Method for automatically generating most unfavorable section of slope of mass shallow soil landslide disaster risk

Abstract

This invention discloses an automatic method for generating the most unfavorable profile of a slope at risk of clustered shallow soil landslides, belonging to the field of geological disaster monitoring and safety early warning technology. This invention automatically extracts multi-dimensional terrain feature parameters along each extension line, including slope length, elevation difference, average slope, steepest slope near the slope, slope-wall distance, and cumulative length of steep slopes. Based on preset multi-parameter combination judgment rules, the risk level of each extension line is automatically determined. For each building, from its highest risk level lines, innovative rules such as a comprehensive slope aspect angle based on slope weighting are used to intelligently select a unique and most representative most unfavorable profile line. This invention achieves full-process automation from risk zone identification to the quantification and visualization of specific threat paths, possessing the advantages of high efficiency, accuracy, and batch processing capability, providing key technical support for early identification, investigation and evaluation, and monitoring and early warning of landslide disasters.

Description

Technical Field

[0001] This invention relates to the field of geological disaster monitoring and safety early warning technology, specifically to an automated method for generating the most unfavorable profile of a slope at risk of clustered shallow soil landslides. Background Technology

[0002] Landslides, a typical geological hazard, usually occur in mountainous, hilly, or sloping areas, characterized by their suddenness and destructiveness. With increasing global climate change and human activities, landslides are becoming more frequent and their harm is increasing year by year, seriously affecting people's lives and property, and causing substantial economic losses and social impacts. Traditional landslide investigation and assessment methods mostly rely on geological exploration, field investigations, and manual analysis, which suffer from problems such as large workload, high cost, low efficiency, and incomplete coverage in regional investigation and evaluation work.

[0003] Currently, with the development of remote sensing and artificial intelligence technologies, landslide disaster investigation, assessment, monitoring, and early warning methods based on remote sensing data and Geographic Information Systems (GIS) have gradually become a research hotspot. These technologies can conduct large-scale real-time monitoring through various means such as drones, satellites, and aerial platforms, providing accurate topographic and soil layer distribution information, and providing strong technical support for the identification and investigation of landslide disasters. The invention patent "CN202510992041.1 An Automated Identification Method for Risk Zones of Clustered Shallow Soil Landslide Disasters" is a new method proposed in this context.

[0004] However, while existing technologies can automatically identify risk areas (slopes) of clustered shallow soil landslides and mark suspected threatened house patches, several problems still exist: 1. After identifying the risk area where a cluster of shallow soil landslides may occur, during the field investigation, due to the high mountains, steep slopes, and dense vegetation, it is still difficult to accurately define the slope distribution and terrain undulation of the main profile of the slope. 2. When technicians want to quantitatively verify slope stability, they need to manually draw lines and read elevation values ​​to form profile maps, which is extremely labor-intensive for regional surveys and evaluations. 3. When it is necessary to install monitoring equipment on high-risk slopes, if there is no slope unit profile diagram, it is impossible to intuitively determine the potential disaster-causing area, which may result in the installation of monitoring equipment not being targeted, which will inevitably affect the monitoring and early warning effect. Summary of the Invention

[0005] The purpose of this invention is to provide an automated method for generating the most unfavorable profile of slopes at risk of clustered shallow soil landslides. This invention not only solves the efficiency bottleneck through automation, but also achieves significant progress in assessment accuracy, model robustness, and application completeness through a series of ingenious algorithm designs (dynamic slope toe determination, weighted direction selection, and inward detection). It produces multiple positive effects, such as improving algorithm intelligence, conforming to engineering thinking, and filling assessment blind spots, and has high practical value and promising prospects for promotion.

[0006] The technical solution adopted in this invention is as follows: An automated method for generating the most unfavorable profile of a slope at risk of clustered shallow soil landslides, based on a digital elevation model and building vector data, is implemented using geographic information system software, including the following steps: S1. Risk Line Generation and Parameter Extraction: Starting from the feature points of the building vector surface boundary, generate multiple extension lines of preset lengths outward; extract terrain elevation information along each extension line, identify and calculate multiple terrain feature parameters of the line; S2. Risk Level Assessment: Based on the preset risk level assessment rules, the terrain feature parameters of each extension line are comprehensively assessed, and a risk level is assigned to each line; S3. Selection of the most unfavorable profile line: For each house, select the line with the highest risk level from all its corresponding extension lines as a candidate line set; according to the selection rules, determine one line from the candidate line set as the most unfavorable profile line representing the threatened situation of the house. S4. Automatic generation of profile map: Based on the most unfavorable profile line determined in step S3, automatically extract its terrain profile information and generate a visual profile map.

[0007] Preferably, the terrain feature parameters mentioned in step S1 include: the elevation difference of the highest slope point (H1), the horizontal distance between the highest slope point and the base of the house wall (L1), the slope of the line connecting the highest slope point and the base of the house wall (S1), the horizontal distance between the slope foot point and the base of the house wall (L3), and the average slope of the slope (△S), where △S = arctan((H1 - H3) / (L1 - L3)), and H3 is the elevation of the point with a vertical elevation difference of 1 meter from the base of the house wall.

