Next Article in Journal
Controlling Product Properties in Forming Processes Using Reinforcement Learning—An Application to V-Die Bending
Previous Article in Journal
Lightweight Insulating Geopolymer/Phase-Change Materials Applied Using an Innovative Spray Method
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Applied Sciences
Volume 15
Issue 10
10.3390/app15105482
applsci-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Monitoring and Analysis of Slope Geological Hazards Based on UAV Images

by
Nan Li
1,
Huanxiang Qiu
1,
Hu Zhai
2,
Yuhui Chen
1 and
Jipeng Wang
1,*
1
School of Civil Engineering, Shandong University, Jinan 250061, China
2
Shandong Bowee Vision Information Technology Co., Ltd., Jinan 250013, China
*
Author to whom correspondence should be addressed.
Appl. Sci. 2025, 15(10), 5482; https://doi.org/10.3390/app15105482
Submission received: 17 March 2025 / Revised: 8 May 2025 / Accepted: 11 May 2025 / Published: 14 May 2025
Browse Figures
Versions Notes

Abstract

Slope-related geological disasters occur frequently in various countries, posing significant threats to surrounding infrastructure, ecosystems, and human lives and property. Traditional manual monitoring methods for slope hazards are inefficient and have limited coverage. To enhance the monitoring and analysis of geological hazards, a study was conducted on the legacy slopes of an abandoned quarry in Jinan, Shandong Province, China. High-resolution images of the slopes were captured using unmanned aerial vehicle (UAV) phase tilt photogrammetry, and three-dimensional models were subsequently constructed. Software tools, including LiDAR360 5.2 and ArcMap 10.8, were employed to extract slope geological information, identify disaster-prone areas, and conduct stability analyses. The Analytic Hierarchy Process (AHP) was employed to further evaluate the stability of hazardous slopes. The results reveal the presence of two geohazard-prone areas in the study area. Geological analysis shows that both areas exhibit instability, with a high susceptibility to small-scale rockfalls and landslides. The integration of UAV remote sensing technology with AHP represents a novel approach, and the combination of multiple analytical methods enhances the accuracy of slope stability assessments.

1. Introduction

Common geological hazards on slopes, such as landslides and avalanches, are sudden, posing significant threats to nearby infrastructure, human lives, and property. Moreover, the prevention and response to these hazards are inherently difficult [1,2,3]. The slope, as a common structural form, plays a critical role in ensuring the overall stability of complex rock and soil structures. Therefore, slope monitoring is essential for safeguarding safety and mitigating risks. With the continuous advancement of technology, traditional monitoring methods can no longer meet industry needs, and their limitations have become increasingly evident. As a result, improving the sophistication of monitoring technologies is essential.
Small low-altitude unmanned aerial vehicle (UAV) photogrammetry is an innovative measurement method, known for its convenience, speed, efficiency, and high precision. It is widely applied in engineering construction, disaster prevention and mitigation, and operations, achieving favorable outcomes [4,5,6,7]. An et al. [8] proposes a monitoring method for the ground subsidence of high-intensity mining with Unmanned Aerial Vehicle LiDAR (UAV-LiDAR) measurement technology. Qiu et al. [9] utilized a UAV equipped with a high-resolution camera to capture images in a “high-density grid imaging” mode, enabling three-dimensional modeling of the study area for geological investigation and risk assessment. Wei et al. [10] summarized and clarified the process of UAV highway aerial photography data acquisition based on years of practice, providing a theoretical foundation for solving UAV operational issues. Pi et al. [11] focused on the Tianxi Expressway, the largest traffic hub, employing UAV technology to identify, measure, and analyze three types of high slopes. Their research provides intuitive, accurate, and timely data support for slope monitoring in the area. Their work plays a vital role in enhancing the accuracy and efficiency of land surveying and mapping, providing a scientific basis for land planning, resource management, and environmental protection. Zolkepli et al. [12] investigated the application of UAV-based photogrammetry for slope mapping and hazard assessment in three high-risk areas at Pahang Matriculation College, Malaysia, and highlighted slope angle analysis as a key criterion for landslide risk evaluation, offering actionable insights for infrastructure planning and disaster prevention.
Slope stability monitoring is crucial for ensuring the safety of engineering projects and the smooth operation of various infrastructures, thereby maximizing their social and economic benefits [13,14,15,16]. Traditional monitoring methods typically involve manually operated equipment such as levels, total stations, and GPS, which are inefficient, labor-intensive, and costly. Consequently, there is a growing consensus within the industry to seek faster and more efficient monitoring solutions for slopes [17,18]. The “sky-earth-time” integrated slope investigation and monitoring system assesses the development characteristics and stability of unstable slopes in the survey area, providing a foundation for the targeted implementation of terrain restoration [19]. Yuan et al. [20] conducted extensive research on intelligent slope monitoring and proposed a monitoring model along with the key data processing elements. Pei et al. [21] identified the key points for intelligently extracting slope features in the study area by investigating the characteristics of three-dimensional point-cloud images, thereby improving the processing of monitoring data. Chen et al. [22] extracted orthophotos and digital surface models of the lagoon slope before and after the landslide using UAV images. They employed COSI-Corr software to calculate the offset information of characteristic points on the landslide surface and determined the vertical displacement, thereby analyzing the motion, material changes, and the formation and evolution processes of the landslide, as well as its damage mechanisms. Kern et al. [23] provided an accurate real-time data processing solution for UAV mapping applications, which supports the widespread adoption and implementation of UAV mapping in practical applications. Sumira et al. [24] highlighted that cloud cover significantly reduces the accuracy of satellite remote sensing, and relevant data cannot be obtained during heavy rainfall. In GIS and cartography, accurate topographic maps are essential. Cirillo et al. [25] demonstrate the application of drone-based high-resolution photogrammetry, using PPK methods to achieve model and rockfall analysis accuracies below 5 cm to assess the geomechanical properties and rockfall potential of rock scarps. It offers a safer, more efficient, and cost-effective alternative to traditional field surveys for analyzing discontinuities, failure mechanisms, and fracturing patterns across large areas. Additionally, susceptibility to rockfall was analyzed using CloudCompare 2.13 and ArcMap 10.8 software. Yu et al. [26] generated digital orthophoto maps, digital elevation models, and three-dimensional geomorphological models based on UAV images. They analyzed the landslide area to identify potential landslides and developed the GeoStudio numerical model to evaluate the seepage field and the stability coefficient change curve of the landslide.
Existing UAV slope monitoring and analysis methods are relatively homogeneous; fewer people use the AHP method for slope assessment. This study combines UAV remote sensing technology with AHP to analyze slope stability and improve the accuracy of slope assessment. This study explores the application of UAV remote sensing technology for identifying and monitoring potential geological hazards on slopes, aiming to develop efficient technical processes for hazard identification, monitoring, and stability analysis [26,27,28]. A slope in Shandong Province, China, was selected for the study, where a UAV-mounted camera was used to capture two sets of high-definition images of the slope. These images were used to construct point-cloud models, three-dimensional real-time models, and digital elevation models. Geological information, including contours, slope gradient, and slope shape, was extracted, and hazardous areas were identified and analyzed for stability. The identification of hazardous areas and the analysis of their stability. This study optimizes and enhances the methods for slope monitoring and stability assessment. This approach will ensure the safety of residents and infrastructure surrounding the slopes while providing technical support for disaster prevention and mitigation efforts in China.

