A method for identifying debris flow in high and steep mountainous area based on vegetation restoration model

Abstract

The application discloses a kind of high and steep mountainous area debris flow remote sensing identification method based on vegetation restoration model, it is related to geological disaster remote sensing identification technical field, the present application includes obtaining the multi-source remote sensing data of target area;Theoretical vegetation restoration potential and residual recovery potential index of each pixel are calculated;Remove vegetation canopy interference to build real ground model, to real ground model is carried out microtopography enhancement processing, according to debris flow accumulation landform feature identification and extract historical debris flow disaster data set;Residual recovery potential index and historical debris flow disaster data set are carried out overlay analysis, establish the quantitative corresponding relationship between residual recovery potential index and vegetation coverage, determine the evolution stage and occurrence age of historical debris flow, form the historical debris flow identification result with spatial position and time attribute.This application can quantitatively determine the evolution stage of historical debris flow by constructing the nonlinear evolution model of vegetation coverage and residual recovery potential index.

Description

Technical Field

[0001] This invention belongs to the field of remote sensing identification technology for geological disasters, and in particular relates to a remote sensing identification method for debris flows in high and steep mountainous areas based on a vegetation restoration model. Background Technology

[0002] Refined identification of debris flow hazards based on remote sensing technology is one of the key means to obtain basic data on regional geological hazards and conduct risk assessments. This technology, with its non-contact observation, wide field of view, and historical data retrospective capabilities, has been widely used in debris flow identification, especially in the investigation of large-area historical disasters, demonstrating typical characteristics such as large detection range and high identification efficiency. However, the application of this technology still faces severe challenges due to the complex terrain and cold climate of high mountain and canyon areas: in steep mountainous areas, debris flow remote sensing identification is often affected by terrain shadows and dense vegetation canopies, resulting in blurred surface texture features, and the spectral characteristics of old deposits tend to converge with the background environment, further increasing the difficulty of refined identification. Furthermore, because the post-disaster ecological restoration process has significant temporal nonlinear characteristics, relying solely on a single topographic indicator or optical characteristics of a single time phase makes it difficult to determine the specific time period of debris flow occurrence, thus failing to accurately assess the activity level and potential recurrence risk of the channel.

[0003] Currently, existing debris flow remote sensing identification methods can be mainly categorized into the following three types based on their technical principles: The first type is the optical texture method, which utilizes the differences in color tone, texture, and vegetation disturbance in stacked fans to establish an interpretation model based on the texture features of optical images. However, this type of method is significantly affected by clouds, fog, and shadows. For medium- to long-term debris flows, the image texture tends to be consistent with the surrounding forest due to vegetation recovery, making the interpretation markers prone to failure and easily leading to missed detections.

[0004] The second category is the micro-topographic method, which utilizes high-precision digital elevation models (DEMs) or lidar data to construct morphological extraction models based on micro-topographic factors such as slope, curvature, and roughness, based on the specific geometric morphology of debris flow deposits. Existing research, in the paper "Identification of debris-flow channels using high-resolution topographic data: A case study in the Quebrada del Toro, NW Argentina," utilizes high-resolution topographic data and connective component analysis to construct a debris flow similarity index, achieving refined identification of debris flow hazards at the watershed scale. This method relies on topographic morphological differences to extract hazards, has a simple technical logic, and performs well in identifying fresh deposits. However, this type of method essentially relies on static analysis of topographic data from a single time phase, lacking a temporal dimension and unable to determine the age of the hazard. Furthermore, in densely vegetated, steep mountainous areas, optical sensors cannot penetrate the canopy to obtain the true elevation of bare ground, resulting in digital surface models that include vegetation canopy height, severely obscuring the true micro-topographic features within the channels. This study clearly confirms that this type of terrain analysis method is only suitable for semi-arid, sparsely vegetated bare areas. In high-vegetation, high-mountain canyon areas, it is very easy to miss or misidentify, thus limiting its applicability.

[0005] The third category is the temporal change detection method, which uses temporal remote sensing data to monitor the post-disaster vegetation recovery process and identifies debris flow disasters by detecting changes in indices such as the normalized vegetation index (NDI) and the normalized combustion index (NCI). Existing research, in the study "UAVSAR and optical analysis of the Thomas Fire scar and Montecito debrisflows: Case study of methods for disaster response using remote sensing products," proposed fusing airborne synthetic aperture radar (UAVSAR) and optical satellite multi-source remote sensing images. This method utilizes the abrupt changes in the state of surface physicochemical characteristics before and after a disaster to rapidly interpret and identify debris flow disasters, making it suitable for rapid response to sudden disasters. However, in high-altitude canyon areas with favorable hydrothermal conditions and rapid vegetation recovery, the spectral perturbation signal attenuates quickly, making it difficult to effectively capture the historical characteristics of long-term debris flows. Furthermore, this method heavily relies on the data reserves of high-resolution pre-disaster reference images; for historical debris flows lacking early observation data, it loses its identification capability due to the lack of temporal comparison benchmarks, failing to meet the needs for refined identification of long-term, large-scale historical disasters.

[0006] In summary, existing remote sensing identification technologies suffer from the following main drawbacks in high and steep mountainous areas: optical texture methods are susceptible to interference from clouds and fog, and have difficulty in mid- to long-term identification; micro-topography methods lack temporal attributes and are severely affected by vegetation cover; and temporal change detection methods experience rapid signal attenuation in areas with rapid vegetation recovery and rely on pre-disaster imagery. Therefore, solving the technical challenge of long-term, accurate identification of historical debris flow events under vegetation cover in high and steep mountainous areas, particularly the difficulty in determining the age of debris flows, has become a bottleneck that urgently needs to be overcome in this field. Summary of the Invention

[0007] The purpose of this invention is to provide a remote sensing identification method for debris flows in high and steep mountainous areas based on a vegetation restoration model. By coupling micro-topographic features with a vegetation restoration potential model, it is possible to quantitatively invert the occurrence time and evolution stage of historical debris flows. This solves the problem that existing remote sensing identification methods in high and steep mountainous areas are unable to achieve long-term, time-attributed, and refined identification of debris flows due to vegetation cover and terrain shadow interference.

[0008] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution: This invention relates to a remote sensing identification method for debris flows in steep mountainous areas based on a vegetation restoration model, comprising the following steps: Acquire multi-source remote sensing data of the target area, generate a high-precision digital elevation model based on the multi-source remote sensing data, and obtain a vegetation coverage dataset; Based on the vegetation restoration potential model of similar habitats, the theoretical vegetation restoration potential and remaining restoration potential index of each pixel are calculated using the vegetation coverage dataset to quantify the degree of ecological restoration after debris flow disasters. Based on the high-precision digital elevation model, vegetation canopy interference is removed to construct a realistic surface model. Micro-topographic enhancement processing is performed on the realistic surface model. Historical debris flow disaster datasets are identified and extracted based on debris flow deposition landform characteristics. By overlaying the remaining recovery potential index with the historical debris flow disaster dataset, a quantitative correspondence between the remaining recovery potential index and vegetation cover is established, the evolution stage and occurrence time of historical debris flows are determined, and a historical debris flow identification result with both spatial location and temporal attributes is formed.

[0009] Furthermore, the multi-source remote sensing data includes airborne LiDAR point cloud data, aerial orthophotos, and multi-temporal optical remote sensing images; The airborne LiDAR point cloud data is preprocessed, including denoising, ground point filtering and spatial interpolation, to generate a digital elevation model with sub-meter resolution; Radiometric calibration and atmospheric correction were performed on the aerial orthophotos and multi-temporal optical remote sensing images. Based on the corrected optical remote sensing images, vegetation cover was calculated, and the erosion texture features of debris flow depositional fans were identified by combining the aerial orthophotos.

[0010] Furthermore, the similar habitat vegetation restoration potential model is based on a sliding window that traverses the entire domain. For each target pixel within the window, pixels with similar terrain conditions and undisturbed vegetation are selected as a set of similar habitats. The similarity of terrain conditions includes the fact that the difference between the two terrain factors, slope and aspect, is less than a preset threshold. The term "undisturbed vegetation" excludes non-natural vegetation pixels that have been damaged or are bare.

[0011] Furthermore, based on the enhanced vegetation index values ​​of all pixels in the set of similar habitats, the highest percentile is taken as the theoretical vegetation restoration potential of the target pixel. The theoretical vegetation restoration potential represents the upper limit of stable growth that vegetation can achieve under the same habitat conditions.

[0012] Furthermore, the remaining restoration potential index is obtained by calculating the difference between the theoretical vegetation restoration potential and the current observed value of the enhanced vegetation index; The higher the residual recovery potential index, the greater the gap between the vegetation and the natural recovery level, and the closer the debris flow occurred; when the residual recovery potential index approaches zero, it indicates that the vegetation has approached the natural community state.

[0013] Furthermore, the micro-topography enhancement processing includes generating multiple mountain shadow images under different lighting conditions based on the high-precision digital elevation model and using multi-directional mountain shadow simulation technology, and fusing the multiple mountain shadow images to highlight the micro-topography details on different slopes and reduce the terrain shading effect caused by a single lighting direction.

[0014] Furthermore, the debris flow depositional landform features include fan-shaped or tongue-shaped planar morphology, boundaries formed by lateral steep slopes or abrupt changes in topographic slope, uneven patchy or strip-shaped micro-topography, and loose deposits. Based on the enhanced micro-topographic images, visual interpretation is performed according to the depositional landform features to map the extent of historical debris flow deposition areas, thus forming a spatial database of historical debris flows.

[0015] Furthermore, the overlay analysis includes extracting the average residual recovery potential index and average vegetation cover within the range of each debris flow depositional fan; Based on the spatial-temporal theory, different depositional fans are used to characterize the sequential state of different recovery times since debris flow, and the quantitative relationship curve between the average residual recovery potential index and the average vegetation cover is fitted by nonlinear regression analysis.

[0016] Furthermore, the determination of the evolutionary stages and occurrence dates of historical debris flows includes: Based on the vegetation cover range, surface texture and erosion characteristics, and the numerical range of the remaining recovery potential index of the depositional fan, a classification standard for the age of debris flow is established. The classification criteria cover three time periods: recent debris flows, intermediate debris flows, and long-term debris flows.

[0017] Furthermore, the historical debris flow identification results include spatial distribution data of debris flows with clear dates of occurrence, which are used to support the graded evaluation of regional debris flow channel activity, the assessment of disaster recurrence cycles, and the disaster risk assessment.

[0018] The present invention has the following beneficial effects: Given that the timing of debris flow disasters is a key factor determining the accumulation cycle of loose sediment within a watershed and directly affects the recurrence probability and potential outbreak scale, this invention constructs a nonlinear evolution model of vegetation cover and residual recovery potential index (VRPI) to quantitatively determine the evolutionary stage of historical debris flows. This time-attributed identification result effectively supports the graded evaluation of regional debris flow channel activity, the scientific assessment of disaster recurrence cycles, and the accurate evaluation of future disaster risks. This provides sufficient data support and decision-making basis for selecting sites for major projects to avoid high-risk channels and for developing targeted disaster prevention and mitigation measures.

[0019] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a distribution map of the remaining vegetation restoration index in the region where this invention is based; Figure 2 This is a schematic diagram of the debris flow deposition fan identification marker based on mountain shadow data according to the present invention; Figure 3 This is a schematic diagram illustrating the quantitative relationship between the remaining vegetation restoration index and vegetation coverage in this invention. Figure 4 This is a technical roadmap for a remote sensing identification method for debris flows in steep mountainous areas based on a vegetation restoration model, as described in this invention. Figure 5This is a schematic diagram illustrating the detailed identification results of historical debris flows that occurred during the specific time periods specified in this invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Please see Figure 1-5 As shown, this invention is a remote sensing identification method for debris flows in steep mountainous areas based on a vegetation restoration model, comprising the following steps: Acquire multi-source remote sensing data of the target area, generate a high-precision digital elevation model based on the multi-source remote sensing data, and obtain a vegetation coverage dataset; Based on the vegetation restoration potential model of similar habitats, the theoretical vegetation restoration potential and remaining restoration potential index of each pixel are calculated using the vegetation coverage dataset to quantify the degree of ecological restoration after debris flow disasters. Based on the high-precision digital elevation model, vegetation canopy interference is removed to construct a realistic surface model. Micro-topographic enhancement processing is performed on the realistic surface model. Historical debris flow disaster datasets are identified and extracted based on debris flow deposition landform characteristics. By overlaying the remaining recovery potential index with the historical debris flow disaster dataset, a quantitative correspondence between the remaining recovery potential index and vegetation cover is established, the evolution stage and occurrence time of historical debris flows are determined, and a historical debris flow identification result with both spatial location and temporal attributes is formed.

[0024] The multi-source remote sensing data includes airborne LiDAR point cloud data, aerial orthophotos, and multi-temporal optical remote sensing images. The airborne LiDAR point cloud data is preprocessed, including denoising, ground point filtering and spatial interpolation, to generate a digital elevation model with sub-meter resolution; Radiometric calibration and atmospheric correction were performed on the aerial orthophotos and multi-temporal optical remote sensing images. Based on the corrected optical remote sensing images, vegetation cover was calculated, and the erosion texture features of debris flow depositional fans were identified by combining the aerial orthophotos.

[0025] The similar habitat vegetation restoration potential model is based on a sliding window that traverses the entire domain. For each target pixel in the window, pixels with similar terrain conditions and undisturbed vegetation are selected as a set of similar habitats. The similarity of terrain conditions includes the fact that the difference between the two terrain factors, slope and aspect, is less than a preset threshold. The term "undisturbed vegetation" excludes non-natural vegetation pixels that have been damaged or are bare.

[0026] Based on the enhanced vegetation index values ​​of all pixels in the set of similar habitats, the highest percentile is taken as the theoretical vegetation restoration potential of the target pixel. The theoretical vegetation restoration potential represents the upper limit of stable growth that vegetation can achieve under the same habitat conditions.

[0027] The remaining restoration potential index is obtained by calculating the difference between the theoretical vegetation restoration potential and the current observed value of the enhanced vegetation index; The higher the residual recovery potential index, the greater the gap between the vegetation and the natural recovery level, and the closer the debris flow occurred; when the residual recovery potential index approaches zero, it indicates that the vegetation has approached the natural community state.

[0028] The micro-topography enhancement process includes generating multiple mountain shadow images under different lighting conditions based on the high-precision digital elevation model and using multi-directional mountain shadow simulation technology. The multiple mountain shadow images are then fused to highlight the micro-topography details on different slopes and reduce the terrain shading effect caused by a single lighting direction.

[0029] The debris flow depositional landform features include fan-shaped or tongue-shaped planar morphology, boundaries formed by lateral steep slopes or abrupt changes in topographic slope, uneven patchy or strip-shaped micro-topography, and loose deposits. Based on the enhanced micro-topographic images, visual interpretation is performed according to the depositional landform features to map the extent of historical debris flow deposition areas, thus forming a spatial database of historical debris flows.

[0030] The overlay analysis includes extracting the average residual recovery potential index and average vegetation cover within the range of each debris flow depositional fan. Based on the spatial-temporal theory, different depositional fans are used to characterize the sequential state of different recovery times since debris flow, and the quantitative relationship curve between the average residual recovery potential index and the average vegetation cover is fitted by nonlinear regression analysis.

[0031] The determination of the evolutionary stages and occurrence dates of historical debris flows includes: Based on the vegetation cover range, surface texture and erosion characteristics, and the numerical range of the remaining recovery potential index of the depositional fan, a classification standard for the age of debris flow is established. The classification criteria cover three time periods: recent debris flows, intermediate debris flows, and long-term debris flows.

[0032] The historical debris flow identification results include spatial distribution data of debris flows with clear dates of occurrence, which are used to support the graded evaluation of regional debris flow channel activity, the assessment of disaster recurrence cycles, and the assessment of disaster risk.

[0033] One specific application of this embodiment is: Step 1: Collect airborne LiDAR point cloud data of the target area and preprocess it to generate a digital elevation model (DEM) with sub-meter resolution. The preprocessing process mainly includes: first, denoising the original point cloud to remove outliers and redundant noise; then, separating ground points from non-ground points (i.e., vegetation points) using a ground point filtering algorithm to obtain accurate surface point clouds; finally, generating a high-precision DEM using spatial interpolation methods. Simultaneously, collect contemporaneous aerial orthophotos and multi-temporal historical optical remote sensing images, and perform radiometric calibration and atmospheric correction to eliminate sensor and atmospheric interference. Based on the corrected optical remote sensing images, calculate vegetation cover (FVC), and combine this with the high spatial resolution of the aerial orthophotos to identify the erosion texture features of debris flow depositional fans, jointly constructing a baseline dataset for subsequent analysis. The data categories, sources, and uses used in this step are shown in Table 1. Table 1 Datasets used in the study Step Two: This step is the core of the method proposed in this invention, and will be discussed in detail here. To separate the disturbance signal caused by debris flow disasters from complex background vegetation, this invention proposes a method for calculating vegetation restoration potential based on a local sliding window. The specific steps are as follows: First, a rectangular sliding window of a fixed size is set (64×64 pixels in this invention), and the entire study area is traversed in steps of one pixel. For each target pixel at the center of the window, similar habitat pixels are screened within the window range. The screening criteria are based on two constraints: 1. Topographic consistency, that is, only select pixels that are relatively similar to the center pixel in terms of the two key topographic factors of slope and aspect; 2. Undisturbed vegetation: This means excluding pixels with non-natural vegetation such as those affected by disasters or bare land by combining vegetation indices. The pixels selected through these two constraints constitute the "similar habitat set" of the target pixel, representing vegetation backgrounds with the same habitat conditions as the target point and in a naturally stable state.

[0034] Subsequently, the theoretical vegetation restoration potential of the microhabitat where the target pixel is located is calculated. The Enhanced Vegetation Index (EVI) values ​​of all pixels in the corresponding "similar habitat set" are statistically analyzed, and the 95th percentile of the EVI value distribution is taken as the potential value. The calculation formula is as follows: in, The pixel index within the sliding window that satisfies similar habitat conditions; This represents the enhanced vegetation index value for undisturbed pixels within the window. The 95th percentile statistical function is used to characterize the upper limit of stable vegetation growth that can be achieved in this habitat. For target pixels The set of pixels within the central sliding window that satisfy similar habitat conditions and are undisturbed.

[0035] Based on this, in order to quantify the degree of vegetation recovery after debris flow disasters, a Residual Vegetation Recovery Potential Index (VRPI) is generated for each pixel by calculating the difference between the theoretical potential and the current actual observed value. This index reflects the gap between the current vegetation state and the theoretical optimal state; the larger the value, the shorter the time since the disaster. The calculation formula is as follows: In the formula, For pixels The remaining recovery potential This represents the upper limit of theoretical potential under the same habitat. This is the current observation value. The higher the value, the greater the gap between the vegetation and its natural recovery level; conversely, A value close to 0 indicates that the vegetation has approached a natural community.

[0036] Finally, by traversing all pixels in the region to complete the above calculations, a map showing the distribution of the remaining vegetation restoration index of the region after removing background vegetation difference interference is generated. Figure 1 This allows for the quantification of the degree of vegetation recovery after a debris flow disaster, providing data support for the determination of the age of the debris flow in step four.

[0037] Step 3: To address the issue of blurred micro-topographic features due to dense vegetation obscuring optical images in steep mountainous areas, this step utilizes the deveined digital elevation model (DEM) generated in Step 1 to visually enhance micro-topographic features through multi-directional mountain shadow simulation technology. Mountain shadows from a single light source direction can create large areas of shadow depending on the terrain's orientation, obscuring details on the shaded slope. To solve this problem, a series of different solar azimuth angle parameters are set, and multiple mountain shadow images under different lighting conditions are generated based on the classic mountain lighting model formula. The model calculation formula is: in, Zenith angle, For slope, The azimuth is the solar azimuth, and the slope is the slope aspect.

[0038] Subsequently, the generated multi-directional mountain shadow images are fused. The fused image effectively highlights terrain details such as gullies, steep slopes, and protrusions on different slopes, greatly reducing the terrain shading effect caused by a single direction of illumination, thus generating a mountain shadow map that presents the most comprehensive representation of landform features.

[0039] Based on this enhanced mountain shadow image, visual interpretation was performed according to the typical geomorphological identification marks of the debris flow deposition area. Figure 2 Interpretation markers include: 1. Fan-shaped or tongue-shaped planar shape, its lateral boundaries often appear as clear lateral steep slopes or abrupt changes in terrain slope; 2. Rough surface texture, which is manifested as uneven, patchy or striped micro-topography on the surface of the deposit due to material sorting and subsequent erosion and modification; 3. Loose deposits: gravel, angular stones, and silt carried by debris flows from the source area and channels to the surface of the deposits.

[0040] Based on the above indicators, the extent of historical debris flow deposition areas was identified and mapped, and background mountains that did not possess the characteristics of disaster morphology were removed, ultimately forming a historical debris flow database for subsequent analysis.

[0041] Step 4: Overlay the Remaining Vegetation Restoration Potential Index (VRPI) calculated in Step 2 with the historical debris flow spatial database extracted in Step 3, and extract the average VRPI value and vegetation cover (FVC) within the range of each debris flow depositional fan.

[0042] Subsequently, based on the "spatial-temporal" theory, different alluvial fans within the study area were used to characterize the sequential state of vegetation recovery over different durations since the debris flow. The average vegetation cover of each alluvial fan was used as a proxy indicator of the recovery time, and the average remaining vegetation recovery potential was used as a quantitative indicator of the current recovery state, both displayed on a scatter plot. A quantitative relationship curve between the two was fitted using nonlinear regression analysis. Figure 3 Based on this, a criterion for determining the age of debris flows was constructed.

[0043] The specific judgment criteria are as follows (Table 2): 1. When the vegetation cover of a debris flow fan is less than 20% and the corresponding surface texture and erosion characteristics are recent erosion texture and sharp edges of the rocks, it is determined to be a recent debris flow that occurred within the last 1-5 years. 2. If the vegetation cover is between 20% and 70% and the corresponding surface texture and erosion characteristics are reduced brightness of the eroded bare land and reduced texture roughness; and gully morphology is preserved in some areas, then it is determined to be a medium-term debris flow in the last 5-20 years. 3. If the vegetation coverage reaches 70%-90% and the corresponding surface texture and erosion characteristics are similar to the background texture, and the natural slope characteristics are significant, then it is determined to be a debris flow in the last 20-30 years.

[0044] Table 2 Criteria for Determining the Age of Debris Flows By using the above-mentioned grading standards, the age of historical debris flows can be quantitatively identified, ultimately resulting in debris flow identification results with a clear date of disaster occurrence.