[0008] Preferably, the terrain feature parameters mentioned in step S1 further include: the maximum value of the angle between the slope point and the line connecting the house wall base to the horizontal line (S2) among the slope points with a horizontal distance L2 between 20 meters and 100 meters and not exceeding L1, and the cumulative horizontal length (LpX) along the horizontal length L1 of the profile line with a slope greater than or equal to X degrees, where X is a preset slope threshold.

[0009] Preferably, the risk level judgment rule described in step S2 defines multiple risk levels by setting threshold combinations for the terrain feature parameters; The criteria for determining extremely high risk include: L1 ≥ 20 meters, H1 ≥ 30 meters, △S ≥ 30 degrees, S2 ≥ 40 degrees, L3 ≤ 6 meters, and Lp35 ≥ 20 meters; The criteria for determining high risk are any combination of the following sub-conditions: High risk 1: L1 ≥ 20 meters, H1 ≥ 30 meters, △S ≥ 30 degrees, S2 ≥ 30 degrees, L3 ≤ 12 meters, and Lp30 ≥ 25 meters; High risk 2: L1 ≥ 20 meters, H1 ≥ 30 meters, L2 < 25 meters, S2 ≥ 40 degrees, L3 ≤ 12 meters, and Lp30 ≥ 25 meters; High risk 3: L1 ≥ 20 meters, H1 ≥ 30 meters, L2 ≥ 25 meters, S2 ≥ 35 degrees, L3 ≤ 12 meters, and Lp30 ≥ 25 meters.

[0010] Preferably, in step S1, the method for determining the horizontal distance (L3) at the toe of the slope includes: S11. Perform smoothing filtering on the original high-order sequence on the extension line; S12. Based on the filtered high program sequence, determine the effective height difference threshold for judging the slope foot at each point on the extension line according to the preset dynamic rules, wherein the effective height difference threshold increases as the horizontal distance between the current point and the starting point increases; S13. Simultaneously analyze the mutation characteristics of the original high-order sequence, identify the points where the elevation difference changes exceed the preset mutation threshold, and mark them as suspected steep slope points; S14. Merge the points that meet the dynamic threshold conditions in step S12 with the suspected steep slope points identified in step S13 to form a candidate point set. S15. From the set of candidate points, select the point that is closest to the starting point of the extension line in terms of horizontal distance, and determine its horizontal distance as the horizontal distance (L3) of the slope toe point of the extension line.

[0011] Preferably, the selection rule in step S3 is as follows: For a given house, the candidate line set for the highest risk level is: If there are candidate lines in the set with a horizontal projection length L1 ≥ 20 meters, then the candidate line with the smallest average angle with the comprehensive slope aspect is selected as the most unfavorable profile line. If the horizontal projection length L1 of all candidate lines in the set is less than 20 meters, then the candidate line with the longest L1 is selected as the most unfavorable profile line.

[0012] Preferably, the average value of the angle between the slope and the overall slope direction is the weighted average value obtained by weighting the angle between the direction of the profile line and the slope direction at each point on the profile line, with the slope at that point as the weight.

[0013] Preferably, in step S1, when generating the extension line, the method further includes: Identify concave vertices in the vector plane of a building whose interior angle is greater than 180 degrees; Starting from the concave vertex, one or more inward detection lines are generated along the direction pointing towards the interior space of the house; The inward probe line participates in the subsequent steps S2 and S3.

[0014] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method for automatically generating the most unfavorable profile of a slope at risk of clustered shallow soil landslides.

[0015] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for automatically generating the most unfavorable profile of a slope at risk of clustered shallow soil landslides.

[0016] Compared with the prior art, the present invention has the following significant advantages: This invention standardizes and algorithms the entire process of "extension line generation—parameter extraction—risk identification—profile selection—graphic drawing" and integrates it into a GIS plugin, achieving one-click, batch processing. For detailed regional geological hazard investigations, this method can complete profile drawing work that previously required months of manpower in just a few hours to a few days, making the impossible regional detailed assessment a reality and greatly freeing up technical manpower.

[0017] This invention automatically generates a precise, quantified "most unfavorable profile line" and its complete topographic profile map for each threatened building. Field technicians can use this map to directly locate key locations requiring focused investigation, such as paths, slope toes, and steep slope sections. This transforms fieldwork from "large-scale general surveys" to "verification along precise paths," significantly improving the targeting, safety, and efficiency of fieldwork.

[0018] This invention introduces a multi-parameter fusion-based quantitative criterion, shifting risk classification from "qualitative experience" to "quantitative models." Instead of simple binary judgments, this invention constructs a comprehensive, multi-level risk assessment model encompassing over ten terrain feature parameters, including slope size (H1, L1), overall slope (ΔS), near-slope steepness (S2), slope-wall distance (L3), and cumulative steep slope length (LpX). By setting precise threshold combinations for "extremely high risk," "high risk," and "medium risk," the risk assessment results become more objective, precise, and repeatable, reducing subjective judgment errors caused by individual variations. The risk level classification has a solid physical meaning and statistical basis.