2. Study Area

The study area is situated in Shandong Province, China, and consists of the remaining slope of an abandoned quarry (Figure 1), covering an area of approximately 42,000 m2. The region is characterized by low hills composed of sedimentary rocks, with altitudes primarily ranging from 300 m to 400 m.
Since the 1960s, rapid urban development and the expansion of built-up areas have led to excessive logging and quarrying, causing significant damage to the region’s mountains. Large areas of exposed rock have resulted in weathering and fragmentation, and human engineering activities have triggered rock collapses, landslides, and other geological hazards. The region features complex and steep terrain with significant undulations, compounded by previous quarrying methods, including step blasting. As a result, the mountain slopes are characterized by stepped cliffs. The area has undergone man-made repairs, but the northwest and northeast slopes remain as stepped exposed cliffs, while the south side still features exposed, broken slopes. Several roads pass through the region, with a production road on the eastern side directly adjacent to the slope. The road slopes on both sides exceed 80°, making the area highly susceptible to slope geological hazards, posing a significant threat to public safety, engineering projects, and the urban ecological landscape.

3. Materials and Methods

3.1. UAV Survey System

The UAV aerial survey system generally consists of three components: the UAV platform, the flight control system, and the aerial camera equipment. The specific parameters of the aerial camera system used in this study are presented in Table 1.
In this study, a DJI Matrice 300 RTK quadcopter UAV produced by DJI Innovation Technology Co., Ltd. in Shenzhen, China was employed for the survey.
The device is equipped with the DJI Zenmuse H20T camera produced by DJI Innovation Technology Co., Ltd. in Shenzhen, China, featuring HD visible light, HD wide-angle, and HD thermal infrared 3-in-1 modules (Figure 2). This configuration enables continuous aerial photogrammetry for up to 55 min in an open 15-km area. The model is equipped with an RTK module, which provides centimeter-level positioning accuracy through real-time dynamic differential technology. Additionally, by combining time synchronization compensation between the flight control system and the camera, it enables image-control point-free photogrammetry. The aerial photography system is equipped with six-direction positioning and obstacle avoidance technology, which enables automatic detection of flight obstacles, thereby improving operational safety.

3.2. UAV Remote Sensing Image Acquisition

Common methods of UAV image acquisition include tilt photogrammetry, orthophoto mapping, strip-area mapping, and close-range photogrammetry. In this study, tilt photogrammetry is employed to capture remote sensing images of the study area, characterized by significant terrain undulations and steep slopes. Tilt photogrammetry offers multiple tilt-angle lenses in addition to vertical downward shots, enabling simultaneous image collection from various angles, such as vertical and oblique. This method allows for the acquisition of side texture and geometric information of features, providing a true representation of the area. It enables the rapid acquisition of large-scale terrain data, overcoming terrain obstacles, and capturing multi-angle image data in a single flight, thereby reducing external work time and efficiently gathering data.
Five routes were planned for tilt photogrammetry, with one route having a gimbal pitch angle of −90° and four routes set to a gimbal pitch angle of −45°. The camera lenses were aligned with the flight direction to ensure the collection of detailed surface morphology information of the objects being photographed. Each route in this study was set at a height of 400 m, approximately 35 m above the highest point of the detected slope, with images captured at equal time intervals. The terrain in this study area is characterized by significant undulations, cliffs, and steep slopes. To ensure adequate coverage of the aerial photography images, it is necessary to increase the route overlap rate. This adjustment ensures comprehensive image acquisition and improves the accuracy of subsequent air-to-ground calculations and three-dimensional model generation, thereby enhancing the overall quality of the work. Therefore, the route overlap rate is set to 75% in the heading direction and 85% in the side direction to balance aerial photography efficiency and accuracy [29,30,31].
In this study, slope image data were acquired during two periods, in December 2023 and December 2024. The two routes used for data acquisition were identical and chosen to be conducted at midday under sunny conditions, effectively minimizing issues such as unclear texture and low brightness in the aerial images. Additionally, the coordinates were automatically adjusted to correct any bias, preventing large-scale shifts in aerial image coordinates due to the limited positioning accuracy of GPS and other factors in mainland China. The two aerial photography missions were conducted with a 33-day interval between them. Each aerial photography mission lasted 33 min, during which 74 photographs of the study area were collected, meeting the quality requirements for the missions.

3.3. Data Processing and Analysis

This study utilized DJI Terra Mapping & Surveying Edition software for the three-dimensional reconstruction of slopes. The software is efficiently integrated with the DJI Matrice 300 RTK UAV and the DJI Zenmuse H20T camera used in this study, significantly improving the efficiency of aerial photography processing while ensuring the accuracy of the data. This study utilized the change detection module in LiDAR software to perform a comparative analysis of slope changes and generate DEMs.
This study employed LiDAR360 5.2 software to extract a contour model from DEMs for analyzing elevation variations in the study area. This study employed the Iterative Closest Point (ICP) algorithm to register two-phase point-cloud datasets and conducted a comparative analysis of variations between two aerial photogrammetric point-cloud models using the change detection plugin within LiDAR360 software, thereby delineating regional changes during the interval between the two aerial surveys. This study employed ArcMap 10.8 software to extract slope gradient and curvature metrics from two-phase DEMs to facilitate slope stability analysis. DJI Terra, LiDAR360, and ArcMap 10.8 software ensure cross-platform data consistency and compatibility by supporting common data formats (e.g., LAS/LAZ, GeoTIFF, Shapefile) and standardized coordinate systems (e.g., WGS84, UTM). The UAV imagery point-cloud and 3D model generated by DJI Terra can be exported to LAS or GeoTIFF format. After point-cloud classification and terrain analysis in LiDAR360, spatial mapping and information extraction are conducted using ArcMap 10.8. This process guarantees that data attributes, accuracy, and spatial references are preserved without loss [29,30]. This paper used the Analytic Hierarchy Process (AHP) to calculate the weights of each evaluation factor of the slope, and then finally adopted the hazard evaluation model to assess the stability of the slope.

