HyMap imagery for copper and manganese prospecting in the east of Ameln valley shear zone (Kerdous inlier, western Anti-Atlas, Morocco)

ABSTRACT Eastern Kerdous demonstrates a polymetallic hydrothermal activity due its association with a paleorelief area. This study focuses on generating a Mineral Potential Map based on Fuzzy Logic Modelling and HyMap-Based input layers. The Relative Absorption Band Depth, Directed Principal Component Analysis, Line Density, and Mixture Tuned Matched Filtering techniques were implemented to perform 20 thematic layers. As a result, three new prospective zones were identified. Fieldwork, petrography, and XRD analysis verified the remote sensing results. Finally, this investigation highlights the significant potential of HyMap imagery for mineral exploration in inaccessible and remote parts of the western Anti-Atlas belt.

Mining companies have shown a great interest in the remote sensing techniques for alteration and lithological mapping since the Landsat-1 program launch in 1970s (Moon et al. 2006).Recently, Hyperion, HyMap, and Airborne Visible/IR Image Spectrometer (AVIRIS) hyperspectral sensors were used for detailed hydrothermal alteration mineral mapping (Kruse et al. 2003, Bedini 2011, 2012, Bedini and Chen 2020), which are valuable for prospecting various ore deposits (Beiranvand Pour et al. 2019, Abdelkareem and Al-Arifi 2021, Beygi et al. 2021).With the utilisation of hyperspectral imaging spectroscopy, as a result of the high number of spectral bands as well as the continuous spectrum, the characterisation of alteration minerals and zones becomes more detectable.Furthermore, tectonic linear features (fault systems) can be mapped using hyperspectral remote sensing analysis to aid in the interpretation of crustal structures of subsurface phenomena that may confine copper/gold mineralisation zones in metallogenic provinces (Beygi et al. 2021).Recently the use of remote sensing-based evidence layers is common in mineral prospectivity mapping (MPM) in GIS framework.The applicability of data-based methods in MPM is limited by the sufficiency of the number of training samples.Therefore, knowledgebased methods (e.g.Fuzzy logic, Boolean logic, analytical hierarchy process) are still widely used (Abdelkareem andAl-Arifi 2021, Nouri 2022), despite the fact that several machine learning and deep learning algorithms are developed (Mohamed Taha et al. 2022).
In the Anti-Atlasic belt, Morocco, some hydrothermal alteration mineral mapping using multispectral and/or hyperspectral datasets were accomplished.El Janati (2019) used ASTER SWIR and VNIR bands to map alteration zones hosting Cu, Ag and Au mineralisations in the Taghdout area (central Anti-Atlas) characterised by a very limited information.Hydrothermal features such as phyllic, carbonate, ferric iron oxide, and ferrous iron silicate were mapped using principal component analysis (PCA), minimum noise fraction (MNF), band ratioing, mafic index and matched filtering (MF) techniques.Adiri et al. (2020a), Adiri et al. (2020b) processed Hyperion, Landsat-8 OLI, ASTER, and Sentinel-2A data to identify new prospecting zones of porphyry copper mineralisation in Sidi Flah Bouskour inlier, Moroccan Anti-Atlas.Colour-normalised (CN) spectral sharpening was applied to map the altered zones.Malachite and propylitic alteration minerals were mapped as high potential zones associated with copper mineralisation in the study area.Atif et al. (2021) used ASTER remote sensing data to map hydrothermal alteration zones related with Ag mineralisation in the Imiter silver mine, eastern Anti-Atlas metallogenic province, Eastern Morocco.Band ratioing, Crosta, and MTMF techniques were applied to ASTER spectral bands for mapping argillic, phyllic and propylitic alteration zones associated with Ag mineralisation.
HyMap data were used for detailed mapping of hydrothermal alteration minerals associated with Cu-Au-Mo porphyry deposits, carbonatite complex and metamorphichydrothermal U-REE deposit (e.g., Bedini et al. 2009, Bedini 2011, 2012;Huo et al. 2014;Salles et al. 2017).Ameln Valley Shear Zone (AVSZ) in the eastern border of Kerdous inlier, western Anti-Atlas was selected as the case study for this investigation (Figure 1a,b).It intersects the paleorelief zone at the limit between the Neoproterozoic formation and the Adoudou basal series, which are correlated with polymetallic hydrothermal activity and contains several copper/gold, tungsten and manganese occurrences as well as stratabound copper mineralisation (Pouit 1966;Elsass 1975;Maddi et al. 2011;Bourque 2016).
The AVSZ is harsh, costly, and challenging area for traditional mineral exploration campaigns.This study develops a novel adaptive methodology to process HyMap data in the context of the paleorelief-related complex copper and manganese mineralisation in eastern Kerdous border, western Anti-Atlas.Therefore, the main aims of this investigation are (1) to process HyMap data for hydrothermal alteration mapping in the study area using Relative Absorption Band Depth (RBD) and Directed Principal Component Analysis (DPCA) and Mixture Tuned Matched Filtering (MTMF) image processing techniques as well as delineating structural lineament high densities and (2) to fuse the input HyMap-Based layers using a fuzzy logic approach for generating Mineral Potential Map (MPM) for the study area.

