Abnormal deformation ground object positioning method based on ground object classification and deformation data coupling

Abstract

The invention relates to an abnormal deformation ground object positioning method based on ground object classification and deformation data coupling, and belongs to the technical field of remote sensing data processing. The method comprises the following steps of: converting a ground feature classification grid into a vector, extracting various ground feature vector surfaces, and extracting corresponding deformation data by taking the vector surfaces as masks; separating settlement and lifting data by using a grid calculator, counting a mean value and a standard deviation of the settlement and lifting data, and obtaining a global deformation reference value through pixel counting; and finally, comparing the single-type ground feature indexes with the global reference, and if the single-type ground feature indexes and the global reference both exceed the reference, determining that the ground features are abnormal ground features with high deformation rate and non-uniform distribution. According to the method, accurate quantification of ground feature deformation features and efficient identification of abnormal ground features are realized, and reliable technical support is provided for geological disaster early warning, engineering safety monitoring and other scenes.

Description

TECHNICAL FIELD

[0001] The present application relates to an abnormal deformation ground object positioning method based on ground object classification and deformation data coupling, and belongs to the technical field of remote sensing data processing. BACKGROUND

[0002] In the fields of geological disaster early warning, national space planning and engineering safety monitoring, ground surface deformation monitoring is the core link for evaluating the stability of ground objects. The current mainstream deformation monitoring technology (such as InSAR) can efficiently obtain large-scale ground surface deformation grid data, but how to associate deformation data with ground object types and accurately identify "high-risk deformation ground objects" is still a technical difficulty.

[0003] Traditional deformation analysis methods mostly focus on global deformation feature statistics or only carry out independent analysis on a single ground object type (such as building area, farmland), lacking deep coupling of "ground object classification-deformation data": on the one hand, ground object classification data often exist in raster format, and direct superposition with deformation raster is easy to cause analysis error due to mismatch of pixel boundaries; on the other hand, existing technologies mostly rely on a single deformation indicator (such as average deformation rate) to determine the risk, ignoring the uniformity of deformation spatial distribution (such as standard deviation), which is easy to cause "deformation rate is moderate but the distribution is scattered" ground object to be missed. In addition, some methods need to manually select the association between ground objects and deformation data, which is low in efficiency and strong in subjectivity, and is difficult to meet the needs of large-scale ground object deformation anomaly identification.

[0004] At present, the channels for obtaining ground object classification data and deformation data have become mature: AI Earth and other platforms can provide high-precision and standardized ground object classification raster data, and Sentinel-1 satellite combined with SARscape software can generate high-resolution deformation data, but how to integrate the two types of data through a standardized process to realize full-chain analysis is still a problem to be solved in current technology application. SUMMARY

[0005] The purpose of the present application is to overcome the deficiencies of the prior art and provide an abnormal deformation ground object positioning method based on ground object classification and deformation data coupling, which realizes accurate quantification of ground object deformation characteristics and efficient identification of abnormal ground objects, and provides reliable technical support for geological disaster early warning, engineering safety monitoring and other scenarios.

[0006] The technical scheme adopted by the present application is: The abnormal deformation ground object positioning method based on ground object classification and deformation data coupling comprises the following steps: S1. Obtain ground object classification data and ground surface deformation data of a target area; S2. Convert the obtained ground object classification raster data into vector surface data, and extract the vector surface data corresponding to each type of ground object; S3. Taking the vector surface data of each type of ground object as spatial constraints, separate the deformation data corresponding to the ground object from the ground surface deformation data; S4. Based on the deformation data corresponding to each type of ground object, classify and calculate the deformation data of each type of ground object, respectively obtain the subsidence data and uplift data of each type of ground object, and synchronously obtain the subsidence mean value, subsidence standard deviation, uplift mean value and uplift standard deviation of each type of ground object; S5. Based on the subsidence / uplift data corresponding to each type of ground object after classification, calculate the subsidence mean value, subsidence standard deviation, uplift mean value and uplift standard deviation of the global ground object as the reference value; S6. By comparing the subsidence / uplift mean value and standard deviation of a single type of ground object with the global reference value, if the subsidence / uplift mean value of a certain type of ground object is greater than the corresponding global reference mean value, and the subsidence / uplift standard deviation is greater than the corresponding global reference standard deviation, it is determined that the type of ground object has the abnormal characteristics of fast deformation rate and uneven spatial distribution.