4. Remote Sensing Image Data Processing

4.1. Construct Models

4.1.1. Aerial Triangulation

The photos containing POS data are imported into DJI Terra Mapping & Surveying Edition software, where feature points are automatically extracted and matched using a feature extraction and matching algorithm. Representative feature points are identified by analyzing grayscale, texture, and other image attributes. Corresponding feature points are then found between different images. The area network is computed using the air-triangulation method, with each image beam serving as a basic unit. The beams intersect in space to perform a parity calculation of encrypted points, with the goal of minimizing reprojection error [32].
min X , R , t ∑ i = 1 n ∑ j = 1 m p i j − π ( R i X j + t i ) 2 ,
where X j represents the coordinates of the 3D point, R i and t i represent the rotation matrix and translation vector of the ith image, and π denotes the camera projection model (e.g., pinhole model). In this study, a high feature point density is used to ensure that enough feature points are obtained for each impact, improving the accuracy of the results. The WGS 84/UTM Zone 50N projection coordinate system is used.
The Air-3 calculation report is shown in Table 2. In the RTK solution of the Air-3 calculation, the positioning accuracy of the floating-point solution is at the decimeter level, and the positioning accuracy of the single-point solution is at the meter level. “No solution” means there is no RTK positioning solution. If all the solutions are fixed, it guarantees that image-control-free accuracy can reach the centimeter level under the POS coordinate system. The unmanned aerial camera system used in this study can achieve image-control-free flight photogrammetry, and the results of the Air-3 calculation indicate that all solutions are fixed. Therefore, it is reasonable to omit the image-control point setting in this study. As all the solutions from the Air-3 calculation are fixed, omitting the image-control point setting in this study is reasonable. The two aerial photography images in this study were calibrated, confirming that the route planning and parameter settings are appropriate. The root-mean-square error (RMSE) of geographic alignment is 0.957 m for the first shot and 1.364 m for the second shot. These values represent the root-mean-square error (i.e., the POS relative accuracy) between the solved image position and the initial image record. Both values are under 2 m, meeting the accuracy requirements for the actual work. An unsimplified model processing enhances the accuracy of the model, making it less prone to sharp corners, damage, and other issues. The processing time for the two aerial photography images was 13 min and 9 s, and 14 min and 42 s, respectively. With the two aerial photography sessions lasting 33 min, the entire process was completed within 50 min, significantly improving efficiency compared to manual inspection and other traditional methods.

4.1.2. 3D Reconstruction

In this study, after the completion of the Air-3 computation, DJI Terra software was used to process the aerial images from the first and second phases, enabling the three-dimensional reconstruction of the study area. This process generated a three-dimensional model of the area (Figure 3) and exported a point-cloud model in LAS format (Figure 4). The first reconstruction took 12 min and 32 s, while the second reconstruction took 13 min and 49 s.

4.2. Point-Cloud Data Processing

In this study, the point-cloud model of the study area was generated during the 3D reconstruction process and further processed in the following sequence.

4.2.1. Point-Cloud Data Pre-Processing

Denoising of point-cloud data [33]. When using a UAV aerial photography system to collect image data of the study area, noise often occurs due to interference from objects in the environment, which can affect the accuracy of the point-cloud model. The standard deviation multiplier is set as mean K, and the software searches for neighboring points for each point, specifying the number of neighboring points. It then calculates the average distance between the point and its neighbors using D as a surrogate and computes the median distance (mean D) and the standard deviation (S). If the maximum distance (Max D) exceeds the threshold (Max D = mean D + mean K ∗ S), noisy points are identified and removed. The smaller the value of S, the more points are classified as noise. The standard deviation multiplier and the number of neighboring points are set to 5 and 10, respectively. Through statistical significance, sensitivity analysis, and engineering validation, it is demonstrated that the combination of a standard deviation multiplier K = 5 and a neighborhood point count n = 10 achieves an optimal balance between denoising rate, false deletion rate, and terrain fidelity. This parameter setting is particularly suitable for medium- to high-density LiDAR point clouds [34,35]. The denoising process is then completed for both sets of point-cloud models. The point-cloud density is approximately 22 points/m2.
The ground point classification process is performed on the point-cloud model of the study area using the Improved Progressive TIN Densification (IPTD) algorithm. This algorithm initially generates a sparse triangular mesh from seed points and then iteratively refines the mesh layer by layer until all the ground points are classified. The largest building in the region is a square pump house located in the southeast of the mountain, with dimensions of 15 m × 15 m × 6 m. Therefore, the maximum building size on the horizontal plane is set to 15 m. The terrain in the region has significant undulation, and thus the iteration angle is set to 15° (greater than the default of 8°). The default iteration angle of 8° may not accurately capture the true ground morphology in steep slope areas, leading to the misclassification of ground points as non-ground points (due to local slopes exceeding the threshold), which in turn causes a fragmented terrain model. Additionally, the default iteration distance of 1.4 m (the maximum distance between a point and the apex of the triangulation plane) may prevent the triangulation mesh from expanding in areas with abrupt terrain changes (e.g., cliffs, steep canyons), resulting in voids. To address this, increasing the iteration distance to 2 m enhances the continuity of the triangulation mesh in fragmented terrain and ensures efficient connection of neighboring ground points [34,35,36], while maintaining other parameters at the default level. The parameters for both phases of the model are set the same. The results of the ground point classification are shown in Figure 5, where orange represents ground points and white represents non-ground points.

4.2.2. Geographic Information Extraction

A digital surface model of the study area was generated using LiDAR360 software to process the point-cloud data. The Digital Elevation Model (DEM) of the study area was constructed, with the XSize and YSize set to their default value of 0.5 m. The buffer size parameter was set to its default of 5. The inverse distance weighting (IDW) interpolation method does not require pre-computation of the spatial variability function, making it especially suitable for the rapid interpolation of high-density LiDAR point clouds. In contrast, the kriging method assumes data smoothness and requires accurate fitting of the semi-variance model. This method may fail if there are local outliers (e.g., building edges or vegetation penetration noise) in the LiDAR point-cloud [37]. Using the inverse distance weighted interpolation method, the raster cell value was calculated based on nearby points. The weighted average was determined based on the distance from the point to the center of the raster cell. In this study, the weight is set to the default value of 2, which effectively balances smoothness and detail preservation in most terrain scenes, such as hills and mountains [38]. A variable radius search mode was selected, with the distance set to 5 and the number of pixel points defaulted to 12. This resulted in a Digital Elevation Model in TIF format (Figure 6) for export.