Geological setting of the study area
The Moroccan Anti-Atlas belt is an ENE-WSW mountain range located in south-western Morocco, it extends south of the High Atlas and north of the West African Craton (WAC) with 800 km in length (Figure 1a).The western Anti-Atlas Precambrian basement crops out in several inliers that are covered by Ediacaran to Palaeozoic cover (Figure 1a).Kerdous is the largest inlier in the western Anti-Atlas, it consists of about 6000 Km 2 .The Kerdous inlier was principally structured during the Eburnean and Pan-African orogeny (Choubert 1964, Clauer 1976, Hassenforder 1978).
The AVSZ is located in the eastern border of Kerdous inlier (Figure 1b).The geological map used in this study was digitalised from the geological map of Tafraout (1:100 000), published in 1983 by the Ministry of Energy and Mines (Figure 1b).In the study area, the Paleoproterozoic is represented by metamorphic series, include Jbel (mountain) Ouiharen gneisses, schist and micaschist.They are intersected by the Paleoproterozoic calc-alkaline granite of Tasserhirt.The Neoproterozoic is deposited in major unconformity, and it is represented by quartzites of Jbel Lkest, Tafraout Pan-African granites, Adrar Mkorn rhyolitic vulcanites and ignimbrites, Tanalt Ait-Baha series and ultimate conglomerates of the Adoudounien base (Hassenforder 1978).Following the lower Cambrian great transgression in the Anti-Atlasic belt (Benssaou and Hamoumi 2001), the Adoudounien formations were deposited by basal series, and lower series mainly formed by carbonates (limestone and dolomite) (Benssaou and Hamoumi 2001).The north part of the study area is traversed by dolerites which are of several generations, these have been set up in the Anti-Atlasic belt during Proterozoic (Gasquet et al. 2004, Thomas et al. 2004).
The east of Kerdous has known old mining activity represented by the existence of the abandoned Manganese Idikel mine, as well as it shows other Mn and W mineral indices that are reported based on the geological map of Tafraout.The Idikel mine ore body is interbedded in a principally conglomeratic series (200m thick) (Choubert and Faure-Muret 1973).The mineralised site is marked by the presence of braunite, pyrolusite, rhodonite, rhodochrosite and, barium-bearing psilomelane.Haematite, baryte, quartz and albite and different micas are also present.Moreover, baryte, oligoclase, quartz as well as Cu minerals such chalcopyrite are found in the veinlets, according to Choubert and Faure-Muret (1973).The exploitation was between 1951 and 1959, where about 100 000tons of minerals were extracted (Choubert and Faure-Muret 1973).Three hypotheses were proposed to explain the genesis of Idikel mineralisation including, syngenetic sedimentation (Bouladon and Jouravsky 1956), hydrothermal syngenetic, and epigenetic hydrothermal origin (Alous type) (Boyer et al. 1978, Leblanc andLancelot 1980).On the other hand, the study area shows a close spatial association with a paleorelief area (Pouit 1966), which can explain several mineral occurrences in the western Anti-Atlasic belt (Supplementary Figure 1).