[0007] In the above method, step S2 is to load the ground object classification raster data in ArcGIS (geographic information system processing software), use the operation of "conversion tool→export from raster→raster surface conversion", obtain the total ground object vector surface data, select the names of each type of ground object in the attribute table, and export as the vector surface data of each type of ground object.

[0008] Step S3 is to load the ground surface deformation data in ArcGIS, use the function of "spatial analysis tool→extraction analysis→mask extraction", input the raster data as the ground surface deformation data, input the feature mask data as the vector surface of each type of ground object, and output the raster as the deformation data corresponding to each type of ground object.

[0009] Step S4 is to load the deformation data of each type of ground object in ArcGIS, use the function of "spatial analysis tool→map algebra→raster calculator", input the map algebra expression, calculate the deformation uplift of each type of ground object, calculate the deformation subsidence of each type of ground object, and output the result as the raster data of the uplift / subsidence of each type of ground object. Open the attribute table of the deformation uplift / subsidence data of each type of ground object, and record the mean value and standard deviation of the raster pixel.

[0010] Step S5 is to load the subsidence / uplift data of each type of ground object in ArcGIS, use the function of "spatial analysis tool→local analysis→pixel statistical data", input the raster as the deformation data of each type of ground object, superimpose statistics, and select the method as "MEAN (mean)" and "STD (standard deviation)", respectively. The output raster is the mean value and standard deviation of the subsidence / uplift data of the global ground object.

[0011] The beneficial effects of the present application are: The application constructs a standardized anomaly identification system deeply coupled with "feature classification-deformation data", converts the feature classification grid into a single feature vector surface through the "raster to vector-classification merging" process, accurately extracts the corresponding deformation data with a vector mask, solves the boundary matching error problem of traditional raster overlay; adopts a "mean + standard deviation" double index system, which quantifies the deformation rate intensity and reflects the spatial distribution uniformity, improves the accuracy of anomaly judgment; the whole process relies on mainstream software tools such as ArcGIS and SARscape, and the data acquisition standard, operation path and calculation expression are clear, with high standardization and easy landing; through automatic extraction, batch statistics and double-index comparison, it replaces the tedious manual operation, realizes parallel analysis of multiple features, balances efficiency and pertinence, and can be directly connected to existing data channels, solving the pain points of low data coupling, single judgment index and unclear technical process in traditional feature deformation analysis, efficiently and accurately identifying abnormal deformation features, providing reliable basis for industry decision-making and reducing the technical application threshold. BRIEF DESCRIPTION OF DRAWINGS

[0012] Figure 1 The method flowchart of the application; Figure 2 The surface deformation data of a river (left) and the feature classification grid data of a river (right) in an embodiment of the application; Figure 3 The feature classification vector data in an embodiment of the application; Figure 4 The feature vector data of each feature in an embodiment of the application; a. cultivated land, b. forest land, c. grassland, d. bare land, e. artificial surface, f. ice and snow; Figure 5 The deformation data corresponding to each feature in an embodiment of the application; a. cultivated land, b. forest land, c. grassland, d. bare land, e. artificial surface, f. ice and snow; Figure 6 The reference data in an embodiment of the application, a. settlement reference mean, reference standard deviation; b. uplift reference mean, reference standard deviation. DETAILED DESCRIPTION

[0013] The application will be further described below in combination with specific embodiments.