5. Geographic Information Interpretation and Analysis

5.1. Identification of Geological Hazard Prone Areas

The contour map, generated based on the Digital Elevation Model (Figure 7), visually represents the surface undulation pattern of the study area. The map indicates significant elevation variation in the study area, ranging from 192 m to 383 m, with numerous cliffs and steep slopes present. A comparison of the contour maps from both time periods (before-and-after aerial photography) reveals a more pronounced contour overlap in the area highlighted by the red box. Since the flight status parameters remained unchanged between the two aerial photographs, it can be qualitatively concluded that the steepness of this section of the area has changed, and there is a high probability that the slope pattern of the mountain has been altered.
The pre-processed point-cloud models from both phases were aligned in the same coordinate system for subsequent analysis. The Iterative Closest Point (ICP) method was used for alignment, selecting the first phase of the point-cloud as the reference datum. The core of the ICP algorithm is to minimize the distance error between two sets of point clouds by iteratively optimizing the transformation matrix (rotation + translation) [39]. The alignment of the second phase was performed with the root-mean-square error (RMSE) mode as the termination criterion, set to 1 × 10−5 [40,41]. RMSE directly reflects alignment accuracy and serves as a key condition for iteration termination. When the change in RMSE between two consecutive iterations falls below a specified threshold, the algorithm halts, and the alignment is considered to have converged. The number of alignment samples was set to 50,000. The LiDAR360 software’s change detection plugin was used to compare the point-cloud models of the two aerial photographs, reflecting the changes in the region between the two periods. The grid edge length for the point-cloud was set to 2, with an elevation tolerance of 0.5 dm and a change distance of 0, generating the differential point-cloud model of the study area (Figure 8). The green dots represent the elevation changes in the comparison point-cloud (Phase II) relative to the reference point-cloud (Phase I). Three areas with significant elevation changes were identified in the study area, as shown in the red boxes in the figure, which are prone to geological hazards such as landslides and rockfalls. These areas should be closely monitored in the future.
Three hazardous areas were identified in the study area through a comprehensive analysis of the 3D models, contour model, and point-cloud differential model (Figure 9). Based on field investigation and close photographic observation, the mountain restoration project in Area 2 involved road construction, and the elevation change is clearly due to the accumulation of soil along both sides of the road, eliminating the potential risk. In Area 1, based on the contour maps mentioned above and field investigation, multi-story stepped cliffs are distributed in this area, and the mountain slopes show severe weathering and fragmentation. Minor rockfall events have already occurred in this area. Additionally, with the entry of heavy machinery for the mountain restoration project, small-scale rock collapses occur intermittently. Area 3 has been affected by heavy machinery, weathering, and freeze-thaw cycles over an extended period, resulting in fragmented rock and steep mountain slopes. Small-scale geotechnical slides have occurred in this area. These conditions make the area highly susceptible to geological disasters, threatening the stability of the slope and the safety of highway traffic below.
In summary, attention must be given to Areas 1 and 3. Their stability should be further analyzed, with long-term dynamic monitoring and risk investigation implemented to prevent destructive geological disasters, thereby avoiding significant social and economic losses.

5.2. Slope Stability Analysis

Slope Gradient Analysis: The slope information of the two DEM phases was extracted using the Spatial Analyst—Surface—Slope module of the ArcMap 10.8 software, and the slopes were classified into 9 categories and set in degrees. The slope maps for both phases were then obtained (Figure 10). The maximum slope value in Phase I of the study area is 87.93°, while the maximum value in Phase II is 87.74°. The change in slope between the two phases is minimal. The topography of this area is steep, with Areas 1 and 3 containing numerous slopes exceeding 50°. Data analysis indicates that soil landslides are likely to occur when the slope gradient exceeds 20°, rocky landslides when the gradient is greater than 30°, and rock avalanches when the gradient exceeds 45°. This suggests that the area is in an unstable state, with a higher likelihood of landslides and rock avalanches as geological hazards.
Slope Curvature Analysis: Slope shapes are generally classified into three basic types: convex, straight, and concave slopes [42]. A commonly used algorithm for calculating curvature involves fitting a surface, consisting of a 3 × 3 window, to a second-order polynomial equation [43], as given by:
Z = A x 2 y 2 + B x 2 y + C x y 2 + D x 2 + E y 2 + F x y + G x + H y + I ,
where Z represents the topographic surface function, x and y are the spatial positions, and coefficients A − I are calculated from the image elevation values and their respective sizes in the 3 × 3 window. Convex slopes are more likely to generate inertial forces, which can lead to landslides under external forces. Concave slope topography facilitates the convergence of rainfall and reduces the strength of geotechnical bodies, while rectilinear slopes are relatively more stable. The curvature values of the DEM were extracted, with standard curvature greater than 0 for convex slopes, less than 0 for concave slopes, and equal to 0 for straight slopes. The curvature maps of the study area were then generated (Figure 11). The curvature maps of the two phases are similar, with convex and concave slopes interspersed in Areas 1 and 3, exhibiting complex and variable slope shapes. This indicates that the slopes in these areas are unstable and prone to geological hazards such as landslides and rockfalls. Areas 1 and 3 are susceptible to slope hazards, particularly under conditions such as heavy rainfall and freeze-thaw cycles. Preventive reinforcement is recommended to mitigate potential risks to nearby pedestrians and road infrastructure.

5.3. Stability Analysis

Investigating a variety of slope stability evaluation models [44,45,46,47,48,49], this paper uses the Analytic Hierarchy Process (AHP) to calculate the weights of each evaluation factor of the slope. AHP is a multi-criteria decision-making tool used for the systematic evaluation and prioritization of complex problems. It is particularly applicable in scenarios that require the integration of both qualitative and quantitative factors, spanning fields such as engineering management, urban planning, environmental sciences, supply chain and logistics, healthcare, finance and investment, marketing, and more. Its core strength lies in combining expert judgment and objective data through a structured model, effectively reconciling multiple conflicting interests [50]. This paper finally adopts the hazard evaluation model to assess the stability of the slope, and its evaluation model is as follows.
D L = ∑ i = 1 n W i ⋅ I i ,
where D L represents the slope hazard index, W i represents the weight vector of the discriminant factor, and I i represents the action index of the discriminant factor.
Based on field investigations and research by other scholars [51,52,53], the slope stability evaluation index hierarchy is constructed according to three levels: the target level (A), the class index level (B), and the base index level (C). The factors influencing slope stability are divided into intrinsic and extrinsic categories. Intrinsic factors include geological conditions (B1) and topography and geomorphology (B2), while extrinsic factors include triggering factors (B3) and human activities (B4), resulting in a total of four primary categories of evaluation indicators. Under geological conditions (B1), secondary evaluation indicators include the type of geotechnical body (C1), structural surface density (C2), and degree of weathering (C3). Under topography and geomorphology (B2), secondary evaluation indicators include slope gradient (C4), slope direction (C5), curvature (C6), and vegetation cover (C7). Under triggering factors (B3), secondary evaluation indicators include average annual rainfall (C8) and groundwater impact (C9). Human activities (B4) include secondary indicators such as anthropogenic disturbances (C11) and vegetation damage (C12).
The Saaty 1–9 scale method was employed to conduct the expert questionnaire [54], with a total of eight experts participating. The Saaty 1–9 scale offers greater sensitivity in distinguishing subtle differences compared to the 1–5 or 1–7 scales, while also avoiding the confusion that may arise from excessive granularity. The consistency ratio, calculated using the maximum eigenvalue method, has been shown to be statistically responsive to inconsistencies in the judgment matrix when using the 1–9 scale, effectively identifying logical contradictions. A comparative study further demonstrated that the correlation coefficient between AHP results using the 1–9 scale and those derived from the Fuzzy Comprehensive Evaluation (FCE) method reached 0.91, confirming the engineering applicability of the approach [55,56]. The eight scoring experts possessed extensive experience and academic expertise in geotechnical engineering, slope stability analysis, and risk assessment. The panel comprised four geotechnical engineers, three surveying and mapping engineers, and one infrastructure management engineer. All experts had prior experience with multi-criteria decision-making frameworks and were well-versed in the Saaty 1–9 scale method. The individual judgment matrices were aggregated using the geometric mean method to construct a matrix of first-level evaluation indicators, as shown in Table 3. The use of geometric means to aggregate expert judgments mitigates the influence of extreme opinions while maintaining the proportional relationships among the indicators. Each expert completed the questionnaire independently, thereby avoiding potential bias or authority dominance that may arise from group discussions. The maximum eigenvalue of the matrix was calculated as λmax = 4.218, the consistency evaluation index (CI) was 0.073, the random consistency index (RI) was 0.89, and the consistency ratio (CR) was 0.082 (<0.1). These results indicate that the slope stability assessment index system has good consistency and reasonable judgment.
By performing step-by-step calculations, the weights of the secondary indicators were obtained, as shown in Table 4, and verified to be in good agreement.
The “Golden Section” method was used to determine the hazard scores of the indicators at all levels of the slope impact factor, minimizing the arbitrariness of scoring [57]. If the indicator danger is large, the score is 1; if the danger is medium, the score is 0.618; and if the danger is small, the score is 0.382. The classification of the secondary evaluation indicators and their quantitative scores are shown in Table 5. The scoring values of the secondary indicators were determined based on the actual situation in Jinan City. According to the literature and the geological survey report issued by the Natural Resources Agency [58,59], it is known that the study area is rich in a large number of hard bedrock, the structural surface density is low, the weathering grade is medium weathering, the annual rainfall is 400–800 mm, the basic seismic intensity is 7 degrees, and the buried depth of groundwater level is greater than 5 m. According to the acquired model of the study area, it is known that the area is dominated by steep slopes with slopes greater than 30°, and the presence of a large number of lateral breaks and more concave shapes. According to the site investigation, the vegetation cover of the area is less than 30%, and the area is a waste quarry. Following the recent construction of roads around the area, the excavation area is more than 30%, and the damage to vegetation is moderate.
As shown in Table 6, the scoring value ( C i ) of the secondary evaluation index for the slope, as well as the comprehensive scoring value ( C i W C i ) considering the weights, are clarified and calculated using Equation (2). Based on this, the hazard index of the slope is R = 0.596 . The slope is classified into four stability grades: stable (I) ( 0 ≤ R < 0.25 ), basically stable (II) ( 0.25 ≤ R < 0.5 ), more unstable (III) ( 0.5 ≤ R < 0.75 ), and unstable (IV) ( 0.75 ≤ R < 1 ). Based on this classification, it can be concluded that the slope is in a more unstable state (III), indicating it is a high-risk slope. As such, corresponding professional monitoring measures and prevention programs must be considered to prevent slope hazards that could negatively impact nearby residents and traffic.