HyMap data characteristics and preprocessing
The HyMap airborne hyperspectral imaging system was developed by Integrated Spectronic, Sydney, Australia, then, was operated by HyVista Corporation (Cocks et al. 1998).It records 128 different spectral bands covering the VNIR and SWIR electromagnetic spectrum regions from 0.45 μm to 2.5 μm (Cocks et al. 1998, Hausknecht 2005).HyMap data provide a spatial resolution of around 5 m, and an average spectral resolution of 15 nm.The main characteristics of HyMap data are summarised in the supplementary Table 1.HyMap data used in this study is provided by the National Office of Hydrocarbons and Mines (ONHYM) of Morocco.
The data contain 125 bands from 0.4522 μm to 2.493 μm and cover a total area of 10000km 2 , which were acquired during a regional airborne survey in the Anti-Atlantic belt.The data had been already geo-referenced in the UTM 29 projection using World Geodetic System WGS-84 datum.The Atmospheric and Topographic Correction Model (ATCOR4) model were applied to the data, which permits converting radiance to surface reflectance data on the basis of the MODTRAN radiative transfer code.It also can suppress the topographic effect of the illumination differences (Richter and Schläpfer 2002).Pre-removal of bad bands is required for hyperspectral datasets before data processing (Ji et al. 2019).Noisy bands and bands covering water absorption features were eliminated during this stage.Thus, only 110 bands are used for further processing (Supplementary Table 2).A vegetation mask was applied on the study area scene, then, a gap filling tool integrated in ENVI (5.3) software was used for the reconstitution of the resulting no data pixels to avoid gaps in the results.Figure 2 represents the flowchart of the methodology used to process HyMap data in this analysis.

Data processing 3.1.1.1. RBD.
The Relative Absorption Band Depth (RBD) employs the ratio of three- point formulation where the sum of the bands exhibiting the shoulders was divided by the band exhibiting the minimum of the absorption feature (Crowley et al. 1989).The RBD technique has the capability to detect targets exposed under different conditions of illumination (Lillesand et al. 2015).The RBD was successfully used to highlight spectral characteristics linked to target alteration mineral zones or mineral assemblages (argillic, phyllic and propylitic) (Pour et al. 2019, Sekandari et al. 2020).Three RBDs were developed using the VNIR and SWIR bands.Within this spectral wavelength range HyMap has the ability to map the manifestations of Fe 3+-Fe 2+ , Al-OH, and Mg-Fe-OH/CO3 minerals based on their distinctive absorption features (Clark and Rencz 1999).

Directed principal component analysis (DPCA).
To handle the problem of data redundancy in hyperspectral imagery, the PCA transform is one of the most effective methods for statistical variance reduction by creating a new set of uncorrelated bands called PC bands (Gupta et al. 2013).In the DPCA technique, as a function of targeted alteration minerals, only a specific input of bands is selected for the PCA transform.The DPCA is broadly applied for alteration zones and mineral mapping studies (Crosta andMoore 1990, Bolouki et al. 2020).

MTMF.
Mixture Tuned Matched Filtering is a sub-pixel classifier, which reveals the target object abundance and eliminates the response of background materials by comparing the reference spectra of the targeted material with the image spectra, without taking into consideration the other endmembers information in the image (Boardman 1998).The MTMF was successfully utilised for mineral alteration mapping in many metallogenic provinces (Adiri et al. 2020a, Pour and Hashim 2015, Jain and Sharma 2019, Rajan Girija and Mayappan 2019, Pour et al. 2020, Atif et al. 2021).It comprises the advantages of the two techniques of MF and MT at the same time.Three steps are normally proceed for mapping using MTMF: (i) Minimum Noise Fraction (MNF) transform calculated from apparent reflectance data; (ii) application of MF method for mineral abundance estimation; and (iii) computation of minerals abundance after identification and rejection of false positives using Mixture Tuning method (Boardman andKruse 2011, Goodarzi Mehr et al. 2013).

Lineament mapping.
Hyperspectral imagery can be suitable for mapping geological lineaments due to its high spatial resolution, where it is applied for manual and automatic extraction of lineaments (Hajaj et al. 2022).In the present study, the automatic approach was adopted for lineament extraction from HyMap data.Recently, the LINE module tool has recognised as a great interest (Enoh et al. 2021, Ghosh et al. 2021).The extraction technique comprises two essential steps.The first one consists of applying an edges detection filter (contours detection), which provides information about areas of brutal changes in the values of neighbouring pixels.Besides, the second step is lines detection where curves are extracted (Hashim et al. 2013).The automatic extraction of structural lineaments using the high resolution PC1 image derived from HyMap data reveals a high number of short lineaments and could be suitable in lineament mapping at local scale (Hajaj et al. 2022).Threshold values for the LINE module parameters are adjusted as follows: RADI: 10, GTHR: 70, DTHR: 20, LTHR: 30, ATHR: 30, and FTHR: 3.

