Method and device for automatic identification of debris flow source and static reserve calculation, and storage medium

Abstract

The application relates to the technical field of debris flow source identification, and specifically discloses a debris flow source automatic identification and static reserve calculation method and device and a storage medium. The method comprises the following steps: S1, obtaining relevant geological data of a study area, reproducing a real field landscape by using remote sensing data, judging the category and boundary of a loose accumulation body according to different surface geometric features, texture features and spectral features, establishing a loose accumulation body database of the study area, and identifying the geometric features and distribution features of the loose accumulation body; S2, automatically identifying the loose accumulation body by using a deep learning model; and S3, classifying the identified loose accumulation body, combining different accumulation forms and different damage modes of different classifications, and establishing a volume calculation model of various debris flow sources.

Description

Technical Field

[0001] This application relates to the field of debris flow source identification technology, specifically to a method, device and storage medium for automatic identification of debris flow sources and static storage calculation. Background Technology

[0002] The solid material source of debris flows is the material basis for their formation. Monitoring its properties, types, spatial distribution, and scale is crucial for predicting debris flow trends. The most traditional method for debris flow source investigation is field surveys. Detailed field surveys can reveal the boundary morphology, distribution area, and soil and rock properties of debris flow sources. Based on this, studies can be conducted on the volume, development and distribution characteristics, and initiation modes of debris flow sources. According to research by multiple scholars and experts, debris flow sources mainly originate from landslide deposits within the watershed, loose deposits on slopes, and pre-existing deposits within gullies. Identifying debris flow sources requires combining the characteristics of various geological hazards with the features of the deposits they directly or indirectly form.

[0003] Currently, the identification of debris flow sources is mainly based on visual interpretation, which is inefficient given the massive amounts of remote sensing data available today. With the increasing application of computer science in geology, many semi-automatic or automatic identification methods have emerged, such as change detection, machine learning, and deep learning.

[0004] The common method for calculating the static storage capacity of debris flow sources involves first classifying the loose deposits within the watershed, then calculating the volume of each type of loose deposit separately, and finally summing them to obtain the static storage capacity of the debris flow source within the watershed. In the absence of subsurface data, many scholars both domestically and internationally rely on morphological methods to calculate the volume of landslide and collapse deposits.

[0005] Whether it's manual visual identification or automatic identification, the quality of the data directly affects the accuracy of the identification results. In vegetated mountainous areas, using only optical satellite imagery to identify debris flow sources has significant limitations; it cannot detect debris flow sources below vegetation. Most debris flow source estimation models are mainly established using relevant parameters that affect debris flow development, and these empirical formulas are mostly regional and lack universality. Summary of the Invention

[0006] The purpose of this application is to provide a method, device and storage medium for automatic identification of debris flow sources and static storage calculation, in order to solve the problem that the existing technology has great limitations in identifying debris flow sources by using only optical satellite imagery, which cannot detect debris flow sources below vegetation, and that most debris flow source estimation models are mainly established by using relevant parameters that affect debris flow development, and most of these empirical formulas lack universality.

[0007] To achieve the above objectives, this application provides an automatic method for identifying debris flow sources and calculating static reserves, including the following steps: S1, acquiring relevant geological data of the study area, reproducing the real landform landscape in the field using remote sensing data, and then determining the type and boundary of loose deposits based on different surface geometric features, texture features and spectral features, establishing a database of loose deposits in the study area, and identifying their geometric features and distribution features.

[0008] S2. Automatic identification of loosely packed structures using deep learning models;

[0009] S3. Classify the identified loose deposits, and establish calculation models for the source volume of debris flows based on the different deposit morphologies and failure modes of different classifications.

[0010] Optionally, step S1 specifically includes:

[0011] S1.1 Obtain relevant geological data for the study area, including at least one of the following: topography, stratigraphy and lithology, geological tectonic evolution history, historical earthquakes, meteorological and hydrogeological conditions, and geological hazards in the study area;

[0012] S1.2 After acquiring survey area data using airborne LiDAR technology, the generated 3D products are visualized using ArcGIS.

[0013] S1.3. Constructing an interpretation environment for loose deposits, classifying, organizing, and comprehensively analyzing existing data, summarizing the LiDAR remote sensing characteristics of loose deposits, and establishing remote sensing identification markers for different types of loose deposits. Step S1.3 specifically includes:

[0014] S1.3.1 Loose deposits are classified into: landslide deposits, collapse deposits, gully deposits, and slope deposits.

[0015] S1.3.2. Based on the characteristics of landslide deposits, collapse deposits, gully deposits, and slope deposits, use ArcGIS software to identify loose deposits on mountain shadow images generated by airborne LiDAR-DEM.

[0016] S1.4 Based on the identification of different types of loose deposits, summarize the geometric and distribution characteristics of loose deposits in the database. The geometric characteristics include at least one of the following: area characteristics, size characteristics, and side length ratio characteristics. The distribution characteristics include at least one of the following: elevation distribution characteristics, slope distribution characteristics, and aspect distribution characteristics.

[0017] Optionally, step S2 specifically includes:

[0018] S2.1 Divide the interpreted loosely packed data into training and testing sets;

[0019] S2.2. Based on the dataset range, convert the image raster and debris flow source vector into sample images and sample labels, respectively. Step S2.2 specifically includes:

[0020] S2.2.1 Determine the pixel size of the sample image based on the size characteristics of the loosely packed volume, and generate the corresponding cropping vector within the dataset vector range.

[0021] S2.2.2. Taking the intersection of the clipping vector and the interpretation vector yields the debris flow source label, i.e., the sample label, for the loose deposit within the image.

[0022] S2.2.3. Use cropping vectors to crop the image raster to obtain sample images;

[0023] S2.3. The average accuracy rate is used to evaluate the performance of the instance segmentation model;

[0024] S2.4. Use the instance segmentation model algorithm to fully learn and train the test set.

[0025] Optionally, step S3 specifically includes:

[0026] S3.1 Calculate the source volume of landslide deposits using the digital elevation model method based on high-precision DEM data;

[0027] S3.2 Based on a high-precision DEM model, by selecting elevation information on the collapse boundary, the original terrain interface is fitted, and then the difference between the original DEM model and the original DEM model is calculated to obtain the volume of the collapse deposit source.

[0028] S3.3 Based on a high-precision DEM model, the volume of the channel deposit source is calculated using a triangular cross section;

[0029] S3.4. Based on a high-precision DEM model, the RUSLE model is used to calculate the volume of the slope deposit source. The RUSLE model expression is as follows:

[0030] A = R * K * LS * C * P,

[0031] Where A is the average annual soil loss, R is the rainfall erosivity factor, K is the soil erosibility factor, L is the slope length factor, S is the slope factor, C is the vegetation cover and management factor, and P is the soil and water conservation measures factor.

[0032] Optionally, step S3.1 specifically includes:

[0033] S3.1.1 Divide the landslide into the source area and the deposition area;

[0034] S3.1.2. Using topographic data from the landslide source area and the boundary line of the deposition area, the failure surface is fitted using the trend surface method with second-order polynomial fitting, specifically including:

[0035] S3.1.2.1 Generate a series of points with elevation information within the slip source area and on the boundary of the deposition area.

[0036] S3.1.2.2: Using the points with elevation information generated in S3.1.2.1, the landslide failure surface is fitted using the least squares method.

[0037] S3.1.3 Calculate the difference between the original DEM data and the fitted failure surface, and then obtain the size of the landslide volume through raster statistics.

[0038] Optionally, step S3.2 specifically includes:

[0039] S3.2.1. Using topographic data along the boundary of the depositional area, the natural neighborhood surface method is used to fit the landslide interface, specifically including:

[0040] S3.2.1.1 Generate a series of points with elevation information on the boundary of the accumulation area.

[0041] S3.2.1.2, Fit the collapsed bottom interface using the elevation information points generated in S3.2.1.1;

[0042] S3.2.2 Calculate the difference between the existing DEM data and the fitted original terrain interface to obtain the size of the source volume of the landslide deposit.

[0043] Optionally, step S3.3 specifically includes:

[0044] S3.3.1. Fit a plane connecting both sides of the trench bed by the boundary line of the deposit source in the trench;

[0045] S3.3.2. Subtract the actual topographic line from the plane fitted in S3.3.1 to obtain the scour depth of the channel deposit surface, and take the maximum value of the scour depth of the channel deposit surface at that location as the thickness of the source material of the channel deposit.

[0046] S3.3.3, Using the formula:

[0047]

[0048]

[0049] The source volume of the channel deposit is obtained, where Δz i To fit the difference between the fitted plane and the actual terrain plane, Z iThe equivalent depth at a certain point in the channel deposit source, where H is the thickness of the channel deposit source, and V is the equivalent depth at a certain point in the channel deposit source. gd S represents the source volume of the channel deposits, and S represents the area of ​​the DEM raster cell.

[0050] Optionally, step S3.4 specifically includes:

[0051] S3.4.1 Obtain precipitation data for the study area, and establish a simplified model using annual rainfall data to estimate the rainfall erosivity factor. The expression for this simplified model is as follows:

[0052]

[0053] Among them, P j R represents the rainfall (mm) in year j. j Let α3 and β3 be the rainfall erosion force in year j, and α3 and β3 be the model parameters.

[0054] S3.4.2. Assess the soil erodibility factor K;

[0055] S3.4.3 Obtain slope and slope length factors through field measurements, or extract slope and slope length factors from DEM data at the watershed scale;

[0056] S3.4.4 Calculate the vegetation cover and management factor C using the Normalized Difference Vegetation Index (NDVI). The relationship between the two is expressed as follows:

[0057]

[0058] S3.4.5. Determine the soil and water conservation measure factor P value by combining field survey data with the method of assigning values ​​to different land use types.

[0059] To achieve the above objectives, this application also provides an automatic debris flow source identification and static storage calculation device, comprising: a memory; and

[0060] A processor connected to the memory, the processor being configured to perform the steps of the method described above.

[0061] To achieve the above objectives, this application also provides a computer storage medium having a computer program stored thereon, wherein the computer program, when executed by a machine, implements the steps of the method described above.

[0062] The embodiments of this application have the following advantages:

[0063] This application provides an automatic method for identifying debris flow sources and calculating static reserves, comprising: S1, acquiring relevant geological data of the study area, reproducing the real landform landscape using remote sensing data, and then determining the category and boundary of loose deposits based on different surface geometric features, texture features, and spectral features, establishing a database of loose deposits in the study area, and identifying their geometric features and distribution features; S2, using a deep learning model to automatically identify loose deposits; S3, classifying the identified loose deposits, and establishing various debris flow source volume calculation models by combining different depositional forms and different failure modes of different categories.

[0064] The above methods not only consider surface deformation caused by human engineering disturbances or natural factors, which is of great significance to the assessment of geological hazard risks, but also better reflect the complex relationships between influencing factors of geological hazards, improving the accuracy and effectiveness of prediction and early warning. This addresses the limitations of existing technologies that rely solely on optical satellite imagery to identify debris flow sources, failing to detect debris flow sources below vegetation cover, and the fact that most debris flow source estimation models rely on parameters affecting debris flow development, resulting in empirical formulas that lack universality. Attached Figure Description

[0065] To more clearly illustrate the embodiments of this application or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.

[0066] Figure 1 A flowchart of a method for automatic identification of debris flow sources and static storage calculation provided for at least one embodiment of this application;

[0067] Figure 2 A flowchart illustrating the establishment of a loose aggregate database provided for at least one embodiment of this application;

[0068] Figure 3 This application provides a DEM-converted mountain shadow effect image for at least one embodiment;

[0069] Figure 4 A landslide identification feature map provided for at least one embodiment of this application;

[0070] Figure 5 This application provides a collapse accumulation identification feature map for at least one embodiment of the present application;

[0071] Figure 6A partial feature map of channel accumulation provided in at least one embodiment of this application;

[0072] Figure 7 A slope deposition identification feature map provided for at least one embodiment of this application;

[0073] Figure 8 Sample images and sample label diagrams provided for at least one embodiment of this application;

[0074] Figure 9 A diagram of the slip region and the accumulation region provided for at least one embodiment of this application;

[0075] Figure 10 This application provides for generating point data maps with elevation information in at least one embodiment;

[0076] Figure 11 A fitted failure surface elevation information map provided for at least one embodiment of this application;

[0077] Figure 12 A difference map between the damaged surface and the landslide surface provided for at least one embodiment of this application;

[0078] Figure 13 A schematic diagram illustrating the generation of point data with elevation information for a landslide deposit source area provided in at least one embodiment of this application;

[0079] Figure 14 A schematic diagram of the fitted collapsed bottom interface provided for at least one embodiment of this application;

[0080] Figure 15 A schematic diagram illustrating the difference between the inferred collapse bottom interface and the collapse surface, provided for at least one embodiment of this application;

[0081] Figure 16 A schematic diagram of a triangular scour cross section provided in at least one embodiment of this application;

[0082] Figure 17 A schematic diagram of surface scouring depth provided for at least one embodiment of this application;

[0083] Figure 18 A schematic diagram of annual erosion volume provided for at least one embodiment of this application;

[0084] Figure 19 A block diagram of an automatic debris flow source identification and static storage calculation device provided for at least one embodiment of this application. Detailed Implementation

[0085] The following specific embodiments illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0086] It should be noted that the steps in the claims and description of this application may be performed substantially in parallel or in reverse order under appropriate circumstances, depending on the function involved.

[0087] Furthermore, the technical features involved in the different embodiments of this application described below can be combined with each other as long as they do not conflict with each other.

[0088] One embodiment of this application provides a method for automatic identification of debris flow sources and static storage calculation, referencing... Figure 1 , Figure 1 The flowchart illustrates an automatic debris flow source identification and static reserve calculation method provided in at least one embodiment of this application. It should be understood that the method may further include additional boxes not shown and / or the boxes shown may be omitted; the scope of this application is not limited in this respect. The method includes the following steps:

[0089] S1. Obtain relevant geological data for the study area, reconstruct the actual landform using remote sensing data, and then determine the type and boundaries of loose deposits based on different surface geometric features, texture features, and spectral features (e.g., through visual interpretation methods or machine learning models). Establish a database of loose deposits in the study area (e.g., Figure 2 (as shown), and identify its geometric and distribution characteristics.

[0090] In some embodiments, step S1 specifically includes:

[0091] S1.1 Obtain relevant geological data for the study area, including at least one of the following: topography, stratigraphy and lithology, geological tectonic evolution history, historical earthquakes, meteorological and hydrogeological conditions, and geological hazards in the study area.

[0092] S1.2 After acquiring the survey area data using airborne LiDAR technology, the generated 3D product (DEM) is visualized using ArcGIS.

[0093] S1.3 Construct an environment for identifying and interpreting loose deposits (e.g., using remote sensing software), classify, organize, and comprehensively analyze existing data, summarize the LiDAR remote sensing characteristics of loose deposits, and establish remote sensing identification markers for different types of loose deposits.

[0094] In some embodiments, step S1.3 specifically includes:

[0095] S1.3.1 Loose deposits are classified into: landslide deposits, collapse deposits, gully deposits, and slope deposits.

[0096] S1.3.2. Based on the characteristics of landslide deposits, collapse deposits, gully deposits, and slope deposits, use ArcGIS software to identify loose deposits on mountain shadow images generated by airborne LiDAR-DEM (e.g., through visual interpretation methods or machine learning models).

[0097] S1.4. Based on the identification of different types of loose deposits, summarize the geometric and distribution characteristics of loose deposits in the database. The geometric characteristics include at least one of the following: area characteristics, size characteristics, and side length ratio characteristics, etc. The distribution characteristics include at least one of the following: elevation distribution characteristics, slope distribution characteristics, and aspect distribution characteristics, etc.

[0098] S2. Use deep learning models (e.g., SOLO v2 architecture) for automatic identification of loosely packed structures.

[0099] In some embodiments, step S2 specifically includes:

[0100] S2.1. Divide the interpreted loosely packed data into a training set and a test set. The training set is used to train the deep learning model, and the test set is used to evaluate and validate the model's performance.

[0101] S2.2. Based on the dataset range, convert the image raster and debris flow source vector into sample images and sample labels respectively (e.g., using ArcGIS software).

[0102] In some embodiments, step S2.2 specifically includes:

[0103] S2.2.1 Determine the pixel size of the sample image based on the size characteristics of the loosely packed body, and generate the corresponding cropping vector within the dataset vector range.

[0104] S2.2.2. Take the intersection of the clipping vector and the interpretation vector to obtain the debris flow source label (i.e., sample label) of the loose deposit in the image.

[0105] S2.2.3. Use cropping vectors to crop the image raster to obtain sample images.

[0106] S2.3. Use AP (Average Precision) to evaluate the performance of the instance segmentation model. AP is calculated for the recall and precision of a certain class at a given IoU (Intersection of Union) threshold. The area enclosed by the recall-precision curve and the coordinate axis is the AP value.

[0107] S2.4. The instance segmentation model algorithm is used to fully learn and train on the test set. In some embodiments, the instance segmentation model algorithm code is derived from the open-source object detection (MMDetection) framework developed by the Visual Intelligence Research Group at the Chinese University of Hong Kong, and implemented using PyTorch. In some embodiments, SOLO v2 is implemented based on the MMDetection framework. In some embodiments, the instance segmentation model uses ResNet-101 as the backbone network. The main hyperparameters used in the instance segmentation model are a learning rate of 0.001 and a batch size of 4. The input size used in the algorithm is 1024 pixels × 1024 pixels.

[0108] S3. Classify the identified loose deposits, and establish calculation models for the source volume of debris flows based on the different deposit morphologies and failure modes of different classifications.

[0109] Specifically, loose deposits are classified as: landslide deposits, collapse deposits, gully deposits, and slope deposits.

[0110] In some embodiments, step S3 specifically includes:

[0111] S3.1. Calculate the landslide deposit volume using the digital elevation model method with high-precision DEM (DEM is a digital elevation model; in this application embodiment, the high-precision DEM model is the aforementioned debris flow source volume calculation model) data.

[0112] In some embodiments, step S3.1 specifically includes:

[0113] S3.1.1 Divide the landslide into the landslide source area and the deposition area.

[0114] S3.1.2 Using the topographic data of the landslide source area and the topographic data of the deposition area boundary line, the failure surface is fitted by the trend surface method of second-order polynomial fitting. The trend surface equation can be expressed by formula (1-1).

[0115] z(x,y)=a1x+a2x 2 +b1y+b2y 2 +c (1-1)

[0116] In some embodiments, step S3.1.2 specifically includes:

[0117] S3.1.2.1 Generate a series of points with elevation information within the slip source area and on the boundary of the deposition area (e.g., using ArcGIS).

[0118] S3.1.2.2: Using the points with elevation information generated in S3.1.2.1, the landslide failure surface is fitted using the least squares method.

[0119] S3.1.3 Calculate the difference between the original DEM data and the fitted failure surface, and then perform raster statistics to obtain the size of the landslide volume.

[0120] S3.2 Based on a high-precision DEM model, the original terrain interface is fitted by selecting the elevation information on the collapse boundary, and then the difference between the original DEM model and the original DEM model is used to obtain the volume of the collapse deposit source.

[0121] In some embodiments, the existing DEM model is downloaded from the SRTM (Shuttle Radar Topography Mission) of the geospatial data cloud platform (http: / / www.gscloud.cn / ) with a spatial resolution of 90m.

[0122] In some embodiments, step S3.2 specifically includes:

[0123] S3.2.1 Using topographic data along the boundary of the deposition area, the natural neighborhood surface method is used to fit the collapse interface.

[0124] In some embodiments, step S3.2.1 specifically includes:

[0125] S3.2.1.1 Generate a series of points with elevation information on the boundary of the accumulation area (e.g., using ArcGIS).

[0126] S3.2.1.2: Fit the collapsed bottom interface using the elevation information points generated in S3.2.1.1.

[0127] S3.2.2 Calculate the difference between the existing DEM data and the fitted original terrain interface to obtain the size of the source volume of the landslide deposit.

[0128] S3.3. Based on a high-precision DEM model, the volume of the sediment source in the channel is calculated using a triangular cross section.

[0129] In some embodiments, step S3.3 specifically includes:

[0130] S3.3.1. Fit a plane connecting both sides of the trench bed by the boundary line of the deposit source in the trench.

[0131] S3.3.2. Subtracting the actual topographic line from the plane fitted in S3.3.1, the scour depth of the channel deposit surface can be obtained, and it is assumed that the maximum value of the scour depth of the channel deposit surface at that location is taken as the thickness of the source material of the channel deposit.

[0132] S3.3.3. The source volume of the channel deposit is obtained using the following formulas (1-2) and (1-3). The formulas for calculating the source volume of the channel deposit are shown in formulas (1-2) and (1-3).

[0133]

[0134]

[0135] In the formula, Δz i To fit the difference between the fitted plane and the actual terrain plane, Z i The equivalent depth at a certain point in the channel deposit source, where H is the thickness of the channel deposit source; V gd S represents the source volume of the channel deposits; S represents the area of ​​the DEM raster pixels.

[0136] S3.4. Based on the high-precision DEM model, the RUSLE model is used to calculate the volume of the slope deposit source, and its expression is shown in equation (1-4).

[0137] A = R * K * LS * C * P (1-4)

[0138] In the formula: A is the average annual soil loss, R is the rainfall erosivity factor, K is the soil erosibility factor, L is the slope length factor, S is the slope factor, C is the vegetation cover and management factor, and P is the soil and water conservation measures factor.

[0139] In some embodiments, step S3.4 specifically includes:

[0140] S3.4.1 Obtain precipitation data for the study area (e.g., obtain precipitation data provided by the National Earth System Science Data Sharing Service Platform), and establish a simplified model using annual rainfall data to estimate the rainfall erosivity factor, as shown in equation (1-5):

[0141]

[0142] In the formula, P j R represents the rainfall (mm) in year j. j Let α3 and β3 be the erosivity of rainfall in year j, and let α3 and β3 be the model parameters. In some embodiments, the erosivity rainfall model of Zhang Wenbo et al. (2003) is used, where α3 = 0.053 and β3 = 1.6548.

[0143] S3.4.2 Using data provided by the National Earth System Science Data Sharing Service Platform and in conjunction with relevant research by Deng Liangji et al. (2003), the soil erodibility factor K was evaluated.

[0144] S3.4.3, slope and slope length factor LS can be obtained through field measurements, and at the watershed scale, they can be extracted from DEM data.

[0145] S3.4.4 Calculate the vegetation cover and management factor C using the normalized difference vegetation index (NDVI). The relationship between the two is shown in equation (1-6).

[0146]

[0147] S3.4.5. The soil and water conservation measure factor P value is determined by assigning values ​​to different land use types based on field survey data. The P value ranges from 0 to 1, where 0 represents areas without soil erosion, and 1 represents areas without any soil and water conservation measures. Water bodies, built-up land, natural forests, shrublands, and unused land generally do not have soil and water conservation measures and are assigned a value of 1; while plantations and orchards can play a better role in preventing soil erosion and are assigned a value of 0.7.

[0148] To make the purpose, technical solution and advantages of this application clearer, the following describes the application in further detail with reference to the accompanying drawings and examples. The example is a method for automatic identification of debris flow sources and static storage calculation in a certain area.

[0149] S1. Collect relevant geological data of the study area, use remote sensing data to reproduce the real landform landscape in the field, and then determine the type and boundary of loose deposits by visual interpretation method based on different surface geometric features, texture features and spectral features, establish a database of loose deposits in the study area, and identify their geometric features and distribution features.

[0150] S1.1 Collect relevant geological data for the study area, including but not limited to: topography, stratigraphy and lithology, geological structure evolution history, historical earthquakes, meteorological and hydrogeological conditions, and geological hazards in the study area.

[0151] S1.2 After acquiring survey area data using airborne LiDAR technology, the generated 3D product (DEM) is visualized using ArcGIS. In this example, the DEM is converted using ArcGIS spatial analysis tools to create a corresponding hillshade effect, as shown below. Figure 3 As shown.

[0152] S1.3. Construct a loose deposit identification and interpretation environment using professional remote sensing software, classify, organize, and comprehensively analyze existing data, summarize the LiDAR remote sensing characteristics of loose deposits, and establish remote sensing identification markers for different types of loose deposits.

[0153] S1.3.1 Loose deposits are classified as: landslide deposits, collapse deposits, gully deposits, and slope deposits.

[0154] S1.3.2. Based on the characteristics of landslide deposits, collapse deposits, gully deposits, and slope deposits, loose deposits are identified through visual interpretation using ArcGIS software on mountain shadow images generated by airborne LiDAR-DEM, such as... Figures 4-7 As shown.

[0155] S1.4. Based on the identification of different types of loose deposits, summarize the geometric and distribution characteristics of loose deposits in the database. Geometric characteristics include, but are not limited to: area characteristics, size characteristics, and side length ratio characteristics. Distribution characteristics include, but are not limited to: elevation distribution characteristics, slope distribution characteristics, and aspect distribution characteristics.

[0156] S2. Automatic identification of loosely packed structures using the SOLO v2 deep learning model.

[0157] S2.1. Divide the interpreted loosely packed data into a training set and a test set. The training set is used to train the deep learning model, and the test set is used to evaluate and validate the model's performance.

[0158] S2.2 In ArcGIS software, based on the dataset range, the image raster and debris flow source vector are converted into sample images and sample labels, respectively.

[0159] S2.2.1. Determine the pixel size of the sample image based on the size characteristics of the loose deposits, and generate the corresponding cropping vector within the dataset vector range. In this embodiment, the layer used to identify the loose deposits is a mountain shadow raster with four different solar azimuth angles (45°, 135°, 225° and 315° respectively). Based on the resolution of the Jiuzhaigou image data, a training set sample image with a resolution of 0.5m and a pixel size of 1024×1024 is obtained by cropping.

[0160] S2.2.2. Take the intersection of the clipping vector and the interpretation vector to obtain the debris flow source label for the loose deposit in the image. In this embodiment, the label pixel values ​​for the debris flow source are "1", "2", "3", and "4". "1" represents landslide deposit, "2" represents collapse deposit, "3" represents gully deposit, and "4" represents slope deposit. The background value is 0. Figure 8The image shown is an example of a sample image of the study area and its label. Because the pixel value of the loosely packed label is very small, the color of the label is very close to the background, appearing as black.

[0161] S2.2.3. Use cropping vectors to crop the image raster to obtain sample images.

[0162] S2.3. Use AP (Average Precision) to evaluate the performance of the instance segmentation model.

[0163] S2.4. Use the instance segmentation model algorithm to fully learn and train the test set.

[0164] S3. Classify debris flow sources and establish volume calculation models for various debris flow sources based on their different depositional forms and different failure modes.

[0165] S3.1 Calculate the source volume of landslide deposits using the digital elevation model method based on high-precision DEM data.

[0166] S3.1.1 Divide the landslide into a source area and a deposition area, such as... Figure 9 As shown.

[0167] S3.1.2. Using the topographic data of the landslide source area and the topographic data along the boundary of the deposition area, the failure surface is fitted using the trend surface method with second-order polynomial fitting. In this embodiment, a series of points with elevation information are generated inside the landslide source area and on the boundary of the deposition area. Figure 10 Then, using these points, the least squares method is used to fit the landslide failure surface, such as... Figure 11 As shown.

[0168] S3.1.3 Calculate the difference between the original DEM data and the fitted failure surface, and then perform raster statistics to obtain the size of the landslide volume, such as... Figure 12 As shown.

[0169] S3.2 Based on a high-precision DEM model, the original terrain interface is fitted by selecting the elevation information on the collapse boundary, and then the difference between the original and existing DEM models can be used to obtain the volume of the collapse accumulation.

[0170] S3.2.1 Using topographic data along the boundary of the deposition area, the natural neighborhood surface method is used to fit the collapse interface.

[0171] S3.2.1.1 Use ArcGIS to generate a series of points with elevation information on the boundary of the accumulation area, such as... Figure 13 As shown.

[0172] S3.2.1.2, Using the elevation information points generated in S3.2.1.1, the collapse floor interface is fitted, such as... Figure 14 As shown.

[0173] S3.2.2 Calculate the difference between the existing DEM data and the fitted original terrain interface to obtain the size of the landslide deposit source volume, such as... Figure 15 As shown.

[0174] S3.3. Based on a high-precision DEM model, the volume of the sediment source in the channel is calculated using a triangular cross section.

[0175] S3.3.1. Fit a plane connecting both sides of the channel bed by the boundary line of the sediment source, such as... Figure 16 As shown, that is Figure 16 The plane containing line AB.

[0176] S3.3.2. Subtracting the actual topographic line from the plane fitted in S3.3.1 yields the scour depth of the channel deposit surface, and it is assumed that the maximum value of the scour depth of the channel deposit surface at that location is taken as the thickness of the source material of the channel deposit. For example... Figure 17 As shown, the maximum surface scour depth of the sediment source in this ditch is 10.45m.

[0177] S3.3.3 The volume of the sediment source in the channel can be calculated according to Equation 5-3.

[0178] S3.4. The volume of slope deposits is calculated using Equations 1-4. In this embodiment, the volume of slope deposits is calculated based on the annual erosion rate. The unit weight of soil is taken as 1.802 t / m³. 3 The erosion volume distribution is as follows Figure 18 As shown.

[0179] The above methods not only consider surface deformation caused by human engineering disturbances or natural factors, which is of great significance to the assessment of geological hazard risks, but also better reflect the complex relationships between influencing factors of geological hazards, improving the accuracy and effectiveness of prediction and early warning. This addresses the limitations of existing technologies that rely solely on optical satellite imagery to identify debris flow sources, failing to detect debris flow sources below vegetation cover, and the fact that most debris flow source estimation models rely on parameters affecting debris flow development, resulting in empirical formulas that lack universality.

[0180] Figure 19 A block diagram of an automatic debris flow source identification and static storage calculation device provided for at least one embodiment of this application. The device includes:

[0181] The memory 201; and the processor 202 connected to the memory 201, the processor 202 being configured to: S1, acquire relevant geological data of the study area, reproduce the real landform landscape in the field using remote sensing data, and then determine the type and boundary of the loose deposits based on different surface geometric features, texture features and spectral features, establish a database of loose deposits in the study area, and identify their geometric features and distribution features.

[0182] S2. Automatic identification of loosely packed structures using deep learning models;

[0183] S3. Classify the identified loose deposits, and establish calculation models for the source volume of debris flows based on the different deposit morphologies and failure modes of different classifications.

[0184] In some embodiments, the processor 202 is further configured such that step S1 specifically includes:

[0185] S1.1 Obtain relevant geological data for the study area, including at least one of the following: topography, stratigraphy and lithology, geological tectonic evolution history, historical earthquakes, meteorological and hydrogeological conditions, and geological hazards in the study area;

[0186] S1.2 After acquiring survey area data using airborne LiDAR technology, the generated 3D products are visualized using ArcGIS.

[0187] S1.3. Constructing an interpretation environment for loose deposits, classifying, organizing, and comprehensively analyzing existing data, summarizing the LiDAR remote sensing characteristics of loose deposits, and establishing remote sensing identification markers for different types of loose deposits. Step S1.3 specifically includes:

[0188] S1.3.1 Loose deposits are classified into: landslide deposits, collapse deposits, gully deposits, and slope deposits.

[0189] S1.3.2. Based on the characteristics of landslide deposits, collapse deposits, gully deposits, and slope deposits, use ArcGIS software to identify loose deposits on mountain shadow images generated by airborne LiDAR-DEM.

[0190] S1.4 Based on the identification of different types of loose deposits, summarize the geometric and distribution characteristics of loose deposits in the database. The geometric characteristics include at least one of the following: area characteristics, size characteristics, and side length ratio characteristics. The distribution characteristics include at least one of the following: elevation distribution characteristics, slope distribution characteristics, and aspect distribution characteristics.

[0191] In some embodiments, the processor 202 is further configured such that step S2 specifically includes:

[0192] S2.1 Divide the interpreted loosely packed data into training and testing sets;

[0193] S2.2. Based on the dataset range, convert the image raster and debris flow source vector into sample images and sample labels, respectively. Step S2.2 specifically includes:

[0194] S2.2.1 Determine the pixel size of the sample image based on the size characteristics of the loosely packed volume, and generate the corresponding cropping vector within the dataset vector range.

[0195] S2.2.2. Taking the intersection of the clipping vector and the interpretation vector yields the debris flow source label, i.e., the sample label, for the loose deposit within the image.

[0196] S2.2.3. Use cropping vectors to crop the image raster to obtain sample images;

[0197] S2.3. The average accuracy rate is used to evaluate the performance of the instance segmentation model;

[0198] S2.4. Use the instance segmentation model algorithm to fully learn and train the test set.

[0199] In some embodiments, the processor 202 is further configured such that step S3 specifically includes:

[0200] S3.1 Calculate the source volume of landslide deposits using the digital elevation model method based on high-precision DEM data;

[0201] S3.2 Based on a high-precision DEM model, by selecting elevation information on the collapse boundary, the original terrain interface is fitted, and then the difference between the original DEM model and the original DEM model is calculated to obtain the volume of the collapse deposit source.

[0202] S3.3 Based on a high-precision DEM model, the volume of the channel deposit source is calculated using a triangular cross section;

[0203] S3.4. Based on a high-precision DEM model, the RUSLE model is used to calculate the volume of the slope deposit source. The RUSLE model expression is as follows:

[0204] A = R * K * LS * C * P,

[0205] Where A is the average annual soil loss, R is the rainfall erosivity factor, K is the soil erosibility factor, L is the slope length factor, S is the slope factor, C is the vegetation cover and management factor, and P is the soil and water conservation measures factor.

[0206] In some embodiments, the processor 202 is further configured such that step S3.1 specifically includes:

[0207] S3.1.1 Divide the landslide into the source area and the deposition area;

[0208] S3.1.2. Using topographic data from the landslide source area and the boundary line of the deposition area, the failure surface is fitted using the trend surface method with second-order polynomial fitting, specifically including:

[0209] S3.1.2.1 Generate a series of points with elevation information within the slip source area and on the boundary of the deposition area.

[0210] S3.1.2.2: Using the points with elevation information generated in S3.1.2.1, the landslide failure surface is fitted using the least squares method.

[0211] S3.1.3 Calculate the difference between the original DEM data and the fitted failure surface, and then obtain the size of the landslide volume through raster statistics.

[0212] In some embodiments, the processor 202 is further configured such that step S3.2 specifically includes:

[0213] S3.2.1. Using topographic data along the boundary of the depositional area, the natural neighborhood surface method is used to fit the landslide interface, specifically including:

[0214] S3.2.1.1 Generate a series of points with elevation information on the boundary of the accumulation area.

[0215] S3.2.1.2, Fit the collapsed bottom interface using the elevation information points generated in S3.2.1.1;

[0216] S3.2.2 Calculate the difference between the existing DEM data and the fitted original terrain interface to obtain the size of the source volume of the landslide deposit.

[0217] In some embodiments, the processor 202 is further configured such that step S3.3 specifically includes:

[0218] S3.3.1. Fit a plane connecting both sides of the trench bed by the boundary line of the deposit source in the trench;

[0219] S3.3.2. Subtract the actual topographic line from the plane fitted in S3.3.1 to obtain the scour depth of the channel deposit surface, and take the maximum value of the scour depth of the channel deposit surface at that location as the thickness of the source material of the channel deposit.

[0220] S3.3.3, Using the formula:

[0221]

[0222]

[0223] The source volume of the channel deposit is obtained, where Δz i To fit the difference between the fitted plane and the actual terrain plane, Z i The equivalent depth at a certain point in the channel deposit source, where H is the thickness of the channel deposit source, and V is the equivalent depth at a certain point in the channel deposit source. gd S represents the source volume of the channel deposits, and S represents the area of ​​the DEM raster cell.

[0224] In some embodiments, the processor 202 is further configured such that step S3.4 specifically includes:

[0225] S3.4.1 Obtain precipitation data for the study area, and establish a simplified model using annual rainfall data to estimate the rainfall erosivity factor. The expression for this simplified model is as follows:

[0226]

[0227] Among them, P j R represents the rainfall (mm) in year j. j Let α3 and β3 be the rainfall erosion force in year j, and α3 and β3 be the model parameters.

[0228] S3.4.2. Assess the soil erodibility factor K;

[0229] S3.4.3 Obtain slope and slope length factors through field measurements, or extract slope and slope length factors from DEM data at the watershed scale;

[0230] S3.4.4 Calculate the vegetation cover and management factor C using the Normalized Difference Vegetation Index (NDVI). The relationship between the two is expressed as follows:

[0231]

[0232] S3.4.5. Determine the soil and water conservation measure factor P value by combining field survey data with the method of assigning values ​​to different land use types.

[0233] For specific implementation methods, please refer to the aforementioned method embodiments, which will not be repeated here.

[0234] This application may be a method, apparatus, system, and / or computer program product. A computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of this application.

[0235] Computer-readable storage media can be tangible devices capable of holding and storing instructions for use by an instruction execution device. Computer-readable storage media can be, for example, but not limited to, electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory sticks, floppy disks, mechanical encoding devices, such as punch cards or recessed protrusions storing instructions thereon, and any suitable combination thereof. The computer-readable storage media used herein are not to be construed as transient signals themselves, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., light pulses through fiber optic cables), or electrical signals transmitted through wires.

[0236] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0237] The computer program instructions used to perform the operations of this application may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuits, such as programmable logic circuits, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), are personalized by utilizing state information from the computer-readable program instructions. These electronic circuits can execute the computer-readable program instructions to implement various aspects of this application.

[0238] Various aspects of this application are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer-readable program instructions.

[0239] These computer-readable program instructions can be provided to a processing unit of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that, when executed by the processing unit of the computer or other programmable data processing apparatus, they create means for implementing the functions / actions specified in one or more blocks of the flowchart and / or block diagram. These computer-readable program instructions can also be stored in a computer-readable storage medium that causes a computer, programmable data processing apparatus, and / or other device to operate in a particular manner. Thus, the computer-readable medium storing the instructions comprises an article of manufacture that includes instructions for implementing aspects of the functions / actions specified in one or more blocks of the flowchart and / or block diagram.

[0240] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions that execute on the computer, other programmable data processing apparatus, or other device to perform the functions / actions specified in one or more boxes of a flowchart and / or block diagram.

[0241] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of an instruction containing one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0242] Note that, unless otherwise explicitly stated, all features disclosed in this specification (including any appended claims, abstract, and drawings) may be replaced by alternative features for achieving the same, equivalent, or similar purpose. Therefore, unless explicitly stated otherwise, each disclosed feature is merely one example of a set of equivalent or similar features. Where used, "further," "preferably," "even further," and "more preferably" are simple starting points for describing another embodiment based on the foregoing embodiments, the combination of which with the foregoing embodiments constitutes the complete configuration of another embodiment. Any combination of several "further," "preferably," "even further," or "more preferably" settings following the same embodiment constitutes yet another embodiment.

[0243] Although this application has been described in detail above with general descriptions and specific embodiments, some modifications or improvements can be made to it, which will be obvious to those skilled in the art. Therefore, all such modifications or improvements made without departing from the spirit of this application fall within the scope of protection claimed in this application.

Claims

1. A method for automatic identification of debris flow sources and static storage calculation, characterized in that, Includes the following steps: S1. Obtain relevant geological data of the study area, use remote sensing data to reproduce the real landform landscape in the field, and then determine the type and boundary of loose deposits based on different surface geometric features, texture features and spectral features, establish a database of loose deposits in the study area, and identify their geometric features and distribution features. S2. Automatic identification of loosely packed structures using deep learning models; Step S2 specifically includes: S2.1 Divide the interpreted loosely packed data into training and testing sets; S2.

2. Based on the dataset range, convert the image raster and debris flow source vector into sample images and sample labels, respectively. Step S2.2 specifically includes: S2.2.1 Determine the pixel size of the sample image based on the size characteristics of the loosely packed volume, and generate the corresponding cropping vector within the dataset vector range. S2.2.

2. Taking the intersection of the clipping vector and the interpretation vector yields the debris flow source label, i.e., the sample label, for the loose deposit within the image. S2.2.

3. Use cropping vectors to crop the image raster to obtain sample images; S2.

3. The average accuracy rate is used to evaluate the performance of the instance segmentation model; S2.

4. Use the instance segmentation model algorithm to fully learn and train the test set; S3. Classify the identified loose deposits, and establish calculation models for the source volume of debris flows based on the different deposit morphologies and failure modes of different classifications. Step S3 specifically includes: S3.1 Calculate the source volume of landslide deposits using the digital elevation model method based on high-precision DEM data; S3.2 Based on a high-precision DEM model, by selecting elevation information on the collapse boundary, the original terrain interface is fitted, and then the difference between the original DEM model and the original DEM model is calculated to obtain the volume of the collapse deposit source. S3.3 Based on a high-precision DEM model, the volume of the channel deposit source is calculated using a triangular cross section; S3.

4. Based on a high-precision DEM model, the RUSLE model is used to calculate the volume of the slope deposit source. The RUSLE model expression is as follows: A = R * K * LS * C * P, Where A is the average annual soil loss, R is the rainfall erosivity factor, K is the soil erosibility factor, L is the slope length factor, S is the slope factor, C is the vegetation cover and management factor, and P is the soil and water conservation measures factor. Step S3.3 specifically includes: S3.3.

1. Fit a plane connecting both sides of the trench bed by the boundary line of the deposit source in the trench; S3.3.

2. Subtract the actual topographic line from the plane fitted in S3.3.1 to obtain the scour depth of the channel deposit surface, and take the maximum value of the scour depth of the channel deposit surface at that location as the thickness of the source material of the channel deposit. S3.3.3, Using the formula: ， ， The source volume of the channel deposits was obtained, of which, To find the difference between the fitted plane and the actual terrain plane, The equivalent depth at a certain point in the channel deposit source, where H is the thickness of the channel deposit source. S represents the source volume of the channel deposits, and S represents the area of ​​the DEM raster cell.

2. The method for automatic identification of debris flow sources and static storage calculation according to claim 1, characterized in that, Step S1 specifically includes: S1.1 Obtain relevant geological data for the study area, including at least one of the following: topography, stratigraphy and lithology, geological tectonic evolution history, historical earthquakes, meteorological and hydrogeological conditions, and geological hazards in the study area; S1.2 After acquiring survey area data using airborne LiDAR technology, the generated 3D products are visualized using ArcGIS. S1.

3. Constructing an interpretation environment for loose deposits, classifying, organizing, and comprehensively analyzing existing data, summarizing the LiDAR remote sensing characteristics of loose deposits, and establishing remote sensing identification markers for different types of loose deposits. Step S1.3 specifically includes: S1.3.1 Loose deposits are classified into: landslide deposits, collapse deposits, gully deposits, and slope deposits. S1.3.

2. Based on the characteristics of landslide deposits, collapse deposits, gully deposits, and slope deposits, use ArcGIS software to identify loose deposits on mountain shadow images generated by airborne LiDAR-DEM. S1.4 Based on the identification of different types of loose deposits, summarize the geometric and distribution characteristics of loose deposits in the database. The geometric characteristics include at least one of the following: area characteristics, size characteristics, and side length ratio characteristics. The distribution characteristics include at least one of the following: elevation distribution characteristics, slope distribution characteristics, and aspect distribution characteristics.

3. The method for automatic identification of debris flow sources and static storage calculation according to claim 1, characterized in that, Step S3.1 specifically includes: S3.1.1 Divide the landslide into the source area and the deposition area; S3.1.

2. Using topographic data from the landslide source area and the boundary line of the deposition area, the failure surface is fitted using the trend surface method with second-order polynomial fitting, specifically including: S3.1.2.1 Generate a series of points with elevation information within the slip source area and on the boundary of the deposition area. S3.1.2.2: Using the points with elevation information generated in S3.1.2.1, the landslide failure surface is fitted using the least squares method. S3.1.3 Calculate the difference between the original DEM data and the fitted failure surface, and then obtain the size of the landslide volume through raster statistics.

4. The method for automatic identification of debris flow sources and static storage calculation according to claim 1, characterized in that, Step S3.2 specifically includes: S3.2.

1. Using topographic data along the boundary of the depositional area, the natural neighborhood surface method is used to fit the landslide interface, specifically including: S3.2.1.1 Generate a series of points with elevation information on the boundary of the accumulation area. S3.2.1.2, Fit the collapsed bottom interface using the elevation information points generated in S3.2.1.1; S3.2.2 Calculate the difference between the existing DEM data and the fitted original terrain interface to obtain the size of the source volume of the landslide deposit.

5. The method for automatic identification of debris flow sources and static storage calculation according to claim 1, characterized in that, Step S3.4 specifically includes: S3.4.1 Obtain precipitation data for the study area, and establish a simplified model using annual rainfall data to estimate the rainfall erosivity factor. The expression for this simplified model is as follows: ， Among them, P j R represents the rainfall (mm) in year j. j For the erosive force of rainfall in year j, , These are model parameters; S3.4.

2. Assess the soil erodibility factor K; S3.4.3 Obtain slope and slope length factors through field measurements, or extract slope and slope length factors from DEM data at the watershed scale; S3.4.4 Calculate the vegetation cover and management factor C using the Normalized Difference Vegetation Index (NDVI). The relationship between the two is expressed as follows: ； S3.4.

5. Determine the soil and water conservation measure factor P value by combining field survey data with the method of assigning values ​​to different land use types.

6. A device for automatic identification of debris flow sources and static storage calculation, characterized in that, include: Memory; as well as A processor connected to the memory, the processor being configured to perform the steps of the method as described in any one of claims 1 to 5.

7. A computer storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a machine, it implements the steps of the method as described in any one of claims 1 to 5.

Images