6. Conclusions

Using UAV inclined photogrammetry technology to obtain high-definition images of the slopes in both phases, we constructed three-dimensional realistic models, point-cloud models, and digital elevation models of the slopes, identified hazardous areas, and analyzed their stability. The following conclusions were drawn:
(1)
UAV remote sensing technology is an efficient tool for identifying and monitoring geological hazards on slopes. The UAV is equipped with a multifunctional camera to capture high-definition images, and various surveying and mapping software are used to construct three-dimensional models. This approach enhances work efficiency and expands the monitoring scope compared to traditional methods of geological hazard monitoring.
(2)
Hazardous slope areas can be rapidly identified using the Digital Elevation Model, point-cloud model, and three-dimensional live model. By extracting the contour and point-cloud differential models of the study area and combining them with field investigation and analysis, two areas prone to geological hazards—susceptible to small debris avalanches and soil landslides—were identified.
(3)
Using the Digital Elevation Model to extract slope gradient and shape information, the stability of the two hazardous areas was analyzed. The maximum slope gradient in Areas 1 and 3 is approximately 87°, with many slopes exceeding 50°. The terrain is steep, and convex and concave slopes appear alternately, making the slope shape highly variable and prone to disasters. A stability evaluation was conducted using AHP and hazard calculation methods, and the slope’s stability status was determined. The slope’s stability status was determined, and it is recommended that greater attention be given to the area, with timely slope management to prevent harm to highways and people’s lives and property. This method is simple and practical, providing new approaches for slope stability assessment.
(4)
The proposed approach has broad applicability in areas such as slope change detection, slope engineering stability assessment, geological hazard identification, and road safety operation and maintenance management. It offers significant practical value by enhancing work efficiency, reducing labor costs, and improving the safety of both people and infrastructure.

Author Contributions

Conceptualization, N.L. and J.W.; methodology, N.L.; software, H.Q.; validation, N.L., H.Q. and H.Z.; formal analysis, Y.C.; investigation, N.L.; writing—original draft preparation, N.L.; writing—review and editing, H.Q.; visualization, H.Z.; supervision, J.W.; project administration, N.L.; funding acquisition, J.W. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the National Key Research and Development Program of China, grant number 2023YFB260400405. This research was funded by the National Key Research and Development Program of China, grant number 2021VI1E0194300. The APC was funded by Shandong University.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The original data pertaining to the research area discussed in this study cannot be disclosed due to contractual obligations with collaborating parties.

Acknowledgments

We acknowledge the scientific contributions of all authors. We would like to express our gratitude to Lin Wu and Xiaohui Chen for their research contributions. We are grateful to Wenbin Li for his contributions during the process of revising the thesis.

Conflicts of Interest

Author Hu Zhai was employed by the company Shandong Bowee Vision Information Technology Co., Ltd. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Abbreviations

The following abbreviations are used in this manuscript:
UAVUnmanned aerial vehicle
DEMDigital elevation model
AHPAnalytic hierarchy process
RTKReal-time kinematic
XMLContextCapture blocks exchange
POSPosition and orientation system
ICPIterative closest point
IPTDImproved progressive TIN densification
LASLASer file format

