Method for identifying coal mining subsidence wetland

Abstract

The invention discloses a coal mining subsidence wetland identification method. The method comprises the following steps: acquiring SLC image data of a Sentinel-1 interference width mode of a target area; processing the SLC image data by using an SBAS-InSAR technology to identify surface deformation characteristics of the target area, the surface deformation characteristics including a subsidence rate; acquiring a Landsat series remote sensing image of the target area based on an optical satellite remote sensing technology; a wetland classification system of the target area and an object feature set of the wetland classification system are established according to the Landsat series remote sensing images, and the object feature set comprises spectral features, texture features, geometric features and earth surface deformation features of the target area; and inputting the object feature set into an object-oriented decision tree classification model corresponding to the target area to identify a coal mining subsidence wetland in the target area, the coal mining subsidence wetland being an area of which the earth surface deformation rate is within a preset deformation rate range. The method can realize high-precision and high-efficiency identification of the high-phreatic-water-level coal mining subsidence wetland.

Description

Technical Field

[0001] The present invention relates to the technical field of mining subsidence, and in particular to a method for identifying coal mining subsidence wetlands. Background Art

[0002] The coal-rich cities of the eastern Huanghuai region are a major coal-grain complex in my country. This region boasts abundant coal reserves and high groundwater levels. Subsidence caused by underground mining leads to seasonal or perennial water accumulation, resulting in large-scale coal mining subsidence wetlands. Coal mining subsidence wetlands, a special type of wetland, play an irreplaceable role in water conservation, flood control and drainage, climate regulation, and biodiversity maintenance. These wetlands are characterized by widespread distribution, large area, fragmentation, and staged development. Furthermore, they are subject to long-term dynamic changes due to human disturbances such as urban development and land reclamation, resulting in a decline in the region's ecological diversity and stability. Therefore, to protect and restore coal mining subsidence wetlands, timely and accurate information on their changes is essential, even when data on the mining area is unavailable.

[0003] Remote sensing data processing and analysis are commonly used for wetland extraction. Remote sensing data processing and analysis offers a wealth of information, a wide monitoring range, and rapid updates. It can also leverage features such as spectral characteristics of land features and employ mathematical statistical clustering models for classification. However, due to the continuous formation and expansion of coal mining subsidence wetlands, existing data struggle to provide up-to-date classification information. Furthermore, the spectral characteristics of coal mining subsidence wetlands are similar to those of lake- and pond-type wetlands, making remote sensing interpretation ineffective in distinguishing them from other wetlands, resulting in suboptimal wetland extraction accuracy. Furthermore, existing remote sensing optical identification methods mostly focus on single mining areas or working faces, with few studies examining the dynamic evolution of subsidence wetlands over large areas and over long periods of time.

[0004] Differential Synthetic Aperture Radar Interferometry (DInSAR) and Short Baseline Array-InSAR (SBAS-InSAR) provide new methods for monitoring ground subsidence in mining areas. However, DInSAR is easily affected by factors such as spatiotemporal incoherence and atmospheric delay, resulting in low detection accuracy. SBAS-InSAR is suitable for monitoring slow and small deformations around mining areas. However, due to the rapid changes in surface cover in coal-rich cities in the eastern Huanghuai region, and the fact that coal mining subsidence wetlands are a special type of wetland caused by underground mining, they suffer from severe incoherence and are unable to capture deformation rates. Therefore, it is difficult to extract coal mining subsidence wetlands using this method, especially for large-scale or large-scale coal mining subsidence wetlands. This problem is particularly prominent, making it difficult to accurately extract coal mining subsidence wetlands.

