Structural-Tectonic Interpretation of Lineaments and Their Role in the Development of Karst-Suffosion Processes in the Mangystau Region Based on Remote Sensing Data

Abstract

This paper presents an integrated approach to the mapping and structural-tectonic interpretation of lineaments in the Mangystau region using multispectral Landsat-8 OLI data and the medium-resolution Airbus WorldDEM4Ortho digital elevation model. Automatic extraction of linear structures has enabled the identification of over 35,000 lineaments of varying length and orientation, forming a network of intersecting zones that influence the distribution of sedimentary thicknesses, drainage directions, and the location of karst-suffosion depressions. The most prominent are the north-western and sub-latitudinal systems, closely correlated with zones of fracturing and faults, which confirms their tectonic origin. The spatial concentration of lineaments coincides with areas of increased permeability in carbonate and gypsum-bearing rocks and localizes the pathways of groundwater circulation, contributing to the development of karst-suffosion processes. The obtained results demonstrate the significance of structural influences on the region’s current geomorphological and hydrogeological conditions and also have practical importance for engineering-geological surveys, the assessment of geological risks, and the planning of sustainable land use.

1. Introduction

Karst-suffosion processes are a complex of interrelated phenomena in which surface and groundwater not only dissolve rock but also mechanically wash out fine-grained material from the overlying loose sediments [1,2,3]. The intensity of these processes depends on both the degree of structural disturbance of the rocks and on their mineral composition: gypsum and anhydrite are the most soluble, limestones have moderate solubility, and dolomites are less susceptible to dissolution [4,5].

The development and spatial organization of karst-suffosion processes are mostly related to lineaments, which indicate zones of increased fracturing and tectonic faults. Lineaments are natural simple or complex linear or curvilinear formations visible on the Earth’s surface [6]. These features may include faults, fractures, dikes, stratigraphic formations, as well as rivers and drainage basins [7,8,9,10]. Other types of lineaments include the boundaries of distinct land cover units and artificial structures such as roads, bridges, and edges [11,12,13].

The interest in studying lineaments can be explained by the fact that surface karst features (sinkholes, depressions) are often oriented along zones of fracturing and tend to be concentrated at lineament junctions, the points where lineaments intersect [14,15]. The most intensive development of underground karst features (channels, fractures) is also related to the areas of lineament intersections [16,17], and they often function as conduits for the migration of groundwater and other fluids [6]. According to research by a number of authors [18,19,20,21,22], lineaments can serve as indicators of an underground network of karst cavities and fractures. They are characterized by increased permeability and concentrate the runoff surface and groundwater.

The mapping of lineaments holds substantial importance across various Earth science disciplines, including mineral and oil exploration [23,24], the analysis of tectonic structures [6], as well as the study and assessment of underground water resources [7,8,9,10]. Furthermore, lineament analysis is widely used in hydrogeology for identifying highly productive boreholes and for detailed mapping of underground structures crucial to groundwater infiltration, migration, and discharge processes [6,7,8,25,26].

Traditional approaches to lineament mapping, such as field surveys of faults and the visual interpretation of aerial photographs, are characterized by high labor intensity, time consumption, and substantial resource requirements. These factors significantly complicate their application over extensive territories. Furthermore, such methods are inherently subjective and depend heavily on the researcher’s experience and expertise, which can lead to the distortion of lineament characteristics or incomplete detection [6,12]. In contrast, Geographic Information Systems (GIS) and remote sensing methods have become essential for extracting and analyzing lineaments, providing scientists with valuable insights into tectonic evolution and fault networks. Numerous studies have highlighted the utility of remote sensing for lineament mapping and the identification of geological structures, citing its synoptic coverage and its automatic, easily reproducible methodology [8,12,13,24,27,28,29,30,31,32]. The availability of multispectral data at various resolutions, combined with advanced image processing techniques, has enabled more accurate extraction of geological lineaments [12]. Various tools for automatic extraction are documented in the literature, such as the LINE module of the PCI Geomatica software [27,33,34]. Additionally, Masoud and Koike [6] developed the LINDA tool, which utilizes elevation data and satellite imagery to characterize and interpret the morphotectonic features of lineaments.

The collection of analyzed works reflects three interconnected classes of research approaches, including comparative analysis of optical and radar sensors, methods of morphometric analysis of relief based on digital elevation models (DEMs), and applied interpretations of lineament structures in the context of tectonics and exogenous processes.

In comparative studies evaluating the effectiveness of optical and radar data, results are generally interpreted using a combination of quantitative and structural metrics. These include the total number and length of lineaments, orientation analysis, spatial density distribution, and the degree of correlation with independent data sources such as geological maps, fault structures, morphometric relief features, and drainage network elements.

Studies [34,35] show that the effectiveness of automated lineament mapping depends significantly on the type of remote sensing data used, the morphological conditions of the area, and the criteria for validating the results. SAR (Synthetic Aperture Radar) radar data (Sentinel-1, ALOS PALSAR) are highly sensitive to areas of increased fracturing and tectonic disturbances, whereas optical data from Landsat-8 OLI (Operational Land Imager), Terra ASTER, and Sentinel-2 better reflect lithological boundaries and surface features. It has been established that the use of only one dataset does not universally yield the best results: SAR data allow for the identification of a greater number of linear structures; however, their geological interpretation requires additional filtering of pseudo-lineaments and verification against independent data [35]. In turn, the combined use of Landsat-8 OLI and ALOS PALSAR enhances the reliability of structural interpretation. Digital elevation models and morphometric parameters often demonstrate greater consistency with known faults, particularly in rugged terrain [34,35].

The results of studies [36,37,38,39,40,41] confirm the validity of integrating multispectral, radar, and digital elevation models. It has been shown that the combination of principal component analysis (PCA), directional filtering, hillshading, and automatic extraction algorithms in the PCI Geomatica environment allows the identification of both known and previously unmapped structural elements associated with fault-fracture zones [37,38]. It has been shown that the number of automatically identified lineaments can significantly exceed the number of mapped faults, which necessitates the mandatory filtering of pseudo-lineaments [37]. The effectiveness of extracting linear structures depends not only on the processing algorithms but also on the characteristics of the digital elevation model, lighting parameters, and the morphometric indices used [39].

A number of studies confirm the close spatial relationship between lineaments and karst features, zones of increased fracturing and areas of active groundwater filtration [38,41]. It has been shown that an anomalous density of lineaments can be regarded as an indicator of the potentially hazardous development of karst-suffosion processes [41]. Furthermore, modern approaches to lineament analysis are largely focused not only on the identification of linear features but also on their morphostructural interpretation in the context of neotectonic and exogenic processes [40].

In studies that are methodologically similar to our approach, the combination of PCA based on Landsat-8 and hillshading based on Shuttle Radar Topography Mission (SRTM) DEM is considered an effective method for jointly enhancing the spectral and morphostructural manifestations of lineaments [42]. For karst areas, it is particularly important not only to map lineaments but also to verify their spatial relationship with depression-type landforms and surface deformations in active depressions [43]. Unlike a number of recent studies, where validation is limited to geological maps or field observations, this study additionally employs a comparison with karst depressions, which allows for a strengthened structural-geomorphological interpretation of the results.