References

  1. Wu, Y.; Tian, A.; Lan, H. Comparisons of Dynamic Landslide Models on GIS Platforms. Appl. Sci. 2022, 12, 3093. [Google Scholar] [CrossRef]
  2. Hussain, Y.; Schlogel, R.; Innocenti, A.; Hamza, O.; Iannucci, R.; Martino, S.; Havenith, H. Review on the Geophysical and UAV-Based Methods Applied to Landslides. Remote Sens. 2022, 14, 4564. [Google Scholar] [CrossRef]
  3. Mcdougall, S. 2014 Canadian Geotechnical Colloquium: Landslide runout analysis—Current practice and challenges. Can. Geotech. J. 2017, 54, 605–620. [Google Scholar] [CrossRef]
  4. Nigro, M.; Pierri, F.; Caccavale, F. Control of an Omnidirectional UAV for Transportation and Manipulation Tasks. Appl. Sci. 2021, 11, 10991. [Google Scholar] [CrossRef]
  5. Zhong, C.; Liu, Y.; Gao, P.; Chen, W.; Li, H.; Hou, Y.; Nuremanguli, T.; Ma, H. Landslide mapping with remote sensing: Challenges and opportunities. Int. J. Remote Sens. 2020, 41, 1555–1581. [Google Scholar] [CrossRef]
  6. Stott, E.; Williams, R.D.; Hoey, T.B. Ground Control Point Distribution for Accurate Kilometre-Scale Topographic Mapping Using an RTK-GNSS Unmanned Aerial Vehicle and SfM Photogrammetry. Drones 2020, 4, 55. [Google Scholar] [CrossRef]
  7. Zieher, T.; Toschi, I.; Remondino, F.; Rutzinger, M.; Kofler, C.; Mejia-Aguilar, A.; Schlögel, R. Sensor- and scene-guided integration of tls and photogrammetric point clouds for landslide monitoring. Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci. 2018, 42, 1243–1250. [Google Scholar] [CrossRef]
  8. An, S.; Yuan, L.; Xu, Y.; Wang, X.; Zhou, D. Ground subsidence monitoring in based on UAV-LiDAR technology: A case study of a mine in the Ordos, China. Geomech. Geophys. Geo-Energy Geo-Resour. 2024, 10, 57. [Google Scholar] [CrossRef]
  9. Qiu, H.; Zhai, H.; Wang, J.; Ge, S.; Wu, L. Geological survey and risk assessment based on UAV photogrammetry. Sci. Technol. Eng. 2024, 24, 7905–7912. (In Chinese) [Google Scholar]
  10. Wei, Y. Application of UAV Inclined Photogrammetry in Highway Survey and Design. Digit. Commun. World 2023, 101–103. (In Chinese) [Google Scholar] [CrossRef]
  11. Pi, H.; Peng, S.; Zeng, X. Expressway high slope monitoring information extraction using UAV oblique photogrammetry. Beijing Surv. Mapp. 2023, 37, 620–624. (In Chinese) [Google Scholar] [CrossRef]
  12. Zolkepli, M.F.; Ishak, M.F.; Talib, Z.A. Unmanned Aerial Vehicle (UAV)-Based for Slope Mapping and the Determination of Potential Slope Hazard. Int. J. Integr. Eng. 2022, 14, 232–239. [Google Scholar] [CrossRef]
  13. Yao, A.; Gong, Y.; Li, Y.; Tian, T.; Xu, C. Mountain-Expressway Slope Safety Based on the Online Monitoring System and Fuzzy Comprehensive Evaluation. J. Test. Eval. 2022, 50, 1906–1919. [Google Scholar] [CrossRef]
  14. Yang, Y.; Wu, W.; Zheng, H. Investigation of slope stability based on strength-reduction-based numerical manifold method and generalized plastic strain. Int. J. Rock Mech. Min. Sci. 2023, 164, 105358. [Google Scholar] [CrossRef]
  15. Zhang, Z.L.; Yang, X.L. Unified solution of safety factors for three-dimensional compound slopes considering local and global instability. Comput. Geotech. 2023, 155, 0266–0352X. [Google Scholar] [CrossRef]
  16. Liang, G.L.; Xu, W.Y.; He, Y.Z.; Zhao, Y.X. Application of catastrophe progression method to comprehensive evaluation of slope stability. Rock Soil Mech. 2008, 29, 1895–1899. [Google Scholar]
  17. Wang, S.; Zhang, Z.; Ren, Y.; Zhu, C. UAV Photogrammetry and AFSA-Elman Neural Network in Slopes Displacement Monitoring and Forecasting. KSCE J. Civ. Eng. 2020, 24, 19–29. [Google Scholar] [CrossRef]
  18. Huang, H.; Long, J.; Lin, H.; Zhang, L.; Yi, W.; Lei, B. Unmanned aerial vehicle based remote sensing method for monitoring a steep mountainous slope in the Three Gorges Reservoir, China. Earth Sci. Inform. 2017, 10, 287–301. [Google Scholar] [CrossRef]
  19. Wu, X.; Yang, Y.; Ye, Z. Stability of rock—Soil slopes under rainfall infiltration and its influencing factors. J. Water Resour. Water Eng. 2022, 33, 189–199. (In Chinese) [Google Scholar]
  20. Yuan, D.; Lu, J.; Liu, D.; Yang, Z.; Wang, J. Visual Query and Analysis System of Slope Monitoring Information and Its Application. J. Eng. Geol. 1999, 7, 355–360. (In Chinese) [Google Scholar]
  21. Li, P.; Wang, Q.; Li, J.; Pei, Y.; He, P. Automated extraction of tunnel leakage location and area from 3D laser scanning point clouds. Opt. Lasers Eng. 2024, 178, 108217. [Google Scholar] [CrossRef]
  22. Chen, Y.; Ye, Z.; Chen, Z.; Wang, G. Morphological deformation characteristics of landslide based on UAV imagery: Taking Xieliupo landslide in Bailong river basin of Longnan as an example. Seience Technol. Eng. 2024, 24, 2665–2672. (In Chinese) [Google Scholar]
  23. Kern, A.; Fanta-Jende, P.; Glira, P.; Bruckmüller, F.; Sulzbachner, C. An accurate real-time UAV mapping solution for the generation of orthomosaics and surface models. Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci. 2021, 43, 165–171. [Google Scholar] [CrossRef]
  24. Mir, S.; Shah, S.A.; Bhat, M.S.; Akhter, S.; Ahad, F.; Rashid, H.; Jeelani, P. Disaster risk assessment of educational infrastructure in mountain geographies using PROMETHEE-II. Int. J. Disaster Risk Reduct. 2024, 107, 104489. [Google Scholar] [CrossRef]
  25. Cirillo, D.; Zappa, M.; Tangari, A.C.; Brozzetti, F.; Ietto, F. Rockfall Analysis from UAV-Based Photogrammetry and 3D Models of a Cliff Area. Drones 2024, 8, 31. [Google Scholar] [CrossRef]
  26. Yu, H.; Li, Z.; Di, T.; Li, Y. Unmanned aerial vehicle images-based landslide geohazards interpretation and stability analysis: An illustrative example of Yanguan landslide. Sci. Technol. Eng. 2019, 19, 84–92. (In Chinese) [Google Scholar]
  27. Ruzic, I.; Benac, C.; Jovancevic, S.D.; Radisic, M. The Application of UAV for the Analysis of Geological Hazard in Krk Island, Croatia, Mediterranean Sea. Remote Sens. 2021, 13, 1790. [Google Scholar] [CrossRef]
  28. Wang, S.H.; Ahmed, Z.; Hashmi, M.Z.; Wang, P.Y. Cliff face rock slope stability analysis based on unmanned arial vehicle (UAV) photogrammetry. Geomech. Geophys. Geo-Energy Geo-Resour. 2019, 5, 333–344. [Google Scholar] [CrossRef]