[0019] This invention employs a "slope-weighted comprehensive aspect angle" method to select master profiles, making the automated results more consistent with geomechanical principles. When determining the "most unfavorable profile line," this invention does not use a simple geometric mean, but innovatively employs a weighted average with slope as the weight to calculate the comprehensive angle between the profile line and the slope aspect. This means that the algorithm assigns higher weight to the direction of steeper, less stable sections. This results in the final selected profile line not being the one that appears most parallel to the slope, but rather the one that is physically most likely to experience instability and slippage. This design ensures that the decision-making logic of the automated program highly aligns with the engineering thinking of experienced geological engineers who "focus on the most unfavorable sections," significantly enhancing the engineering credibility and scientific value of the automated results.

[0020] To address the issue of DEM data noise interference in determining slope toe when using a fixed elevation threshold, this invention effectively filters out misjudgments of subtle terrain undulations in the distance by setting a dynamic threshold that increases with distance. Simultaneously, it detects abrupt changes in the original elevation data to capture steep slopes formed by nearby artificial cuts. Finally, the two criteria are combined, and the point closest to the building is selected as the slope toe. This combined approach, with very low computational cost, greatly enhances the algorithm's adaptability to different terrains (natural gentle slopes, artificial steep slopes) and varying data quality, making analysis results based on conventionally accurate DEMs more stable and reliable.

[0021] Traditional risk assessments assume threats originate from outside the building, with all analysis lines extending outwards. This invention, through polygonal geometric analysis of the building, automatically identifies "concave angles" and generates "inward probe lines" pointing towards the building's interior spaces (such as courtyards and patios). This design, at virtually zero cost, expands the assessment scope from "slopes behind the house" to "slopes within the building site," effectively covering the equally frequent but often overlooked risk of landslides within the yard caused by slope-cutting construction. This significantly improves the completeness of the risk assessment model and the lead time for early warnings, representing an important and practical supplement to existing assessment frameworks.

[0022] Without precise cross-sectional diagrams, the deployment of monitoring equipment often becomes haphazard. The most unfavorable cross-sectional diagram generated by this invention clearly indicates the extent of the potential landslide, the main sliding direction, and key components. This allows for targeted deployment of monitoring equipment (such as GNSS, crack gauges, and rain gauges), directly installed at the most likely deformation locations and arrayed along the main sliding direction. This ensures sensitive and effective monitoring data, significantly improving the accuracy and timeliness of monitoring and early warning systems.

[0023] The standard cross-sectional diagram automatically generated by this invention, with key parameter annotations, can be directly used as an input model for quantitative stability calculations such as limit equilibrium methods and numerical simulations, avoiding secondary errors in manual data preparation. Simultaneously, it provides clear operational locations and design parameters for subsequent design of engineering mitigation measures such as anti-slide piles, retaining walls, and drainage ditches, promoting the digitalization and integration of the entire process of geological disaster prevention and control, from risk identification to engineering design.

[0024] In summary, this invention not only solves the efficiency bottleneck through automation, but also achieves significant progress in evaluation accuracy, model robustness, and application completeness through a series of ingenious algorithm designs (dynamic slope determination, weighted direction selection, and inward detection). It produces multiple positive effects, such as improving algorithm intelligence, conforming to engineering thinking, and filling evaluation blind spots, and has high practical value and promising prospects for promotion. Attached Figure Description

[0025] Figure 1 This is a schematic flowchart of the method of the present invention; Figure 2 This is a schematic diagram of the results of the generation of building patch extension lines and extraction of risk lines in this invention; Figure 3 This is a schematic diagram of the most unfavorable profile line extraction results of the present invention; Figure 4 This is a cross-sectional view automatically generated by the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0027] See Figures 1 to 4 This invention discloses an automated method for generating the most unfavorable profile of a slope at risk of clustered shallow soil landslides. The implementation of this invention relies on a professional geographic information processing platform and high-precision data. Typically, GIS software with powerful spatial analysis and secondary development capabilities, such as ArcGIS Pro and QGIS, is chosen as the operating environment. The required core data includes: High-precision digital elevation model (DEM): This is the cornerstone of all terrain parameter calculations. DEM data should be acquired using technologies such as airborne lidar and UAV photogrammetry to ensure sufficient vertical and horizontal resolution (e.g., grid spacing better than 1 meter) even in complex terrains such as hills and mountains, in order to accurately reflect micro-topographic features such as steep slopes, toes, and ridgelines.

[0028] Rural housing cadastral vector data: provided in the form of an isometric vector layer, containing the precise spatial boundaries of all buildings within the study area. Each housing patch will serve as the starting point for risk assessment (the target point of the threat source) and the basic unit for spatial association.

[0029] Derived terrain parameter layers: Based on the above DEM, two important raster layers need to be pre-calculated and generated: Slope layer: The value of each raster cell represents the slope angle of the ground at that point.

[0030] Slope layer: The value of each raster cell represents the direction of the ground slope at that point (0-360 degrees).

[0031] These two layers will play a crucial role in the subsequent calculation of the match between the profile line direction and the main dip of the terrain.

[0032] The core process of this invention, a method for monitoring and early warning of three-dimensional deformation of slopes based on cascaded single-line multi-point crack gauges, is as follows: S1: Risk Detection Line Generation and Multidimensional Terrain Parameter Extraction This step is the data preparation phase for all subsequent analyses. Its purpose is to construct a "detection line network" covering the potential threat area behind each building and to quantify the terrain features of each detection line, specifically including: S1.1 Saturated Extension Line Generation Mechanism Using a single building surface vector pattern as the processing unit, a series of straight line segments radiating outward from the building boundary are systematically generated, called "extension lines" or "detection lines".