Despite a significant number of studies devoted to lineament analysis, the task of comprehensively assessing the structural influence on karst-suffosion processes based on automated methods for extracting linear features, followed by validation using geological and morphostructural data, remains relevant. Research aimed at identifying the spatial relationship between the density and orientation of lineaments, modern relief depressions, and zones of active groundwater filtration is particularly in demand.

The aim of this study is to apply multispectral satellite data from Landsat 8 OLI and digital elevation models for the extraction, analysis, and interpretation of lineaments. The relevance of this research is driven by the necessity to develop and validate an integrated approach to mapping and structural interpretation of lineaments using remote sensing data and digital elevation models. Additionally, it aims to analyze the tectonic control of karst-suffosion processes and the hydrogeological conditions of the area. The results obtained can serve as a scientific basis for assessing geological risks, planning engineering investigations, and implementing rational management of natural resources.

2. Study Area

The Mangystau Region is situated in the west of the Republic of Kazakhstan, in a desert zone between the Caspian and Aral Seas (Figure 1). The region covers an area of 165,600 km². The territory belongs to the arid zone of Central Asia and is characterized by desert landscapes, low surface runoff, and the widespread occurrence of exposed carbonate, sulphate, and saline rocks, which are highly conducive to karst formation.

Geologically, the area represents the western part of the Turanian Plateau within the Epipaleozoic Platform. Its structure was formed by the interaction of the south-eastern margins of the East European and Central Eurasian platforms, which resulted in a complex tectonic structure and inherited deformation patterns. Within the sedimentary cover, the following major structural elements are distinguished: the Caspian depression, the North Ustirt system deflections, the Bozashchy system of uplifts and depressions, the South Mangystau–South Ustirt system of deflections, and the Central Mangystau–Tuarkyr system of rises (Figure 2) [44].

The section comprises Palaeozoic, Permian–Triassic, Jurassic–Lower Miocene, and Middle Miocene–Quaternary structural units, reflecting the region’s long history of tectonic and sedimentary development [44,45]. The most significant role in karst formation is played by widespread Neogene carbonate and sulphate deposits, overlain by thin Quaternary sediments, which facilitate active infiltration and the surface manifestation of karst processes [46,47,48].

A system of sub-latitudinal faults, intersected by sub-meridional faults, forms a complex block structure and a network of lineaments. These fault zones play a key role in the spatial organization of karst-suffosion processes, controlling rock permeability, the direction of filtration flows, and the localization of dissolution zones. Large endorheic basins, such as Karakiya and Karynzharyk, are widespread within the region. Their formation is associated with a combination of tectonic deformations, karst-erosional processes, and deflation. These forms represent unique geomorphological objects and reflect the close interrelationship between tectonics, lithology, and karst.

Modern tectonic movements, inherited from the neotectonic stage, remain active. Vertical displacements of the Earth’s crust range from 1 to 4 mm/year, with uplift predominating, while in the north-east, local subsidence zones of up to 2 mm/year are observed [44].

Rock fracturing is one of the key factors in karst formation. Tectonic fractures determine the position, orientation, and dimensions of karst features of various orders, whilst lithogenic and exogenic fracturing contributes to the formation of micro-relief, including karren and corrosion-enlarged fractures [20,49]. Increased fracturing in zones of neotectonic uplift enhances secondary porosity and intensifies chemical denudation processes [50].

The region’s hydrological conditions are characterized by the absence of permanent surface runoff. Temporary watercourses form episodically and terminate in closed depressions, where water is rapidly lost through evaporation and infiltration. Groundwater is the primary agent of karst formation. The chemical composition of the waters varies from weakly mineralized hydrocarbonate waters in zones of active water exchange to highly mineralized sulphate-chloride brines in deeper horizons [51,52,53]. Such hydrochemical heterogeneity contributes to the development of carbonate, sulphate, and salt karst.

Groundwater dynamics in the study area are uneven and depend on the tectonic structure, rock fracturing, and climatic conditions. Under conditions of low atmospheric precipitation, the main recharge of groundwater occurs through infrequent but intense torrential rains, typical of arid regions [53,54]. In a significant part of the region under consideration, where water exchange is slow, there is an accumulation of dissolved salts and a reduction in the rate of karst formation. However, a change in hydrodynamic conditions (for example, due to human activity or changes in the groundwater level) may lead to a sharp intensification of karst processes, thereby increasing geo-environmental risks [55].

The climatic conditions of the Mangystau Region are distinctly continental and arid, characterized by low precipitation and high evaporation rates. Despite this, episodic rainfalls and infiltration in zones of tectonic disturbance play a significant role in activating exogenous processes [56].

The Mangystau Region was chosen as the study area due to the high prominence of lineament structures, which are clearly identifiable from Earth remote sensing data, as well as the presence of numerous karst-suffosion features. The combination of geological, tectonic, and natural-climatic factors makes this region representative for analyzing the relationship between lineament systems and the development of karst-suffosion processes using remote sensing data.

3. Materials and Methods

3.1. Data Sources

As initial data for this study, optical satellite images from Landsat-8 (NASA/USGS, Colorado Springs, CO, USA), acquired by the OLI sensor, as well as a radar digital elevation model (DEM) Airbus WorldDEM4Ortho (Airbus Defence and Space GmbH, Taufkirchen, Germany), were used. The integration of optical and radar data allowed minimizing the influence of atmospheric conditions and surface illumination, thereby increasing the reliability of identifying geomorphological and structural-tectonic elements of the territory.

3.1.1. Optical Data: Landsat-8 OLI

Landsat-8 OLI acquires imagery in nine spectral bands (

Table 1. Characteristics of Landsat 8 OLI bands.

Bands	Wavelength (µm)	Spatial Resolution (m)
Band 1: Aerosols	0.43–0.45	30
Band 2: Blue	0.45–0.51	30
Band 3: Green	0.53–0.59	30
Band 4: Red	0.64–0.67	30
Band 5: Near Infrared (NIR)	0.85–0.88	30
Band 6: Shortwave Infrared 1 (SWIR 1)	1.57–1.65	30
Band 7: Shortwave Infrared 2 (SWIR 2)	2.11–2.29	30
Band 8: Panchromatic	0.50–0.68	15
Band 9: Cirrus clouds	1.36–1.38	30

), with a spatial resolution of 30 m for multispectral bands and 15 m for the panchromatic band and a radiometric resolution of 12 bits [57]. For this study, 15 Level-2 Type 1 (L2SP) Landsat-8 OLI scenes acquired during August 2025 were downloaded from the USGS Earth Explorer platform (https://earthexplorer.usgs.gov/) and used for lineament extraction (Figure 3). The L2SP data were selected due to their excellent pre-processing, including geometric, radiometric, and atmospheric corrections, which improve the accuracy of surface reflectance retrieval [58,59]. For the PCA-based lineament extraction, seven multispectral bands (B1–B7, covering the spectral range from visible to shortwave infrared, 0.43–2.29 pm) were used as input. The panchromatic band (B8) and the cirrus band (B9) were excluded: B8 has a different spatial resolution (15 m) that would require resampling, while B9 targets thin cirrus clouds and carries no lithological information relevant to structural mapping.