[0014] The abnormal deformation feature positioning method based on feature classification and deformation data coupling in the embodiment includes the following steps: S1. Obtain feature classification data and surface deformation data of a target area: (1) Feature classification data acquisition In the "Data Products" module of the AI Earth platform (website: https: / / engine-aiearth.aliyun.com / ), first select "China 10-meter feature classification dataset"; then input the target area information in the search interface and initiate feature classification data retrieval; after the retrieval is complete, select and download the feature classification data for the desired year.

[0015] (2) Surface deformation data acquisition ① Data download Log in to the European Space Agency Copernicus SciHub platform (https: / / scihub.copernicus.eu / ), define the study area in the platform, set the time range, data type (Sentinel-1 SLC), orbital parameters, etc., select two scenes of data that cover the entire area and meet the interference conditions (same satellite, same polarization, same orbit type), and add them to the download queue to obtain the data (including manifest.safe metadata file).

[0016] ② SARscape InSAR deformation processing Use SARscape software, load "SENTINEL TOPSAR" preset parameters in "Preferences", and select TOPSAR mode suitable for sentinel-1 data. Import the manifest.safe file of the two scenes of data through the "Import Data→SAR Spaceborne→SENTINEL 1" tool to generate intensity maps, SLC index files, etc.; use the vector boundary file (.shp) of the desired area to crop the study area data through the "Sample Selection SAR Geometry Data" tool. Call the "Interferometry→Interferometric Tools→Baseline Estimation" tool, input the SLC index files of the main and slave images, calculate the baseline parameters to verify the suitability of the interference pair. Run the "Interferometry→DInSARDisplacement Workflow" tool to sequentially process interferogram generation, interferogram filtering and coherence calculation, phase unwrapping, control point selection, orbital refinement and re- flattening, and finally output the surface deformation results (deformation raster data in mm / year).

[0017] S2. Convert the obtained feature classification raster data to vector surface data, and extract the vector surface data corresponding to each type of feature: Load the classification grid data in ArcGIS, and use the "conversion tool → out from grid → grid conversion" operation to get the total ground object vector surface data. Select the names of various ground objects in the attribute table and export them as separate vector surface data for each type of ground object.

[0018] S3. Use each type of ground object as a spatial constraint to separate the corresponding ground object deformation data from the ground surface deformation data: Load the ground surface deformation data in ArcGIS, and use the "spatial analysis tool Spatial Analyst Tools → extraction analysis → mask extraction" function to input the grid data as the ground surface deformation data and the feature mask data as the vector surface of each type of ground object. The output grid is the deformation data corresponding to each type of ground object.

[0019] S4. Based on the deformation data corresponding to each type of ground object, classify and calculate the deformation data of each type of ground object to obtain the subsidence data and uplift data of each type of ground object, and simultaneously obtain the subsidence mean, subsidence standard deviation, uplift mean, and uplift standard deviation of each type of ground object: Con(“some type of ground object deformation data”>0, “some type of ground object deformation data”, 0) Equation (1) Con(“some type of ground object deformation data”<0, “some type of ground object deformation data”, 0) Equation (2) Load each type of ground object deformation data in ArcGIS, and use the "Spatial Analyst Tools → Map Algebra → Grid Calculator" function to input the map algebra expression. Calculate the deformation uplift of each type of ground object as equation (1), and calculate the deformation subsidence of each type of ground object as equation (2). The output result is the grid data of the uplift / subsidence of each type of ground object. Open the attribute table of each type of ground object deformation uplift / subsidence data to record the mean and standard deviation of the grid pixels.

[0020] S5. Based on the classified subsidence / uplift data corresponding to each type of ground object, calculate the subsidence mean, subsidence standard deviation, uplift mean, and uplift standard deviation of the global ground object as the reference value: Load each type of ground object subsidence / uplift data in ArcGIS, and use the "Spatial Analyst Tools → Local Analysis → Pixel Statistical Data" function to input the grid as the deformation data of each type of ground object. Superimpose the statistics and select the methods as "MEAN (mean)" and "STD (standard deviation)". The output grid is the mean and standard deviation of the subsidence / uplift data of the global ground object, which serves as the reference value.

