Image processing for ASTER remote sensing data to map hydrothermal alteration zones in East Kazakhstan

Full Text

Introduction Remote sensing is a widely used tool in mineral exploration, as it has replaced the physical approach to discovering deposits. The physical approach required spending much time, effort, and money looking for geological studies. Since 1920, the use of aerial photographic interpretation in the field of Earth sciences has become a fast and effective tool for the exploration of natural resources [1]. Therefore, the launch of Landsat-1 in 1972 and the continued development of new sensors have increased the spatial-temporal-spectral resolution of Earth observation data [2]. This made the digital imagery of the electromagne- tic spectrum available for interpretation and use in mineral explorations in a short time. Remote sensing is a comprehensive method that enables scientists to identify an object, by collecting all needed information about it. The interpretation of satellite images requires applying two basic paradigms, namely, data-driven, and knowledge-driven models. Both models are the dominant paradigms for spatio-temporal modeling and spatio-temporal decision-making [2]. The main idea behind this is that everything on Earth has its unique spectral signature, which provides the ability to identify features or abstract information about what is displayed on Earth's surface. Spectral signature is the energy reflected from features on earth and stored as bands. Mostly, bands will capture the visible, NIR, and SWIR regions which tend to contain more useful information about the earth's surface. Minerals have distinctive spectral reflectance patterns at visible wavelengths and especially at reflected IR wavelengths [3]. Multispectral image data has been used for mapping hydrothermal alteration zones. Since 2000, and after the launch of Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) in 1999, it has become more applicable for mineralogical and lithological studies to be run using the multispectral images provided by ASTER on a wide range of samples. ASTER covers a wide spectral region of the electromagnetic spectrum, from visible near-infrared (VNIR) to thermal infrared (TIR) [4]. The spectral range in ASTER consists of three main subsystems with different spatial resolutions and wavelengths. the subsystems are Visible near-infrared (VNIR), shortwave infrared (SWIR), and thermal infrared (TIR) [4]. SWIR spectral bands were designed to identify reflected radiation to distinguish Al-OH, Fe, Mg-OH, Si-O-H, and CO3 absorption features [5]. Therefore, scientists could identify specific hydrothermal alteration minerals like alunite, kaolinite, calcite, dolomite, chlorite, talc, and muscovite, as well as mineral groups. Hence, the SWIR properties make it suitable for mapping alteration zones in mineral exploration [6]. Applying statistical methods to the produced maps from remote sensing has been helping scientists with different approaches to analyzing point data but also filtering the data (removing missing pixels or filling the voids). Also, combining statistical methods with GIS layers obtained from remote sensing helps improve the generation of DEMs, simulate them, and optimize spatial sampling, the selection of spatial resolution for image data, and the selection of support size for ground data. Geostatistics is a subset of statistical methods used to analyze and interpret geographical data. Geostatistics enables mapping environmental variables using different techniques [7]. Advanced Spaceborne Thermal Emission and Reflection Radiometer give us the potential to map mineralogical alteration zones at low cost with high accuracy. Mapping these zones is important to distinguish high-potential areas of economical minerali- zation such as epithermal gold and hydrothermal por- phyry copper deposits. Hydrothermal porphyry depo- sits consist of alteration mineral zones (Figure 1) [8]. These zones (phyllic, argillic, and propylitic) contain minerals that can be distinguished from each other using SWIR from ASTER data [9-11]. a b Figure 1. Hydrothermal alteration zones are associated with porphyry copper deposits: a - a schematic cross-section of hydrothermal alteration mineral zones, which consist of propylitic, phyllic, argillic, and potassic alteration zone; b - a schematic cross-section of ores associated with each alteration zone 1. Geological settings Hydrothermal deposits of porphyry copper are usually formed in areas of magmatic rock development. The deposits are usually associated with calc-alkaline plutons. Each hydrothermal copper-porphyry deposit is characterized by hydrothermal alteration mineral zones. The area under consideration is characterized by low vegetation, which could mask part of the data, causing problems in image processing. It is located within the Zhilanda-Aygyz subzone of the Predchingiz zone. The study is in the Eastern Pribalkhash region and represents a fragment of the Kazakhstan Copper Belt. To locate areas with copper, molybdenum, lead, and zinc anomalies, as well as to locate pink rhyolite porphyry in the central zone, where contact changes are apparent and suggest the presence of a copper-molybdenum porphyry mineralization process, Viktor V. Diakonov and Alexander E. Kotelnikov, 2016 conducted geological and geochemical analysis in the study area in 2016. They also linked these data to geophysical anomalies to further their understanding of the research area [12; 13]. 