3.1.2. Radar Data (Digital Elevation Model (DEM))

To complement the optical data and reduce susceptibility to atmospheric artefacts, illumination effects, and cloud-related noise, the DEM was integrated into the analysis. Unlike multispectral imagery, DEM-based derivatives are unaffected by surface albedo or atmospheric conditions, which is particularly advantageous in the high-albedo arid terrain of the Mangystau region. The DEM provides geological, geomorphological, and structural information necessary for a detailed characterization of landforms [60].

The DEM Airbus WorldDEM4Ortho was used as the elevation dataset for morphometric lineament extraction and karst depression mapping. The product is created based on the global WorldDEM™ model, obtained through the TanDEM-X mission, carried out in partnership with the German Aerospace Center (DLR) and Airbus Defence and Space. The model has a spatial resolution of approximately 24 m and an absolute vertical accuracy of 2.5 m (LE90%), making it suitable for detecting subtle topographic expressions of faults and fractures in the low-relief arid terrain of the Mangystau region [61].

3.1.3. Geological Reference Data

To ensure the accurate interpretation of spatial analysis results, this study incorporated tectonic fault data digitized from 1:200,000 scale geological maps. The vectorization of these faults was based on official geological maps that accurately reflect the regional tectonic setting. The resulting fault vector layer served as a crucial reference for constructing a fault density map and for validating automatically extracted lineaments. This approach enabled a direct comparison between the spatial distribution and orientation of lineament systems identified using remote sensing data and DEMs with known tectonic features. Consequently, it significantly improved the reliability of the structural-tectonic interpretation of the results.

3.2. Methodology

The methodological framework of this study is based on the integration of two complementary analytical streams: (1) spectral analysis of multispectral optical imagery (Landsat-8 OLI, processed via PCA) for the detection of surface spectral contrasts associated with linear structures and (2) morphometric analysis of a medium-resolution DEM (Airbus WorldDEM4Ortho) for the detection of topographically expressed lineaments and closed karst depressions. The convergence of results from both streams, and their joint validation against independently digitized geological faults, constitutes the basis for the structural-tectonic interpretation.

• Stage 1. Identification of Karst Depressions

To identify potential karst features, the DEM Airbus WorldDEM4Ortho with a spatial resolution of 24 m was used. Data processing was carried out in ArcGIS Pro 3.4 using spatial analysis tools.

Karst landforms are characterized by distinct morphometric features that enable their identification through DEM analysis. Key diagnostic characteristics include the presence of closed negative landforms and local depressions in elevation relative to the surrounding surface [62,63,64,65]. Therefore, DEM analysis is widely employed to identify potential karst depressions and to assess their spatial distribution [66].

In the initial stage of analysis, the Fill tool from the Hydrology toolset of the Spatial Analyst module was employed for the automatic detection and filling of local relief depressions. Next, the filled terrain model was subtracted from the original DEM, resulting in a difference raster reflecting the depth of closed depressions. Negative values in the difference raster were interpreted as potential depression landforms. To assess the spatial distribution of potential karst features, closed depressions with a depth of ≥5 m were identified, which allowed for the exclusion of minor local relief irregularities and improved the reliability of identifying morphologically distinct depression forms. The resulting areas were reclassified and converted into vector polygonal objects for subsequent morphometric analysis [43,67].

The subsequent stage involved a visual verification of the identified depressions using the ESRI World Imagery base map (sources: Esri, Maxar, GeoEye, Earthstar Geographics, CNES/Airbus, DS, USDA, AeroGRID, IGN, and the GIS user community) and 1:200,000 scale topographic maps, which allowed us to exclude anthropogenic and false landforms. Additionally, axis lines (elongations) were manually digitized for each depression. An elongation is a straight line characterizing the elongated shape of each karst depression and reflecting its predominant direction. These axial lines were critical for the subsequent structural-morphological interpretation and analysis of the spatial relationship between karst features and lineament systems. As a result of this process, 365 closed depressions with a depth of more than 5 m and an area of over 0.5 km² were identified.

• Stage 2. Optical Image Pre-Processing: Principal Component Analysis

Seven multispectral bands of Landsat-8 OLI (B1-B7) were processed using principal component analysis (PCA) in ArcGIS Pro 3.4 (Spatial Analyst module) to enhance the expression of structural features (Figure 4). PCA transforms the original correlated spectral bands into uncorrelated principal components, thereby reducing redundancy and amplifying structural features that are difficult to detect in the original imagery [13,68,69,70,71,72]. In the arid, sparsely vegetated conditions of the Mangystau region, the first two principal components (PC1 and PC2) together accounted for 98.88% of the total data variance (PC1 = 89.83%; PC2 = 9.05%); components PC3–PC7 collectively contributed less than 1.12% and were not used further. PC1 and PC2 were therefore used as the input images for automated lineament extraction in Stage 4 [12,23,28,73,74,75].

• Stage 3. DEM Derivative Generation: Multi-Azimuth Shaded Relief

Shaded relief models derived from DEM data provide a clear visualization of geological linear structures and allow assessment of their structural relationships identified through the processing of Landsat-8 OLI imagery. To generate these images, the analytical hillshading method was applied, a widely used approach for simulating topographic illumination. This method is based on varying the elevation and azimuth of the light source, which simulates different directions of artificial illumination and thereby enhances the visibility of surface morphostructural elements [12,36,76]. This approach is essential for the automated extraction and analysis of geological lineaments and involves selecting azimuthal and elevation angles that optimize the distinction between shaded and illuminated areas [36]. Using a solar illumination angle of 30°, four shaded relief images were created with four contrasting illumination directions (0°, 45°, 90°, and 135°) (Figure 5). According to Bentahar et al. [77], boundaries between shaded and unshaded areas of the Earth’s surface indicate the presence of lineaments, which require subsequent verification [37].

• Stage 4. Automated Lineament Extraction (PCI Geomatica 2018, LINE Module)

Lineaments were extracted using an automated method (LINE module, PCI Geomatica) and refined based on geological maps. The LINE algorithm was applied with specific parameter settings. This module supports multiple satellite datasets, facilitating the extraction of lineaments from both optical and radar imagery. Its adaptability allows for more detailed structural analysis by combining datasets with varying spectral and spatial characteristics. By providing an automated and efficient approach to lineament detection, the LINE module serves as a reliable tool for geological and geomorphological studies. Studies indicate that the filter radius (RADI) should not exceed 10 pixels, enabling effective discrimination between significant and insignificant geological features. Additionally, the gradient threshold (GTHR), which defines the contrast between a pixel and its surroundings, is recommended to be set between 10 and 70 pixels for optimal lineament detection [27]. The contour detection process often employs edge-filtering techniques to enhance pixel-level spectral contrast. For line detection, the curve length threshold (LTHR) is typically set to 20 pixels, ensuring that only significant linear structures are extracted while minimizing noise. The fit threshold (FTHR), which controls the allowed deviation when approximating a curve, is usually maintained in the range of 2–5 pixels to balance accuracy and computational efficiency. Furthermore, the angular difference threshold (ATHR), defining the maximum angle at which two polylines can be joined, should be within 3–20° to account for natural variations in lineament orientations. The linking distance threshold (DTHR), indicating the maximum permissible distance between two polylines for connectivity, is recommended in the range of 10–45 pixels to ensure reliable structural continuity of the extracted features [27].