[0021] S6. By comparing the mean and standard deviation of subsidence / uplift of a single land feature with the benchmark value of the entire region, if the mean subsidence / uplift of a certain land feature is greater than the corresponding benchmark mean value of the entire region, and its subsidence / uplift standard deviation is greater than the corresponding benchmark standard deviation of the entire region, then it is determined that the land feature has the abnormal characteristics of fast deformation rate and uneven spatial distribution.

[0022] In this embodiment, land cover classification data and surface deformation data are obtained by cropping the vector boundary of a river basin to obtain land cover classification data and surface deformation data for the study area. The land cover classification types include cultivated land, forest land, grassland, bare land, artificial surfaces, and snow / ice. The unit of surface deformation data is mm / year. Figure 2 As shown; the land cover classification raster data is converted into vector polygon data, and the vector polygon data of each type of land cover is extracted, such as... Figure 3 , Figure 4 As shown; based on the obtained vector surface data of various land features, the deformation data corresponding to each land feature is extracted by masking, such as... Figure 5 As shown; the above data sources are attribute sources for settlement and uplift data of various land features, such as... Figure 6 As shown.

Claims

1. A method for locating abnormal deformation ground objects based on coupling of ground object classification and deformation data, characterized in that, The steps include: S1. Obtain the ground object classification data and surface deformation data of the target area; S2. Convert the obtained ground object classification raster data into vector surface data, and extract the vector surface data corresponding to each type of ground object; S3. Spatially constrain each type of ground object vector surface data to separate the deformation data corresponding to the ground object from the surface deformation data; S4. Based on the deformation data corresponding to each type of ground object, classify and calculate the deformation data of each type of ground object to obtain the subsidence data and uplift data of each type of ground object, and simultaneously obtain the subsidence mean, subsidence standard deviation, uplift mean, and uplift standard deviation of each type of ground object; S5. Based on the classified subsidence / uplift data corresponding to each type of ground object, calculate the subsidence mean, subsidence standard deviation, uplift mean, and uplift standard deviation of the global ground object as the reference values; S6. Compare the subsidence / uplift mean and standard deviation of a single type of ground object with the global reference values. If the subsidence / uplift mean of a certain type of ground object is greater than the corresponding global reference mean, and its subsidence / uplift standard deviation is greater than the corresponding global reference standard deviation, it is determined that this type of ground object has abnormal characteristics of fast deformation rate and uneven spatial distribution.

2. The abnormal deformation ground object positioning method based on ground object classification and deformation data coupling according to claim 1, wherein step S2 is loading the ground object classification raster data in ArcGIS, using the "conversion tool→export from raster→raster to face" operation to obtain the total ground object vector surface data, selecting the names of each type of ground object in the attribute table, and exporting the vector surface data of each type of ground object.

3. The method according to claim 1, characterized in that, Step S3 is loading the surface deformation data in ArcGIS, using the "spatial analysis tool→extraction analysis→mask extraction" function, inputting the raster data as the surface deformation data, inputting the vector surface of each type of ground object as the mask data, and outputting the raster data corresponding to each type of ground object.

4. The method according to claim 1, characterized in that, Step S4 is loading each type of ground object deformation data in ArcGIS, using the "spatial analysis tool→map algebra→raster calculator" function, inputting the map algebra expression, calculating the deformation uplift of each type of ground object, calculating the deformation subsidence of each type of ground object, and outputting the raster data of the uplift / subsidence of each type of ground object. Open the attribute table of the deformation uplift / subsidence data of each type of ground object to record the mean and standard deviation of the raster pixels.

5. The abnormal deformation ground object positioning method based on ground object classification and deformation data coupling according to claim 1, wherein step S5 is loading each type of ground object subsidence / uplift data in ArcGIS, using the "spatial analysis tool→local analysis→pixel statistical data" function, inputting the raster data as the deformation data of each type of ground object, superimposing the statistics, selecting the "mean" and "standard deviation" methods, and outputting the raster data of the mean and standard deviation of the subsidence / uplift data of the global ground object.

Images