number of deposits, large-scale, and associated polymetallic mineralization that have attracted geological research both locally and internationally (FENG et al., 2011;QIAO et al., 2015QIAO et al., , 2016WANG H., 2016;LI et al., 2010). After years of field work, the present author has dis-covered magnetite deposits, such as Laohe and Zankan, in the Bulunkuole group of the Palaeoproterozoic in the Taxkorgan area, along with the haematite-pinerite containing copper, lead, and zinc, in the Wenquangou group of Lower Silurian age in the Heiqia area (WANG H., 2016). The iron ore of Taxkorgan and Heiqia in the mountains of West Kunlun is used herein as a case study and a major data source for WorldView-2 and IKONOS remote sensing images, in the fabrication of a standard image map and adoption of methods of image enhancement so as to extract information on ore-controlling factors and mineralization via interpretation of the remotely-sensed mineral resources. Data provided by ASTER, WorldView-2, and IKONOS are exploited for the extraction of alteration anomaly information. A remote sensing geology prospecting model is established with an appropriate amount of field sampling and a verification test, to provide the basis for future remote sensing on the metallogenic belt of West Kunlun, in the prospect of discovering similar mineral deposits.

Geological characteristics of iron deposits of the Taxkorgan area
The Taxkorgan area is bounded by the Kangxiwa-Muzitage-Animaqing Late Palaeozoic junction belt, with the Upper Carboniferous (undivided) in the north, with a sporadic distribution of Ordovician-Silurian, and the widely exposed Palaeoproterozoic Brungula Group in the south. The contact relationships between different strata are tectonic. Magmatic activity is very strong and an intermediate-acid rock mass is widely distributed north of the junction zone. Similarly, south of the junction zone, magmatic activity is frequent although there is less distribution of the intermediate-acid rock mass. Relatively, there are strong folds and faults in the area, accompanied by an overall structural orientation to the NW. The strata exposed in this area mainly belong to a ferrosilicon formation of high green schist facies of the Blenkole Group in the Palaeoproterozoic, distributed mainly in the in NW and SEE directions, and mostly in fault contact with the surrounding strata.
Metallogenic geological background: The study area was located in the Taxkorgan landmass, with the junction zones of Taasi-Sekblak and Kangxiwa-Wacha as north and south boundaries, respectively. The outcropping strata belong to the Brungula Group of Palaeoproterozoic age (Pt 1 B). Faults and folds were well developed in the area. The dominant structural orientation was NW, consistent with the strike of the north-south boundary faults. Magmatic activity existed from the Proterozoic-to-Himalayan tectonic movement. Ultrabasic to acidic magmatic rocks were exposed, of which intermediate-acidic granite and granodiorite were dominant during the Yanshanian tectonic movements (YAN et al., 2012;WAND, J.F. ,2013;WANG et al., 2013;WANG et al., 2017).
Ore body characteristics: Most magnetite ores were ferrous black and gray-black, irregular (thick, massive, layered and lenticular) bodies. The occurrence of the ore bodies was similar to those of the roof and floor wall rocks, striking NWSE and dipping NE. The surface was layered, banded, and irregularly distributed. The lithology of the roof was mostly (iron-stained) black cloud quartz schist (WANG et al., 2013;WANG et al., 2017).

Geological characteristics of iron-polymetallic mineralization belt of Heiqia
The ironpolymetallic mineralization belt of Heiqia is located in the D Formation of the Wenquangou Group of Early Silurian age (S 1 W d ), whereas its northeast side is a Formation of the Huangyanling Group from the Early Middle Permian (PH a ) along the Karatag fault boundary. The lithology of S 1 W d is composed mainly of cinerous-gray sandy slate with argillaceous slate and metamorphic sandstone; the top contains several layers of carbonate rocks, such as dolomite, iron dolomite, silicified dolomite, and a small amount of limestone (locally marbled) with siderite. In contrast, the lithology of PH a is characterized by gray-black carbonaceous slate or spotted (pyrite-phenocryst) slate. It was truncated northward by the Kangxiwa fault and extended southeastward beyond the zone for approximately 60-km. Many new metallogenic clues (7 mineralization points in the northwest section and 3 mineralization points in the southeast section) have been discovered in the belt through geochemical surveying of stream sediments, remote sensing anomaly verification, and detailed surface tracing, together with an exploration trench-revealing project (Fig. 2).