Fuzzy logic modelling.
Fuzzy logic modelling (FLM) has a great interest in mineral exploration using GIS and remotely sensed data, it is adopted for generating mineral favourability maps in metallogenic provinces (Kim et al. 2019, Moradpour et al. 2022, Wambo et al. 2020).The FLM is based on fuzzy sets theory introduced by (Zadeh 1965).It is a form of multi-valued logic, where the real values of the variables are included in the interval from 0 to 1, which allows the characterisation of the membership degree in the set.Each category is given a membership value, 1 corresponds to full membership, and zero corresponds to non-membership.The n-fuzzy sets Ai (i = 1, 2, 3 . .., n) of the evidence layer x is defined in equation (1): Where, x is the combination of all the thematic layers xi (i = 1, 2, 3 . . .n), each layer presents m levels indicated as (j = 1, 2, 3 . . .m). Calculated membership function (mA) 0.5 < μ A < 1, x ij is promising for mineralised sites, while, the 1 < μ A < 0.5, x ij is not (Ma et al. 2020).The μ A (x) is considered as the membership function or the membership degree of x in the A ij fuzzy set.
Mineral Potential Map (MPM) in GIS can be computed using the following equation ( 2), producing the final score of each category (Ma et al. 2020):

Fieldwork and laboratory analysis.
To approve the results of HyMap data image processing, geological fieldwork and laboratory analysis were performed.Two field surveys were conducted in the study area during May 25th to 29th, 2021 and, August 22nd to 26th, 2022.Several photographs of hydrothermal alterations were taken illustrating the macroscopic and microscopic aspects of the altered zones.Rock samples were carefully collected from the hydrothermally altered zones and oxidation zones for laboratory analysis.A handheld GPS (Global Positioning System) with ~5 m accuracy was used to validate the spatial distribution of the collected samples.Furthermore, X-ray diffraction (XRD) analysis was applied to powdered samples.XRD data were collected from 5° to 90° 2θ using the EMPYREAN diffractometer in the Characterisation and Analysis Center (CAC) of Cady Ayyad University, Morocco.Thereafter, the X'Pert HighScore Plus tool was used for further analysis of the resulting patterns in order to identify the existent mineral phases.
Figure 3 shows the RGB False Colour Composite image of RBD1, RBD2 and RBD3, respectively.The Magenta colour highlights the zones of Mg-Fe-OH/CO3 and Fe 2+ /Fe 3+ alteration minerals within the carbonates formations (Figure 3).The RBDs RGB confirms that RBD1 detected the abundance of Fe 2+ /Fe 3+ alteration minerals in the lower limestone and dolomite (Figure 3).Magenta hue manifests the areas relatively weak of Fe 2+ /Fe 3+ alteration minerals as observed in basal series and north-eastern the lower limestone and dolomite lithological units.The orange tone within the ultimate conglomerate as well as in quaternary sediment units highlights the Al-OH and Fe 2+ /Fe 3+ alteration minerals co-occurrences.Green shade reflects the dominance of Al-OH minerals in Jbel Lkest Quartzites around the doleritic dykes, and southern the Jbel Ouiharen gneisses (Figure 3).In the north-western part of study area, white hues illustrate a highly altered zone containing Al-OH, Fe 2+ /Fe 3+ alteration minerals with some admixture of Mg-Fe-OH/CO3 minerals (Figure 3).
DPCA transform was applied to some selected bands of HyMap, the band selection was based on spectral features of Fe 2+ /Fe 3+ , Al-OH, and Mg-Fe-OH/CO3 alteration minerals.Bands 21, 31 and 51 were used to map Fe 2+ /Fe 3+ alteration minerals that exhibit absorption features at 0.90 μm (Hunt and Ashley 1979).The absorption feature at 0.90 μm is matched with band 31 (900 nm).Eigenvector loadings matrix analysis (Table 2) indicated that PC3 is suitable to map Fe 2+ /Fe 3+ alteration minerals following the high negative contribution of band 31 (−0.79419593) and the high positive contributions of bands 21 and 51 (reflectance bands) with loadings of 0.57260547 and 0.20341044, respectively.The PC3 image highlights Fe 2+ /Fe 3+ alteration minerals as bright pixels (Figure 3).
Bands 101, 108, and 111 were selected for the DPCA to map Al-OH alteration minerals that exhibit specific absorption features at ~2200 nm (Hunt 1977), which is matched with band 108 (2217 nm).The analysis of eigenvector loadings showed the PC2 has strong positive loading in band 108 (0.80924933) and strong negative loadings in bands 101 (−0.55269015) and 111 (−0.19912086), respectively (Table 3).Thus, the negation of the PC2 image will depict Al-OH minerals as bright pixels (Figure 3).