[0033] S1.1.1 Generation start point and direction: Boundary edge method: Set starting points at fixed intervals along each outer boundary edge of the building facade. Starting from each starting point, generate an extension line along the normal direction of that edge (i.e., perpendicular to the building walls and pointing outwards).

[0034] Inflection point densification method: At each convex corner of the building's facade, in addition to generating a line along the normal direction of each side edge, multiple extension lines are generated at fixed angle intervals within the fan-shaped area formed by these two lines. This ensures that all possible slopes from the sides and rear of the building are detected without omission.

[0035] Special treatment for concave points: For the apex of the house boundary that is concave inward (internal angle greater than 180 degrees), an additional "inward probe line" needs to be generated from this point towards the interior space of the house to assess the risks that courtyards, patios, and other internal areas may face due to steep internal slopes or instability of retaining walls. This line is involved in all subsequent analysis processes.

[0036] S1.1.2 Initial Length and Intelligent Termination: The initial horizontal projection length of all extension lines is preset to a sufficiently large value, such as 300 meters. However, the actual calculated length is not fixed at 300 meters, but is dynamically determined by an intelligent algorithm. The program starts from the starting point, traverses the DEM elevation points along the extension line, and records the maximum elevation value Hmax encountered in real time. Once it detects that the current point elevation Hx satisfies Hmax - Hx > 1 meter, the program immediately stops searching backwards. At this time, the point where Hmax is located is identified as a "local ridge point" or "highest terrain point" in that direction, its elevation is recorded as H1, and its horizontal distance from the starting point is recorded as L1. If this condition is not triggered within 300 meters, the effective calculated length of the extension line is 300 meters. This mechanism cleverly and automatically determines the slope range corresponding to each detection line, avoiding invalid long-distance calculations and focusing on the nearest slope that poses a direct potential threat to buildings.

[0037] S1.2 Calculation of refined terrain feature parameters For each valid extension line, the system automatically extracts and calculates a standardized set of terrain feature parameters, which are risk assessment indicators with clear physical meaning.

[0038] S1.2.1 Macroscopic morphological parameters (describing the basic scale of the entire potential landslide body): H1 (Maximum Relative Elevation Difference): The vertical elevation difference between the highest point (ridge point) on the extension line and the base of the building wall (starting point). Unit: meters. Reflects the potential energy of the potential landslide body.

[0039] L1 (Maximum Horizontal Distance): The horizontal distance between the point corresponding to H1 and the starting point. Unit: meters. Reflects the horizontal scale of the potential landslide.

[0040] S1 (Overall Apparent Slope): S1 = arctan(H1 / L1). Unit: degrees. Reflects the average steepness of the line connecting the house and the ridge.

[0041] S1.2.2 Characteristic parameters of the near-slope section (focusing on the local terrain near houses that has a direct impact on the initiation and movement of disasters): S2 (Maximum Slope Near the Starting Point): Calculate the slope of the line connecting each point to the starting point from any point within a horizontal distance L2 between 20 and 100 meters (and L2 ≤ L1), and take the maximum value. This parameter aims to identify whether there are particularly steep slope sections within close proximity to the building; such steep slopes are often the starting point or acceleration zone of shallow landslides. Rule: If L1 < 20 meters, the slope is considered too small, and S2 and related parameters are not calculated.

[0042] H2: The vertical height difference between the point corresponding to S2 and the starting point.

[0043] L2: The horizontal distance between the point corresponding to S2 and the starting point.

[0044] S1.2.3 Slope-wall relationship parameters (describing the spatial proximity relationship between the slope and the building): L3 (Horizontal distance at the toe of the slope): Starting from the starting point, along the extension line, the horizontal distance from the first point whose vertical elevation difference with the starting point is ≥ 1 meter. This point is approximated as the "toe of the slope". Setting a 1-meter threshold is to effectively filter out misjudgments caused by DEM errors, small mounds of soil behind buildings, or field ridges. On natural slopes or artificially cut slope toes, the terrain changes drastically, and the error between L3 determined by this method and the actual toe of the slope is usually within 1 meter, meeting engineering requirements.

[0045] S1.2.4 Average Slope Parameter: ΔS (Effective Average Slope): The average slope of the slope segment from the toe (horizontal distance L3, height approximately H3 ≈ starting elevation + 1 meter) to the highest point (L1, H1). The calculation formula is: ΔS = arctan((H1 - H3) / (L1 - L3)). This parameter eliminates the influence of potentially flat ground between buildings and the toe, and better represents the average inclination angle of the slope itself, which is more likely to slide.

[0046] S1.2.5 Cumulative parameters of steep slope scale (quantifying the development scale of steep sections and reflecting the volume potential of unstable soil): Lp25, Lp30, and Lp35 represent the cumulative horizontal projection lengths of sections with slopes p greater than or equal to 25 degrees, 30 degrees, and 35 degrees, respectively, within a horizontal distance L1. For example, Lp30 ≥ 25 meters means that the total length (horizontal projection) of steep slope sections with a slope exceeding 30° on this profile reaches more than 25 meters, indicating the existence of a considerable steep slope area.