These parameter ranges are widely applied in remote sensing and geospatial analysis for the automated extraction of lineaments, in accordance with best practices reported in previous studies [78,79,80,81,82].

To select the optimal input data and LINE parameters, several lineament extraction maps were generated using various threshold values and combinations of parameters. The test ranges were set within limits consistent with the primary and validation literature on LINE, adapted to the conditions of the arid plain-plateau terrain: RADI 5–10 pixels, GTHR 40–70, LTHR 20–30 pixels, FTHR 2–5 pixels, ATHR 5–15°, and DTHR 10–30 pixels. For each combination, a quantitative and qualitative assessment of the results was carried out, taking into account the spatial distribution and orientation of the identified lineaments, as well as their consistency with digitized faults and the axial lines of karst depressions. The final analysis included a set of parameters that ensured the most reliable identification of geologically valid linear structures with a minimum number of pseudo-lineaments (

Table 2. Used parameter values for LINE.

Parameter	Unit	Used Value
RADI	Pixel	10
GTHR	Unitless	60
LTHR	Pixel	25
FTHR	Pixel	3
ATHR	Degrees	10
DTHR	Pixel	20

).

• Stage 5. Post-Processing and Filtering

Following the automatic extraction of linear features using Landsat-8 OLI (PCA) data and WorldDEM4Ortho DEM derivatives, a working layer was generated and subjected to multi-stage filtering. Post-processing is a crucial stage in the extraction of linear features, involving their verification and validation. In the first stage, geometric duplicates and repetitive segments, which arose from extraction using different DEM derivatives, were systematically removed. The second stage involved the systematic exclusion of non-tectonic linear features using the following hierarchical procedure applied consistently across the entire dataset. First, a GIS masking step was performed: lineaments were overlaid with vector layers of roads, tracks, canals, and pipelines obtained from OpenStreetMap and official topographic datasets. All lineaments spatially coinciding with these layers (within a 30 m buffer corresponding to the spatial resolution of the Landsat-8 OLI imagery) were flagged and removed. Second, the remaining features were visually verified against the ESRI World Imagery base map to identify and eliminate additional anthropogenic structures (building edges, agricultural field boundaries) not captured by the vector masks. Third, lineaments associated with drainage, erosion, or lithological–geomorphological boundaries were identified by cross-referencing with the DEM-derived hydrographic network (stream channels and dry riverbeds) and with the boundaries of lithological units; features spatially coinciding with these elements and lacking corroboration in the PCA-derived dataset were excluded. The remaining features were then manually verified for overall spatial consistency with the vector fault layer. Only this systematically cleaned layer was subsequently included in the analysis.

• Stage 6. Geometric Segmentation and Metric Characterization

At the next stage, using ArcGIS 10.x software, composite linear features were segmented into individual segments at their vertices. The resulting vector polyline elements were then analyzed to determine their metric and statistical characteristics—including length, minimum and maximum values, mean, and standard deviation—using the Geometry Calculation tool [66]. To visualize the dominant orientations of lineaments, the vector data were converted to AutoCAD 2023 DXF format and imported into RockWorks 17. Following successful lineament extraction, rose diagrams and lineament density maps were generated using both datasets, along with digitized faults from geological maps.

• Stage 7. Validation

Validation of the extracted lineaments was performed by comparing their orientations with those of existing faults and the directions of karst depressions. Additionally, validation was conducted through interpretation of the lineament density maps in relation to mapped faults. In lineament studies, lineament density and the orientation of existing faults in the study area are widely used parameters for validating lineaments automatically extracted from remote sensing data [7,12,27,29].

The integration of spectral (Landsat-8 OLI/PCA) and morphometric (DEM WorldDEM4Ortho/shaded-relief) lineament sets produced a consolidated structural dataset validated at two levels: (i) directional consistency with digitized geological faults and (ii) spatial correspondence with the axial orientations of karst depressions identified through DEM-based depression modelling. This dual-source, dual-validation design minimizes the inherent limitations of each individual data source—notably, the susceptibility of optical imagery to atmospheric and illumination artefacts and the dependency of DEM-based extraction on relief expression—while maximizing the reliability of the final structural interpretation.

4. Results

4.1. Results of PCA Analysis of Landsat-8 OLI Data and Automatic Lineament Extraction

The PCA method, applied to the seven multispectral channels of Landsat-8 OLI, effectively highlighted spectral contrasts associated with lithological boundaries and zones of tectonic fracturing. The first principal component (PC1) accounted for 89.83% of the total variance in the raw data, and the second (PC2) for 9.05%; their combined contribution amounted to 98.88%, which corresponds to the 85–97% range typical for PCA processing of Landsat-8 OLI in arid regions with sparse vegetation cover. Components PC3–PC7 collectively account for less than 1.12% of the variance and do not carry significant structural information; therefore, interpretation and visualization were focused on PC1 and PC2.

Based on the results of automatic extraction using Landsat-8 OLI data (PCA method), 18,977 linear features with a total length of 21,525 km were identified. The average length of a single lineament was 1.13 km with a standard deviation of 0.63 km, a minimum value of 0.75 km, and a maximum of 22.6 km. Using WorldDEM4Ortho (Hillshade method), 35 629 lineaments were obtained with a total length of 34 930 km, an average length of 0.98 km, and a similar standard deviation of 0.63 km (minimum—0.59 km, maximum—19.5 km). Overall, the DEM method identified 1.88 times more linear features, with an average length 15% shorter than that of the PCA dataset (

Table 3. Lineament statistics.

Data	Number	Minimum, km	Maximum, km	Sum, km	Medium, km	Standard Deviation
PCA	18,977	0.75	22.6	21,525.3	1.13	0.63
DEM (Hillshade)	35,629	0.59	19.5	34,930.8	0.98	0.63

).

This relationship has a methodological basis. Shaded relief is sensitive to short morphological features—the edges of terraces, linear thalwegs, and the contours of denudation basins—which are abundant in the arid platform relief of the Mangystau Region. In contrast, the PCA method primarily detects extensive spectral contrasts at the boundaries of lithological bodies and zones of tectonic weathering. A similar pattern has been described in comparative studies of the mountainous regions of Morocco [12,36].

4.2. Orientations of Lineaments, Faults, and Karst-Suffosion Depressions

4.2.1. Lineaments Identified by PCA (Landsat-8 OLI)

The rose diagram of lineaments identified by PCA using Landsat-8 OLI data reflects a polymodal distribution of orientations with two dominant systems. The sub-latitudinal system (E-W, 80-100°) is characterized by the highest frequency and forms well-defined symmetrical lobes at 90° and 270°. The north-western system (NW-SE, 295-325°) appears as a stable lobe in the north-western sector and, despite its lower frequency, corresponds to the most extensive linear structures. The submeridional (N-S, 350-10°) and north-eastern (NE-SW, 40-60°) systems are identified as secondary trends (Figure 6a).