MTMF mapping results
The MTMF sub-pixel classifier is adopted to identify alteration minerals in the study area using HyMap data.The VNIR and SWIR bands of HyMap were used after the pre-removal of noisy bands.Spectral results were derived from 110 spectral bands (Supplementary Table 2) after removal of bad bands.MNF band images (14 first bands) were selected based on MNF plot to avoid noise.The n-dimensional visualiser was used to identify the end-member (alteration mineral) spectra.Thereafter, the spectral analyst was applied to 14 MNF bands and the endmembers derived from the n-dimensional visualiser technique were compared to the USGS spectral library.Therefore, the characteristics of absorption depth as well as the scores generated in 'Hitlist' (Supplementary Table 3) were used for identifying the endmembers.The extracted Endmembers spectra are shown in supplementary figure 2, where the continuum in the spectra was removed by the division of the convex hull into the original (Clark et al. 2003).The continuum removed spectra (Clark et al. 1990) of the 14 classes were generated and compared to the USGS spectral library, as it is represented in The n-d class#1 is attributed to a mixture between pyrophyllite and low remaining vegetation after applying the 'Gap-fill'.Pyrophyllite was also detected in endmembers 6, 11, and 14, where it presents the typical absorption features at ∼2170 and ∼2324 nm (Hunt 1977, Carrino et al. 2018).In endmembers 6, and 14, pyrophyllite was detected in association with iron oxides especially haematite with its characteristic features at 0.9 and 0.5 um (Hunt and Ashley 1979).While the pure pixels of pyrophyllite could be mapped using the Em#11.Illite presents numerous absorption features related to alkaline K and Na at 2188.2 (Na) to 2205.5 (K) nm 2340.9 nm (K, Na) ∼2000, ∼2100, ∼2200, and ∼2330, and 2440 nm (Carrino et al. 2018).Illite was depicted in Em#10 and Em#12.Kaolinite shows two (Al-OH) absorption picks centred at ∼2150 and ∼2200 nm, respectively (Hunt 1977, Rajesh 2004, Jain and Sharma 2019).Endmember 13 represents topaz, which is demonstrated by diagnostic absorption feature at 2080 nm narrow Al-OH absorption (Clark et al. 1990, Bedini 2012).Montmorillonite exhibits absorption features pinpointed at 1400 and 1900 nm for (OH), and at ∼2205.5 nm for (Al-OH) (Hunt 1977).However, only (Al-OH) absorption is considered in the HyMap image after water absorptions removal, while endmember 7 belongs to carbonate minerals (dolomite).The montmorillonite has been revealed as Em#2 and Em#8 associated with illite.Dolomite mineral was characterised by (CO3) pick absorption at 2310 nm (Hunt 1977), corresponding to band 114 centred at 2318.1 nm.Muscovite mineral was revealed as Em#9, presenting Al-OH absorption features at ∼2200 nm and 2355 nm (Hunt 1977) (Figure 4).It is important to note that some detected endmember spectra could exhibit a weak characteristic absorptions, which can be explained with their low abundance, it is the case of Em#1, 5, and 6.The rule images of sub-pixel abundance of alteration minerals derived from the MTMF method are shown in supplementary Figure 3 (the n-D class #1 was nominated as 'Em#1' and so on for all the classes, accordingly).Supplementary figure 4 shows alteration mineralogical map of study area after post classification.

