Lithological and Alteration Mapping at the Um El-Rus Area, Central Eastern Desert, Egypt, Using Remote Sensing Techniques

The present study aims at discriminating the rock units and mapping the alteration zones from their host rocks, as promised targets for mineral exploration in Um El-Rus area using the widely considered ASTER data. Several remote sensing techniques were integrated for the lithological discrimination and alteration zone delineation. The approach methodology was able to map the various rock units and define the contacts between them precisely. Moreover, the spatial abundances of the alteration minerals (e.g., alunite, iron oxides, chlorite, kaolinite, illite, montmorillonite, and sericite) were mapped as an indication of the occurrence of alteration zones. More addition, the lineaments in the study area were extracted automatically using the PCI Geomatics software. Areas with considerable lineament density may indicate high potentiality for mineralization as these lineaments may represent faults and fractures that are the main channel ways for hydrothermal solutions to migrate through. This study yielded a modified geological map with enhanced lithological boundaries depending on the remote sensing data. Five localities were delineated and regarded as alteration zones with high potentiality for mineralization. A field visit was conducted to the study area through which the obtained results were validated. Field observations maintained the high accuracy of the obtained enhanced lithological map. During the field study, several alteration indicators were observed, including, quartz veins/veinlets, kaolinitization, chloritization, sericitization, and carbonatization types of alteration. The adopted alteration mapping methodology enabled the delineation of the already existing Um El-Rus gold mine; this validates the results as well.

The aim of the present study is to differentiate and map the lithological units and the alteration zones at Um EL-Rus area using advanced remote sensing techniques and field validation.

Geological Setting
The study area comprises different main rock units of ultramafic, island arc assemblage, and late to post-orogenic rocks (Fig. 2). The ultramafic rocks are represented by ophiolitic dismembers of serpentinites and talc carbonates, found thrusting over the metavolcanics in the southwestern part of the study area. The serpentinites are mostly altered to talc carbonate and iron carbonate silica (listwanite ridges) (Kafafy & Abd El Dayem, 1992). The island arc assemblages are represented by metavolcanics and the associated metavolcaniclastics, which are mostly acidic to intermediate varieties, including meta-dacite and metaandesites (Akaad & Noweir, 1996). They are highly jointed, sheared, and traversed by some acidic and basic dykes and quartz veins (Abu El-Leil et al., 2008). Um Rus gabbros occur as moderate hills forming a semicircular body (&34 km 2 ) surrounding the tonalite granitic pluton of Gabal Um El-Rus, which directly intrude these gabbros by sharp intrusive, contacts (Tolba, 2007). EI-Mezayen & Abu El Leil (1996) differentiated four varieties of gabbroic rocks in the study area including, melanocratic olivine gabbro, gabbro norite, troctolite, and anorthosite.

Data Used in This Study
For the present study, several datasets were used, including (1) ASTER level-1B image (visible near infra-red (VNIR), short wave infra-red (SWIR), and (thermal infra-red (TIR) bands) acquired on April 4th, 2004, (2) recently updated geological map of the study area (after Abu El-Leil et al., 2008), and (3) Field data for validation of the remote sensing results.

Methodology
Several techniques were carried out on the ASTER image data, including the preprocessing required for correcting the geometric and atmospheric errors encountered in the raw ASTER level1-B data, processing techniques for mapping the lithological units and delineating the potentially mineralized alteration zones, and other techniques for automatically extracting the lineaments in the study area (Fig. 3).