4.2.2. Lineaments Based on DEM (Hillshade)

The rose diagram of DEM lineaments generally reproduces the same dominant orientations but is more diffuse: the lobes are distributed more evenly around the azimuthal circle due to the inclusion of numerous short morphological lineaments of various directions. The sub-latitudinal system (E-W) retains its dominant position, while the north-western system (NW-SE, 295–320°) is clearly expressed and corresponds to the strike of the region’s main structural elements (Figure 6b). The lobes of the DEM diagram in the sub-latitudinal sector are somewhat wider compared to the PCA, which is explained by the increased sensitivity of the shading method to sub-horizontal plateaus and the edges of chinks.

4.2.3. Mapped Faults

The rose diagram of faults digitized from 1:200,000 scale geological maps show a distinct bimodal structure. The north-western system (NW-SE, approximately 295-330°) is dominant in terms of total length. The submeridional system (N-S, 0-15°) is represented as a secondary trend, more pronounced on the fault map than on the lineament diagrams. It is likely that submeridional transverse faults are reliably mapped during geological surveys but are less well captured by automated algorithms. The sublatitudinal system (E-W) is represented on the fault diagram to a smaller extent than on the lineament diagrams, indicating the predominantly morpholithological (rather than tectonic) nature of a significant portion of the sublatitudinal features in the satellite data (Figure 6c).

The alignment of the dominant peak of the fault diagram (NW–SE) with similar peaks in the PCA and DEM diagrams confirms that the north-western lineament system, as derived from remote sensing data, largely reflects actual tectonic disturbances. The divergence in the sublatitudinal sector is due to the contribution of morpholithological boundaries and highlights the need for expert post-processing of automatically identified lineaments.

4.2.4. Axis Lines of Karst-Suffosion Depressions

A rose diagram of the axis lines (extensions) of 365 karst-suffosion basins with a depth of more than 5 m and an area exceeding 0.5 km² reveals a clear dominance of the NW-SE direction (290-330°), coinciding with the main tectonic maximum of faults and lineaments. A secondary maximum is observed in the sub-latitudinal direction (E-W, 80-100°), corresponding to the second leading lineament system (Figure 6d). The correspondence between the orientations of the long axes of the depressions and the dominant lineament systems indicates that structural control extends not only to spatial distribution but also to the morphology of karst features: most elongated depressions inherit the geometry of fault-fracture zones, which act as preferred pathways for the circulation and dissolving action of groundwater in carbonate and gypsum-bearing strata.

4.3. Spatial Distribution of Lineament Density

4.3.1. Map of Lineament Density Based on DEM (Hillshade)

The map of lineament density, calculated using DEM derivatives (maximum value 1.47 m/km²), shows marked spatial heterogeneity. Areas of maximum density (exceeding 0.86 m/km²) form an elongated NW–SE belt covering the central part of the Mangystau Peninsula (Figure 7a). This belt spatially coincides with the zone of development of the Central Mangystau anticlines and the South Mangystau deflection, where Neogene carbonate and gypsum-bearing deposits lie subhorizontally beneath a thin Quaternary cover. Moderately elevated density is recorded in the northern part of the territory, associated with the fault structures of the North-Ustirt deflection system. The peripheral areas of the territory – the Caspian Plain and the eastern regions of Ustirt – are characterized by lower density values, reflecting the weak structural disturbance of the young Quaternary deposits.

Overlaying the vector of mapped faults onto the density map shows that zones of maximum concentration of DEM lineaments generally coincide spatially with the zones of mapped faults or are directly adjacent to them. This spatial coincidence confirms that the main contribution to the formation of density anomalies is made by zones of tectonic fracturing, rather than random morphological artifacts.

4.3.2. PCA-Based Lineament Density Map (Landsat-8 OLI)

The PCA lineament density map, with a maximum value of 0.44 m/km², reproduces the same general spatial structure but with a number of diagnostically significant differences. The absolute density values are consistently lower, which directly reflects the smaller number of PCA lineaments combined with their greater average length (Figure 7b). The nature of the spatial distribution is more discrete: anomalies are point-like and are confined mainly to narrow bands along fault lines and lithological contacts, whereas the DEM map forms a more extensive continuous background of elevated values. Areas of highest density of PCA-lineaments more closely ‘trace’ fault lines, which makes them particularly valuable for identifying lineaments on geological maps.

4.3.3. Fault Density Map

The fault density map of the Mangystau Region shows spatial heterogeneity, with a maximum density of 0.14 faults per km². The areas of highest values form two distinct clusters: the first and most intense is confined to the western part of the Mangystau Peninsula, where there is a high concentration of north-west trending faults; the second cluster is located in the central and eastern parts of the territory within the Zhanaozen–Kendirlisor belt, extending as an elongated tongue of increased density with a NW-SE strike. The northern part of the territory and peripheral areas are characterized by low background density values, reflecting the low density of mapped faults. The spatial structure of the map confirms that zones of increased tectonic disturbance are confined to the axial parts of the Central Mangystau Uplift and the South Mangystau deflection, which is consistent with the distribution of karst-suffosion depressions within the study area.

Comparison of lineament density maps with models of subsurface depressions (deeper than 5 m) revealed a strong spatial correlation between linear structures and areas of karst-suffosion development (Figure 8). Along the north-west-oriented zones, depressions tend to cluster, suggesting the influence of tectonic factors on the formation of present-day depressions. These fault-fracture zones likely function as discharge channels for groundwater and pathways for material removal, promoting the formation of karst and suffosion depressions within carbonate and gypsum-bearing strata.

4.4. Analysis of the Spatial Relationship Between Lineaments and Karst-Suffosion Depressions

4.4.1. Density Overlay Analysis

A density overlay analysis was performed to quantitatively assess the spatial association between lineaments and karst-suffosion depressions. The lineament density rasters (Hillshade and PCA) were reclassified using the Jenks Natural Breaks method into five classes: Very Low, Low, Moderate, High, and Very High. For each class, the number of karst-suffosion depressions falling within the corresponding zone was calculated. The results are presented in

Table 4. Distribution of karst-suffosion depressions by lineament density class.

Density Class	Hillshade		PCA
	Number of Depressions	%	Number of Depressions	%
Very low (0–0.075)	205	56.2%	138	37.8
Low (0.076–0.242)	119	32.6%	189	51.8
Moderate (0.243–0.489)	28	7.7%	38	10.4
High (0.49–0.863)	11	3.0%	-
Very high (0.864–1.468)	2	0.5%	-

.

An analysis of the data in Table 4 reveals a significant difference in the distribution of depressions between the two sets of lineaments. According to the Hillshade data, the vast majority of depressions (56.2%) are concentrated in the very-low-density zone, with a further 32.6% in the low-density zone. Only 11.2% of depressions fall within the medium-, high-, and very-high-density classes.