Lineament extraction results
Lineament extraction revealed 2090 lineaments in the study area, ranging in size from 100 m to 1193.56 m, with an average length of 191.4 m (Figure 5a).The small size lineaments show relatively more abundance.A lineaments density map for the study area is presented in Figure 5b.The analysis of the directional rose diagram of the study area shows that the NNE-SSW, NE-SW, and ENE-WSW are the main three lineament trends.
The high lineament density zones were mapped in the north-east in the lithological units of quartzites, and ultimate conglomerates.The southern flank of Jbel Ouiharen formed by gneisses shows a medium to high lineaments density.Near the Pan-African granites the gneiss shows a considerable lineament density.The Quartzites unit at the north-western part of the study area is marked by a high lineament density (Figure 5b).In the east the lithological units of ultimate conglomerates as well as the lower limestone and dolomite demonstrate a medium density of lineaments with directions ranging from N-S to NE-SW.
The medium and high density of lineaments exhibits a close association with pyrophyllite mineral in many locations (Figure 5a,b).

Mineral potential mapping (MPM) using fuzzy logic modelling
An MPM for the study area was generated using 20 input layers derived from HyMap data.Fuzzy logic modelling was implemented for the integration of RBD and DPCA-based hydrothermal alteration layers, MTMF-based mineralogical layers, and structural lineaments density layer (Table 5).The feed-forward stages for the application of FLM in mineral prospecting were detailed by Carranza (2008).By applying the FLM method, the reclassification of multiclass evidential thematic maps was performed in ten classes from 0 to 1 (Table 5).An MPM was generated using the fuzzy gamma operator (γ = 0.9).The MPM highlighted many potential zones in the study area (Figure 6).Some zones (A, B and C) with the highest potential are delineated in the MPM using yellow rectangles (Figure 12).Zone (A) shows a high prospective zone oriented N-S within the ultimate conglomerate unit.Zone (B) shows a high prospective zone in quartzite and dolerite units marked by the intersection of the lineaments NW-SE related to the AVSZ and other NE-SW lineaments.In the north-western part of study area, zone (C) displays another potential zone in the quartzites unit characterised by the abundance of pyrophyllite and iron oxides/hydroxides minerals and the intersection with lineaments with NE-SW and NW-SE trends (Figure 5a).

Petrographic analysis
Some hydrothermal alteration minerals are revealed based on the microscopic study of the representative samples including, silicification, dolomitisation, sericitisation, and kaolinitization.An intense silicification was noted in the lithofacies of the study area where quartzites formation are marked by a new generation of abundant anhedral crystals, while the first large size grains exhibit an undulose extinction (Supplementary figure 6A,B).Kaolinitization occurs in feldspar crystals transformed to kaolinite, which is found in the ultimate conglomerates represented by microconglomerate in supplementary figure 6C, as well as in some doleritic dykes (Supplementary figure 6D).The dolomitisation was depicted in lower series basal limestone showing hydrothermal dolomite crystals within some silica-rich veins (Supplementary figure 6E) or with more fine dolomite crystals (Supplementary figure 6F).Sericitisation alteration is replaced plagioclase in a doleritic dykes, in association with kaolinitization alteration (Supplementary figure 6D).Furthermore, geologic and petrographic data reveals the development of silicic alteration in the study area, which is generally associated with the argillic alteration.The illite mineral occurs within fractures in some cases, as shown in supplementary figure 6A, the preferential orientation of illite mineral spread in silicified zones is observable.Supplementary figure 6g shows an interstitial pyrophyllite in a quartzite sample around a doleritic dyke.As observed in southeastern Kerdous inlier at Agoujgal ore deposit, the texture of the mineralisation does not show synsedimentary characteristics.It shapes the fractures in the dolomites and constitutes the cement in the terrigenous host rocks (Maddi et al. 2011).Some parts in the gneiss unit reveal argillic alteration (Supplementary figure 6 H, I).