[0045] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0046] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A remote sensing identification method for debris flows in steep mountainous areas based on a vegetation restoration model, characterized in that, The method includes the following steps: Acquire multi-source remote sensing data of the target area, generate a high-precision digital elevation model based on the multi-source remote sensing data, and obtain a vegetation coverage dataset; Based on the vegetation restoration potential model of similar habitats, the theoretical vegetation restoration potential and remaining restoration potential index of each pixel are calculated using the vegetation coverage dataset to quantify the degree of ecological restoration after debris flow disasters. Based on the high-precision digital elevation model, vegetation canopy interference is removed to construct a realistic surface model. Micro-topographic enhancement processing is performed on the realistic surface model. Historical debris flow disaster datasets are identified and extracted based on debris flow deposition landform characteristics. The remaining recovery potential index is overlaid with the historical debris flow disaster dataset to establish a quantitative correspondence between the remaining recovery potential index and vegetation cover, determine the evolution stage and occurrence time of historical debris flows, and form historical debris flow identification results with both spatial location and temporal attributes. The multi-source remote sensing data includes airborne LiDAR point cloud data, aerial orthophotos, and multi-temporal optical remote sensing images. The airborne LiDAR point cloud data is preprocessed, including denoising, ground point filtering and spatial interpolation, to generate a digital elevation model with sub-meter resolution. Radiometric calibration and atmospheric correction were performed on the aerial orthophotos and multi-temporal optical remote sensing images. Based on the corrected optical remote sensing images, vegetation cover was calculated, and the erosion texture features of debris flow depositional fans were identified by combining the aerial orthophotos. The similar habitat vegetation restoration potential model is based on a sliding window that traverses the entire domain. For each target pixel in the window, pixels with similar terrain conditions and undisturbed vegetation are selected as a set of similar habitats. The similarity of terrain conditions includes the fact that the difference between the two terrain factors, slope and aspect, is less than a preset threshold. The term "undisturbed vegetation" excludes non-natural vegetation pixels that have been damaged or are bare land. Based on the enhanced vegetation index values ​​of all pixels in the set of similar habitats, the highest percentile is taken as the theoretical vegetation restoration potential of the target pixel. The theoretical vegetation restoration potential represents the upper limit of stable growth that vegetation can achieve under the same habitat conditions.

2. The remote sensing identification method for debris flows in steep mountainous areas based on a vegetation restoration model according to claim 1, characterized in that, The remaining restoration potential index is obtained by calculating the difference between the theoretical vegetation restoration potential and the current observed value of the enhanced vegetation index; The higher the residual recovery potential index, the greater the gap between the vegetation and the natural recovery level, and the closer the debris flow occurred; when the residual recovery potential index approaches zero, it indicates that the vegetation has approached the natural community state.

3. The remote sensing identification method for debris flows in steep mountainous areas based on a vegetation restoration model according to claim 1, characterized in that, The micro-topography enhancement process includes generating multiple mountain shadow images under different lighting conditions based on the high-precision digital elevation model and using multi-directional mountain shadow simulation technology. The multiple mountain shadow images are then fused to highlight the micro-topography details on different slopes and reduce the terrain shading effect caused by a single lighting direction.

4. The remote sensing identification method for debris flows in steep mountainous areas based on a vegetation restoration model according to claim 3, characterized in that, The debris flow depositional landform features include fan-shaped or tongue-shaped planar morphology, boundaries formed by lateral steep slopes or abrupt changes in topographic slope, uneven patchy or strip-shaped micro-topography, and loose deposits. Based on the enhanced micro-topographic images, visual interpretation is performed according to the depositional landform features to map the extent of historical debris flow deposition areas, thus forming a spatial database of historical debris flows.

5. The remote sensing identification method for debris flows in steep mountainous areas based on a vegetation restoration model according to claim 1, characterized in that, The overlay analysis includes extracting the average residual recovery potential index and average vegetation cover within the range of each debris flow depositional fan. Based on the spatial-temporal theory, different depositional fans are used to characterize the sequential state of different recovery times since debris flow, and the quantitative relationship curve between the average residual recovery potential index and the average vegetation cover is fitted by nonlinear regression analysis.

6. The remote sensing identification method for debris flows in steep mountainous areas based on a vegetation restoration model according to claim 5, characterized in that, The determination of the evolutionary stages and occurrence dates of historical debris flows includes: Based on the vegetation cover range, surface texture and erosion characteristics, and the numerical range of the remaining recovery potential index of the depositional fan, a classification standard for the age of debris flow is established. The classification criteria cover three time periods: recent debris flows, intermediate debris flows, and long-term debris flows.

7. The remote sensing identification method for debris flows in steep mountainous areas based on a vegetation restoration model according to claim 1, characterized in that, The historical debris flow identification results include spatial distribution data of debris flows with clear dates of occurrence, which are used to support the graded evaluation of regional debris flow channel activity, the assessment of disaster recurrence cycles, and the assessment of disaster risk.

Images