[0047] S1.3 Robustness testing algorithm for slope toe distance L3 To improve the accuracy and anti-interference capability of L3 identification, this invention employs a composite algorithm combining smoothing filtering and terrain change detection, the specific steps of which are as follows: S1.3.1 Data Smoothing: The original high-order sequence of the extension line is smoothed and filtered to suppress high-frequency noise that may exist in the DEM data, so as to obtain a high-order sequence that better reflects the terrain trend.

[0048] S1.3.2 Dynamic Threshold Detection: Based on the smoothed sequence, a dynamically increasing effective elevation difference threshold is designed. This threshold gradually increases as the distance L from the current calculation point to the starting point increases. For example, when L < 5 meters, the threshold can be set to 0.5 meters to capture small slopes nearby; when L > 20 meters, the threshold can be increased to 2 meters to avoid misjudging distant natural undulations as slopes. The point A that first exceeds the dynamic threshold corresponding to the current point is recorded.

[0049] S1.3.3 Abrupt Change Detection: Simultaneously analyze the local change rate of the original elevation sequence to identify abrupt terrain changes. Calculate the elevation difference change rate (approximate slope) between adjacent points. When this value exceeds a high preset "abrupt change threshold" (e.g., corresponding to a slope > 60°), the point is marked as "suspected steep slope point B". This effectively captures obvious terrain transitions such as artificially cut slopes and steep cliffs.

[0050] S1.3.4 Candidate Point Set Merging: The dynamic threshold point A found in step 2 and all suspected steep slope points B found in step 3 are merged to form a candidate point set. This set contains information on both "soft slope toes" based on cumulative elevation changes and "hard slope toes" based on abrupt slope changes.

[0051] S1.3.5 Determining the final L3: From the above candidate point set, select the point closest to the starting point; its horizontal distance is determined as the horizontal distance L3 at the toe of the extension line. This method comprehensively considers both gradual slope changes and abrupt terrain changes, enabling more robust identification of the true and effective toe location.

[0052] S2: Intelligent risk level determination based on multi-indicator combination After obtaining the complete parameter set for all extension lines, the system determines the risk level of each extension line based on a pre-set rule base with clear physical meaning and statistical experience. The risk levels are divided into three categories: "extremely high risk," "high risk," and "medium risk," with decreasing priority in that order. If a line meets the rules for multiple risk levels, the highest risk level is applied.

[0053] The risk level determination rules are as follows: S2.1 [Extremely High Risk] Judgment Criteria (All 6 conditions must be met simultaneously): L1 ≥ 20 meters. (Scale requirement: The slope must have a certain horizontal extension length to ensure that the potential landslide body has sufficient scale). H1 ≥ 30 meters. (Potential energy condition: A large relative elevation difference provides sufficient gravitational potential energy for the landslide). ΔS ≥ 30 degrees. (Overall slope condition: The average slope of the entire effective sliding surface is very steep, which is conducive to sliding). S2 ≥ 40 degrees. (Steep slope conditions: There is an extremely steep slope near the house, which can easily cause surface instability). L3 ≤ 6 meters. (Adjacent conditions: The toe of the slope is extremely close to the house; once a landslide starts, there is almost no buffer space and it may impact the house.) Lp35 ≥ 20 meters. (Steep slope scale condition: Extremely steep slope sections (≥35°) have a large development scale, which means that there are a large number of unstable soil masses).

[0054] A slope that meets all the above conditions, from a topographical perspective, possesses all the favorable topographical factors for a large-scale, powerful shallow landslide, and the disaster location is directly adjacent to the disaster-bearing body, therefore it is judged as "extremely high risk".

[0055] S2.2 [High Risk] Judgment Criteria (meeting any one of the following three sub-combinations is sufficient): High risk 1: L1 ≥ 20 meters, H1 ≥ 30 meters, ΔS ≥ 30 degrees, S2 ≥ 30 degrees, L3 ≤ 12 meters, and Lp30 ≥ 25 meters.

[0056] High risk 2: L1 ≥ 20 meters, H1 ≥ 30 meters, L2 < 25 meters, S2 ≥ 40 degrees, L3 ≤ 12 meters, and Lp30 ≥ 25 meters.

[0057] High risk 3: L1 ≥ 20 meters, H1 ≥ 30 meters, L2 ≥ 25 meters, S2 ≥ 35 degrees, L3 ≤ 12 meters, and Lp30 ≥ 25 meters.

[0058] High-risk conditions have slightly relaxed thresholds for certain parameters compared to extremely high-risk conditions (such as S2, L3, or ΔS), but still strictly limit the scale, elevation difference, local steepness, and proximity. The three sub-combinations characterize various terrain patterns of high risk from different perspectives (overall steep, extremely steep nearby but slightly distant, and relatively steep but at a certain distance). If any one of these conditions is met, the likelihood of a landslide occurring and threatening houses is considered very high.

[0059] S2.3 [Medium Risk] Judgment Criteria (meeting any one of the following four sub-combinations is sufficient): Medium risk 1: L1 ≥ 20 meters, H1 ≥ 15 meters, S2 ≥ 25 degrees, ΔS ≥ 25 degrees, L3 ≤ 6 meters.