KEY TECHNOLOGIES AND METHODS
The WorldView-2 satellite could provide an 8 band, 1.84-m reso lution, multispectral image and 0.46-m panchromatic image, with 4 industry standard bands (red, green, blue, near-infrared) and 4 additional bands (blue indigo, yellow, red edge, near-infrared 2). Its diverse spectrum contains abundant spectral information (0.400-1.040 um), while its extremely high spatial resolution can accurately reflect the structure, shape, texture, and other characteristics of objects; thus, the WorldView-2 satellite is a fine data source for geological remote sensing (ZHU, 2003;JIN et al., 2013). The WorldView-2 remote sensing image data used in this paper was from May 2010.
The IKONOS satellite, manufactured by LOCKHEED MARTIN, could collect 1-m resolution full colour and 4-m resolution multispectral images; the full colour and multispectral images could be blended into the 1-m resolution colour images. In this paper, the IKONOS remote sensing image data used was from June 2010.
ASTER is a multispectral imager mounted on the Terra satellite, the received information of which includes the spectral reflectance of the ground in the VNIR (visible and nearinfrared) and SWIR (thermal infrared) bands, and thermal radiation of the ground in the TIR range (thermal infrared). Due to the wide wavelength range, several bands, and reasonable cost performance for the data, ASTER is widely used in extracting remote sensing (mineralization) alteration anomaly information .
The high spatial resolution data of WorldView-2 and IKO-NOS were taken as the main information sources to establish standard remote sensing image maps through image preprocessing, geometric correction, image fusion, image enhancement, and image mosaic. On this basis, image enhancement was processed by band combination transform, principal component transform information decomposition, multiple principal component analysis (PCA), and differential stretching.

Production of standard remote sensing images
According to the principle of band selection, the bigger the variance of band radiation and the smaller the band correlation is, the better. Through calculation of the best index, or optimum index factor for B8 (R) B4 (G) B3 (B) of WorldView-2, and B3 (R) B2 (G) B1 (B) and B4 (R)B2 (G) B1 (B) of IKONOS, image fusion was performed using intensity-hue-saturation (IHS) transform, principal component transform, and PANSHARP. Afterwards, a 1:50000 topographic map and digital elevation model (DEM) data were used for selecting the rational polynomial coefficient parame ter correction model to carry out orthorectification and cubic convolution resampling. Finally, the inlay mosaic method was used to create an image mosaic, and the base image for remote sensing interpretation was produced.

Recognition of lithologic information by band combination transformation
The characteristics of different lithologies that correspond to the absorption/reflection bands of WorldView-2 and IKONOS were applied into a series of band algebra operations to enhance the lithological information. For example, a combination of band8, band4/band1, and band5/band3 from the WorldView-2 data facilitated identification of diorite, marble, and schist, or highlight the different lithological differences and improve the accuracy of lithological interpretations (Fig. 3a).

Recognition of lithological information by information decomposition technology of principal component transform
The information of 532 and 681 principal component transformation was decomposed to fuse the two results and obtain a new image. It contains multiple bands of information and makes the correlation of different bands smaller, thus, enhancing the contrast between the different lithologies. For example, the different mineral contents of iron oxide, biotite, muscovite, feldspar, and quartz in schist yielded striplike patterns of different tones, of which the orange-red tone with high iron content was either a magnetitebearing quartz schist or a magnetite belt (Fig. 3b).

Multiplex principal component analysis
PCA, band difference, ratio, and the like were used for the extraction of weak geological information in the image, and further, extraction of the most abundant combination of information, secondary processing, or selection of the most favorable result on the thematic information. Specifically, PCA was repeated for the last two objectives and the geological information was enhanced twice.