[0005] Based on the analysis of the above related technologies, single-method wetland classification and extraction methods ignore the specific characteristics of wetland ecosystems and fail to meet the urgent needs for accurate identification and planning and management of coal mining subsidence wetlands. Therefore, developing a highly accurate method for identifying coal mining subsidence wetlands is a pressing technical challenge. Accurately identifying and extracting coal mining subsidence wetlands can reveal their temporal and spatial variations and patterns of occurrence and development. This, in turn, provides a scientific basis for the ecological protection and restoration of coal mining subsidence wetlands. Summary of the Invention

[0006] The present invention aims to address, at least to some extent, one of the technical problems in the related art. To this end, a first object of the present invention is to propose a method for identifying coal mining subsidence wetlands. This method utilizes SBAS-InSAR technology to identify coal mining subsidence areas, combines object-oriented decision tree classification to classify wetlands in the study area, and identifies coal mining subsidence wetlands to address the aforementioned problem, achieving high-precision and high-efficiency identification of high-water-level coal mining subsidence wetlands.

[0007] To achieve the above objectives, a first embodiment of the present invention provides a method for identifying coal mining subsidence wetlands, the method comprising:

[0008] S1, acquires SLC image data of the target area in the wide-band mode of Sentinel-1 interferometry;

[0009] S2, using SBAS-InSAR technology to process SLC image data to identify surface deformation characteristics of the target area, where surface deformation characteristics include subsidence rate;

[0010] S3, obtains Landsat series remote sensing images of the target area based on optical satellite remote sensing technology;

[0011] S4, establishing a wetland classification system for the target area and an object feature set of the wetland classification system based on Landsat series remote sensing images, wherein the object feature set includes spectral features, texture features, geometric features, and surface deformation features of the target area;

[0012] S5, inputting the object feature set into an object-oriented decision tree classification model corresponding to the target area to identify coal mining subsidence wetlands in the target area, wherein the coal mining subsidence wetlands are areas where the surface deformation rate is within a preset deformation rate range.

[0013] In addition, the method for identifying coal mining subsidence wetlands according to the above embodiment of the present invention may also have the following additional technical features:

[0014] According to one embodiment of the present invention, step S2 includes:

[0015] S21, preprocessing the SLC image data according to the precise track data, the digital elevation model, and the vector file of the target area corresponding to the SLC image data to obtain an image set of the target area;

[0016] S22, selecting a super master image from the image set, and performing interferometric image pair pairing on the image set according to a spatial baseline threshold and a temporal baseline threshold, so as to generate temporally consecutive interferometric image pair combinations within the image set, wherein each interferometric image pair consists of two SAR images;

[0017] S23, performing interference pattern generation, interference pattern flattening, adaptive filtering, coherence calculation, and phase unwrapping on the interference image pair combination to generate a phase map of the target area;

[0018] S24, performing SBAS inversion processing on the phase map of the target area to obtain the surface deformation characteristics of the target area in the SAR slant range coordinate system. The SBAS inversion includes the first inversion and the second inversion. The first inversion calculates the residual terrain phase and deformation rate through a linear model, removes the residual terrain phase from the interferometric phase, and re-performs phase unwrapping and filtering. The second inversion calculates the displacement in the time series.

[0019] S25, converting the surface deformation characteristics of the target area in the SAR slant range coordinate system into the geographic coordinate system WGS1984 for geocoding processing, and converting the subsidence results in the radar line of sight direction into subsidence results in the direction perpendicular to the ground, and calculating the subsidence rate based on the subsidence results in the direction perpendicular to the ground.

[0020] According to one embodiment of the present invention, step S21 includes: positioning the SLC image data based on the precise orbit data to reduce the orbit error; removing the terrain phase and terrain correction of the SLC image data through the digital elevation model; using the vector file of the target area to crop the SLC image data, setting the VV polarization mode, and obtaining an image set of the target area.

[0021] According to one embodiment of the present invention, step S4 includes:

[0022] S41, preprocessing of Landsat series remote sensing images;

[0023] S42, performing multi-scale segmentation on the pre-processed Landsat series remote sensing images according to preset segmentation conditions to generate a segmented object layer;

[0024] S43, establishing a wetland classification system for the target area based on the segmented object layer, wherein the wetland classification system includes natural wetlands and artificial wetlands, wherein natural wetlands include rivers, lakes, and mudflats; artificial wetlands include reservoirs, ponds, coal mining subsidence wetlands, and urban water surfaces;

[0025] S44, establishing object feature sets corresponding to each subset in natural wetlands and artificial wetlands respectively.