According to the PCA data, the picture is fundamentally different. The vast majority of depressions are distributed between the low- (51.8%) and very-low-(37.8%) density classes, while 10.4% of depressions are concentrated in the moderate-density class. The ‘high’ and ‘very high’ density classes are absent below the lower absolute maximum density of the PCA raster (0.44 m/km²). Nevertheless, the PCA data show a higher proportion of depressions in the moderate-density zones (10.4% compared to 7.7% for Hillshade), indicating a closer association of PCA lineaments with sites of karst-suffosion processes at the level of average density values.

4.4.2. Near Analysis

The Near Analysis enabled us to quantitatively assess the distance from each of the 365 karst-suffosion depressions to the nearest lineament for both datasets—Hillshade and PCA. The results are presented in

Table 5. Distribution of karst-suffosion depressions by distance to the nearest lineament.

Density Class	Hillshade		PCA
	Number of Depressions	%	Number of Depressions	%
up to 1 km	17	4.7%	134	36.7%
up to 3 km	78	21.4%	243	66.6%
up to 5 km	108	29.6%	307	84.1%

.

The results of the Near Analysis reveal a fundamental difference between the two methods, which has direct diagnostic significance. For the set of PCA lineaments: 36.7% of all depressions (134 out of 365) are located within a 1 km radius of the nearest PCA lineament; 66.6% of depressions (243 out of 365) are within a 3 km radius; 84.1% (307 out of 365) are within a 5 km radius.

For the set of DEM lineaments (Hillshade), the corresponding figures are significantly lower: only 4.7% of depressions (17 out of 365) are within a 1 km radius; 21.4% (78 out of 365) within a 3 km radius; and 29.6% (108 out of 365).

At first glance, the Near Analysis data may be interpreted as evidence of a weak spatial relationship between DEM lineaments and depressions. However, this apparent contradiction is explained by the structural characteristics of the two datasets. The set of Hillshade lineaments contains 35,629 elements with a total length of 34,930 km, uniformly covering virtually the entire study area as a result, the distance from any point to the nearest DEM lineament is generally small (less than 1 km).

The fact that 84.1% of depressions are located within 5 km of the nearest PCA lineament, whereas the PCA dataset covers the area much less densely than the DEM dataset, indicates a statistically significant and geologically meaningful spatial association between karst-suffosion depressions and tectonically controlled lineaments. A similar result, 62.5% of depressions in close spatial association with lineaments, was obtained in a study of the Ziria karst system (Peloponnese, Greece) [83].

4.4.3. Regression Analysis

The regression analysis was performed using a regular 20 km × 20 km grid covering the entire study area. For each cell, the average lineament density (m/km²) was determined using the relevant raster (Hillshade, PCA, and the density of mapped faults)—the Zonal Statistics as Table tool, ArcGIS Pro; and the number of karst-suffosion depressions whose centroids fall within that cell—the Spatial Join tool. The resulting pairs of values were used to construct scatter plots and calculate linear regression models of the form y = a·x + b, where y is the number of depressions in the cell, and x is the average density in the cell.

The coefficients of determination were as follows: for mapped faults—R² = 0.0519; for DEM (Hillshade) lineaments—R² = 0.0011; for PCA lineaments—R² = 2·10⁻⁸. All three values are close to zero, indicating the absence of a significant linear relationship between the average density of structural elements in a cell and the number of depressions.

The structural control of karst-suffosion processes is focal rather than diffuse in nature and cannot be described by a simple linear model at the regional level of aggregation. The identification of non-linear spatial patterns requires the use of spatially weighted regression (GWR) or probabilistic models (frequency ratio, weights of evidence), which may form the subject of further research.

4.5. Comparative Assessment of the Effectiveness of the PCA and Hillshade Methods

A comparative analysis of the results obtained using the two methods for lineament extraction allows for a differentiated methodological assessment of their applicability to tasks of structural-geological interpretation and karst geomorphological forecasting.

The PCA method (Landsat-8 OLI) provides fewer numbers but longer average length of lineaments (1.13 km versus 0.98 km for DEM), which corresponds to its ability to identify extensive first-order structures. More precise spatial correlation with mapped faults (narrow density anomalies along fault lines), a smaller contribution from morphological features of non-tectonic origin, and higher association indices with karst-suffosion depressions in the proximity analysis (84.1% of depressions within a 5 km radius) confirm the preference for this method in regional tectonic interpretation. The low R² value in the regression (2·10⁻⁸) is consistent with the conclusion that PCA lineaments mark specific structural zones but do not form a diffuse high-density field that would correlate with the number of depressions at the 400 km² grid cell level. The effectiveness of the method is further enhanced by the arid climatic conditions of the Mangystau region: minimal vegetation cover ensures a clear lithological-structural spectral signal.

The Hillshade method (WorldDEM4Ortho DTM) offers additional advantages: a significantly greater number of identified lineaments (35,629 vs. 18,977), covering a wide range of morpholineaments – short fracture systems, erosional terraces, and contours of suffosion subsidence. The high density of the lineament network is potentially informative for detailed morphostructural mapping and the assessment of neotectonic deformations of the relief. At the same time, it is precisely the high density of this network that reduces the method’s discriminatory power in proximity analysis: the majority of points in the area, including those randomly located, would fall within a few kilometers of some DEM lineament. The low R² in the regression (0.0011) further indicates that the DEM dataset, which includes a wide range of morphological features, has no explanatory power for predicting the location of depressions within the linear model.

Therefore, both methods should be regarded as hierarchically complementary tools: the PCA dataset is preferable for regional tectonic interpretation and validation against geological maps, whereas the DEM set is preferable for detailed morphostructural mapping and local prediction of karst and suffosion hazards. This conclusion is consistent with the methodological recommendations of a range of comparative studies [12,36].

4.6. Summary Interpretation of the Results and the Nature of Structural Control over Karst-Suffosion Processes

A comprehensive analysis of lineaments, mapped faults, and karst-suffosion depressions shows that the structural control of karst formation within the Mangystau Region is distinct but heterogeneous. The coincidence of the dominant orientations of lineaments, faults, and the elongations of karst depressions indicates that the development of depressions is governed by the general tectonic organization of the territory. The north-western system of faults, corresponding to the main structural trend of the region, plays the most significant role.

The results obtained suggest that fault-fracture zones act as channels of increased permeability, through which groundwater circulates, and the dissolution of carbonate and gypsum-bearing rocks is intensified. However, tectonic control manifests itself not as a uniform relationship between the density of lineaments and the number of depressions but through the localized influence of individual structural zones and their intersection points. It is precisely such areas that are likely to be characterized by the highest intensity of infiltration and karst-suffosion processes.

A comparison of the PCA and Hillshade methods reveals their different but complementary informative value. PCA lineaments predominantly reflect regional tectonic structures and are better consistent with mapped faults, whereas DEM-derived features capture a wide range of local morphostructural elements, including erosional and denudational forms. Consequently, the PCA dataset is more effective for identifying the structural framework of karst formation, while the Hillshade dataset is more effective for detailing the morphostructural heterogeneity of the area.

