Located on the border of the Adamawa-Yadé shear zone, Chad is part of the Central Pan-African Belt (CPAB) (Figure 1(a) and Figure 1(b); [44][49]). The Oubanguides [50] are referred to as the CPAB, which is located north of the Congo craton. This belt extends across Chad, Cameroon, and the Central African Republic, continuing on to Sudan. The development of the CPAB is seen as a point of convergence and collision between the São Francisco-Congo cratons, the West African craton (WAC) and the Sahara metacraton as defined by [44][49] Abdelsalam et al. (2002), Toteu et al. (2004) and [51]. In Chad, the CPAB extends over the Baïbokoum massif in the south, the Ouaddaï in the east, the Tibesti massifs in the north, the Mayo Kebbi massif in the southwest, and the Guéra massif in the center. According to [52], the Tibesti is composed of metasediments, metavolcanics,

Figure 1. (a) Map of Africa showing the different orogens, craton and central African mobile zone (after [39][40]; (b) Geological map of Central Africa (after [41]-[48]) showing the main litho-tectonic units and domains of the Central African Orogenic Belt (SZ—Shear Zone; CCSZ—Central Cameroon shear zone; MNSZ—Mayo Nolti shear zone; SSZ—Sanaga shear zone; MBSZ—M’Béré shear zone; BOSZ—Bozoum-Ndélé shear zone; TBF—Tcholliré Banyo fault; CAR—Central African Republic; R. Congo—Republic of Congo; D.R. Congo—Democratic Republic of the Congo.

and a variety of amphibolites, with local occurrences of migmatites in enclaves and lenses within gneisses and granitoids. The Tibesti massif has been dated to between 1250 and 820 million years old using Rb-Sr techniques on whole rocks, as mentioned by [53] and [54]. The Ouaddaï massif consists of metasedimentary series crossed by huge peraluminous leucogranitic batholiths and small plutons of pyroxene monzonite, hornblende monzogabbro, and biotite granite, giving rise to a strongly potassic to shoshonitic calc-alkaline series [55]-[57]. U-Pb dating on zircon has determined that the age of the peraluminous granites is 635 ± 3 Ma [57] for biotite leucogranites, while it is 613 ± 8 Ma for muscovite and garnet leucogranites [57]. The age of the metaluminous, calc-alkaline, and strongly potassium-rich to shoshonitic series has been given as 540 ± 5 Ma in biotite granites, as shown by [57]-[60] in their studies on the U-Pb of zircons. The Baïbokoum massif, which extends to the southern border of Chad and continues into the Central African Republic and Cameroon, represents an extension of the Adamwa-Yadé territory. The Baïbokoum massif, which extends along the southern border of Chad and continues into the Central African Republic and Cameroon, represents an extension of the Adamwa-Yadé territory. The Baïbokoum massif has been little explored, with only general aspects having been identified in the research of Jérémie and [9]-[11][58][61]-[64]). Several rocks in the massif date from the Neoproterozoic, including syenites (654 Ma, determined by the K-Ar method on whole rock), as well as granites and a syenite (632 and 568 Ma, established by K-Ar and Rb-Sr on whole rock [65][58]). The Mayo-Kébbi Massif is a structure extending in a northeast-southwest direction between the extensions of the Tcholliré faults in Chad. It is distinguished by its juvenile Neoproterozoic crust [66]-[69]. This region represents a magmatic arc zone that formed between 800 Ma and 550 Ma following successive collisions with the Central Cameroon and North Cameroon domains, respectively [66]-[68]. It encompasses the greenstone belt, the batholith, and post-tectonic intrusions [36][55][66]-[68][70]. The Guéra massif, which has around ten charnockitic formations ranging from noritic to granitic in composition, with a predominance of syenite [71][72], as well as small massifs with gabbroic and granodioritic compositions, often oriented and present as enclaves in heterogeneous granite, could be evidence of the oldest rocks in the Massif Central [71][73].

Baïbokoum is located in the Adamawa-Yadé zone, also known as the Central Cameroon Shear Zone or the Adamawa Shear Zone ([74]; Figure 1(b)). Vast areas of Paleoproterozoic metasedimentary rocks and orthogneisses ([44][75]) outcrop in the Yadé massif. Based on Sm-Nd model ages and inherited zircon ages, [44] proposed that the metasediments and orthogneisses of the Adamawa-Yadé territory show significant influence from the Archean crust, comparable to that of the Congo craton [48]. Initial geological work carried out by [9]-[11][65][66] identified the following in the Baïbokoum area: (i) syenites; (ii) monzonites; (iii) hornblende-biotite orthogneisses; (iv) biotite gneisses and syenodiorites (Figure 2).