[0060] Medium risk 2: L1 ≥ 20 meters, H1 ≥ 15 meters, L2 < 21 meters, S2 ≥ 35 degrees, L3 ≤ 6 meters.

[0061] Medium risk 3: L1 ≥ 20 meters, H1 ≥ 15 meters, L2 ≥ 21 meters, S2 ≥ 30 degrees, L3 ≤ 6 meters.

[0062] Medium risk 4: 5 ≤ L1<20 meters, H1 ≥ 15 meters, S1 ≥ 45 degrees, L3 ≤ 6 meters.

[0063] The medium-risk conditions have further relaxed the requirements for scale, elevation difference, or steepness, but still emphasize a certain degree of terrain steepness and spatial proximity. In particular, medium-risk level 4 is aimed at special cases that are small in scale (L1<20 meters) but extremely steep (S1≥45 degrees) and close to houses (L3≤6 meters), such as small artificial slopes with high and steep slopes behind houses.

[0064] S2.4 House Risk Level Consolidation: A house may correspond to multiple extension lines with different risk levels. The final risk level of the house is the highest risk level among all its corresponding extension lines. For example, if a house is crossed by 5 extension lines, of which 1 is "extremely high risk", 2 are "high risk" and 2 are "medium risk", then the house is marked as an "extremely high risk house".

[0065] S2.5 Risk Line Spatial Deduplication: To prevent the same dangerous slope segment from being repeatedly assigned to multiple adjacent houses, resulting in redundant results, a spatial filtering rule is set: if the endpoint (or segment) of a risk level extension line is within 30 meters of the boundary of another house patch, this extension line is deleted. This ensures that each identified risk line is primarily associated with its nearest and most significant threat target.

[0066] S3: Intelligent selection of the most unfavorable section line for a single building For each house marked with a risk level, this step intelligently selects the most representative "most unfavorable profile line" from its numerous associated extension lines as a digital representation of the most important and typical threat path faced by the house, which is then used for the final mapping.

[0067] The selection process is divided into two levels: S3.1 Initial Screening: Constructing the Highest Risk Candidate Set Extract the lines with the highest risk level from all extension lines associated with the house, forming a "candidate line set for the highest risk level". For example, if a house is "high-risk", then all lines marked "high-risk" are selected from all its extension lines to form a candidate set. If the house is "extremely high-risk", then the candidate set consists of all "extremely high-risk" lines.

[0068] S3.2 Final Selection: Determine the unique worst-case scenario from the candidate set. From the obtained set of highest-risk candidate lines, the following rule is applied to determine the unique most unfavorable profile line: Scenario A: There are large candidate lines (L1 ≥ 20 meters).

[0069] In this scenario, the slope exhibits a certain level of hazard. From all candidate lines satisfying L1 ≥ 20 meters, the line with the smallest "average angle with the overall slope aspect" is selected. The "average angle with the overall slope aspect" parameter measures the consistency between the profile line direction and the natural slope aspect of the terrain.

[0070] For each grid point on the candidate line, obtain the ground slope aspect (from the slope aspect layer) and profile line direction for that point.

[0071] Calculate the angle between these two directions (range 0-180 degrees). A 0-degree angle indicates that the profile line direction is completely consistent with the slope direction (downhill); 90 degrees indicates that it is perpendicular (dipping perpendicular to the slope direction); and 180 degrees indicates that it is completely opposite (uphill).

[0072] Calculate the weighted average of the included angles of all points on the profile, with the weights being the slope values ​​of each point (from the slope layer). This means that the steeper the location, the stronger the control of its slope aspect over the direction of landslide movement, and the greater its weight in the calculation.

[0073] The profile with the smallest value means that its extension direction most closely matches the direction of maximum slope inclination (i.e., the direction in which material is most easily transported). The landslide body is most likely to move along this path, therefore its threat to houses is most "targeted" and "direct" in direction.

[0074] Scenario B: All candidate lines are relatively small (L1 < 20 meters).

[0075] In this scenario, the slope size is limited. Therefore, the line with the largest L1 value in the candidate line set is directly selected as the most unfavorable profile line. The logic here is: under the premise of the same risk level, choose the relatively longest path to reflect the relatively largest potential threat range.

[0076] S4: Automated Profile Generation and System Integration Implementation Once the most unfavorable profile line is determined, the system can automatically generate standardized terrain profile maps and integrate and tool the entire process.

[0077] S4.1 Sectional View Automatically Generated: Data extraction: Based on the selected most unfavorable profile line, the system automatically overlays it with the DEM for analysis, extracts the elevation value of the center point of each grid point through which the line passes, and forms a set of "horizontal distance-elevation" data sequence.

[0078] Graphic drawing: The drawing engine is invoked to draw terrain profile curves with horizontal distance as the x-axis and elevation as the y-axis. Different scales can be used for the x and y axes to highlight the terrain undulations, such as "horizontal 1:200, vertical 1:200".

[0079] Intelligent annotation: The system automatically annotates all key topographic parameters corresponding to the generated profile, such as H1, L1, S1, ΔS, L3, S2, L2, and the "comprehensive aspect angle" value used in the calculation, around the profile or in the blank areas of the map. These annotations make the profile not only a topographic display map, but also a technical analysis map containing quantitative risk assessment criteria.