The absence of a clear linear correlation between the average density of structural elements and the number of depressions indicates the limitations of simple regression models in the analysis of karst-suffosion processes. The formation of depressions is determined by a combination of several factors, including lithology, hydrogeological conditions, the thickness of the Quaternary cover, and local fracturing. Consequently, tectonic structures should be regarded as necessary but not the sole condition for karst development.

Overall, the results of the study confirm that karst-suffosion processes in the Mangystau region are structurally determined, and the most promising areas for the development of depressions are the intersections of northwest- and sublatitudinal-trending fault systems. For further verification of the identified patterns, it is advisable to apply multi-factor spatial models and field geophysical surveys within structural nodes.

5. Discussion

The integrated approach applied in this study, combining spectral analysis (PCA of Landsat-8 OLI data) and morphometric analysis (hillshading of the WorldDEM4Ortho DEM), has demonstrated high effectiveness for lineament mapping under the arid conditions of the Mangystau region. The similarity of dominant orientations (NW-SE and E-W) identified from both data types confirms the tectonic origin of most linear structures and helps minimize the subjectivity inherent in visual interpretation. At the same time, the observed differences in the number and average length of lineaments (18,977 features from PCA versus 35,629 from the DEM) have a methodological explanation. PCA-derived data primarily capture surface spectral contrasts associated with lithological boundaries, weathering zones, and vegetation cover [37,43]. In contrast, shaded relief generated from the high-accuracy WorldDEM4Ortho DEM is more sensitive to terrain morphology, revealing both pronounced scarps and subtle linear relief elements that may remain undetected in optical imagery [78]. Thus, the DEM-based method proved more productive for identifying morphostructural lineaments, which is consistent with studies emphasizing the key role of geomorphometric analysis in structural geology [36,68].

Structural-tectonic interpretation of the lineament network. The identified lineament network, dominated by NW-SE (290-320°) and E-W (90-270°) trends, reflects the principal stages of the tectonic evolution of the Mangystau region. The northwestern orientation is inherited in nature and is likely associated with the strike of folded structures and basement faults formed during the Hercynian stage of tectogenesis [57]. The widely developed eastern component may be interpreted as the result of later neotectonic movements characteristic of the Turan Plateau or as a system of transverse and oblique faults cutting across the major structural elements. The spatial correspondence between zones of high lineament density and the boundaries of major tectonic domains (Central Mangyshlak and Southern Mangyshlak), as well as contacts between different lithological complexes (carbonate and terrigenous deposits), indicates their deep structural control. These zones represent weakened segments of the Earth’s crust that govern not only the distribution of sedimentary basins but also contemporary hydrogeodynamic processes [7,16].

Role of lineaments in the development of karst-suffosion processes.

The region’s current structure has been shaped by a multi-stage tectonic evolution. The division into structural stages allowed the identified lineaments to be correlated with deformation phases of different ages. The most pronounced linear structures, recorded within the lineament field, correlate with zones of faulting formed during the Palaeozoic stage and preserved through subsequent phases of tectonic activity.

The distribution and development of karst features on the plateau surface are primarily determined by the structural plan of the territory. Certain patterns in the density, depth, and morphology of karst features are identified, reflecting the neotectonic activity of local and larger structures in the study area. It has been observed that within the South Mangyshlak deflection, the density of karst features and their depth increase significantly in areas of local anticline uplifts, characterized by increased rock fracturing. The fracture permeability of carbonate rocks in sections of brachyanticline structures is twice the normal value. Fracturing, however, increases if the structures are characterized by long-term development, including in recent times.

Two chains of large, dry depressions are associated with the areas of local uplift in the South Mangyshlak depression: the northern chain: Karamandybas, Assar, Korganoy, Ozen, Tungrakshyn, and the southern chain: Koshkarata, Karakiya, Ashchysor, Kauyndy, and Zhazygurly-Basygurly. These basins are characterized by well-defined shapes, significant depth, and steep, often stepped slopes. The presence of such depressions on the plateau indicates the existence of local structures that are reflected in the relief. They are characterized by shallow bedrock and large amplitudes and have experienced the most intense recent positive movements [84]. Shallow karst valleys (10–15 m) and corrosion sinkholes (2–5 m deep) are associated with the crests and, less frequently, the flanks of local anticline structures that have suffered less active recent movements and are directly reflected in the relief as ridges 10–25 m high. Often, corrosion karst sinkholes 1–5 m deep are arranged in a circle, outlining a group of isolated ridges that reflect in the relief local anticline structures that have experienced insignificant positive movements in recent times. In such cases, the plane view of karst sinkholes, sometimes connected by dry ravines, also allows the delineation of areas of local uplift.

The long axes of individual karst depressions and sinkholes often reflect the extension of persistent tectonic fractures. The linear arrangement of karst features reveals flexures and disjunctive faults in the relief [84,85].

The close spatial correlation between the increased density of north-west trending lineaments and concentrations of karst-suffosion depressions deeper than 5 m and 10 m is a key finding of the study. Fault-fracture zones act as the main pathways for the infiltration of atmospheric precipitation and the migration of groundwater in arid climates [17]. Within the Mesozoic–Cenozoic carbonate and gypsum-bearing strata that constitute the region, groundwater circulating along fractures dissolves the rock, leading to the formation of subsurface conduits and subsequent roof collapse. Increased suffosion activity is also associated with these zones, where fine-grained material is being removed from the overlying terrigenous deposits. Thus, lineaments not only indicate areas of fracturing but may also function as active hydrogeological conduits, determining the location and intensity of destructive processes [9,16,26].

Study limitations and prospects for further research. Despite the high efficiency of automated methods, the study has several limitations. Firstly, some of the extracted linear features, particularly those derived from PCA, may be of anthropogenic (roads, ditches) or lithological-geomorphological origin (coastlines, vegetation boundaries) rather than tectonic [12]. Secondly, the spatial resolution of the data used (30 m for Landsat imagery and approximately 24 m for the DEM) sets a threshold that does not allow the detection of small but potentially significant fracture systems. To overcome these limitations, it is necessary to perform field verification with GPS-based georeferencing of key lineaments and karst features; to use high-resolution remote sensing data, such as WorldView and Pleiades imagery, as well as radar interferometry for monitoring contemporary fault movements and subsidence above karst cavities [24,29]; and to integrate with geophysical data (electromagnetic sounding, seismic surveys), which would allow assessment of the depth extent of the identified structures and their relationship to aquifers [10,25].

Lineaments automatically identified using Landsat-8 OLI data and WorldDEM derivatives should be considered not as a direct representation of all tectonic faults but as a set of probable linear morphostructural indicators that are sensitive to the spatial resolution of the source data, the quality of the DEM, the direction of pseudo-illumination, the parameters of the extraction algorithm, and subsequent expert filtering. Comparative studies show that different sensors and different DEM derivatives can lead to differences in the number, length, and predominant orientations of automatically extracted features, whereas the Hillshade raster can introduce an azimuthal bias favoring structures perpendicular to the illumination direction [12,36,86,87,88]. Therefore, we interpret the resulting network of lineaments as a representative but not exhaustive set of structural features, suitable primarily for the analysis of regional spatial patterns.