  29. James, M.R.; Robson, S. Straightforward reconstruction of 3D surfaces and topography with a camera: Accuracy and geoscience application. J. Geophys. Res. Earth Surf. 2012, 117, 17. [Google Scholar] [CrossRef]
  30. Nex, F.; Remondino, F. UAV for 3D mapping applications: A review. Appl. Geomat. 2014, 6, 1–15. [Google Scholar] [CrossRef]
  31. Eltner, A.; Kaiser, A.; Castillo, C.; Rock, G.; Neugirg, F.; Abellan, A. Image-based surface reconstruction in geomorphometry—Merits, limits and developments. Earth Surf. Dyn. 2016, 4, 359–389. [Google Scholar] [CrossRef]
  32. Andrew, A.M. Multiple View Geometry in Computer Vision. Kybernetes 2001, 30, 1333–1341. [Google Scholar] [CrossRef]
  33. Zhang, J.; Lin, X. Filtering airborne LiDAR data by embedding smoothness-constrained segmentation in progressive TIN densification. ISPRS-J. Photogramm. Remote Sens. 2013, 81, 44–59. [Google Scholar] [CrossRef]
  34. Zhang, W.; Qi, J.; Wan, P.; Wang, H.; Xie, D.; Wang, X.; Yan, G. An Easy-to-Use Airborne LiDAR Data Filtering Method Based on Cloth Simulation. Remote Sens. 2016, 8, 501. [Google Scholar] [CrossRef]
  35. Meng, X.; Currit, N.; Zhao, K. Ground Filtering Algorithms for Airborne LiDAR Data: A Review of Critical Issues. Remote Sens. 2010, 2, 833–860. [Google Scholar] [CrossRef]
  36. Chen, Q.; Gong, P.; Baldocchi, D.; Xie, G. Filtering airborne laser scanning data with morphological methods. Photogramm. Eng. Remote Sens. 2007, 73, 175–185. [Google Scholar] [CrossRef]
  37. Reese, R.A. Geostatistics for Environmental Scientists. J. R. Stat. Soc. Ser. A-Stat. Soc. 2009, 172, 700. [Google Scholar] [CrossRef]
  38. Lu, G.Y.; Wong, D.W. An adaptive inverse-distance weighting spatial interpolation technique. Comput. Geosci. 2008, 34, 1044–1055. [Google Scholar] [CrossRef]
  39. Besl, P.J.; Mckay, N.D. A method for registration of 3-D shapes. IEEE Trans. Pattern Anal. Mach. Intell. 1992, 14, 239–256. [Google Scholar] [CrossRef]
  40. Rusinkiewicz, S.; Levoy, M. Efficient variants of the ICP algorithm. In Proceedings of the Third International Conference on 3-D Digital Imaging and Modeling, Quebec City, QC, Canada, 28 May–1 June 2001; pp. 145–152. [Google Scholar] [CrossRef]
  41. Chen, Y.; Medioni, G. Object modeling by registration of multiple range images. Image Vis. Comput. 1992, 10, 145–155. [Google Scholar] [CrossRef]
  42. Xue, Q.; Zhang, M.; Gao, B. Hazard assessment of loess landslide based on soil moisture content and supported by slope unit in Qingjian City, Shaanxi Province. Geol. China 2020, 47, 1904–1914. (In Chinese) [Google Scholar]
  43. Zevenbergen, L.W.; Thorne, C.R. Quantitative-analysis of land surface-topography. Earth Surf. Process. Landf. 1987, 12, 47–56. [Google Scholar] [CrossRef]
  44. Liu, J.; Wang, L. Investigation and Risk Assessment of a Loess Landslide Based on UAV Tilt Photogrammetry. Shuitu Baochi Tongbao 2023, 43, 139–147. [Google Scholar]
  45. Vandenberge, D.R.A.D.; Mcguire, M.P.A.M. Practical Use of Modified Hoek–Brown Criterion for Soil Slope Stability Analysis. Geotech. Geol. Eng. 2019, 37, 5441–5455. [Google Scholar] [CrossRef]
  46. Ni, K.; Yang, J.; Ning, Y.; Kang, G. A modified sub-block DDA fracturing modelling method for rock. Eng. Anal. Bound. Elem. 2020, 111, 154–166. [Google Scholar] [CrossRef]
  47. Kohno, M.; Ono, Y. Evaluation of earthquake-induced slope failure distribution using the AHP method and GIS. Int. J. GEOMATE 2021, 20, 74–81. [Google Scholar] [CrossRef]
  48. Ye, W.; Gao, C.; Liu, Z.; Wang, Q.; Su, W. A Fuzzy-AHP-based variable weight safety evaluation model for expansive soil slope. Nat. Hazards 2023, 119, 559–581. [Google Scholar] [CrossRef]
  49. Leong, E.C.; Rahardjo, H. Two and three-dimensional slope stability reanalyses of Bukit Batok slope. Comput. Geotech. 2012, 42, 81–88. [Google Scholar] [CrossRef]
  50. Vasquez, J.; Estrada, M. A comparative study of the bivariate statistical methods and the Analytical Hierarchical Process for the assessment of mass movement susceptibility. A case study: The LM-116 Road-Peru. Rud.-Geol.-Naft. Zb. 2023, 38, 149–166. [Google Scholar] [CrossRef]
  51. Cengiz, L.D.; Ercanoglu, M. A novel data-driven approach to pairwise comparisons in AHP using fuzzy relations and matrices for landslide susceptibility assessments. Environ. Earth Sci. 2022, 81, 222. [Google Scholar] [CrossRef]
  52. Abay, A.; Barbieri, G.; Woldearegay, K. GIS-based Landslide Susceptibility Evaluation Using Analytical Hierarchy Process (AHP) Approach: The Case of Tarmaber District, Ethiopia. Momona Ethiop. J. Sci. 2019, 11, 14–36. [Google Scholar] [CrossRef]
  53. He, H.; Hu, D.; Sun, Q.; Zhu, L.; Liu, Y. A Landslide Susceptibility Assessment Method Based on GIS Technology and an AHP-Weighted Information Content Method: A Case Study of Southern Anhui, China. ISPRS Int. J. Geo-Inf. 2019, 8, 266. [Google Scholar] [CrossRef]
  54. Saaty, T.L. A scaling method for priorities in hierarchical structures. J. Math. Psychol. 1977, 15, 234–281. [Google Scholar] [CrossRef]
  55. Vargas, L.G. An overview of the analytic hierarchy process and its applications. Eur. J. Oper. Res. 1990, 48, 2–8. [Google Scholar] [CrossRef]
  56. Zhang, L.; Wang, Y.; Zhang, J.; Zhang, S.; Guo, Q. Rockfall hazard assessment of the slope of Mogao Grottoes, China based on AHP, F-AHP and AHP-TOPSIS. Environ. Earth Sci. 2022, 81, 377. [Google Scholar] [CrossRef]
  57. Huang, J.L.; Zhao, J.G.; Zhang, T.; Li, J.L. AHP-based hazard degree assessment of high-speed landslide of reservoir bank. J. Nat. Disasters 2011, 20, 95–99. [Google Scholar]
  58. Shang, H.; Wang, Y.; Li, G.; Guo, Y.; Ma, A.; Yan, X.; Luo, Y. Research on Geological Environment Evaluation of Jinan Based on 3D Model. In Proceedings of the 3rd Guangdong-Hong Kong-Macao Greater Bay Area Artificial Intelligence and Big Data Forum, AIBDF 2023, Guangzhou, China, 22–24 September 2023. [Google Scholar]
  59. Zhu, M.; Li, L. Correlation Analysis of Geological Hazards and SoilRock Types in Taishan Scenic Area. Energy Technol. Manag. 2023, 48, 178–180. (In Chinese) [Google Scholar]
Applsci 15 05482 g001
Figure 1. The map of the study area.