Discussion
In the western Anti-Atlasic belt numerous copper occurrences reveal a close spatial association with the paleorelief areas, which can be explained by a polymetallic hydrothermal activity (Pouit, 1966).This investigation targets the identification of prospective zones for copper/gold, tungsten and manganese mineralisations using HyMap remote sensing imagery in the AVSZ of eastern border of Kerdous inlier, Morocco.As primary mapping, the use of RBD and DPCA allowed the identification of the distribution of Fe 2 + /Fe 3+ , Al-OH, and Mg-Fe-OH/CO3 alteration minerals.The analysis of RBD and DPCA results demonstrates a high similarity in terms of the spatial distribution of alteration minerals (Figure 3).Also, an RGB image of HyMap was developed using the developed RBDs in this study (Figure 3g).A widespread distribution of hydrothermal alterations was depicted in the lithological units of quartzites, ultimate conglomerate, basal series (limestone and dolomite), and lower series (limestone and dolomite).High Al-OH alteration is detected around doleritic dykes, which are mapped using MTMF method as argillic (pyrophyllite, illite, and montmorillonite).In the ultimate conglomerate a manganese index fits the zone of both Fe 2+ / Fe 3+ and Al-OH alteration minerals.The n-dimensional visualiser analysis allowed the extraction of 14 endmembers that indicate the presence of seven mineral phases, including pyrophyllite, montmorillonite, haematite, kaolinite, illite, dolomite, and muscovite.The extracted endmembers using MTMF classification helped to map spatial distribution of Fe 2+ /Fe 3+ , Al-OH, and Mg-Fe-OH/CO3 alteration minerals in the geological background of the study area, comprehensively (Supplementary figures 3,4).XRD analysis, fieldwork, and microscopic study demonstrate the presence of hydrothermal alteration zones promising to host Cu and Mn mineralisations.Validation data confirm the spatial association of Fe 2+ /Fe 3+ and Mg-Fe-OH/CO3 alteration minerals in the eastern part of the study area, where some dolomitised samples hosting haematite and Fe-hydroxides occur.Furthermore, the detected Al-OH alteration minerals within ultimate conglomerate using RBD1 were also mapped in detail using MTMF method as kaolinite and illite and confirmed with laboratory analysis.The Al-OH alteration presents a subequatorial trend, which is observed in the quartzites and in the south of Jbel Ouiharen gneisses.This trend fits the structures attributed to the pan-African major phase.Topaz when associated with other clay minerals (alunite) can be a sign of a relative high temperature (>∼250-260°C) (White and Hedenquist 1995).The FLM was applied to 20 HyMap-based input layers.FLM allows the generation of a high spatial resolution MPM, where three Cu and Mn potential zones were highlighted in the eastern and northwestern parts of study area.Moreover, the importance of the conglomerate of the Ouarzazate Group as a potential exploration zone was also identified in other inliers of the western Anti-Atlas as the Igherm inlier by applying an aeromagnetic analysis (Ouchchen et al. 2021).The identification of a NE-SW and NW-SE structural lineaments intersection in some highly altered lithofacies demonstrated that the alteration could be structurally controlled.The FLM-based mineral mapping results play a significant role to minimise the cost of further mineral prospecting surveys by pinpointing potential zones (Lindsay et al. 2016).

Conclusions
The findings of this research demonstrate that the use of Fuzzy modelling for integrating thematic layers (i.e.RBD, DPCA, Line density, and MTMF) derived from HyMap imagery is a pertinent approach to generate high-resolution mineral potential maps.It provides an accurate map of high prospective zones in the Ameln valley region, Kerdous inlier, western Anti-Atlas, Morocco.The lithological units of ultimate conglomerate, and quartzites display three locations that are suitable for mineral exploration campaigns.The close spatial association between the NE-SW and NW-SE lineaments and alteration zones highlighted an important tectonic control of mineralisation in the study area.Hence, a detailed structural investigation will be very useful for the better understanding of the complexity of copper mineralisation in the western Anti-Atlas province.Furthermore, the findings of this investigation principally based on HyMap data processing could be supported by stable isotopes study that is needed to highlight hydrothermal fluid sources.The present image processing methodology applied to HyMap data allows accurate mapping of hydrothermal alteration zones in the study area, then, it can be applied in other regions with similar geological conditions in the western Anti-Atlas province.

Figure 1 .
Figure 1.Geological sitting of the Kerdous inlier in the south-western Anti-Atlas; a. Location in the map of the Anti-Atlasic chain, modified from (Gasquet et al. 2008).b.Geological map of the study area.

Figure 2 .
Figure 2. The flowchart of methodology applied for HyMap data processing.

Figure 5 .
Figure 5. (a) Mineralogical thematic map overlain by the extracted lineaments derived from HyMap data (The lineaments rose diagram is presented in the bottom left corner).(b) Lineaments and faults density thematic map of the study area.

Figure 6 .
Figure 6.MPM generated from FLM for the study area.High potential zones are delimited using yellow rectangle.

Table 1 .
Spectral details of alteration minerals for developing RBDs using HyMap Bands (B).

Table 5 .
Fuzzification parameters applied to the input data in this investigation.