In this study, the equipment and methods used and monitored in the Baïbokoum area are illustrated in the following flow chart (Figure 3). Landsat 9 OLI bands (LC09_L1TP_183055_20250218_20250218_02_T1) obtained on February 18, 2025, and downloaded from USGS site web (https://doi.org/10.5066/P975CC9B), cloud cover 0.21. Geodetic parameters (Map projection = “utm”, datum = “wgs84”, ellipsoid = “wgs84”, utm zone = 33). Landsat 9 OLI data consist of a collection of nine bands (bands 1-9) and two TIR (thermal infrared) bands (bands 10-11). In this work, the Landsat 9 OLI bands visible and near infrared (VNIR), shortwave infrared (SWIR) and panchromatic (b8) bands were used. Principal

Figure 2. Geological sketch map of the study area showing sample station.

Figure 3. Methodological flowchart.

component analysis (PCA) was applied to the spectral bands of Landsat-9 data to highlight the exposed rock units of the study area using petrography.

Envi 5.3 software was used to apply radiometric correction to Landsat 9 OLI bands. Radiometric calibrations and fast atmospheric corrections (QUAC) were applied to the Landsat 9 VNIR-SWIR bands for high-precision detection. Indeed, QUAC performs atmospheric correction better than other methods such as Fast Line-of-sight Atmospheric Analysis of Spectral Hypercube (FLAASH), generally producing a reflectance spectrum that is between 10% and 15% of the ground truth [76]. In addition, lineaments were extracted using PCI software version 9.1 with two different datasets, namely the high-resolution digital model. The rosette and lineament statistics were obtained from RockWorks software and finally the layout was done with ArcGIS 10.8 software. field campaigns collected samples on different lithological types and in the laboratory thin section making and description were carried out.

The geochemical data (major elements, trace elements, and rare earth elements) used in this study were obtained from [9] and analyzed for whole rock chemistry at the laboratories of ALS Chemex South Africa (Pty), Ltd. in Johannesburg, South Africa. Major element analysis was performed by inductively coupled plasma atomic emission spectrometry (ICP-AES); 0.2 g of rock powder was fused with 1.5 g of LiBO2 and then dissolved in 100 mm of 35% HNO3. Trace and rare earth element content was determined by ICP-MS from pastes after dissolving 0.25 g of rock powder using four acid digests. Analytical precisions range from 0.1% to 0.02% for major elements, 0.01 to 0.5 ppm for trace elements, and 0.01 to 5 ppm for rare earth elements.

These geochemical data were plotted on discrimination diagrams to determine the productivity of each sample in certain strategic metals, as is generally recommended in granitoid metallogeny [35][77]-[79]. In addition, the geochemical data were statistically processed using the GeoChemical Data toolkit (GCDkit.6.0; [80]) supported by R.3.6.0 software (for Windows) to identify positive or negative anomalies in the data. The normal distribution of the selected metals was tested using histograms and box plots in accordance with [81]. In the box plots, the bold line in the middle of the boxes is the median or 50^th percentile, and is given such that Q2 = 50%. The first quartile Q1 and third quartile Q3 (25%) and (75%) are the lower percentile and upper percentile, respectively. However, the difference between Q1 and Q3 is known as the interquartile range (IQR), where IQR indicates the extreme values calculated from a window of 1.5 * IQR. The statistical threshold for element concentration was calculated using the following formula: Threshold = Q3 + 1.5 * (IQR) for the upper threshold [81]), with IQR = Q3 − Q1. Values above or below the threshold are outliers and represent positive and negative anomalies, respectively.

FCCs using combinations like (Band 7, 5, 3) visually enhanced differences between vegetation, soil, and rock types. Igneous terrains (syenites) appeared in light pink tones, while metamorphic (gneiss) and sedimentary (shale) rock units showed green and white hues respectively (Figure 4). Also, the study area river display blue hues. FCCs offered a qualitative but quick overview for initial lithological discrimination (Figure 4).

Figure 4. FCC of Baïbokoum area.