The spatial relationship between lineaments and karst depressions indicates a probable structural control over the development of some negative landforms but in itself is not unequivocal evidence of their genetic connection to faults. The final result is influenced not only by the stability of the lineament network but also by the method of identifying the depressions themselves using DEM, which also depends on the type of DEM and the thresholds used.

In this regard, the conclusions drawn should be understood as probabilistic: they confirm the existence of a statistically and geologically meaningful spatial relationship and do not exclude the influence of lithology, the drainage network, sediment thickness, and local morphogenetic factors. Promising directions for further work include sensitivity tests of automatic feature extraction parameters, multi-azimuthal analysis of DEM derivatives, and a quantitative assessment of the uncertainty in the spatial relationship between “lineament systems and karst depressions” [57,81,89].

Comparison of the results of lineament analysis with additional geomorphological and structural methods. The lineaments obtained in this study should be considered as remotely identified linear morphostructural elements, some of which may correspond to faults, zones of increased fracturing, and other structural heterogeneities, while others may reflect lithological boundaries, drainage features, or relief [90,91]. In this sense, our approach based on remote sensing and GIS is an effective tool for the initial regional identification of structural elements, but its interpretive reliability increases when combined with additional geomorphological and structural methods [81,92].

First and most importantly, the results of lineament mapping can be correlated with DEM derivatives such as slope, exposition, various types of curvature, and topographic position index (TPI), as these parameters allow for better identification of linear relief inflections, linear elements of the drainage network, and zones potentially associated with tectonic faults [39,93]. Additional independent information is provided by morphotectonic indices and hypsometric characteristics of the relief, allowing for the analysis not only of the geometry and spatial organization of lineaments but also of their relationship to the relative activity of modern relief-forming processes [39,94].

A significantly more reliable geological interpretation is achieved by comparing remotely derived lineaments with mapped faults and field structural measurements. At the local scale, high-resolution digital elevation models (DEMs) derived from LiDAR or UAV photogrammetry allow for the refinement of structural geometry, while the use of ground-penetrating radar (GPR) and electrical resistivity tomography (ERT) at deeper levels helps verify their subsurface continuity and depth-related characteristics [95,96,97].

Our results should be interpreted as a basic framework for the structural organization of the territory, which can be progressively refined through geomorphometric analysis, correlation with known faults, selective field verification, and geophysical surveys. Such a complex approach appears to be particularly important for further discussion of the possible structural causes of karst and suffosion processes in the region under study [90,91,96].

According to the reviewed literature, the structural patterns identified in the Mangystau region generally align with findings from other karst terrains. In these regions, tectonic faults and fracture systems play a major role in karst development and groundwater circulation. In karstic regions such as Iran, China, the USA, Puerto Rico, Russia, and Greece [1,19,22,25,38,41,83], the predominant orientations of lineaments are typically associated with regional tectonic stress fields and inherited fault structures. Analogous to these observations, a distinct spatial correlation has been established in Mangystau between zones of high lineament density and the distribution of karst features, underscoring the significant role of structural discontinuities in karst development.

Concurrently, Mangystau possesses several specific characteristics attributed to its unique geological evolution, arid climate, and the presence of hydrocarbon-bearing carbonate sequences. The combination of low precipitation, intense physical weathering processes, and structurally controlled fluid circulation results in karst-suffosion processes being predominantly localized along fault and fracture zones. These characteristics generally conform to broader regional and global trends observed in fractured carbonate massifs, thereby confirming the applicability of remote sensing and lineament analysis methods for studying karst in similar geological settings.

Practical significance of the results. The results obtained have a wide range of practical applications in various fields of geosciences and natural resource management. In engineering geology, the analysis of lineaments can be used to identify zones of increased fracturing and faulting, characterized by reduced rock strength and increased permeability. Such zones pose a potential hazard during the design and construction of linear and area-based structures (motorways and railways, pipelines, power lines, industrial facilities), as they may be associated with areas of karst sinkholes, suffosion deformations, and uneven ground subsidence. The use of maps showing the density and orientation of lineaments enables preliminary engineering–geological zoning of the territory and optimizes the selection of routes and construction sites.

In hydrogeology, lineaments are regarded as indicators of zones of increased permeability associated with the development of fracture-fault systems. Zones where lineaments of different orientations intersect may be considered as promising sites for the location of water supply wells. Lineament analysis can be effectively used in the search for and exploration of groundwater, particularly in arid regions where water resources are limited and their location is of critical importance.

The results of this study have direct practical relevance for the sustainable development of the Mangystau region. Lineament density and orientation maps, overlaid with karst depression layers, can serve as a basis for engineering–geological zoning in the planning of linear infrastructure (pipelines, roads, power transmission lines) and industrial facilities geological hazard assessment, including the prediction of sinkholes and ground subsidence; groundwater exploration, and prospecting, since lineament intersection zones are often associated with areas of increased aquifer productivity.

Thus, the integrated approach based on remote sensing data has proven its effectiveness in identifying the tectonic structures and assessing their role in the development of karst-suffosion processes, thereby providing a scientific basis for decision-making in territorial planning and economic activities.

6. Conclusions

The novelty of the research lies in the development and implementation of an integrated approach to the structural-tectonic interpretation of lineaments based on the combined use of Landsat-8 OLI multispectral data and Airbus WorldDEM4Ortho digital elevation model for the Mangystau region. For the first time in this area, a comprehensive automated extraction of linear structures has been conducted, followed by statistical, morphometric, and spatial analysis of their role in the development of karst-suffosion processes.

The work implements a quantitative assessment of the parameters of the lineament network (number of objects, total length, average length, density, and orientation), which has enabled a transition from qualitative interpretation to reproducible analysis of the structural organization of the territory. The comparison of results obtained from PCA data and shaded relief models allowed for the differentiation of spectrally conditioned and morphostructural lineaments and provided a basis for their tectonic origin.

For the first time in the Mangystau region, a statistically and spatially confirmed relationship has been established between zones of increased linear element density and areas of development of karst-suffosion depressions of varying depths. It has been demonstrated that fault-fracture zones of northwestern and sub-latitudinal orientation form the structural framework of the territory and serve as hydrogeological conduits, controlling infiltration, circulation, and discharge of groundwater under arid climate conditions. A close spatial correlation has been identified between zones of elevated linear element density and regions of karst-suffosion depression development in the relief. This confirms the dominant role of tectonic factors in creating conditions for groundwater circulation, the development of underground cavities, and subsequent surface expression of karst features. Intersections of linear elements of different orientations are considered the most permeable areas, possessing increased hydrogeological potential, as well as heightened engineering and geological risks.

The obtained results indicate that lineament analysis, based on remote sensing data and digital elevation models, is an effective tool for studying the tectonic organization of the territory, assessing hydrogeological conditions, and identifying zones with potential development of hazardous geological processes. The presented approach can be employed in engineering and geological surveys, planning the placement of water intake facilities, evaluating karst risks, and developing measures for sustainable territorial development.

Future research could be directed towards integrating SAR radar data, detailed geophysical materials, and field structural observations, which would refine the geodynamic interpretation of the identified lineament systems and enhance the predictive value of the results.