[0026] According to one embodiment of the present invention, step S41 includes: performing radiometric calibration on the Landsat series remote sensing images to eliminate errors in the sensor itself; performing FLAASH atmospheric correction on the Landsat series remote sensing images to eliminate the effects of atmospheric refraction and scattering on the reflection of surface landscape elements; using the WGS84 coordinate system, performing orthorectification on the Landsat series remote sensing images and resampling the remote sensing images into orthophoto images; and using the vector file of the target area to crop the orthorectified Landsat series remote sensing images to remove invalid data and improve processing efficiency.

[0027] According to one embodiment of the present invention, in step S5, an object-oriented decision tree classification model corresponding to the target area is constructed, including: identifying and extracting the wetland distribution of the target area through AWEInsh spectral index extraction; distinguishing vegetation and non-vegetation from wetlands through NDMI and NDVI indices; distinguishing water bodies and ponds from non-vegetation based on brightness and mNDWI index; distinguishing linear water bodies and non-linear water bodies from water bodies through geometric features, wherein the geometric features include aspect ratio and shape index, and linear water bodies are rivers; distinguishing artificial water bodies and natural water bodies from non-linear water bodies through artificial trace features, wherein artificial water bodies include reservoirs, coal mining subsidence wetlands, and urban water surfaces, and natural water bodies include lakes; combined with the surface deformation rate, the area in the artificial water body where the surface deformation rate is within a preset deformation rate range is identified as a coal mining subsidence wetland.

[0028] According to one embodiment of the present invention, step S5 includes: inputting the object feature set into the object-oriented decision tree classification model corresponding to the target area, and the object-oriented decision tree classification model executes the decision tree algorithm to identify the object feature set to identify the coal mining subsidence wetland in the target area.

[0029] Compared with the prior art, the method for identifying coal mining subsidence wetlands of the present invention has the following beneficial effects:

[0030] 1. Combining remote sensing technology and SBAS-InSAR technology to identify coal mining subsidence wetlands, this model fully utilizes remote sensing information acquisition technology and combines optical remote sensing and radar measurement technology to provide a new method for studying large-scale coal mining subsidence wetlands in coal-rich cities;

[0031] 2. This method is easy to implement. It only requires obtaining remote sensing images and radar images of the study area and combining them with the object-oriented decision tree classification method to achieve accurate identification of coal mining subsidence wetlands, providing a good data basis for analyzing the changing patterns of coal mining subsidence wetlands.

[0032] Additional aspects and advantages of the present invention will be set forth in part in the description which follows and, in part, will be obvious from the description which follows, or may be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 Flowchart of a method for identifying coal mining subsidence wetlands according to an embodiment of the present invention;

[0034] Figure 2 FIG. 4 is a schematic diagram of the structure of an object-oriented decision tree classification model according to an embodiment of the present invention. DETAILED DESCRIPTION

[0035] The following describes embodiments of the present invention in detail, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and are not to be construed as limiting the present invention.

[0036] The following describes a method for identifying coal mining subsidence wetlands according to an embodiment of the present invention with reference to the accompanying drawings.

[0037] like Figure 1 As shown, the method for identifying coal mining subsidence wetlands according to the embodiment of the present invention may include the following steps:

[0038] S1, acquire SLC image data of the target area in Sentinel-1 interferometric wide-band mode.

[0039] Specifically, Sentinel-1, a satellite launched by the European Space Agency (ESA), carries a SAR sensor capable of multiple operating modes. Interferometric wide-swath mode is an imaging mode capable of acquiring a wide coverage area. SLC (Single-Look Complex) imagery is a pre-processed SAR data product that retains complete phase and amplitude information. This imagery is fundamental to InSAR processing because it contains phase information used for interferometric measurements, which can reveal subtle changes in the Earth's surface.

[0040] S2, using SBAS-InSAR technology to process the SLC image data to identify the surface deformation characteristics of the target area, where the surface deformation characteristics include the surface deformation rate.