[0080] S4.2 System Integration and Tool-based Implementation: To enable batch and efficient business operations, this invention is typically integrated into the GIS platform as a professional plugin. For example, an add-in plugin can be developed using C# based on ArcGIS Pro.

[0081] Functional Flow: The plugin provides a graphical interface to guide users in inputting building vector layers, DEMs, and slope / aspect layers. Upon startup, the plugin automatically executes all the aforementioned steps: batch generation of extension lines, parallel parameter calculation, application of a rule base for risk assessment, and selection of the most unfavorable profile line for each building.

[0082] Output: The final output includes: A new vector line layer containing all risk extension lines and their risk level attributes.

[0083] A new vector point layer that marks all risky houses and their final risk level.

[0084] A single vector line layer containing only the “most unfavorable profile line” selected for each at-risk house.

[0085] Generate and export all terrain profile maps (such as images or PDFs) corresponding to the most unfavorable profile lines in one batch with a single click.

[0086] This integrated tool encapsulates complex geographic calculations, logical judgments, and graphic generation processes in the background. Users only need to perform simple data preparation and click operations to obtain a complete set of professional results, which greatly reduces the technical threshold and improves the operational efficiency of detailed investigations of regional landslide risks.

[0087] Traditional geological hazard investigation projects typically involve significant investment of human and material resources, resulting in investigation outcomes that are widely recognized in the industry and possess a "standard answer" nature (including verified hazard point ledgers, investigation reports, monitoring point layout maps, etc.). This provides an excellent benchmark for evaluating the effectiveness of the method of this invention.

[0088] In a mature project area where traditional detailed surveys have been completed, the automated method of this invention is applied to process basic data from the same area and the same period. The results automatically identified by this invention are then compared item by item with the final results of historical projects that have been validated in the field, thereby quantitatively evaluating the invention in the real world. Recall rate: How many known potential hazards can be detected; Precision: How many of the newly discovered "risk points" are effective; Efficiency gain: How much manual survey work was saved compared to the original project; Technical advantages and limitations: In which types of terrain or hidden dangers does it perform well, and in what situations may it be overlooked or misjudged; This experiment selected a typical hilly and mountainous township that had previously completed a "1:10,000 Township Geological Hazard Risk Survey and Assessment." The project had passed review and acceptance, and the list of hazard points and risk zoning maps were recognized as "ground reality." High-precision DEM (such as aerial or LiDAR data) and vector data of rural houses, using the same version as those used in the project, were also obtained. This is crucial for ensuring fairness in the comparison. The area is dominated by clustered shallow soil landslides, which highly aligns with the scope of this invention.

[0089] The testing process is as follows: Using the method of this invention, input the prepared DEM and house data, run the plugin, and obtain the "risk house layer" and the "most unfavorable section line layer".

[0090] In GIS, the "high-risk / extremely high-risk houses" output by this invention are spatially overlaid with "houses threatened by actual hidden danger points" for analysis.

[0091] Set matching rules: For example, if the risky house marked by this invention is the same building as the threatened house of a real hidden danger point, or the distance is within 10 meters, and the angle between the direction of the most unfavorable profile line generated by this invention and the main sliding direction determined in the history is less than 30 degrees, then it is considered "successfully identified".

[0092] Categorized statistics: True positive: A potential risk factor for a successful match.

[0093] False negative: Historical potential hazards were not identified by this invention (the houses they threaten were not marked with the corresponding risk level).

[0094] False positive: The high-risk houses marked by this invention have no corresponding hidden danger points in the historical records.

[0095] On-site verification: Sampling and field verification were conducted on "false positive" points and some "true positive" points to verify the accuracy of the results of this invention and to analyze the reasons for misjudgment.

[0096] The following table shows the quantitative comparison results:

[0097] Based on this experiment, the following conclusions can be drawn: With a recall rate exceeding 95%, this invention demonstrates its highly reliable ability to capture traditional hazard points that conform to the topographical characteristics of "group-onset shallow soil landslides." Its rule base, based on quantitative parameters, possesses strong generalization capabilities.

[0098] In the test, approximately 43% of the "false positives" (10 / 23) were confirmed to be valid risk points. This is not merely a "false alarm" of the method, but rather a manifestation of its core advantage—overcoming the fatigue, blind spots, and subjectivity of manual investigation, and being able to systematically and comprehensively scan all areas that conform to the dangerous terrain model, truly realizing "machine supplementing human" and potentially discovering risks that were previously overlooked.

[0099] This shifts the fieldwork from comprehensive patrols to precise verification of high-risk target areas identified by machines, increasing efficiency several times over. This makes rapid re-inspection and dynamic updating of large areas possible.

[0100] Machines excel at processing massive amounts of data quickly, objectively, and tirelessly, performing standardized calculations and rule-based judgments to complete the initial screening "from data to potential risk targets".

[0101] Human experts conduct final geological assessments of the machine's output (soil / rock quality, hydrological influences, existing signs of deformation), verify "true and false positives," handle complex situations beyond the rules (such as lateral slippage and compound disasters), and make engineering decisions.

[0102] The test revealed false negatives and false positives, providing a clear path for algorithm iteration. For example, we can explore dynamic threshold adjustment, introduce optical imagery to help remove buildings, and integrate simplified geological map unit information to distinguish soil-rock interfaces, thereby continuously optimizing model performance.