Figure 1. The map of the study area.
Applsci 15 05482 g001
Applsci 15 05482 g002
Figure 2. DJI Matrice 300 RTK UAV.
Figure 2. DJI Matrice 300 RTK UAV.
Applsci 15 05482 g002
Applsci 15 05482 g003
Figure 3. 3D model: (a) Phase I 3D model; (b) Phase II 3D model.
Figure 3. 3D model: (a) Phase I 3D model; (b) Phase II 3D model.
Applsci 15 05482 g003
Applsci 15 05482 g004
Figure 4. 3D reconstruction of mining area: (a) Phase I point-cloud model; (b) Phase II point-cloud model.
Figure 4. 3D reconstruction of mining area: (a) Phase I point-cloud model; (b) Phase II point-cloud model.
Applsci 15 05482 g004
Applsci 15 05482 g005
Figure 5. Classification of ground points in point-cloud models.
Figure 5. Classification of ground points in point-cloud models.
Applsci 15 05482 g005
Applsci 15 05482 g006
Figure 6. Digital Elevation Model.
Figure 6. Digital Elevation Model.
Applsci 15 05482 g006
Applsci 15 05482 g007
Figure 7. Contour map: (a) Phase I contour map; (b) Phase II contour map.
Figure 7. Contour map: (a) Phase I contour map; (b) Phase II contour map.
Applsci 15 05482 g007
Applsci 15 05482 g008
Figure 8. Point-cloud difference model.
Figure 8. Point-cloud difference model.
Applsci 15 05482 g008
Applsci 15 05482 g009
Figure 9. Distribution map of hazardous areas.
Figure 9. Distribution map of hazardous areas.
Applsci 15 05482 g009
Applsci 15 05482 g010
Figure 10. Slope map: (a) Phase I slope map; (b) Phase II slope map.
Figure 10. Slope map: (a) Phase I slope map; (b) Phase II slope map.
Applsci 15 05482 g010
Applsci 15 05482 g011
Figure 11. Curvature diagram: (a) Phase I curvature diagram; (b) Phase II curvature diagram.
Figure 11. Curvature diagram: (a) Phase I curvature diagram; (b) Phase II curvature diagram.
Applsci 15 05482 g011
Table 1. Main parameters of the UAV photogrammetry platform.
Table 1. Main parameters of the UAV photogrammetry platform.
Equipment ParametersNorm
Weight (with battery)6.3 kg
Maximum horizontal flight speed23 m/s
Navigation systemGPS + GLONASS + BeiDou + Galileo
RTK position accuracy1.5 cm + 1 ppm (vertical);
1 cm + 1 ppm (horizontal)
Camera PixelsWide-angle 12 megapixels;
Zoom 20 megapixels
Pitch of the headPitch: −120° to +30°
Horizontal: ±320°
Maximum wind speed15 m/s
Maximum flight time per trip55 min
Table 2. Aerial triangulation for quality assessment report.
Table 2. Aerial triangulation for quality assessment report.
Parameter ClassificationParameter NameParametric Indicators
Image InformationNumber of images74
With positional image74
Calibrated images74
Root-mean-square error of geographic alignmentPhase I 0.957 m;
Phase II 1.364 m
timesairtimePhase I 37 s; Phase II 53 s
RTK solution valuefixed solutionPhase I 74; Phase II 74
floating solutionPhase I 0; Phase II 0
single-point solutionPhase I 0; Phase II 0
Other solutionsPhase I 0; Phase II 0
Table 3. Matrix of indicators for first-level evaluations.
Table 3. Matrix of indicators for first-level evaluations.
AB1B2B3B4WB
B113570.513
B21/31350.289
B31/51/3130.135
B41/71/51/310.054
Table 4. Weights of secondary evaluation indicators.
Table 4. Weights of secondary evaluation indicators.
BC1C2C3C4C5C6C7C8C9C10C11
WC0.2770.1520.0840.1630.0560.0290.0500.0190.0390.0770.036
Table 5. Classification and scoring of secondary evaluation indicators.
Table 5. Classification and scoring of secondary evaluation indicators.
Evaluation ProjectsClassification of Indicators and Quantitative Scores Ci
10.6180.382
C1: Rock and Soil Typeloose stacked layer (geology)semi-hard rocksolid bedrock
C2: Structural Plane Densitylow densityMedium densitylow density
C3: Weathering DegreeStrong weathering: weathering index > 0.6Medium weathering:
0.2 < weathering index ≤ 0.6		Unweathered: weathering index ≤ 0.2
C4: Slope GradientSteep slopes: slope > 30°Medium gradient: 15–30° gradientGentle slopes: slopes < 15°
C5: Slope Aspectdownhilltransverse slopereverse slope
C6: Curvatureconvex slopestraight slopeconcave slope
C7: Vegetation CoverageLow coverage: <30% coverageMedium coverage: 30–70% coverageHigh coverage: >70% coverage
C8: Annual Average RainfallHigh rainfall areas:
annual rainfall > 800 mm		Medium rainfall zone: annual rainfall 400–800 mmLow rainfall zone:
annual rainfall < 400 mm
C9: Groundwater InfluenceLong-term saturation:
water table depth < 3 m		Seasonal impact:3 m ≤ water level depth ≤ 5 mNo groundwater: depth to water table > 5 m
C10: Seismic IntensityHigh intensity (>VII)Medium intensity (degrees VI–VII)Low intensity (<VI)
C11: Human-Induced DisturbanceLarge-scale excavation: >30% of excavated areaPartial excavation: 10% ≤ excavated area ≤ 30%Undisturbed: <10% of excavated area
C12: Vegetation Destruction ConditionSevere damage: recovery rate < 30%Moderate damage: 30–60% recovery rateMinor damage: recovery rate > 60%
Note: The underlined content in the table represents the secondary classification indicators of the study area.
Table 6. Calculation results of slope risk assessment.
Table 6. Calculation results of slope risk assessment.
i1234567891011
C i 0.2770.1520.0840.1630.0560.0290.0500.0190.0390.0770.036
W C i 0.3820.3820.61810.6180.38210.6180.3820.6181
C i W C i 0.1060.0580.0520.1630.0350.0110.0500.0120.0150.0480.036
i1234567891011
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Li, N.; Qiu, H.; Zhai, H.; Chen, Y.; Wang, J. Monitoring and Analysis of Slope Geological Hazards Based on UAV Images. Appl. Sci. 2025, 15, 5482. https://doi.org/10.3390/app15105482

AMA Style

Li N, Qiu H, Zhai H, Chen Y, Wang J. Monitoring and Analysis of Slope Geological Hazards Based on UAV Images. Applied Sciences. 2025; 15(10):5482. https://doi.org/10.3390/app15105482

Chicago/Turabian Style

Li, Nan, Huanxiang Qiu, Hu Zhai, Yuhui Chen, and Jipeng Wang. 2025. "Monitoring and Analysis of Slope Geological Hazards Based on UAV Images" Applied Sciences 15, no. 10: 5482. https://doi.org/10.3390/app15105482

APA Style

Li, N., Qiu, H., Zhai, H., Chen, Y., & Wang, J. (2025). Monitoring and Analysis of Slope Geological Hazards Based on UAV Images. Applied Sciences, 15(10), 5482. https://doi.org/10.3390/app15105482

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop