Mountain geological disaster hidden danger identification system based on spatial clustering algorithm

Abstract

The present application relates to mountain geological disaster hidden danger identification technical field, disclose a mountain geological disaster hidden danger identification system based on spatial clustering algorithm, it includes: obtain synthetic aperture radar interferometry deformation point, digital elevation model, orthographic projection and research area boundary, construct unified grid;The validity screening and observation quality record are carried out to deformation point, construct slope unit number and slope drainage connected network;The slope deformation anomaly density value is calculated, and the adaptive spatial clustering of slope drainage connected constraint is carried out;Then combine invalid data mask, slope boundary trace and elevation turning line and carry out boundary integration, generate mountain geological disaster hidden danger identification map.The present application is used for the spatial identification and mapping of mountain geological disaster hidden danger area.

Description

Technical Field

[0001] This invention relates to the field of mountain geological hazard identification technology, and more specifically, to a mountain geological hazard identification system based on spatial clustering algorithms. Background Technology

[0002] Identifying potential geological hazards in mountainous areas typically requires the comprehensive utilization of synthetic aperture radar (SAR) interferometric deformation points, digital elevation models (DEMs), orthophotos, study area boundaries, and existing geomorphological interpretation data. SAR interferometrics provides surface deformation information, DEMs offer topographically derived data such as slope, aspect, flow direction, runoff accumulation, watershed lines, and gully lines, while orthophotos assist in identifying landslide trails, lateral edges, toe lines, and changes in land cover. In engineering applications, this data usually needs to be converted to a unified projected coordinate system and a unified grid before spatial overlay, neighborhood analysis, and candidate area mapping.

[0003] In existing work, density clustering, hierarchical density clustering, slope unit division, remote sensing interpretation, and deformation anomaly screening have been used for landslide hazard identification or susceptibility analysis. In practical applications, the distribution of deformation points measured by synthetic aperture radar interferometry on the slope surface is affected by radar line-of-sight direction, side-view imaging geometry, coherence, land cover, terrain occlusion, and phase unwrapping quality; densely populated areas do not necessarily correspond to hazard zones, and sparsely populated areas do not necessarily indicate no deformation. If clustering is performed directly using planar distance, a fixed neighborhood radius, or a fixed number of points threshold, differences in point distribution caused by observation conditions can easily be introduced into the hazard zone boundaries.

[0004] In mountainous scenarios, the spatial extension of landslide hazard zones is typically constrained by slope units, confluence paths, watersheds, gully lines, and elevation transition zones at the toe of the slope. Even if two deformation points are close in distance on the projection plane, they should not be directly merged into the same candidate area if they cross a watershed, gully mainline, or an invalid area in the digital elevation model. Conversely, continuous deformation areas within the same slope unit may create observational voids due to vegetation, snow cover, shadows, or low-coherence areas, requiring processing based on slope connectivity and boundary cues. Summary of the Invention

[0005] This invention provides a mountain geological hazard identification system based on spatial clustering algorithm, which solves the technical problems mentioned in the background art.

[0006] This invention provides a mountain geological hazard hazard identification system based on spatial clustering algorithm, configured to execute sequentially: Acquire multi-source spatial data, standardize the multi-source spatial data, and construct a unified raster; The effectiveness of the deformation points measured by synthetic aperture radar interferometry is screened and the observation quality is recorded to obtain a set of effective deformation points and an invalid data mask. Construct slope unit numbering and slope drainage connectivity network based on digital elevation model; Based on the set of effective deformation points, the slope unit number, and the slope drainage network, calculate the slope deformation anomaly density value; Adaptive spatial clustering of slope drainage connectivity constraints is performed based on the slope deformation anomaly density values ​​to obtain the initial hazard clustering area; Based on the initial hazard clustering area, the invalid data mask, the slope unit number, the orthophoto slope boundary trace, and the digital elevation model elevation transition line, the hazard candidate area boundary is integrated to obtain a polygon set of hazard candidate areas; A mountain geological hazard identification map is generated based on the polygon set of the hazard candidate areas.

[0007] Preferably, acquiring multi-source spatial data, standardizing the multi-source spatial data, and constructing a unified raster includes: The time-series synthetic aperture radar data, synthetic aperture radar interferometry deformation points, digital elevation model, orthophoto image and study area boundary are used as the multi-source spatial data. The multi-source spatial data is cropped according to the boundary of the study area, and the cropped multi-source spatial data is transformed to the same projection coordinate system; The unified grid is generated based on the geocoding parameters of the time-series synthetic aperture radar data, the ground spacing between the deformation points of the synthetic aperture radar interferometry, and the ground resolution of the digital elevation model. Area-preserving resampling is performed on continuous variables, and nearest-neighbor resampling is performed on categorical and masked variables. Establish the point-to-grid mapping relationship between the synthetic aperture radar interferometry deformation points and the unified grid.

[0008] Preferably, the deformation points of synthetic aperture radar interferometry are screened for validity and the observation quality is recorded to obtain a set of valid deformation points and an invalid data mask, including: Read the processing results corresponding to the synthetic aperture radar interferometry deformation points, and extract the line-of-sight deformation rate, coherence or stability coefficient, geographical location, and radar line-of-sight direction for each synthetic aperture radar interferometry deformation point; The grid number to which each synthetic aperture radar interferometry deformation point belongs is determined based on the point-to-grid mapping relationship. The coherence or stability coefficients are normalized, and the normalized coherence or stability coefficients are used as observation quality information. Records with missing coordinates, missing rates, missing coherence, missing orbital geometry, failed phase unwrapping, non-finite values, or located outside the boundary of the study area are written into the invalid data mask; The synthetic aperture radar interferometry deformation points that are not written into the invalid data mask are taken as the set of valid deformation points.

[0009] Preferably, constructing slope unit numbering and slope drainage connectivity network based on the digital elevation model includes: The slope, aspect, flow direction, and cumulative runoff volume are calculated based on the digital elevation model, and a unit downhill direction is generated based on the slope and aspect. The slope surface is divided according to the watershed line, confluence line, slope aspect change line and slope toe elevation change line, and the slope surface unit number is assigned to the unified grid. The unified grid is used as a drainage connection grid, and adjacent connection relationships are established between adjacent drainage connection grids within the same slope unit. Establish adjacent connections between drainage connectivity grids that span slope units but share a confluence boundary and have continuous flow direction; It is prohibited to establish adjacent connections between drainage connectivity grids that cross watersheds, closed ridgelines, or invalid areas of the digital elevation model; Record the adjacent connection length, grid slope direction, grid slope angle, number of slope unit connections, and water catchment area affiliation information to obtain the slope drainage connectivity network.

[0010] Preferably, based on the set of effective deformation points, the slope unit number, and the slope drainage connectivity network, the slope deformation anomaly density value is calculated, including: The grid to be calculated is determined in a unified grid that has slope unit numbers and is connected to effective deformation points within the same catchment area; The downslope deformation radar observation sensitivity is calculated based on the unit downslope direction of the grid to which the effective deformation point belongs and the radar line of sight direction. Calculate the radar observation correction value based on the downslope deformation radar observation sensitivity, the normalized coherence or stability coefficient, and the median value of downslope deformation radar observation sensitivity within the same slope unit. Based on the line-of-sight deformation rate, the median line-of-sight deformation rate within the same slope unit, and the radar observation correction value, calculate the observation correction deformation anomaly value. Based on the adjacent connection relationships, adjacent connection lengths, grid slope aspect and grid slope angle in the slope drainage connectivity network, calculate the slope connectivity distance and determine the slope search distance range; Based on the slope connectivity distance and the slope search distance range, the observed and corrected deformation anomaly values ​​are weighted by distance to obtain the slope deformation anomaly density value.

[0011] Preferably, adaptive spatial clustering of slope drainage connectivity constraints is performed based on the slope deformation anomaly density values ​​to obtain initial hazard clustering areas, including: The uniform grid that has calculated the slope deformation anomaly density value and has a slope search distance range is used as the candidate grid. For any candidate grid, select candidate grids that are connectable in the slope drainage connectivity network within the same catchment area and do not cross the watershed or invalid areas of the digital elevation model to form a set of candidate neighboring grids; Combine the slope units corresponding to the slope unit numbers of the two candidate grids into a local slope unit combination. Calculate the density change benchmark value based on the abnormal density value of slope deformation within the local slope unit combination; The density difference value is calculated based on the difference in slope deformation anomaly density values ​​between the two candidate grids and the density change benchmark value. The clustering judgment distance value is calculated based on the slope connectivity distance between the two candidate grids, the slope search distance range, and the density difference value. The local clustering search range is determined based on the clustering judgment distance value within the candidate neighboring grid set, and the effective neighboring grid set is determined based on the local clustering search range; The minimum number of neighboring grids for the core grid is determined based on the number of connected slope units. The candidate rasters that meet the minimum number of neighboring rasters required for the core raster and are within the set of effective neighboring rasters are merged to obtain the initial hazard clustering area.

[0012] Preferably, based on the initial hazard clustering area, the invalid data mask, the slope unit number, the orthophoto slope boundary trace, and the digital elevation model elevation transition line, the hazard candidate area boundary is integrated to obtain a hazard candidate area polygon set, including: Each of the initial hazard clustering areas is rasterized into an initial region surface; Inspect the void areas formed by the invalid data mask within the same slope unit; When the cavity area is surrounded by the same initial hidden danger cluster area, belongs to the same slope unit, and does not cross the main line of the gully, the cavity area is incorporated into the initial area surface; Using the elevation turning lines of the digital elevation model and the locatable landslide trailing edge, side edge, or toe traces in the orthophoto slope boundary traces, the boundary of the initial area surface is snapped. Disconnect the parts that cross the watershed, invalid areas of the digital elevation model, or where no adjacent connection relationship has been established in the slope drainage connectivity network, and generate the polygon set of the hazard candidate area.

[0013] Preferably, generating a mountain geological hazard identification map based on the set of candidate hazard areas includes: Perform a topological validity check on each candidate polygon in the set of candidate hazard regions, remove self-intersecting edges and close dangling edges; The boundaries of the candidate hazard areas polygons after topological validity checks are snapped to the grid edges of the unified grid and the boundary lines of the slope units corresponding to the slope unit numbers. Generate vector layers and corresponding raster masks; The vector image layer is overlaid on the orthophoto or digital elevation model shaded base map; The coordinate reference, raster resolution, invalid data mask, generation time, unique number of the hidden danger area, boundary coordinates, slope unit number and water catchment area attribution information are written into the vector layer to obtain the mountain geological hazard identification map.

[0014] Compared with the prior art, the beneficial effects of the present invention include: By converting synthetic aperture radar interferometric deformation points, digital elevation models, orthophotos, and study area boundaries into a unified raster, observation quality information is preserved during validity screening. The slope deformation anomaly density is calculated using slope unit numbers and slope drainage connectivity networks, ensuring that the clustering of deformation points is simultaneously constrained by observation quality, downslope observation sensitivity, catchment area attribution, watershed line, gully main line, and slope connectivity distance. This reduces the splitting of the same hazard area due to lack of coherence, lowers the impact of radar-observable area point density on the hazard range, and limits erroneous merging of areas crossing watershed lines, gullies, or invalid areas in the digital elevation model. Simultaneously, during the hazard candidate area boundary integration stage, invalid data masks, digital elevation model elevation transition lines, and orthophoto slope boundary traces are used to generate closed polygons, ultimately outputting a mountain geological hazard identification map with coordinate reference, raster resolution, boundary coordinates, slope unit numbers, and catchment area attribution information. Attached Figure Description

[0015] Figure 1 This is a system module diagram of the mountain geological hazard identification system based on spatial clustering algorithm of the present invention. Detailed Implementation

[0016] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.

[0017] Reference Figure 1 The mountain geological hazard identification system based on spatial clustering algorithm includes a data processing device 100, which includes a processor 101, a memory 102, a data input interface 103, and a layer output interface 104. The data input interface 103 receives temporal synthetic aperture radar (SAR) data 110, SAR interferometric deformation points 111, a digital elevation model 112, an orthophoto 113, and the study area boundary 114. The memory 102 stores the temporal SAR data 110, SAR interferometric deformation points 111, digital elevation model 112, orthophoto 113, study area boundary 114, a unified grid, point-to-grid mapping relationships, a set of valid deformation points, an invalid data mask, slope unit numbers, slope drainage connectivity networks, slope deformation anomaly density values, initial hazard clustering areas, a set of hazard candidate polygons, a mountain geological hazard identification map 191, and programs for executing each processing step. Processor 101 calls the programs and data in memory 102, sequentially driving the standardization processing module 120, the unified raster construction module 130, the validity screening module 140, the slope drainage connectivity network construction module 150, the slope deformation anomaly density value calculation module 160, the adaptive spatial clustering module 170, the boundary integration module 180, and the identification map generation module 190. Layer output interface 104 is used to output vector layers, corresponding raster masks, base map overlay results, attribute information, and a mountain geological hazard hazard identification map 191.

[0018] The standardization processing module 120 performs cropping, coordinate unification, and resampling on the temporal synthetic aperture radar data 110, synthetic aperture radar interferometric deformation points 111, digital elevation model 112, orthophotos 113, and study area boundary 114, outputting standardized multi-source spatial data. The unified raster construction module 130 constructs a unified raster based on the standardized multi-source spatial data and establishes a point-to-raster mapping relationship from the synthetic aperture radar interferometric deformation points 111 to the unified raster. The validity screening module 140 performs validity screening and observation quality recording on the synthetic aperture radar interferometric deformation points 111 based on the point-to-raster mapping relationship, outputting a set of valid deformation points and an invalid data mask. The slope drainage connectivity network construction module 150 constructs slope unit numbers and a slope drainage connectivity network based on the digital elevation model 112. The slope deformation anomaly density calculation module 160 calculates the slope deformation anomaly density value based on the set of valid deformation points, slope unit numbers, and the slope drainage connectivity network. The adaptive spatial clustering module 170 performs adaptive spatial clustering based on slope deformation anomaly density values ​​and slope drainage connectivity constraints, outputting an initial hazard clustering area. The boundary integration module 180 integrates the hazard candidate area boundaries based on the initial hazard clustering area, invalid data mask, slope unit number, orthophoto slope boundary traces, and digital elevation model elevation transition lines, outputting a polygon set of hazard candidate areas. The identification map generation module 190 generates a mountain geological hazard identification map 191 based on the polygon set of hazard candidate areas.

[0019] The slope boundary traces in the orthophoto are extracted by the boundary integration module 180 based on the input orthophoto 113 and the digital elevation model 112. The boundary integration module 180 first uses the Canny edge detection operator to detect the grayscale gradient of the orthophoto 113 to obtain an initial set of edge pixels. Then, it superimposes the extreme value positions of the slope profile curvature extracted by the digital elevation model 112 for constraint screening. Edge segments that simultaneously satisfy the following conditions are retained: the gradient magnitude is greater than 0.3 times the maximum gradient magnitude of the study area image, and the absolute value of the slope profile curvature at the corresponding position is greater than the upper quartile value of the slope profile curvature of the study area. These are classified into landslide trails, lateral trails, and toe trails according to the terrain morphology characteristics and used for boundary adsorption calculation. The landslide trail corresponds to the upslope turning point where the slope changes from gentle to steep, the lateral trail corresponds to the slope abrupt change position of the lateral boundary of the slope, and the toe trail corresponds to the downslope turning point where the slope changes from steep to gentle.

[0020] The modules executed by processor 101 are in accordance with Figure 1 Data is transferred from data input to layer output. The output of the previous module serves as the input of the next module; when intermediate data fails to meet the input conditions of the next module, the intermediate data is not included in the corresponding subsequent calculation, and the corresponding processing state is retained in memory 102.

[0021] The standardization processing module 120 uses temporal synthetic aperture radar data 110, synthetic aperture radar interferometric deformation points 111, digital elevation model 112, orthophotos 113, and the study area boundary 114 as multi-source spatial data. The standardization processing module 120 trims the multi-source spatial data according to the study area boundary 114 and transforms the trimmed multi-source spatial data to the same projected coordinate system. During trimming, the study area boundary 114 is used as the final processing range, and the digital elevation model 112 retains the adjacent data outside the study area boundary 114 in the terrain derivation calculation. In the terrain derivation calculation, the digital elevation model 112 retains adjacent data outside the study area boundary 114. The retention width is set to the ground distance corresponding to 30 original digital elevation model 112 grids. This width covers the neighborhood analysis range required for terrain derivation calculation, avoiding boundary effects in the calculation at the study area boundary 114. After all terrain derivation calculations are completed, the derivation results are then cropped according to the precise range of the study area boundary 114, and the terrain derivation data consistent with the range of the study area boundary 114 is output.

[0022] When generating a unified raster, the unified raster construction module 130 uses the geocoding origin of the temporal synthetic aperture radar data 110 as the reference origin and the row and column directions of the digital elevation model 112 as the row and column directions of the unified raster. The unified raster resolution is the median of the ground resolution of the temporal synthetic aperture radar data 110, the average ground spacing between the synthetic aperture radar interferometric deformation points 111, and the ground resolution of the digital elevation model 112. The spatial extent of the unified raster is determined by the outward expansion of the study area boundary 114 by three unified raster resolutions, ensuring that all positions inside the study area boundary 114 correspond to a complete unified raster. The generated unified raster origin, row and column directions, raster resolution, and spatial extent remain fixed, serving as the unified carrier for all subsequent spatial calculations.

[0023] Each grid cell in the unified grid has a row number, column number, and center coordinates, and serves as a common spatial carrier for subsequent slope unit numbering, invalid data masks, slope deformation anomaly density values, and corresponding grid masks. For any synthetic aperture radar interferometry deformation point 111, the unified grid construction module 130 determines the corresponding unified grid cell based on its planar coordinates. The column number is determined according to the following formula: ; in, For column number, The eastern coordinates of the deformation point 111 in the synthetic aperture radar interferometry are given. To unify the coordinates of the western boundary of the raster, To standardize raster resolution.

[0024] Line numbers are determined according to the following formula: ; in, For line numbers, Here are the north coordinates of the deformation point 111 in the synthetic aperture radar interferometry. To unify the coordinates of the northern boundary of the raster. (From...) and The determined uniform grid is used as the grid to which the corresponding synthetic aperture radar interferometry deformation point 111 belongs.

[0025] When the synthetic aperture radar interferometric deformation point 111 is located on the boundary of a unified grid, the unified grid construction module 130 assigns it to the unified grid with the closest center coordinate distance. If the center coordinate distances of multiple unified grids are the same, the grid to which it belongs is determined in the order of smaller column number first, followed by smaller row number. This ensures that a synthetic aperture radar interferometric deformation point 111 corresponds to only one grid number.

[0026] The standardization module 120 performs area-preserving resampling on continuous variables and nearest-neighbor resampling on categorical and masked variables. Continuous variables include elevation values ​​of the digital elevation model 112, line-of-sight deformation rates of synthetic aperture radar interferometry deformation points 111, and coherence or stability coefficients; categorical variables include slope cell numbers; and masked variables include invalid data masks. The resampled data maintains consistency with the uniform raster in spatial extent, row and column numbers, and raster resolution.

[0027] For continuous variables, area-preserving resampling employs an area-weighted average resampling algorithm. For each raster cell to be resampled in the target uniform raster, all original data sub-cells falling within the target raster cell's range are counted. The attribute values ​​of these original data sub-cells are then weighted and summed based on their coverage area ratio within the target raster cell, yielding the resampling result for the target raster cell. The area-weighted average resampling is performed according to the following formula: ; in, The resampling results of the raster cells to be resampled in the target unified raster. The set of original data sub-cells falling within the range of the raster cell to be resampled. For raw data sub-units The coverage area within the raster cell to be resampled. For raw data sub-units The attribute values. The resampling process first completes the projected coordinate transformation, and then performs the resampling operation to ensure that the resampled data is registered with the unified raster space.

[0028] The validity screening module 140 reads the processing results corresponding to the synthetic aperture radar interferometry deformation point 111, extracts the line-of-sight deformation rate, coherence or stability coefficient, geographical location and radar line-of-sight direction of each synthetic aperture radar interferometry deformation point 111, and determines the grid number to which each synthetic aperture radar interferometry deformation point 111 belongs based on the point-to-grid mapping relationship.

[0029] When the validity screening module 140 determines invalid records, missing orbit geometry refers to records where the satellite orbit state vector parameters corresponding to the synthetic aperture radar interferometry deformation point 111 are incomplete, making it impossible to calculate the radar line-of-sight direction vector; failed phase unwrapping refers to records where the phase unwrapping quality factor corresponding to the synthetic aperture radar interferometry deformation point 111 is less than 0.2, or the unwrapped phase residual is greater than the phase residual threshold corresponding to the average coherence coefficient of the corresponding interferometric pair; records with missing coordinates, missing rates, missing coherence, missing orbit geometry, failed phase unwrapping, non-finite values, or located outside the study area boundary 114 are all excluded from the set of valid deformation points.

[0030] The validity screening module 140 normalizes the coherence or stability coefficients and uses the normalized coherence or stability coefficients as observation quality information. The statistical range for the normalization of the coherence or stability coefficients is the original coherence or stability coefficients corresponding to all synthetic aperture radar interferometry deformation points 111 involved in the calculation within the study area. The minimum value among all original coherence or stability coefficients is taken as the minimum value. The maximum value is used as Based on the extreme values ​​of global statistics, the original values ​​of each synthetic aperture radar interferometry deformation point 111 are normalized to ensure that the normalization results are uniformly comparable throughout the entire study area. The normalization process is performed according to the following formula: ; in, For the first The normalized coherence or stability coefficients of the deformation points from the synthetic aperture radar interferometry measurement. For the first The original coherence or stability coefficient of a synthetic aperture radar interferometric deformation point 111. This represents the minimum value among the original coherence or stability coefficients corresponding to all synthetic aperture radar interferometric deformation points 111 involved in the calculation within the study area. This represents the maximum value among the original coherence or stability coefficients corresponding to all synthetic aperture radar interferometric deformation points 111 involved in the calculation within the study area. When At that time, all coherence or stability coefficients involved in the normalization process are normalized to a value of 1. A normalization result less than 0 is set to 0, and a result greater than 1 is set to 1.

[0031] The marking rule for invalid data masks is as follows: if a single uniform grid contains at least one synthetic aperture radar interferometry deformation point 111 that is determined to be invalid, and the invalid synthetic aperture radar interferometry deformation point 111 can be determined to belong to the grid based on the point-to-grid mapping relationship, then the uniform grid is marked as invalid in the invalid data mask; if a single grid contains both valid deformation points and invalid synthetic aperture radar interferometry deformation points 111, the grid is still marked as invalid according to the rule for the existence of invalid points. The marking result is only used for subsequent boundary integration steps and does not change the observation quality information already retained in the set of valid deformation points.

[0032] Records with missing coordinates or located outside the study area boundary 114 and whose grid number cannot be determined are not written to the grid position of the invalid data mask; they are simply recorded as invalid records and excluded from the set of valid deformation points. Synthetic Aperture Radar Interferometry Deformation Point 111, which is not written to the invalid data mask and is not a valid record, is included as a valid deformation point in the set of valid deformation points.

[0033] The slope drainage connectivity network construction module 150 calculates the slope, aspect, flow direction, and cumulative runoff volume based on the digital elevation model 112, and generates a unit downhill direction based on the slope and aspect. When calculating the flow direction, the module 150 uses the D8 single-flow-direction algorithm. First, it fills all closed depressions in the digital elevation model 112 with a depth less than twice the resolution of the digital elevation model 112. Then, for each effective elevation grid, it calculates the elevation difference between it and its eight adjacent grids, directing the flow direction towards the adjacent grid with the largest elevation difference. The cumulative runoff volume is calculated using a cumulative counting method based on the flow direction results. The cumulative runoff volume of each grid is the sum of the number of upstream grids to which all flow directions point. In flat areas, the flow direction is allocated according to a preset dominant slope direction, which is the median value of the slope directions of the surrounding effective elevation grids. If there are no surrounding effective elevation grid slope directions, no flow direction is established for that flat area to avoid flow stagnation.

[0034] Slope aspect in the formula for unit downhill direction Using the geographical azimuth definition, with true north as the starting reference, the angle increases clockwise, and the range of angle values ​​is [range missing]. Slope angle The angle between the grid slope and the horizontal plane, with a range of values. The slope angle corresponding to the horizontal position is 0. The three components of the unit downslope direction vector correspond to the east, north, and vertical coordinate axes, respectively. The east and north axes form a horizontal plane coordinate system, and the vertical axis corresponds to the elevation upward direction, consistent with the elevation coordinate system of Digital Elevation Model 112. The unit downslope direction is generated according to the following formula: ; in, For grid The unit is downhill. For grid slope direction, For grid The slope angle. For a grid with a slope angle of 0, if it is located in a flat area and there is a preset dominant slope direction, then the preset dominant slope direction and the slope angle of 0 are used to generate a unit downhill direction; if there is no preset dominant slope direction, then the grid will not participate in the subsequent calculations that require a unit downhill direction.

[0035] The slope drainage connectivity network construction module 150 divides the slope surface based on watershed lines, runoff lines, aspect change lines, and slope toe elevation change lines, and assigns slope unit numbers to a unified grid. The quantitative judgment criteria for the four types of line elements used in slope unit division are as follows: watershed lines are defined as ridge locations where the accumulated runoff is less than the 10th percentile of the accumulated runoff in the study area and the planar curvature is less than -0.02; runoff lines are defined as gully locations where the accumulated runoff is greater than the 90th percentile of the accumulated runoff in the study area and the planar curvature is greater than 0.02; aspect change lines are defined as those where the difference in aspect between adjacent grids is greater than... The location of the slope; the slope toe elevation transition line is determined as the transition point where the slope profile curvature is greater than 0.03, and the slope becomes gentler. The four types of line elements together constitute the dividing boundary of the slope unit, and each connected slope area after division corresponds to a slope unit number.

[0036] The catchment area attribution information is generated using a watershed extraction method based on the flow direction and cumulative runoff results. First, the cumulative runoff threshold is set as the 80th percentile value of the cumulative runoff in the study area. Grids with cumulative runoff greater than this threshold are extracted as valley networks. Then, starting from the outlet location of the valley network, all grid ranges that flow to the outlet are traced back along the flow direction. The runoff range corresponding to each independent outlet is a catchment area. Each uniform grid with a valid elevation corresponds to unique catchment area attribution information, which is used to constrain subsequent density calculations and clustering calculations to be performed only within the same catchment area.

[0037] The slope drainage connectivity network construction module 150 uses a uniform grid as the drainage connectivity grid, establishing adjacency connections between adjacent drainage connectivity grids within the same slope unit. For any drainage connectivity grid, if its 8-neighborhood adjacent drainage connectivity grids belong to the same slope unit and there is a flow direction relationship between them, then an adjacency connection is established between the two drainage connectivity grids. This adjacency connection is stored as a bidirectional connection in the slope drainage connectivity network.

[0038] The criteria for determining the adjacent connection relationship across slope units are as follows: two drainage connected grids belonging to different slope units share a grid edge or a grid vertex, the flow direction of the upstream grid points towards the slope unit where the downstream grid is located, and the slope aspect difference between the two grids is less than 1. This means that the flow direction is determined to be continuous; cross-slope cell grids that meet the above conditions of shared confluence boundary and flow direction continuity are established as adjacent grids. Cross-cell grids that cross watersheds, closed ridgelines, or invalid areas of Digital Elevation Model 112 are prohibited from establishing adjacent grids, and any such connections that have already been established are deleted from the slope drainage connectivity network.

[0039] The slope drainage connectivity network construction module 150 records the adjacent connection length, grid slope aspect, grid slope angle, number of slope unit connections, and catchment area affiliation information to obtain the slope drainage connectivity network. The adjacent connection length is the straight-line distance between the centers of two adjacent drainage connectivity grids on the ground, and the connection length for diagonally adjacent grids is... The connection length of adjacent grid cells in the row and column directions is The number of slope unit connections is the number of cross-unit adjacent connections that have been established between the corresponding slope unit and other slope units.

[0040] The slope deformation anomaly density calculation module 160 determines the grid to be calculated from a unified grid that has slope element numbers and is connectable to effective deformation points within the same catchment area. The grid to be calculated must have a non-empty slope element number and a reachable path in the slope drainage connectivity network to the grid belonging to at least one effective deformation point within the same catchment area. Unified grids that do not meet the above conditions are not included in the slope deformation anomaly density calculation.

[0041] The radar line-of-sight direction vector is first calculated from the satellite orbit state vector to obtain a unit vector in the satellite orbit coordinate system. Then, it is rotated and transformed to the East-North-Sky local horizontal coordinate system consistent with the unit downslope direction. During the coordinate system transformation, the meridian and zonal directions are first calculated based on the geodetic latitude and longitude of the study area center. The line-of-sight vector in the satellite orbit coordinate system is decomposed into radial and tangential components. Then, it is rotated and mapped to the east, north, and vertical components to obtain the radar line-of-sight direction unit vector in the same coordinate system as the unit downslope direction vector. Finally, it is substituted into the formula to calculate the radar observation sensitivity of downslope deformation.

[0042] The slope deformation anomaly density calculation module 160 calculates the downslope deformation radar observation sensitivity based on the unit downslope direction of the grid to which the effective deformation point belongs and the radar line-of-sight direction. The downslope deformation radar observation sensitivity is calculated according to the following formula: ; in, For the first Sensitivity of downslope deformation radar observation at effective deformation points For the first The unit downslope direction of the grid to which each effective deformation point belongs. For the first The radar line-of-sight direction unit vector of each effective deformation point.

[0043] The slope deformation anomaly density calculation module 160 calculates the radar observation correction value based on the downslope deformation radar observation sensitivity, the normalized coherence or stability coefficient, and the median value of the downslope deformation radar observation sensitivity within the same slope unit. The radar observation correction value is calculated according to the following formula: ; in, For the first Radar observation correction values ​​for each effective deformation point For the first The slope cell number corresponding to the grid to which each effective deformation point belongs. Indicates the relationship with the first A set of effective deformation points that are within the same slope element number. This represents the median value. When there are no valid deformation points within the same slope element number that can be used to calculate the median value, the [number]th [unit] is used. Each valid deformation point is not included in the calculation of the observed correction deformation anomaly value; when At that time, the first The effective deformation points are not included in the calculation of the observed correction deformation anomaly value.

[0044] The slope deformation anomaly density calculation module 160 calculates the observation-corrected deformation anomaly value based on the line-of-sight deformation rate, the median line-of-sight deformation rate within the same slope unit, and the radar observation correction value. The observation-corrected deformation anomaly value is calculated according to the following formula: ; in, For the first The observed and corrected deformation anomaly values ​​of each effective deformation point For the first The line-of-sight deformation rate of each effective deformation point Within the same slope unit number, the first The line-of-sight deformation rate of each effective deformation point.

[0045] The slope deformation anomaly density calculation module 160 calculates the slope connectivity distance based on the adjacent connection relationships, adjacent connection lengths, grid aspect, and grid slope angle in the slope drainage connectivity network. (Grid to be calculated) To the grid of the effective deformation point The slope connectivity distance is calculated using the following formula: ; in, For the raster to be calculated To the grid of the effective deformation point The slope connectivity distance, Indicates the raster to be calculated To the grid of the effective deformation point reachable path, For adjacent connections in reachable paths, The length of the adjacent connection. Connecting adjacent drainage grids and The difference in slope aspect between them Connecting adjacent drainage grids and The slope angle difference between them. If no reachable path exists, the grid to which the valid deformation point belongs is not calculated. With the raster to be calculated The distance between the slopes.

[0046] Adjacent drainage connecting grids and The slope difference between them is calculated using the minimum circumferential difference, according to the following formula: ; in, Connecting adjacent drainage grids slope direction, Connect adjacent drainage grids Slope direction. Adjacent drainage connecting grids. and The difference in slope angle between them is calculated using the following formula: ; in, Connecting adjacent drainage grids The slope angle, Connect adjacent drainage grids The slope angle.

[0047] The slope deformation anomaly density value calculation module 160 determines the slope search distance range based on the slope connectivity distance. (Grid to be calculated) The slope search distance range is calculated using the following formula: ; in, For the raster to be calculated The range of slope search distances, To be calculated with the raster The set of effective deformation points located within the same catchment area and with effective slope connectivity. If the set used in the calculation is empty, the raster to be calculated is used. The minimum positive value among all adjacent connection lengths within the catchment area is taken as When the raster to be calculated corresponding When, take the raster to be calculated. The minimum positive value among all adjacent connection lengths within the catchment area is taken as The value of ; if there is no positive adjacent connection length in the catchment area, the grid to be calculated will not participate in the calculation of the slope deformation anomaly density value.

[0048] The slope deformation anomaly density calculation module 160 calculates the slope deformation anomaly density value by weighting the observed and corrected deformation anomaly values ​​according to the slope connectivity distance and the slope search distance range. The slope deformation anomaly density value is calculated according to the following formula: ; in, For the raster to be calculated The slope deformation anomaly density value. The effective deformation points involved in the summation are those corresponding to the grid to be calculated. Located in the same catchment area, with effective slope connectivity, and with effective observation and correction of deformation anomalies and corresponding Valid deformation points. If the denominator is 0, then the raster to be calculated... It is not included in the calculation of abnormal density values ​​of slope deformation.

[0049] The adaptive spatial clustering module 170 uses uniform rasters with calculated slope deformation anomaly density values ​​and a slope search distance range as candidate rasters. For any candidate raster... The adaptive spatial clustering module 170 selects candidate grids from the same catchment area that are connectable in the slope drainage connectivity network and do not cross the watershed or invalid areas of the digital elevation model 112, forming a candidate neighbor grid set. Candidate neighbor grid set The selection criteria are: candidate grid With candidate grid They belong to the same catchment area, and there is a slope drainage network reachable path between them. The slope connectivity distance between them is... Valid and meets the requirements Candidate graticles that do not meet any of the conditions will not be included in the candidate neighbor graticle set. .

[0050] The adaptive spatial clustering module 170 combines the slope elements corresponding to the slope element numbers of the two candidate grids into local slope element combinations, and calculates the density change benchmark value based on the slope deformation anomaly density value within the local slope element combination. The density change benchmark value is calculated according to the following formula: ; in, Candidate grid and candidate grid The corresponding density change baseline value, Candidate grid and candidate grid The candidate raster set of calculated slope deformation anomaly density values ​​within the local slope unit combination composed of slope units corresponding to their respective slope unit numbers. and All values ​​represent the slope deformation anomaly density values ​​for the corresponding candidate grid.

[0051] The adaptive spatial clustering module 170 calculates the density difference value based on the difference in slope deformation anomaly density values ​​between two candidate graticets and the baseline value of density change. The density difference value is calculated according to the following formula: ; in, Candidate grid and candidate grid Density difference between them Candidate grid The density value of slope deformation anomaly, Candidate grid The anomaly density value of slope deformation. When and hour, Take 0; when and hour, Take 1.

[0052] The adaptive spatial clustering module 170 calculates the clustering decision distance based on the slope connectivity distance, slope search distance range, and density difference value between two candidate graticets. The clustering decision distance value is calculated according to the following formula: ; in, Candidate grid and candidate grid Clustering distance values ​​between them Candidate grid and candidate grid The distance between the slopes Candidate grid The range of slope search distances, Candidate grid The slope search distance range. If or Invalid, candidate grid and candidate grid Clustering distance values ​​are not calculated between them.

[0053] The adaptive spatial clustering module 170 determines the local clustering search range based on the clustering judgment distance values ​​within the candidate neighboring grid set. The local clustering search range is calculated according to the following formula: ; in, Candidate grid The local clustering search range, Indicates the absolute deviation of the median. When When empty, candidate grid Not used as a core grid; when When there is only one candidate raster, Take 0.

[0054] The adaptive spatial clustering module 170 determines the effective neighboring raster set based on the local clustering search range. The effective neighboring raster set is determined according to the following formula: ; in, Candidate grid The effective neighbor grid set, Candidate grid The local clustering search range. If or If invalid, then candidate grid Do not enter the candidate grid The effective neighboring raster set.

[0055] The adaptive spatial clustering module 170 determines the minimum number of neighboring cells for the core grid based on the number of slope cell connections. The minimum number of neighboring cells for the core grid is determined by the following formula: ; in, Candidate grid The minimum number of neighboring cells for the core grid. Candidate grid The corresponding slope unit number, This represents the number of slope unit connections corresponding to the slope unit number, where 4 is the minimum number of neighboring grids for the basic core grid. Candidate grids. The number of candidate rasters in the effective neighbor set is greater than or equal to At that time, candidate grid The core grid conditions are met.

[0056] The execution steps of the adaptive spatial clustering module 170 in generating the initial hazard clustering area are as follows: First, traverse all candidate graticets, determine whether each candidate graticet meets the core graticet condition, and mark all core graticets; Second, select any unassigned core graticet as the seed point, include itself and all core graticets in its effective neighboring graticet set into the same initial hazard clustering area, and then recursively expand along the effective neighboring graticet set of the included core graticet until all core graticets connected to the seed core graticet through the core graticet adjacency relationship are included in the same cluster; Third, include non-core graticets that contain at least one core graticet in the effective neighboring graticet set into the initial hazard clustering area to which the corresponding core graticet belongs, as the cluster boundary graticet; Fourth, mark candidate graticets that neither meet the core graticet condition nor have an effective neighboring relationship with any core graticet as noise graticets, and do not merge them into any initial hazard clustering area.

[0057] The boundary integration module 180 rasterizes each initial hazard cluster into an initial region surface. The initial region surface is formed by merging eight adjacent candidate grids within the same initial hazard cluster. The eight adjacent grids include grids that are adjacent in the four row and column directions (up, down, left, right) and the four diagonal directions. As long as two candidate grids have an adjacent relationship with shared grid edges or shared grid vertices and belong to the same initial hazard cluster, they are merged into the same initial region surface. The outer boundary of the initial region surface is generated by connecting the grid edges of the outermost candidate grids sequentially.

[0058] The boundary integration module 180 checks for void areas formed by invalid data masks within the same slope unit. The rules for determining whether a void area is included in the initial area surface are as follows: being surrounded by the same initial hazard cluster means that all boundary grids of the void area are adjacent to the grids of the initial hazard cluster, and the void area is completely within the closed outer boundary of the initial hazard cluster; the gully main line is the grid line connecting the main gullies whose cumulative runoff is greater than the 95th percentile of the cumulative runoff in the study area; a void area is included in the corresponding initial area surface when it meets the three conditions of belonging to the same slope unit, not crossing the grid range of the gully main line, and being completely surrounded by the same initial hazard cluster. Void areas that do not meet any of the conditions are not included.

[0059] The elevation transition lines used in the boundary integration module 180 include two types: slope toe elevation transition lines and ridge elevation transition lines. These are extracted from the extreme values ​​of profile curvature of the digital elevation model 112. The extraction criterion is to take the position where the absolute value of the profile curvature along the flow direction of the grid slope is greater than the 85th percentile value of the profile curvature of the study area as the candidate position of elevation transition. Then, the slope change direction is combined for classification. The slope toe elevation transition line is the transition position of the downslope where the slope changes from steep to gentle, and the ridge elevation transition line is the transition position of the upslope where the slope changes from gentle to steep. This elevation transition line uses the same calculation standard as the slope toe elevation transition line used for slope unit segmentation, and only the extraction range covers the entire study area.

[0060] The boundary integration module 180 utilizes the elevation transition lines of the digital elevation model and the locatable landslide trailing edge, lateral edge, or toe traces in the orthophoto slope boundary traces to snap the boundary of the initial area surface. During boundary snapping, the boundary vertices of the initial area surface are taken as the objects to be snapped, and the surrounding areas are also snapped. The system searches for elevation transition lines in the digital elevation model and slope boundary traces in the orthophoto image within the specified range. If multiple line segments are found simultaneously, the target for adsorption is determined in the order of landslide trailing edge, side edge, slope toe trace, and elevation transition line in the digital elevation model. If adsorption would cross the watershed, invalid area of ​​digital elevation model 112, or a location in the slope drainage connectivity network where no adjacent connection relationship has been established, the adsorption of the target object is cancelled.

[0061] The determination of the boundary break operation is based on the internal range and boundary line of the initial area surface. When the internal grid of the initial area surface crosses the watershed line, the invalid area of ​​the digital elevation model 112, or the location in the slope drainage connectivity network where no adjacent connection relationship has been established, the initial area surface is cut and split along the boundary line of the prohibited crossing location. The splitting operation is performed along the grid edge line of the prohibited crossing location. Each independent closed area after splitting is subjected to boundary closure processing. Each closed area retains the slope unit number and catchment area affiliation information corresponding to its internal grid, and each is regarded as an independent polygon in the set of candidate hazard areas.

[0062] The identification image generation module 190 performs a topological validity check on each candidate polygon in the candidate hazard area polygon set. The topological validity check includes self-intersecting edge checks, hanging edge checks, duplicate vertex checks, and overlapping edge checks. The repair operation method for the topological validity check is as follows: For candidate polygons with self-intersecting edges, the original polygon is split into multiple non-self-intersecting sub-polygons along the intersection points. Each sub-polygon is closed according to the boundary direction to ensure that each sub-polygon is a simple closed polygon. For candidate polygons with hanging edges, the free endpoint of the hanging edge is taken, and a perpendicular line is drawn from the direction perpendicular to the hanging edge to the nearest valid boundary line segment. The foot of the perpendicular is taken as the new boundary vertex, and the endpoint of the hanging edge is connected to the foot of the perpendicular to complete the hanging edge closure. For duplicate vertices and overlapping edges, the boundary vertices are traversed in order, and duplicate vertices with completely overlapping coordinates and overlapping edge segments with overlapping beginning and end are deleted, while retaining the continuous direction of the boundary.

[0063] The identification map generation module 190 snaps the boundaries of the candidate hazard area polygons after topological validity checks to the grid edges of the unified grid and the boundary lines of the slope units corresponding to the slope unit numbers. When the identification map generation module 190 performs the final boundary snapping, if the distance between the boundary vertex and the grid edge of the unified grid and the boundary line of the slope unit is less than 0.5 times the unified grid resolution, it is determined to be close and snapping is performed. This boundary snapping step is performed after the boundary snapping step of the boundary integration module 180. If the same boundary vertex is close to both the grid edge of the unified grid and the boundary line of the slope unit, it is snapped to the boundary line of the slope unit first. If snapping to the boundary line of the slope unit would cause the boundary to cross the watershed, the invalid area of ​​the digital elevation model 112, or the location where no adjacent connection relationship has been established, then the snapping of the slope unit boundary line is canceled and snapped to the grid edge of the unified grid instead.

[0064] The recognition image generation module 190 generates a vector layer and a corresponding raster mask based on the set of candidate hazard areas polygons after topological validity checks and final boundary snapping. The corresponding raster mask and the unified raster have the same origin, row and column directions, raster resolution, and spatial range; the unified raster located inside the candidate hazard area polygons is written with the unique hazard area number, and the unified raster located outside the candidate hazard area polygons is written with the background value.

[0065] The identification map generation module 190 overlays a vector layer onto the orthophoto 113 or the digital elevation model 112 shaded base map, and writes coordinate reference, raster resolution, invalid data mask, generation time, unique number of the hazard area, boundary coordinates, slope unit number, and catchment area attribution information into the vector layer. After writing, the identification map generation module 190 outputs the mountain geological hazard identification map 191 through the layer output interface 104.

[0066] The embodiments of this example have been described above. However, this example is not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms based on the guidance of this example, and all of them are within the protection scope of this example.