[0103] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for automatically generating the most unfavorable profile of a risk slope of a group of shallow soil landslide disasters, characterized in that, Based on digital elevation models and building vector data, and implemented using geographic information system software, the following steps are included: S1. Risk Line Generation and Parameter Extraction: Starting from the feature points of the building vector surface boundary, generate multiple extension lines of preset lengths outward; extract terrain elevation information along each extension line, identify and calculate multiple terrain feature parameters of the line; S2. Risk Level Assessment: Based on the preset risk level assessment rules, the terrain feature parameters of each extension line are comprehensively assessed, and a risk level is assigned to each line; S3. Selection of the most unfavorable profile line: For each house, select the line with the highest risk level from all its corresponding extension lines as a candidate line set; according to the selection rules, determine one line from the candidate line set as the most unfavorable profile line representing the threatened situation of the house. S4. Automatic generation of profile map: Based on the most unfavorable profile line determined in step S3, automatically extract its terrain profile information and generate a visual profile map.

2. The method according to claim 1, characterized in that, The terrain feature parameters mentioned in step S1 include: the elevation difference of the highest slope point (H1), the horizontal distance between the highest slope point and the base of the house wall (L1), the slope of the line connecting the highest slope point and the base of the house wall (S1), the horizontal distance between the slope foot point and the base of the house wall (L3), and the average slope of the slope (△S), where △S = arctan((H1 - H3) / (L1 - L3)), and H3 is the elevation of the point with a vertical elevation difference of 1 meter from the base of the house wall.

3. The method according to claim 2, characterized in that, The terrain feature parameters mentioned in step S1 also include: the maximum value of the angle between the slope point and the line connecting the house wall base to the horizontal line (S2) in the slope points with a horizontal distance L2 between 20 meters and 100 meters and not exceeding L1, and the cumulative horizontal length (LpX) along the horizontal length L1 of the profile line with a slope greater than or equal to X degrees, where X is a preset slope threshold.

4. The method according to claim 3, characterized in that, The risk level judgment rule described in step S2 defines multiple risk levels by setting threshold combinations for the terrain feature parameters; The criteria for determining extremely high risk include: L1 ≥ 20 meters, H1 ≥ 30 meters, △S ≥ 30 degrees, S2 ≥ 40 degrees, L3 ≤ 6 meters, and Lp35 ≥ 20 meters; The criteria for determining high risk are any combination of the following sub-conditions: High risk 1: L1 ≥ 20 meters, H1 ≥ 30 meters, △S ≥ 30 degrees, S2 ≥ 30 degrees, L3 ≤ 12 meters, and Lp30 ≥ 25 meters; High risk 2: L1 ≥ 20 meters, H1 ≥ 30 meters, L2 < 25 meters, S2 ≥ 40 degrees, L3 ≤ 12 meters, and Lp30 ≥ 25 meters; High risk 3: L1 ≥ 20 meters, H1 ≥ 30 meters, L2 ≥ 25 meters, S2 ≥ 35 degrees, L3 ≤ 12 meters, and Lp30 ≥ 25 meters.

5. The method according to claim 2 or 3, characterized in that, In step S1, the method for determining the horizontal distance (L3) at the toe of the slope includes: S11. Perform smoothing filtering on the original high-order sequence on the extension line; S12. Based on the filtered high program sequence, determine the effective height difference threshold for judging the slope foot at each point on the extension line according to the preset dynamic rules, wherein the effective height difference threshold increases as the horizontal distance between the current point and the starting point increases; S13. Simultaneously analyze the mutation characteristics of the original high-order sequence, identify the points where the elevation difference changes exceed the preset mutation threshold, and mark them as suspected steep slope points; S14. Merge the points that meet the dynamic threshold conditions in step S12 with the suspected steep slope points identified in step S13 to form a candidate point set. S15. From the set of candidate points, select the point that is closest to the starting point of the extension line in terms of horizontal distance, and determine its horizontal distance as the horizontal distance (L3) of the slope toe point of the extension line.

6. The method according to claim 1, characterized in that, The selection rules mentioned in step S3 are as follows: For a given house, the candidate line set for the highest risk level is: If there are candidate lines in the set with a horizontal projection length L1 ≥ 20 meters, then the candidate line with the smallest average angle with the slope aspect is selected as the most unfavorable profile line. If the horizontal projection length L1 of all candidate lines in the set is less than 20 meters, then the candidate line with the longest L1 is selected as the most unfavorable profile line.

7. The method according to claim 6, characterized in that, The average value of the angle between the slope and the overall slope direction is the angle between the direction of the profile line and the slope direction at each point on the profile line, and the weighted average value is obtained by weighting the slope at that point.

8. The method according to claim 1, characterized in that, In step S1, when generating the extension line, the following is also included: Identify concave vertices in the vector plane of a building whose interior angle is greater than 180 degrees; Starting from the concave vertex, one or more inward detection lines are generated along the direction pointing towards the interior space of the house; The inward probe line participates in the subsequent steps S2 and S3.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method for automatically generating the most unfavorable profile of a slope at risk of clustered shallow soil landslides as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the method for automatically generating the most unfavorable profile of a slope at risk of clustered shallow soil landslides as described in any one of claims 1 to 8.

Images