Image Preprocessing
Optical satellite images generally contain radiometric and atmospheric errors caused by various effects, including the interaction with the atmosphere, the topography, and overheating of the instrument itself. Several preprocessing steps were performed to retrieve the surface reflectance data without no interference from the atmospheric interactions; the carried out preprocessing steps include; Fig. 1 Location map of the study area; the left-hand image is a base map of Egypt, and the right-hand image is a false-color composite of ASTER bands 7, 3, and 1 in RGB (1) Cross-talk correction to overcome the signal scattering problem in the SWIR subsystem sensors. (2) Radiometric calibration and atmospheric correction using FLAASH models of the ENVI 5.1 software. Surface reflectance data of the two ASTER satellite images were obtained after removing the atmospheric effects.
(3) Spatial resolution merging of the SWIR 30 m bands with the VNIR 15 m bands to obtain SWIR bands with 15 m spatial resolution. (4) Layer stacking of the VNIR and SWIR nine 15 m bands in one dataset. (5) Cropping the area of interest. Image Processing for Geological Mapping Several image processing techniques have been approached for discriminating the various rock types. These are; false color composites (FCC), color ratio composites (CRC), principal component analysis (PCA), support vector machine (SVM), and maximum likelihood classification (MLC). Several authors worldwide used these techniques successfully for lithologic mapping on multispectral satellite image data (e.g., Boardman, 1998;Salem et al., 2013;Zadeh et al., 2014;Ghoneim et al., 2018).
(1) False Color Composite (FCC) To display a colored image, only three bands are required in a band combination, where ASTER VNIR-SWIR data are nine bands so, it has a wide range of possible triplet band combinations (84 possible combinations); To select the best triplet band combinations for visualizing a false color composite (FCC) images we used the optimum index factor (OIF) method that is a statistical-based algorithm used to determine the optimum combination of three bands in a satellite image from-which one can create a color composite image (Chavez et al., 1982). Ilwes software was used to obtain the optimum combinations automatically ( Table 1) that enhance the desired target and includes the most informative bands with less information redundancy (Drury, 1993).
(2) Color Ratio Composite (CRC) Band rationing is an effective method for distinguishing among rock types because the rationing process removes first-order brightness effects due to topographic slopes and enhances subtle color variations between materials (Abrams et al., 1983).
(3) Principal Component Analysis (PCA) The principal component analysis (PCA) technique was used to produce uncorrelated output bands, segregate noise components, and reduce the dimensionality of data sets. PCA technique was applied on the nine bands of the VNIR-SWIR subsystems of the ASTER data. This technique gained wide acceptance for lithological mapping purposes (e.g., Ghoneim et al., 2018;Ready & Wintz, 1973).
(4) Supervised Classification Two supervised classification techniques were used: the maximum likelihood classification (MLC) and the support vector machine (SVM). The accuracy of the mapped classes using the two techniques was evaluated in terms of overall accuracy and the Kappa coefficient. The overall accuracy is the average of the accuracy of the individual classes. The Kappa coefficient is a measure of how the classification results compare to values assigned by chance. It can take values from zero to one. If the Kappa coefficient equals zero, this means very low accuracy of the classification results. If the Kappa coefficient equals to one, then the classified image and the ground truth image are identical. So, the higher the Kappa coefficient, the more accurate the classification is (Mondal et al., 2012;Otukei & Blaschke, 2010).

(a) Maximum Likelihood Classification (MLC)
The MLC technique is used for determining and discriminating various classes depending on the maximum similarity to user-provided validated training areas for each single rock type (Scott & Symons, 1971). It is widely accepted for lithological mapping from ASTER data (e.g., Zhang et al., 2007;Sun et al., 2013;Zamzam, 2017;Fatima et al., 2017).
The support vector machine (SVM) is a machinelearning algorithm used for classification. Several training sample sites were provided for each of the different rock unit classes and the algorithm was allowed to map the spatial extent of the similar varieties to the provided ones. The theory of the SVM was initially proposed by Chervonenkis and Vapnik (1971) and later discussed in detail by Vapnik (1999). The same regions of interest used in the MLC technique were used for this technique as well. Image Processing for Alteration Mapping ASTER image processing approaches including; the Crosta PC method, constrained energy minimization (CEM), and spectral ratio indices techniques were used to delineate the alteration zones in the study area; (1) Crosta Principal Component (Crosta PC) Crosta and Moore (1989); studied the spectral specifications of the alteration minerals from the United States Geological Survey (USGS) spectral library (Fig. 4). They defined the best four bands for each alteration mineral, with two bands representing the maximum reflectance and the other two bands representing the absorption features (Table. 1). Crosta procedure was validated by several authors around the world (e.g., Crosta & Moore, 1989;Loughlin, 1991;Ruiz-Armenta & Prol-Ledesma, 1998;Tangestani & Moore, 2002;Carranza & Hale, 2002;Crósta, 2003).
Per each mapped mineral, a fraction image is obtained as a grayscale image in which the brightest (white) color represents a high match and the matching possibility is reduced gradually to zero in the pixels with black color. Each of these images represents the spatial distribution of the possible match of each pixel to the specific alteration mineral. The density slicing technique was used to delineate the pixels with an optimum match to the reference spectral signature of each of the provided minerals through a thresholding process; in which the threshold for each mineral abundance was estimated as a value equal to three standard deviations added to the average of the pixels values calculated using the ''compute statistics'' tool in ENVI software. The threshold values for each of the mapped minerals are illustrated in the corresponding image.
(2) CEM CEM technique maximizes the target spectrum response and suppresses the other features response considering all other features than the target as the unknown background. It is regarded as one of the best and most effective alteration mapping techniques (e.g., Mohey, 2014, Abou El-Magd et al., 2015Ghoneim et al., 2022;Gabr et al., 2022). The spectral signature of the same eight minerals used in the Crosta method were used in this technique to determine their spatial distribution over the study area and to confirm the harmony with the Crosta method.