[0041] It's important to understand that SBAS-InSAR is an improved synthetic aperture radar interferometry (SAR) technique. It's primarily used for surface deformation monitoring. Compared to traditional InSAR, SBAS-InSAR effectively mitigates the effects of atmospheric delay and terrain undulation on interferometric measurement accuracy. By processing a time series of SAR images, it can capture surface deformation information.

[0042] According to one embodiment of the present invention, step S2 may include the following steps:

[0043] S21 , preprocessing the SLC image data according to the precise track data, the digital elevation model, and the vector file of the target area corresponding to the SLC image data to obtain an image set of the target area.

[0044] According to one embodiment of the present invention, step S21 includes: positioning the SLC image data based on the precise orbit data to reduce the orbit error; removing the terrain phase and terrain correction of the SLC image data through the digital elevation model; using the vector file of the target area to crop the SLC image data, setting the VV polarization mode, and obtaining an image set of the target area.

[0045] Specifically, the Sentinel-1A SLC imagery is imported into processing software (such as SARPROZ or ISCE). During the import process, the image file path, format, and other information must be specified. After the imagery is imported, the corresponding precise orbit data is applied to the imagery. This step significantly improves the image's positioning accuracy and reduces the impact of orbital errors on subsequent interferometric processing.

[0046] Next, crop the image: Use a vector file that covers the entire target area to crop the imported image. The purpose of cropping is to remove invalid data outside the study area, reduce data volume, and improve processing efficiency. During the cropping process, ensure that the cropped image completely covers the study area. For example, in GIS software, you can use a vector file as a mask to crop the image.

[0047] The Sentinel-1A satellite supports multiple polarization modes, including HH, HV, VH, and VV. VV polarization is typically selected for surface deformation monitoring. This is because VV polarization is more sensitive to the scattering properties of the surface and better reflects surface deformation. In processing software, it is necessary to explicitly specify VV polarization for image processing.

[0048] Through the above steps, an image set of the target area can be obtained.

[0049] S22, selecting a super master image from the image set, and pairing the image set into interferometric image pairs according to a spatial baseline threshold and a temporal baseline threshold, so as to generate temporally consecutive interferometric image pair combinations within the image set, wherein each interferometric image pair consists of two SAR images.

[0050] Specifically, the most important task in the processing process is to generate temporally consecutive connected pairs within the image collection. Each connected pair consists of two SAR images. When N images are involved in constructing the dataset, M connected pairs can be formed. The number M satisfies the following relationship:

[0051]

[0052] The input data were paired with interferometric image pairs, and the system automatically selected the super master image, set the spatial baseline threshold to 6%, and the temporal baseline threshold to 90 days.

[0053] For example, assume there are the following five SAR images, acquired at the following times:

[0054] Image 1: 2023-01-01, Image 2: 2023-02-01, Image 3: 2023-03-01, Image 4: 2023-04-01, Image 5: 2023-05-01; based on the time baseline threshold of 90 days, the following interferometric image pairs can be formed: Image 1 and Image 2, Image 1 and Image 3, Image 2 and Image 3, Image 2 and Image 4, Image 3 and Image 4, Image 3 and Image 5, Image 4 and Image 5.

[0055] The system then further checks the spatial baseline of these pairs to ensure that it does not exceed 6%. The resulting connectivity map will only contain pairs that meet these two conditions.

[0056] S23, performing interference pattern generation, interference pattern flattening, adaptive filtering, coherence calculation, and phase unwrapping on the interference image pair combination to generate a phase map of the target area.

[0057] Specifically, the number of multi-views is set according to the satellite incidence angle and the original sampling interval, and the multi-view ratio is set to 4:1, so that the resolution in the azimuth direction and the range direction are consistent, reducing the impact of speckle noise on the image; to enhance the filtering effect, the Goldstein filtering method is adopted, 0.15 is set as the minimum threshold during unwrapping, and the small cost flow method is used for phase unwrapping. In order to further improve the accuracy of the unwrapping results, it is necessary to remove images with low coherence, and remove interference pairs with poor coherence or jumps in the unwrapped phase, and recalculate.