Claims

1. A mountain geological hazard identification system based on spatial clustering algorithm, characterized in that, Configured to execute sequentially: Acquire multi-source spatial data, standardize the multi-source spatial data, and construct a unified raster; Establish a point-to-grid mapping relationship between the deformation points of synthetic aperture radar interferometry and the unified grid; The effectiveness of the synthetic aperture radar interferometry deformation points is screened and the observation quality is recorded to obtain a set of effective deformation points and an invalid data mask. During the effectiveness screening and observation quality recording process, the processing results corresponding to the synthetic aperture radar interferometry deformation points are read, and the line-of-sight deformation rate, coherence or stability coefficient, geographical location, and radar line-of-sight direction of each synthetic aperture radar interferometry deformation point are extracted. The grid number to which each synthetic aperture radar interferometry deformation point belongs is determined based on the point-to-grid mapping relationship. The coherence or stability coefficients are normalized, and the normalized coherence or stability coefficients are used as observation quality information. Based on the digital elevation model, a slope element numbering system and a slope drainage connectivity network are constructed, including: The slope, aspect, flow direction, and cumulative runoff volume are calculated based on the digital elevation model, and a unit downhill direction is generated based on the slope and aspect. The slope surface is divided according to the watershed line, confluence line, slope aspect change line and slope toe elevation change line, and the slope surface unit number is assigned to the unified grid. The unified grid is used as a drainage connection grid, and adjacent connection relationships are established between adjacent drainage connection grids within the same slope unit. Establish adjacent connections between drainage interconnect grids that span slope units but share a confluence boundary and have continuous flow direction; It is prohibited to establish adjacent connections between drainage connectivity grids that cross watersheds, closed ridgelines, or invalid areas of the digital elevation model; Record the adjacent connection length, grid slope direction, grid slope angle, number of slope unit connections, and water catchment area affiliation information to obtain the slope drainage connectivity network; Based on the set of effective deformation points, the slope unit number, and the slope drainage connectivity network, the slope deformation anomaly density value is calculated, including: The grid to be calculated is determined in a unified grid that has slope unit numbers and is connected to effective deformation points within the same catchment area; The downslope deformation radar observation sensitivity is calculated based on the unit downslope direction of the grid to which the effective deformation point belongs and the radar line of sight direction. The radar observation correction value is calculated based on the downslope deformation radar observation sensitivity, the normalized coherence or stability coefficient, and the median value of downslope deformation radar observation sensitivity within the same slope unit. Based on the line-of-sight deformation rate, the median line-of-sight deformation rate within the same slope unit, and the radar observation correction value, calculate the observation correction deformation anomaly value. Based on the adjacent connection relationships, adjacent connection lengths, grid slope aspect and grid slope angle in the slope drainage connectivity network, calculate the slope connectivity distance and determine the slope search distance range; Based on the slope connectivity distance and the slope search distance range, the observed and corrected deformation anomaly values ​​are weighted by distance to obtain the slope deformation anomaly density values. Based on the slope deformation anomaly density values, adaptive spatial clustering with slope drainage connectivity constraints is performed to obtain initial hazard clustering areas, including: The uniform grid that has calculated the slope deformation anomaly density value and has a slope search distance range is used as the candidate grid. For any candidate grid, select candidate grids that are connectable in the slope drainage connectivity network within the same catchment area and do not cross the watershed or invalid areas of the digital elevation model to form a set of candidate neighboring grids; Combine the slope units corresponding to the slope unit numbers of the two candidate grids into a local slope unit combination. Calculate the density change benchmark value based on the abnormal density value of slope deformation within the local slope unit combination; The density difference value is calculated based on the difference in slope deformation anomaly density values ​​between the two candidate grids and the density change benchmark value. The clustering judgment distance value is calculated based on the slope connectivity distance between the two candidate grids, the slope search distance range, and the density difference value. The local clustering search range is determined based on the clustering judgment distance value within the candidate neighboring grid set, and the effective neighboring grid set is determined based on the local clustering search range; The minimum number of neighboring grids for the core grid is determined based on the number of connected slope units. The candidate graticules that meet the minimum number of neighboring graticules required for the core graticule and are within the set of effective neighboring graticules are merged to obtain the initial hazard clustering area. Based on the initial hazard clustering area, the invalid data mask, the slope unit number, the orthophoto slope boundary trace, and the digital elevation model elevation transition line, the hazard candidate area boundary is integrated to obtain a polygon set of hazard candidate areas; A mountain geological hazard identification map is generated based on the polygon set of the hazard candidate areas.

2. The mountain geological hazard identification system based on spatial clustering algorithm according to claim 1, characterized in that, Acquiring multi-source spatial data, standardizing the multi-source spatial data, and constructing a unified raster includes: The time-series synthetic aperture radar data, synthetic aperture radar interferometry deformation points, digital elevation model, orthophoto image and study area boundary are used as the multi-source spatial data. The multi-source spatial data is cropped according to the boundary of the study area, and the cropped multi-source spatial data is transformed to the same projection coordinate system; The unified grid is generated based on the geocoding parameters of the time-series synthetic aperture radar data, the ground spacing between the deformation points of the synthetic aperture radar interferometry, and the ground resolution of the digital elevation model. For continuous variables, area-preserving resampling is performed; for categorical variables and masked variables, nearest-neighbor resampling is performed.

3. The mountain geological hazard identification system based on spatial clustering algorithm according to claim 2, characterized in that, The effectiveness of synthetic aperture radar interferometric deformation points is screened and the observation quality is recorded to obtain a set of effective deformation points and an invalid data mask, which also includes: Records with missing coordinates, missing rates, missing coherence, missing orbital geometry, failed phase unwrapping, non-finite values, or located outside the boundary of the study area are written into the invalid data mask; The synthetic aperture radar interferometry deformation points that are not written into the invalid data mask are taken as the set of valid deformation points.

4. The mountain geological hazard identification system based on spatial clustering algorithm according to claim 3, characterized in that, Based on the initial hazard clustering area, the invalid data mask, the slope unit number, the orthophoto slope boundary trace, and the digital elevation model elevation transition line, the hazard candidate area boundaries are integrated to obtain a set of hazard candidate area polygons, including: Each of the initial hazard clustering areas is rasterized into an initial region surface; Inspect the void areas formed by the invalid data mask within the same slope unit; When the cavity area is surrounded by the same initial hidden danger cluster area, belongs to the same slope unit, and does not cross the main line of the gully, the cavity area is incorporated into the initial area surface; Using the elevation turning lines of the digital elevation model and the locatable landslide trailing edge, side edge, or toe traces in the orthophoto slope boundary traces, the boundary of the initial area surface is snapped. Disconnect the parts that cross the watershed, invalid areas of the digital elevation model, or where no adjacent connection relationship has been established in the slope drainage connectivity network, and generate the polygon set of the hazard candidate area.

5. The mountain geological hazard identification system based on spatial clustering algorithm according to claim 4, characterized in that, A mountain geological hazard identification map is generated based on the polygon set of the hazard candidate areas, including: Perform a topological validity check on each candidate polygon in the set of candidate hazard regions, remove self-intersecting edges and close dangling edges; The boundaries of the candidate hazard areas polygons after topological validity checks are snapped to the grid edges of the unified grid and the boundary lines of the slope units corresponding to the slope unit numbers. Generate vector layers and corresponding raster masks; The vector image layer is overlaid on the orthophoto or digital elevation model shaded base map; The coordinate reference, raster resolution, invalid data mask, generation time, unique number of the hidden danger area, boundary coordinates, slope unit number and water catchment area attribution information are written into the vector layer to obtain the mountain geological hazard identification map.

Images