Lineament Extraction and Integration with Alteration Data
In this study, the first principal component band (PC1) was used effectively to extract the lineaments automatically using PCI the Geomatics software (Burdick & Speirer, 1980;Karnieli et al., 1996;Hung et al., 2002Hung et al., , 2003Gupta, 2003;Ghoneim et al., 2022;Gabr et al., 2022). The used PC1 band was selected for this process as it enhances the response of the structural elements and enhances the edge contrast and the topographic variations. The extracted Fig. 4 The USGS library spectra of the studied minerals superimposed on ASTER data band intervals in the VNIR-SWIR region  Abrams et al., 1983) lineaments were then converted into lineament density contours to be integrated with the obtained alteration data.

Lithological Mapping Results
(1) False Color Composite (FCC) The OIF method yielded the six band triplets with the highest information content without redundancy (Table. 2).
By the visual inspection of the highest OIF ranked FCC image of bands 1, 5, and 8 in RGB, respectively (OIF rank = 1), it is obtained that this band combination provided a good general view of the various exposed lithological units in the study area (Fig. 5).
(2) Color Ratio Composite (CRC) On the resulted first ratio composite image (Fig. 6); the tonalite rocks of Gabal Um El-Rus along with the wadi deposits that are mainly composed of granitic fragments derived from Um El-Rus (in the middle of the study area)  and Kadabora (northwest of the study area) granitic plutons are shown in blue color. The gabbroic varieties are marked by green color, while the contact zone between tonalite and gabbro rocks is illustrated as dark green color. Serpentinite rocks are shown in yellow, metavolcanics and metavolcaniclastics are in yellowish green color mainly in the southern parts of the study area, and the altered granitic rocks mainly altered tonalities are illustrated by magenta color and found scattered in and around the highly deformed contact zone around Gabal Um El-Rus. On the resulted second ratio composite image (Fig. 7), the tonalite rocks of Gabal Um El-Rus along with the wadi deposits are shown in green color. The gabbroic rocks are marked by brown color, while the contact zone is illustrated as blue color. Serpentinite rocks are shown in red, metavolcanics and metavolcaniclastics are in reddish pink color, and the altered tonalities are illustrated by bright color.
Both the obtained ratio composites were found effective in discriminating all the rock units in the study area.
(3) Principal Component Analysis (PCA) From the analysis of the eigenvector matrix (Table. 3), PC1, 2, 3, and 4 contain 99.05% of the information content of the data while all the other PC bands contain only 0.95%. PCA color composite PC3, PC2, and PC1 (Fig. 8) are found the most informative with a maximum variance; hence it was used for delineating the contacts between the rock units. All the mapped units are shown with their respective symbols plotted in Fig. 8.

(4) Supervised Classification (5) Maximum Likelihood Classification (MLC)
This technique enabled grouping of all pixels with the same characteristics as a class. Consequently, nine classes were obtained representing the exposed rock units in the study area (Fig. 9). The accuracy results of this technique revealed precision with an overall accuracy of 94.1187% and Kappa coefficient of 0.9316 (Table. 4).

(b) Support Vector Machine (SVM)
Similar to the MLC, SVM technique showed perfect lithological discrimination between the rock units in the study area ( Fig. 10) with an overall accuracy of 90.814% and a Kappa coefficient of 0.8926 (Table. 5).
High accuracy and Kappa values obtained using both the supervised classification techniques (MLC and SVM) indicated the suitability of such classification technique on ASTER data for areas with similar characteristics as that of the study area. Integrated interpretation of the results of FCC, CRC, PCA, SVM, and MLC techniques showed significant identification and delineation of the exposed rock units and enhanced contacts between the rocks in the study area (Fig. 11).

(a) Crosta PC
The results of the Crosta technique are shown in Fig. 12. It is found that all the alteration minerals are mainly associated with the areas of tonalite, altered tonalite, and in the highly deformed contact zone between the gabbro and tonalite rocks of Gabal Um El-Rus.

(b) CEM
By applying this method, the most dominant and concentrated locations for these minerals were obtained (Fig. 13). Similar to the Crosta technique; the detected spatial abundance of the used alteration minerals was encountered in the tonalites, altered tonalites, and the highly deformed contact zone surrounding the tonalite pluton.

(iii) Spectral Ratio Indices
As a result of applying this method, the spatial distribution of the iron oxides, kaolinite, calcite, and the OHbearing minerals was mapped (Fig. 14). It is found that the spatial distribution of kaolinite, OH-bearing minerals, and  the iron oxides are highly comparable to the areas of altered tonalite, and in the contact zone area, while, the distribution of calcite was found mainly associated with the areas of serpentinites; this sounds logic because calcite is a carbonate mineral that is an alteration product of serpentinite rocks. By comparing the results of the three techniques (Crosta, CEM, and spectral ratio indices) used for alteration mapping, it is obtained that the locations of alterations delineated by all the three techniques are more or less the same and this supports the results one by another.

Lineaments Extraction and Integration with Alteration Data
A lineaments map was obtained showing lineaments distribution in the study area (Fig. 15). Moreover, a lineament density contour map was yielded showing the high and low-density lineaments localities (Fig. 15). Where the high lineaments localities refer to the possible hydrothermal activities and formation of mineral bearing alteration zones as valid targets for mineral exploration.
The high-density lineaments localities and the alteration zones obtained from the CEM approach are mostly matching; this is approved during the fieldwork as promised sites for mineral exploration and possible mineralization. The presence of Um El-Rus gold mine in an altered  area appreciates the mapped alteration zones to be valid localities for mineral occurrences and exploration. Five localities with similar conditions and alterations were suggested to be potential for mineral prospecting (Fig. 16).

Validation of the Results
A field study was conducted in the study area during which several observation points (Fig. 17) were recorded along the different exposed rock units in the study area. Several rock units were identified including; the tonalite, gabbro, the highly deformed contact zone, the metavolcanics, and the serpentinites along with their associated talc carbonates and quartz iron carbonate ridges (listwanite) (Fig. 18). Moreover, several alteration marks were recorded including; chloritization, carbonatization, quartz veins/ veinlets, sericitization, and kaolinitization (Fig. 19). All the recorded observations regarding the lithological rock units and the recorded alteration marks came coincident with the remote sensing results that give high confidence of the obtained results.

Discussion
The present study demonstrates the capabilities of remote sensing data in mapping the lithological units and the alteration zones in Um El-Rus area in the central eastern desert of Egypt. Several techniques were applied to achieve the objective of the study, here given a discussion of the results of the adopted techniques;

Lithological Mapping
FCC one can think about it as a simple technique; however, it requires intense statistical work through the OIF method to determine the most informative triplet band combinations in order not to choose any other bands instead. Based on the OIF calculations, the triplet band combination of b1, b5, and b8 (Fig. 5) was found to have a higher rank. This combination gave a general overview of the exposed rock units in the study area. CRC Two color ratio composites were applied; equivalent to Abram's ratio (1983) and Sultan's ratio (1986). These two color ratio composites are the most widely used ratio composites for lithological discrimination in the Arabian-Nubian shield (e.g., Ghoneim et al., 2018;Salem et al., 2013Salem et al., , 2016. These CRC's (Figs. 6, 7) proved high efficiency in discriminating the various rock units in the study area including; older granites, different gabbro varieties, volcanics, and volcaniclastic rocks, and the serpentinites that area efficiently separated from Sultan's ratio (Fig. 7) as red color.
PCA is a widely accepted technique for lithological mapping. Its idea depends on removing and isolating the redundant data from all the original spectral bands into the last PC bands and maintaining the data with maximum variance in the first PC bands. Consequently, the PC band combination of PC1, PC2, and PC3 (Fig. 8) was used and found to have more than 99% of the information content of the entire original. This combination provided perfect discrimination of all the rock units in the study area. Supervised classification the adopted two techniques (MLC and SVM) (Figs. 9, 10) both depended on a userprovided field validated regions of interest. Accuracy assessment was made and the Kappa values indicate high accuracy of both the techniques with 94.12% for MLC and 90.81% for SVM.
The integration of all these techniques provided a very accurate geological map in which all the exposed rock units are mapped accurately.

Alteration Mapping
Crosta PC this technique depends mainly on specifying the spectral characteristics of specific alteration minerals in terms of high and low reflectance bands for each mineral, then, mapping the abundance of these minerals through the PCA transformation of the high reflectance two bands and the low reflectance two bands for each mineral. Because of applying this technique, eight mineral abundance maps were obtained for chlorite, kaolinite, sericite, hematite, illite, alunite, montmorillonite, and epidote being the most common mineral sets in the alteration zones (Fig. 12). The mapped alteration minerals by the means of this technique was found associated with the areas of the contact zone and the areas dominated by altered older granites (tonalite). comparable to the corresponding Crosta maps giving alteration minerals almost in the same locations (Fig. 13).
Spectral ratio indices this technique depends mainly on mathematical calculations derived from the characteristic spectral bands for specific alteration minerals. The OH-bearing minerals, kaolinite, iron oxides, and calcite minerals were mapped using different spectral indices developed by Ninomyia (2003) representing the main alteration indicators (Fig. 14). All the mapped minerals except calcite are in complete harmony with the abundance maps of both Crosta and CEM technique. Calcite abundance came associated with the areas of serpentinites and their related talc and quartz carbonates. Lineaments mapping the lineaments being possible channel ways for hydrothermal solutions were mapped automatically using the first PC band in which all the structure variations are highly enhanced (Fig. 15).
Overall, the adopted alteration mapping techniques provided good clues about the potential locations of alteration zone abundance. By integrating the results of the alteration zones with the lineament density map, five localities were delineated as the potential for mineralization (Fig. 16). The field work (Fig. 19) proved the occurrence of the alteration indicators in the same localities delineated by remote sensing. Investigations that are more detailed are recommended for quantitative evaluation of the delineated areas.

Conclusion
Several remote sensing techniques were used for lithological and alteration mapping in Um El-Rus area, Central Eastern Desert, Egypt. The adopted techniques for lithological mapping included; FCC, CRC, PCA, SVM, and MLC. These techniques showed high efficiency in discriminating the various lithological units in the study area. Consequently, an enhanced remote sensing-based geological map of the study area was obtained. Moreover, other techniques were used for alteration mapping including; Crosta PC, Spectral ratio indices, and CEM. These techniques were effective in mapping the mineral-bearing alteration zones from their host rocks and provided good clues about the distribution, and the dominant minerals characterizing the alteration zones of the study area. The obtained alterations were integrated with the lineaments density map for delineating the highest potential areas for ore deposits. As a result of this integration, five potential localities were delineated including the locality of Um El- Rus mine. The localization of the alteration zones extracted by the adopted remote sensing techniques is compatible with the fieldwork verification, where the Um El-Rus gold mine was mapped within an alteration zone hosted in the contact zone between the tonalite and gabbro rocks.
Funding Open access funding provided by The Science, Technology & Innovation Funding Authority (STDF) in cooperation with The Egyptian Knowledge Bank (EKB).

Declarations
Conflicts of interest The authors declared that they have no conflict of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons. org/licenses/by/4.0/.