PC1 and PC2 accounted for broad lithological variation. PC3 isolated specific spectral anomalies linked to mineralized zones. PCA maps (Figure 5) revealed rock units as syenites (in cyan tone), gneisses (in orange tone), and shales (in yellow tone) and contacts not easily detected in raw bands or FCCs.

According to Table 1 analysis of the eigenvector and eigenvalue loadings derived from PCA analysis of the Landsat 9 OLI bands (B2, 3, 4, 5, 6, and 7), PC1, PC2 and PC3 comprises a satisfactory loading from each of the 06 PC bands and makes up 92.86%, 5.94% and 0.84% respectively of the total data. It is therefore perfect for focusing on geological formations.

The detection of rock types and alteration regions was done using the MNF1, 2, and 3 of OLI bands that were derived using MNF analysis. Syenites are identified as yellow to light-green hues, gneisses as pink, and shale as blue coloration in the resulting MNF pictures, which have been subjected to an RGB composite analysis (Figure 6).

Figure 5. PCA map of Baïbokoum.

Table 1. PC values of OLI bands 2, 3, 4, 5, 6 and 7.

According to Table 2 analysis of the eigenvectors and eigenvalues loadings derived from MNF analysis of the Landsat 9 OLI bands (B1, 2, 3, 4, 5, 6, and 7), MNF1, MNF2 and MNF3 comprises a satisfactory loading from each of the 07 MNF bands and makes up 12.89%, 9.08% and 6.12% respectively of the total data. It is therefore perfect for focusing on geological formations.

Band ratios such as (4/2), (6/7), and (6/5) were particularly useful for: Highlighting the study area lithological units (syenites, gneisses and shales), iron oxide (e.g., hematite/goethite, Figure 7(a)), ferrous (Figure 7(b)), clay minerals (Figure 7(c)) and hydrothermal alteration zones (Figure 7(d)). Syenites display purple tones, shales exhibit pink tones while gneisses are distinguished by light green

Figure 6. MNF map of Baïbokoum.

Table 2. MNF values of OLI bands 1, 2, 3, 4, 5, 6 and 7.

and yellow (Figure 7(d)).

We used the USGS laboratory mineral spectra’s (biotite, chlorite, goethite, hematite, illite, ilmenite, kaolinite, magnetite, pyrite and rutile) to run the CEM algorithm on 753 OLI band because of their spectral properties (Figure 8). Hence, the minerals in the study area shows up several tones (Figure 8).

Using endmembers extracted from known mineral spectra (e.g., USGS library;Figure 9(a)), SAM accurately identified rocks such as syenite, shale and gneiss (Figure 9(b)) based on a fixed threshold (i.e. angle radian = 0.1). The syenitic, gneissic and shale rocks exhibit green, red and light-blue pixels (Figure 9(b)).

Therefore, following multiple tests, the accuracy evaluation findings (Table 3) of the SAM classification utilizing the confusion matrix methodology showed that the classification method overall accuracy is 100% with a Kappa coefficient of 1.00.

Figure 10(a) and Figure 10(b), which reveal that each rock unit retains its lineament

Figure 7. PCA map of Baïbokoum.

Figure 8. CEM map of Baïbokoum.

Figure 9. SAM classification map of Baïbokoum.

Table 3. Confusion matrix values.

Figure 10. Lineament map of Baïbokoum area with rose diagram.

trend according to the created lineament map (Figure 10(a)), are the findings of lineament studies (manual and automated extraction) within the Baïbokoum area by processing band 8 (panchromatic band) of Landsat 9 OLI data. Two primary trends were discovered by the rose diagram of the lineaments: a NE-SW trend and E-W trend. Additionally, an extra NNE-SSW trend is noted (Figure 10(b)).

The ability to map lithology and minerals was greatly improved by integrating FCC, PCA, MNF, BR, CEM, and SAM using Landsat 9 OLI data. PCA and MNF offered dimensional reduction and noise suppression, which improved feature extraction, while FCC and BR offered visual and semi-quantitative improvements. Accurate mineral identification and supervised lithological and mineral classification were made possible by SAM and CEM respectively.

[94] suggests that FCC pictures may enhance understanding of lithological units and hydrothermally altered rocks on a local or regional scale [24][25][35]. Furthermore, some mineral groups and minerals in changed lithologies exhibit unique electromagnetic absorption capabilities [35][25].

PCA reduced data redundancy by concentrating spectral variability into the first few components [24][25]. [95] explains that the sign and magnitude of eigenvectors are dictated by the spectral properties of minerals and rocks that contribute to statistical variance in each PC. Table 1 shows that PC1, PC2, and PC3 have substantial eigenvector loadings for bands 7, 5, and 3.

MNF transformed the OLI data to maximize signal while minimizing noise [25]. Helped extract subtle spectral differences within homogeneous lithologies. MNF images were particularly effective in improving the signal for low-reflectance units such as syenites and shale.

Suppressing topographic effects and enhancing subtle mineralogical differences [24][25]. These BR helped delineate alteration halos and weathered rock units, aiding early mineral exploration targeting. All hydrothermal minerals display unique spectral signatures on multi-spectral remote sensing systems, determined by their texture and mineralogy [24][96]. Focusing on the research area’s hydrothermally changed regions and rock units allowed for the discovery of alteration mineral deposits, such as iron oxides, ferrous, and clay minerals.

According to the USGS spectral analysis database, the analyzed minerals show higher reflectance and absorption values than it usually does. The CEM method used OLI bands 753 to solve common problems with adaptive beam formation in signal processing [35][96][97] to detect the aforementioned minerals using the USGS spectral library.

We used the SAM method on the FCC 753 to sort the lithological units we found in the study area (Figure 8). We used an angle radian of 0.1 for correlation and high exactness [25][35][96][98] with a value of 100% and a Kappa coefficient of 1.0 (Table 3). This shows that the method works well and is accurate for lithological classification.

Processing the panchromatic band (B8) brought out important lineaments that help us figure out how hydrothermal minerals move through the study area.

Using GIS tool to combine FCC, PCA, MNF, BR, CEM and SAM data led to the creation of a synthesis geological-mineralization map of the study area (Figure 16).

Figure 16. Baïbokoum synthesis litho-mineral map.

Because of the limited surface exposures and subtle mineralogical contrasts found in Precambrian terrains, this multi-technique approach proved particularly effective. Better discrimination over older sensors was further supported by the use of Landsat 9’s enhanced radiometric resolution.

In the mineral exploration sector, attention is generally focused on positive anomalies, assuming that mineral-rich regions and their subsequent leaching have amplified the concentration of elements beyond the usual standards of the Earth’s crust. However, negative anomalies can also be important, particularly when they illustrate a decrease in certain components/metals during the modification of surrounding rocks that accompanies the mineralization process [7][8][99].

The selected elements, although they have different chemical properties, behave similarly during fractional crystallization. According to [100], chalcopyrite is the main element in copper ore, while zinc ore consists mainly of ferromagnetic silicates. Copper deposits usually form either in volcanic arc zones above subduction zones or in post-collisional magmatism that forms at the end of the subduction process [101][102]. This is the case in the Baïbokoum region, which is located within the Pan-African chain in Central Africa. In both environments, the distinction between productive and barren rocks with respect to selected major and rare earth elements is regularly based on the overall geochemistry of the rocks [31][103]-[106]. According to [92], granitoids containing mineralization of the selected elements have Clarke values between 0.3 and 144.3. In the samples analyzed, Clarke values range from 0.36 to 511.53.

Vegetation poses a considerable challenge for mineral remote sensing, mainly in the visible and near-infrared range, as it camouflages the underlying rock while introducing its own spectral signature. Precise mineral signatures (such as those of clay alteration, which stimulates the growth of certain types of vegetation) can only be identified indirectly. MNF and PCA methods can help distinguish between certain spectral variations related to noise and vegetation, but they cannot eliminate them entirely. Erosion, by exposing new rocks or altered areas, can be beneficial because it makes the surface more accessible for remote sensing [17][25][32]. However, excessive erosion can complicate the detection of the initial spectral signatures of the source rock or primary mineralization, as it has the potential to reveal deeper alteration strata or amalgamate various minerals.

The orientations of the detected lineaments (E-W, NNE-SSE to NW-SE, and NE-SW) in the studied region are due to shear zones, which are frequently permeable and can serve as channels for the movement of metal-laden hydrothermal fluids [38]. It is possible that these fluids transported and deposited mineralization along these gold, copper, or polymetallic veins. Tourmalinization, for example, may be common along these zones, indicating high-temperature hydrothermal processes.