Differential stretching
Initially, the high-resolution remote sensing image of World-View-2 did not show considerable distinction between the diffe rent lithologies, such as whether phyllite and metasandstone were of light blue appearance. Upon application of differential stretching, the distinction became clear, such as the phyllite being light blue and the sandstone being white.

Extraction of information of mineralization and alteration based on ASTER
Several methods can be used to perform extraction of anomaly information in remote sensing (mineralization) alteration, including PCA, ratio analysis, and spectral angle analysis. If the principle considers large-scale ore (often containing high-value anomaly) as a main objective, PCA is the primary choice. Thus, iron staining, hydroxyl, and carbonate anomalies were extracted by principal component transformation, considering the relationship between the absorption spectra of various anomalies and the ASTER band. The main minerals in the area were sedimentary metamorphic magnetite, and the iron stain abnormality was the most direct remote sensing anomaly information with the iron ore. Nonetheless, the relationship between hydroxyl and carbona te anomalies was less related to the iron ore, providing the basis for a focused discussion on iron-stained remote sensing anomaly extraction herein.
Fe 2+ generated a strong and broad band at approximately 1.0-1.1 μm, whereas Fe 3+ exhibited strong absorption at 0.45, 0.55, 0.85, 0.90, and 0.94 μm, specifically indicating a strong absorption band between 0.9-1.0μm and a relatively strong reflection between 0.6-0.8 μm. According to the relationship between the abnormal absorption spectrum of iron staining and the AS-TER data band (Table 1), the iron stain information was reflected mainly in B1 to B4 bands, with strong absorption in B1 and B3, and strong reflection in B2 and B4.
Iron alteration information (FCA) extraction: Principal component transformation was performed for B1, B2, B3, and B4 bands of the ASTER data, with the mean + 4 s (standard deviation) as the dynamic range of the principal component output. The iron dye information was absorbed in ASTER data bands 1 and 3 and reflected in bands 2 and 4, thereby, the eigenvectors of the anomalous principal components were characterized by the opposite sign of the contribution coefficients of bands 1 and 3 and bands 2 and 4. After the principal components of 1, 2, 3, and 4 were transformed to obtain the feature matrix (Table 2), the fourth principal component band 2 was opposite the band 1 and band 3 symbols, but was the same as the band 4 symbol, and therefore, could be used as the main component of the iron-staining anomaly.

Extraction of information of mineralization and alteration based on WorldView-2
The centres of the unusual absorption spectra of iron staining were 0.45, 0.55, 0.85, and 0.90 μm, corresponding to B1, B2, B3, B7, and B8 of the WorldView-2 data. Accordingly, the reflective characteristic spectra were within 0.60-0.80 μm, corresponding to B4, B5, and B6 of the WorldView-2 data, of which, B1 and B8 showed strong absorption while B4 showed strong reflection. As a method of orthogonally transforming several specific bands, PCA could be employed to remove the correlation between bands   and reduce the dimension of data, for concentration of as much useful information as possible, into a small number of band images. In PCA, each principal component often represents a certain unique geological significance.
After the principal components B1, B4, B6, and B8 of the Worldview-2 data were transformed to obtain the feature matrix (Table 3), the coefficient of B4 was found opposite to the coefficient of B1 and B8, but was the same as the B6 coefficient, and therefore, could be used as a principal component of the ironstaining anomaly.
In this study, PCA was applied for extraction of the iron ore mineralization (iron staining) remote sensing anomaly information. PC3 was the characteristic principal component of such an anomaly, in which the low-value area became the high-value area of the anomaly. The lower limit of abnormality was determined according to the mean +3δ (standard deviation), with the apparent extracted iron staining anomaly and higher degree of coincidence with the known iron ore body (Fig. 4).

Extraction of information of mineralization and alteration based on IKONOS
The centres of absorption spectra of iron staining were 0.45, 0.55, 0.85 and 0.90 μm, corresponding to B1 of the IKONOS data.    Accordingly, the reflective characteristic spectra were within 0.60-0.80μm, corresponding to B3 of the IKONOS data.
After the principal components B1, B3, B4, and B2 were transformed to obtain the feature matrix (Table 4), the coefficient of B3 was found opposite those of B1 and B4, but was the same as the B2 coefficient, and, therefore, could be used as the principal component of the iron-staining anomaly.
The lower limit of abnormality was determined according to the mean +3δ (standard deviation), with the banded extracted iron staining anomaly (Fig. 5).

Remote sensing geological characteristics of iron ore in the Taxkorgan area
The WorldView-2 data were combined with the B8 (R) B4 (G) B3 (B) band, whereas the DEM data were employed for orthorectification. A full colour band fusion process was used to achieve a 0.46-m fusion image resolution, later enhanced and transformed to highlight the lithology, structure, and iron ore (body) information.

Stratigraphic lithology
Within the study area, magnetite was localized in the iron-bearing section of the Brongol Group (Pt 1 B). The main lithology included (ironstained) biotite quartz schist, followed by gneiss with hornblende and biotite, magnetite quartz schist, two mica quartz schist, sericite quartz schist with magnetite, a small number of gneiss with plagioclase and hornblende, chlorite schist, and marble. On the combined image of B8 (R) B4 (G) B3 (B) band from the WorldView-2 data, biotite quartz schist was dark gray-black with a striped pattern and medium weathering resistance, and mostly formed a gentle slope microtopography that was most closely related to the magnetite space, as the wall rock of the ore body (Fig. 6a). Biotite plagioclase gneiss (including magnetites?) was dark gray to dark gray in tone with a striped pattern and strong weathering resistance, and mostly formed a striped steep ridge microtopography that was closely related to the magnetite space as the important ore hosting wall rock. Moreover, magnetite quartz schist was graygray in tone with a narrow strip-like pattern and strong weathering resistance, and mostly formed a normal topography that was closely related to the magnetite space, often located in the magnetite mineralization zone. The dimica quartz schist was bright gray to dark gray or variegated tones with a banded pattern and medium weathering resistance, and mostly formed a gentle slope microtopography. Sericite quartz schist was bright graywhite tones with a fine striped pattern and medium weathering resistance, and mostly formed a similar gentle slope microtopography. Plagioclase amphibolite schist (gneiss) was dark gray-dark gray-greengray-black tones with a dark tone, banded or massive schistosomiasis, weak or non-schistosomiasis. Chlorite schist was dark graygreen to dark gray in tone, with fine striplike patterns and weak weathering resistance, and mostly formed gentle slopes or negative topographic micro-landforms. Marble was bright grayyellow to gray-white in colour, with banded and irregular blocklike patterns and strong weathering resistance, and was generally dominated by a normal topography (Fig. 6b).

Structure
The study area was located in the Taxkorgan landmass, with the Taasi-Sekblak and Kangxiwa-Wacha junction zones as the northern and southern boundaries, respectively. Faults and folds were well developed in the area, at a NW dominant structural orientation, consistent with the strike of the north-south bounda ry faults. Most of the faults were compressive (torsional) structures. The image showed apparent linear structural features, which could form linear tone shadow abnormal zones and a linear negative topography. Moreover, hues and textures on both sides of the fracture were significantly different, often forming dif ferent image volume boundaries (Fig. 7a). Some magnetite bodies in the area were reformed by secondary fracture structures, as manifested in two situations. First, under the influence of hydrothermal fluid (> 300 ℃), the thickness of the ore body increased and its grade became rich. Second, under no hydrothermal coordination or low hydrothermal temperature (< 300 ℃), the ore fabric changed, mostly in the form of fragmentation and breccia, and the ore body continuity was destroyed to varying degrees. The fold structure showed the same set of banded image bodies on the image map, or an image marker layer with mirror symmetry and repeated distribution in space, continuous along the upwardly-inclined or back-tilted end. Such distinct features were easily identifiable (Fig. 7b). As an integral part of the stratum, the iron ore body was deformed synchronously with the ore-bearing surrounding rock, under the action of longitudinal bending and folding. Additionally, under the action of flattening, the thickness of the wing ore body was reduced while that of the turning end body was increased.

Ore body, mineralized zone
The field observation results indicated that most magnetite ores are ferrous black and gray-black and have irregular (thick, massive, layered, and lenticular) bodies. Specifically, the occurrence  of the ore bodies was similar to those of the roof and floor wall rocks. The ore hosting rocks were mainly biotite quartz schist, followed by biotite plagioclase gneiss, magnetite quartz schist, and quartz schist. In the Laobing area, the iron ore bodies were patchy (dotted), beaded, and distributed intermittently along the foliation direction (Fig. 8b). Here, magnetite was a bright orange yellowhue on the B8(R)B4(G)B3(B) band, combined with a highscore image of Quickbird (Fig. 8a). The Kebengou iron ore belt with good continuity was distributed in regular strips ( Fig. 8d). Due to differences in surface scale, occurrence, and outcropping degree of the iron ore bodies, their shape and shadow structures were also different.

Iron ore belt distribution along the area of ore concentration
Based on the extracted remote sensing (mineralization) alteration information and remote sensing interpretation of the main ore controlling elements, the distribution of mineralization zones in the iron ore belt was preliminarily obtained through a comprehensive remote sensing interpretation of their lithology, structure, intrusive rocks, ore bodies, and mineralization zones in the area of ore concentration. Specifically, the iron ore (mineralization) zones in the study area were distributed as follows (Fig. 9): (1) The Zankan iron ore (mineralization) belt extends 5 km NW to SE, from the Mokart Snow Mountain (Pl. 1a, b). The Laohe-Yelite-Taasi iron ore (mineralization) belt extends over a 21-km-long and 1-2-km-wide NW-SE distribution. There were 2 to 4 mineral veins in this belt, occurring as stratified or stratoid in form (Pl. 1c, d; Fig.8b). Similarly, the Zoukeben-Taaxi iron ore deposit (mine ralization) belt lies along a NW-SE distribution that extends to 18 km and that occurs as discontinuous stratified, stratoid, or lenticular morphologies (Fig. 8d). Moreover, the Yukuli Kukat iron ore (mineralization) belt in the northern Maryang Ocean extended approximately East-West, over a 2km distance.

Remote sensing image interpretation markers
The iron polymetallic mineralization zone of Heiqia in the high resolution remote sensing images of IKONOS (band combination B3 (R) B2 (G) B1 (B)) showed a light brown banded pattern, which was slightly different from the image characteristics of the two sides of the rock layer (Fig. 10a). In the false-colour synthetic image of the enhanced thematic mapper (ETM; band combination B7(R) B4 (G) B3 (B)), the mineralized zone showed a yellowgreen tone and regular strip pattern. On both sides of the rock strata, a blue stripe pattern was apparent; although the colour difference between these rocks was apparent and easily distinguished, the boundary was barely discernible (Fig. 10b).
The effect of remote sensing interpretation was enhanced by decorrelating the B3(R) B2 (G) B1 (B) band combination of the IKONOS images. The iron polymetallic mineralization belt of Heiqia was brown in varying depths and regular striplike pattern in the enhanced image, which was apparently different from the image characteristics of the two lateral strata outside the belt. Nonetheless, the boundary was clear and easily identifiable. Iron ore bodies were located mostly in the middle and lower part of the mineralization zone, with dark reddish-brown tone, narrow strip, lenticular, beaded, and intermittent extension, whereas the boundary with the mineralization zone was clear. Through detailed interpretation, the mineralized bodies were distributed mainly in the northwest part of the mineralized zone, although other areas were difficult to interpret due to the small-scale single mineralized bodies or the serious surface coverage. The tones of floor wall rocks were mainly dark grayblue dark brown-black with banded shape, whereas the tones of roof wall rocks were light gray-white-light yellowish-brown with banded extension (Fig. 11a). Information decomposition based on the PCA was the common method of enhancing the weak information of geological lithology. Using the IKONOS data in B1, B3, B4, and B2 bands for the PCA, band 4 was combined with the original bands 1 and 3 to obtain a new image. The iron polymetallic mineralization belt of Heiqia was a pink tone and banded shadow, obviously different from the image characteri stics of the surrounding rock (Fig. 11b).

Geological characteristics of the ore (mineralization) body in the study area
The ore bodies are of irregular stratoid, lenticular, or podiform appearance, with a few veins having bedding or through the  strata, had a sudden change of relationship with the surrounding rocks. It has an inclination of 40°-50°and an angle of 68°-81°, consistent with the occurrence of the surrounding rocks. The length of a single ore body was typically tens to hundreds of metres, and its surface outcropping thickness was within tens of centimetres to several metres (Table 5). The boundary between the ore body and the upper and lower footwall was distinctive. The surrounding rock was mainly iron (iron-containing) dolomite, followed by dolomitic limestone, silicified dolomite marble (mostly located in the ore body), and a small amount of other metamorphic clastic rocks (metamorphic sandstone and sandy slate). The wall rocks of the ore body roof were dolomite, iron dolomite, silicified dolomite, marble dolomite, (strip) marble, and a small amount of limestone. The wall rocks of the ore body floor were sandstone, metasandstone, sandy slate, and carbonate rock.

Ore characteristics
The surface iron ore minerals were mainly limonite, followed by haematite and a small amount of siderite. Galena, blue copper ore, malachite (Pl.2d), and occasional? sphalerite, antimony lead ore, white lead ore, and black lead ore, were found in the lead (copper, zinc) associated with the iron ore. Gangue minerals included quartz, muscovite, iron dolomite, and barite, and occasionally graphite, tourmaline, apatite, among others. The iron ore structure was composed mainly of two types: fine grain and the medium grain coarse-grained (variable crystal). Furthermore, the main iron ore structure was of four types: (a) a compact block (Pl.2a) and (hidden) lamellar structure formed during the syndepositional period; (b) banded and wrinkled structures formed during the post-metamorphic stage; (c) veins, cavities, clusters, and breccia structures formed during hydrothermal superimposition in the late tectonic stage (Pl.2b); and (d) honeycomb, earth like, colloidal, and tuberculous structures formed after weathering and leaching of the earth's surface. Galena, which forms in the cracked carbonate rocks of the roof of an iron ore body, was mainly of semiebony and finegrained structure (Pl.2c). Moreover, it was filled with irregular structural fractures (Pl.2e). Likewise, fractured carbonate was filled with structural fractures (Pl.2f), showing disseminated and membranous structures. The Pb (Zn) ore was closely related to the iron (iron-bearing) dolomite, whereas Cu mineralization was related closely to dolomitic marble.

Ore grade
The results of the surface sampling and laboratory test showed a total iron? (TFe) grade within the range of 29.90% to 51.03% and an average grade of 36.94%, whereas Pb+Zn in the lead-zinc ore body had a grade within 0.24% to 2.37%, and that of Cu in the copper ore body was within 0.27% to 0.86%.

Metallogenic regularity and zoning characteristics of the metallogenic elements
The results of the systematic field investigation showed that the NW-SE development of the iron ore bodies deteriorated gradually, and the surface thickness of the mineralized bodies decreased gradually, together with the strike extension length of the single ore bodies. Spatially, the iron polymetallic mineralization zone was closely related to the carbonate rocks on the roof of the ore body. Carbonate rocks were developed in all occurrences of the iron ore body, although its thickness often had insignificant or negative correlation with the thickness of the carbonate rocks. Along the strike of the mineralization zone, metallogenic elements NW-SE generally showed the zoning rule of Pb-Zn-Cu elements.
As the iron ore body's thickness reduced, the Pb content decreased gradually, while that of Zn gradually increased. Moreover, as the Cu content increased in the pinch-out area of the iron ore bodies, the element's independent ore bodies appeared.

REMOTE SENSING INTEGRATED PROSPECTING MODEL
Based on the analysis of the metallogenic geological characteristics of the Taxkorgan and Heiqia iron deposits, interpretation of high-resolution remote sensing of ore controlling elements, and extraction of anomalous information of remote sensing mineralization and alteration, a comprehensive remote sensing prospecting model based on high-resolution remote sensing images was established and supported by field validation and correction (Table 6, Table 7).

CONCLUSIONS
(1) WorldView−2 and IKONOS remote sensing images were used as major data sources for the creation of a standard image map and for the adoption of image enhancement methods to ex- Surrounding rock alteration that is generally weak; sericitization, chloritization, carbonation, and silicification that are common only in the surrounding rocks with strong tectonic reformation in the latter stage; apparent malachitization, blue copper mineralization, or lead-zinc mineralization in the mineralization zone Criteria of field geological prospecting An ore body closely related to the iron dolomite in space; iron dolomite that is dark-yellow brown after oxidation on the surface and is easily identifiable, which is a good prospecting indicator; exposure of the primary siderite on the surface, followed by its transformation into iron oxides, such as hematite and limonite under oxidation to form an "iron cap, " which can directly guide the search for the primary siderite

Remote sensing characteristics
High-score image features Mineralized zones turning brown in different depths and regular strip-like pattern, after decorrelation analysis and enhancement of the B3(R)B2(G)B1(B) band combination of IKONOS; clear and easily identifiable boundary; iron ore bodies of dark reddish brown tone, narrow strip, lenticular, beaded, and intermittent extension, and whose boundary with the mineralization zone is clear; carbonate rocks of light gray-white-yellowish-brown tones with striped patterns Remote sensing anomaly information Iron alteration anomaly Geophysical and geochemical anomaly information Geophysical information showing magnetic anomalies and geochemical information showing comprehensive anomalies of elements, such as Pb, Zn, Cu and Au tract the information of ore-controlling factors and mineralization and to provide an interpretation of remote sensing for mineral resources. Furthermore, ASTER, WorldView-2, and IKONOS data were utilized to carry out alteration anomaly information extraction. Given appropriate amounts of field sampling and a verification test, large iron ore belts were discovered in the study areas of Taxkorgan and Heiqia.
(2) PCA was used for the extraction of iron-staining anomalies from the highfraction data (WorldView-2, IKONOS). The results indicated that the high-fraction data contained abundant information, high spatial resolution, good effect, and high accuracy of remote sensing anomalies extraction. Thus, it used to locate iron ore bodies worthy of extraction and has further applications.
(3) High-resolution remote sensing technology is suitable in iron ore geological and mineral remote sensing surveys and is similarly advantageous in both high-ground resolution of optical characteristics and particular spectral recognition capability. Moreover, it is effective not only for information extraction from a large area, but also for the recognition of local outcrops of mineralization. Therefore, it shouble be extended to other regions. The rapidly developing remote sensing technology, coupled with the successful launch of domestic high-resolution and hyperspectral satellites, is expected to provide a prospecting effect and wider application prospects for the improved exploration of mineral resources.

ACKNOWLEDGEMENT
This paper is supported by China's Geological Survey Project "Remote sensing geological survey of mineral resources in the West Kunlun Metallogenic Belt" (No. 1212011120888), Remote sensing interpretation and application of resources and environment satellite in "one belt and one road" (No.DD20160117), and National key basic research program of "973" project "Sedimentary environment and large-scale mineralization in continental basin of giant sandstone uranium metallogenic belt in North of China" (2015CB453000). Thanks to the anonymous reviewers and editors of TEA FLUKSI, LIDIJA GALOVIĆ and JULIE ROBSON for their suggestions on the paper. Plate 1. Field photography of the iron ore belt in Taxkorgan. a jarosite on the surface of the Zankan iron ore body; b Zankan iron ore body outcrop; c Laobing iron ore body outcrop and d Yerite Valley iron ore body outcrop.