2. Multispectral properties of hydrothermal alteration zone by ASTER data The Advanced Spaceborne Thermal Emission and Reflection Radiometer is a multispectral remote sensing instrument that is a highly spatial, spectral, and radiometric instrument. ASTER is a cooperative effort between the Japanese Ministry of Economic Trade and Industry (METI) and the National Aeronautics and Space Administration (NASA). It was launched in December 1999. ASTER consists of three main subsystems with a total of 14 bands that provide observation in these three different spectral regions of the electromagnetic spectrum: visible near-infrared (VNIR), shortwave infrared (SWIR), and thermal infrared (TIR), which contain 3, 6, and 5 bands, respectively, with different ranges of wavelength. In the VNIR subsystem, bands’ ranges differ (from 0.52 to 0.86 μm) with a spatial resolution of 15 m. While the SWIR subsystem's bands' ranges differ (from 1.6 to 2.43 μm) with a spatial resolution of 30 m, TIR, the last subsystem, has bands’ ranges (from 0.1.6 to 2.43 μm) with a spatial resolution of 90 m [6]. ASTER provides data that can be useful in a wide range of scientific investigations and applications, including (a) geology studies, (b) climatology stu- dies, (c) volcano monitoring, (d) hydrothermal and water resource applications, and in other different fields of science [14]. It has significant properties widely applied in geology: (1) it allows the dis- crimination and identification of hydrothermal alteration minerals in the SWIR electromagnetic region; (2) it gives the ability to identify the vegetation and iron oxide minerals on the surface and map carbonates and silicates [15; 16]. ASTER generates two data products: Level-1A, which is raw image data, and Level-1B, which is a data product gene- rated from Level-1A by applying radiometric and geometric correction coefficients [17]. 3. Image analysis Different image-processing techniques can be used on ASTER data, such as principal component analysis (PCA), band ratio, and minimum noise fraction (MNF) [18; 19]. The alteration zone as described previously is separated into three main parts; each of these zones is distinguished by speci- fic minerals that work as indicators as they all have different spectra (Figure 2). Изображение выглядит как диаграмма Автоматически созданное описание Figure 2. Laboratory spectra of common hydrothermal alteration minerals [18] ASTER minerals' spectra are important indicators for different hydrothermal alteration zones, as summarized in Figure 2, and can indicate the zone as follows: (1) muscovite as an indicator for phyllic alteration zones with a 2.20 μm absorption feature shown in the 6th ASTER band; (2) kaolinite and alunite as indicators for argillic alteration zones with a 2.20 and 2.17 μm, respectively, absorption feature shown in the 5th ASTER band; (3) epidote, chlorite, and calcite are associated with propylitic alteration zones with 2.31-2.33 μm absorption features shown in the 8th ASTER band. Therefore, these unique absorption features for minerals led to many useful approaches for mapping and discriminating hydrothermal alteration zones [3]. 4. Principal component analysis PCA used the principal component transformation technique to reduce the dimensionality of the correlated multispectral data. The PCA method is widely used to map alteration zones [18]. The PCA technique aims to extract specific spectral responses, as in the case of hydrothermal alteration minerals. The likelihood of having a specific spectral contrast increases as the number of input channels decreases. In this study, the bands that have been used are those that have the potential to show more common spectral features of the alteration mineral. To confirm the occurrence of minerals, a PCA technique was applied to find the relationship between the spectral responses of target minerals or rocks. The relationship is used to determine which of the PCs contain the spectral information due to the minerals and whether the pixels have high or low values related to the presence of the target mi- neral in that pixel or the absence of it [20]. Applying PCA to map hydrothermal alteration zones has been widely used as an advanced tool for statistical data reduction and satellite image proces- sing. As it was recommended in the articles [21; 22] to map alteration minerals to indicate different alteration zones, for example, using a subset of ASTER bands (1, 4, 6, and 7) to map Kaolinite Also, band subsets (1, 3, 5, and 7) and (1, 3, 5, and 6) for mapping Alunite and Illite, respectively. Figure 3. A curve showing that the first few bands contain most of the data, and the signal decreases with increasing noise towards the the curve tail (the graph is made using ENVI 5.3) The eigenvalues of the 14 ASTER bands show that PC1, PC2, and PC3 have over 97% of the spectral information displayed in Figure 3; the rest of the low-order PCs have less than 3%; they usually contain low signal-to-noise ratios. PCs that contain more than 97% are widely used for lithological mapping [20]. 5. Results and discussion By applying the principal component analysis (PCA) to ASTER bands, we can highlight different areas of the hydrothermal alteration zones, as each zone has its rocks with specific minerals. Different minerals can be identified, like kaolinite and alunite, which show an absorption behavior in band 6 due to Al-OH; these two also show a reflection behavior that corresponds with the argillic zone [23]. Illite, smectite, and sericite minerals give an ab- sorption behavior in band 6 and a reflection behavior in band 5, which correspond with the phyllic zone. The propylite zone is shown as a response to the re- flecting of chlorite, epidote, and calcite, which shows absorption behavior in the 8th band and reflecting behavior in the 5th band [23]. PCA is a statistical tool used to extract specific spectral features. In 1989, Crosta and Moore deve- loped a PCA technique using Landsat TM to map oxide/hydroxide iron minerals related to sulfide ore bodies in the granite-greenstone belt. PCA is calculated by forming a relationship between the spectral responses of minerals under consideration and va- lues extracted from the eigenvector matrix. Using a few selected bands to avoid mapping certain materials like vegetation and applying PCA to them helps extract information about targeted materials (hydrothermal alteration). This procedure is called the “Crosta technique” and has been widely used for mineral exploration due to its ease [24]. Choosing subsets of ASTER bands proposed by Loughlin (1991) according to spectral features rela- ted to hydrothermal alteration minerals in VNIR and SWIR and applying PCA to each chosen subset gives information about the target mineral. To identify which PC has the target information, we choose the PC that has the highest eigenvector value difference among bands. In the application of PCA to ASTER bands, subsets 1, 4, 6, and 7 were used to successfully map the argillic alteration zone as bright pixels into the PC-3 image shown in Figure 4, a. This is evident by the low negative contribution of band 7 and the high positive contribution of band 4 (Table 1). To map the phyllic alteration zone, we used subsets 1, 3, 5, and 6. The phyllic zone shows a dark color pixel value in the PC-4 image (Figure 4, b), due to the high positive contribution of band 5 and the low negative contribution of band 6 (Table 2). Implementation of PCA in ASTER bands 1, 3, 5, and 7 suggests information on the propylitic alteration zone. The spatial map shows the alteration zone as bright pixels in PC-3 (Figure 4, c), due to the high positive contribution of band 3 and the low negative contribution of band 5 (Table 3). Table 1 Eigenvectors values for PC bands 1, 4, 6, and 7 Input layer Eigenvectors PC1 PC2 PC3 PC4 Band 1 0.69778 -0.7139 0.05644 0.01627 Band 4 0.35619 0.41171 0.83287 -0.09976 Band 6 0.44468 0.42624 -0.31436 0.72233 Band 7 0.43416 0.37303 -0.45202 -0.68413 Table 2 Eigenvectors values for PC bands 1, 3, 5, and 6 Input layer Eigenvectors PC1 PC2 PC3 PC4 Band 1 0.7678 -0.27059 -0.58071 -0.00636 Band 3 0.42198 -0.4684 0.77623 -0.00297 Band 5 0.3278 0.56166 0.16356 0.74184 Band 6 0.3535 0.62603 0.18303 -0.67054 Table 3 Eigenvectors values for PC bands 1, 3, 5 and 7 Input Layer Eigenvectors PC1 PC2 PC3 PC4 band 1 0.77196 -0.23351 -0.59107 0.01353 band 3 0.42365 -0.50435 0.7524 -0.00656 band 5 0.32243 0.57527 0.21036 0.7217 band 7 0.34732 0.60014 0.20068 -0.69204 Map Description automatically generated Изображение выглядит как карта Автоматически созданное описание a b Map Description automatically generated Map Description automatically generated c d Figure 4. Applying the Crosta technique indicates places that show the presence or absence (the map is made using QGIS 3.18.3): a - Kaolinite as an indicator of Argillic zone; b - Illite indicates Phyllic’s zone; c - Alunite indicates Propylitic’s zone; d - false-color composite image of PC 3, PC 4, PC 3 image Map Description automatically generated Figure 5. Predictions of hydrothermal alteration zones assigned as Argillic, Phyllic, and Propylitic on a geological base map [1; 2] Conclusion Analysis of ASTER satellite images using PCA and applying the Crosta technique gives promising findings and an understanding of the area under consideration. The satellite image interpretation and integration with the geological map of the area show that PCA is an applicable technique to be used in our study area to map hydrothermal alteration zones. Also, the result shows that our study area is suitable for this kind of image pro- cessing as it is characterized by a low vegetation mask.

About the authors

Hamza A. Mahmoud

RUDN University

Email: 1032205919@rudn.ru
ORCID iD: 0000-0002-2946-7144
SPIN-code: 1929-6130

master's student, Department of Subsoil and Petroleum Engineering, Academy of Engineering

6 Miklukho-Maklaya St, Moscow, 117198, Russian Federation

Elena V. Karelina

RUDN University

Author for correspondence.
Email: karelina-ev@rudn.ru
ORCID iD: 0000-0003-4691-4855
SPIN-code: 4919-8300
Scopus Author ID: 57215413670

PhD of Geology, Associate Professor of the Department of Mineral Developing and Oil & Gas Business, Academy of Engineering

6 Miklukho-Maklaya St, Moscow, 117198, Russian Federation

Vladimir E. Markov

RUDN University

Email: markov-ve@rudn.ru
ORCID iD: 0000-0001-6594-0763
SPIN-code: 5882-5663

senior lecturer, Department of Mineral Developing and Oil & Gas Business, Academy of Engineering

6 Miklukho-Maklaya St, Moscow, 117198, Russian Federation

Viktor V. Diakonov

Sergo Ordzhonikidze Russian State University for Geological Prospecting

Email: mdf.rudn@mail.ru
ORCID iD: 0000-0002-9153-6489
SPIN-code: 8780-8588
Scopus Author ID: 57200068947

Doctor of Science in Geology, Professor

23 Miklukho-Maklaya St, Moscow, 117997, Russian Federation

Ilya V. Vikentyev

Institute of Geology of Ore Deposits, Petrography, Mineralogy, and Geochemistry of the Russian Academy of Sciences

Email: viken@igem.ru
ORCID iD: 0000-0001-9133-7562
SPIN-code: 2456-3030
Scopus Author ID: 6506542626

Doctor of Geology, leading researcher, Laboratory of Ore Deposits

35 Staromonetnyi Pereulok, Moscow, 119017, Russian Federation

References