[0058] It's important to understand that to reduce speckle noise and improve interferogram quality, SLC images are typically subjected to multi-look processing. Multi-look processing resamples the original image in azimuth and range, reducing image resolution while also reducing noise. The multi-look ratio is set based on the satellite's angle of incidence and the original sampling interval. For example, a 4:1 multi-look ratio means resampling by a factor of 4 in azimuth and a factor of 1 in range. This ensures consistent resolution in azimuth and range, reducing the impact of speckle noise on the image. Paired SLC images are interferometrically processed to generate an interferogram. The interferogram contains information about the phase difference between the two images, which primarily reflects surface deformation and topographic relief.

[0059] Deflating is the process of removing the phase variations in the interferogram caused by terrain undulations to more accurately reflect surface deformation. This step is usually done using a Digital Elevation Model (DEM) to calculate the terrain phase and subtract this phase from the interferogram.

[0060] To enhance filtering effectiveness, the Goldstein filter method is used. Goldstein filtering is an adaptive filtering method that adjusts filter strength based on local coherence, thereby preserving surface deformation information while reducing noise. Goldstein filter parameters, such as the filter window size, are adjusted based on the characteristics and requirements of the interferogram to achieve optimal filtering results.

[0061] Coherence is an important indicator of interferogram quality, reflecting the similarity between two images. Regions with high coherence indicate more reliable phase information in the interferogram, while regions with low coherence may exhibit noise or excessive distortion. Coherence maps are typically calculated by calculating the complex correlation coefficient of the interferogram.

[0062] Phase unwrapping involves expanding the phase in the interferogram from the range of -π to π to the entire real number range, thereby obtaining the true deformation phase. This is a key step in SBAS-InSAR processing. During the unwrapping process, a minimum coherence threshold is set to 0.15. This means that only regions with a coherence greater than 0.15 will undergo phase unwrapping, avoiding erroneous unwrapping of low-coherence regions. The minimum cost flow method (MCF) is used for phase unwrapping. The MCF method is a graph-theory-based unwrapping method that can effectively handle complex phase unwrapping problems and improve the accuracy of the unwrapping results.

[0063] The above steps complete the interferometric workflow for SBAS-InSAR processing. These steps ensure the quality of the interferogram and the accuracy of the unwrapping results, providing a reliable data foundation for surface deformation monitoring. In practice, the parameters in each step must be flexibly adjusted according to the specific study area and data characteristics to achieve optimal processing results.

[0064] S24, SBAS inversion processing is performed on the phase map of the target area to obtain the surface deformation characteristics of the target area in the SAR slant range coordinate system. The SBAS inversion includes the first inversion and the second inversion. The first inversion calculates the residual terrain phase and deformation rate through a linear model, removes the residual terrain phase from the interferometric phase, and re-performs phase unwrapping and filtering. The second inversion calculates the displacement in the time series.

[0065] Specifically, in the first inversion, a linear model is used to calculate the residual terrain phase and deformation rate. The residual terrain phase is then removed from the interferometric phase, and phase unwrapping and filtering are performed again. The second inversion aims to calculate the displacement over the time series. This inversion employs atmospheric high-pass filtering to estimate and remove the influence of the atmospheric phase on the results, thereby obtaining a more accurate true surface deformation phase.

[0066] S25, converting the surface deformation characteristics of the target area in the SAR slant range coordinate system into the geographic coordinate system WGS1984 for geocoding processing, and converting the subsidence results in the radar line of sight direction into subsidence results in the direction perpendicular to the ground, and calculating the subsidence rate based on the subsidence results in the direction perpendicular to the ground.

[0067] Specifically, the SAR image's slant range coordinate system is converted to a geographic coordinate system (such as WGS1984) for integration with geographic information systems (GIS) and other geographic data. The deformation results in the radar's slant range line of sight are converted to deformation results perpendicular to the ground to facilitate subsequent surface deformation analysis.

[0068] Specifically, the deformation results obtained after processing SAR image data are usually expressed in a slant range coordinate system. Then, the slant range coordinates are converted into geographic coordinates using DEM and orbit data. The following steps are usually involved: terrain correction: mapping the slant range coordinates to the geographic coordinate system based on the DEM data; geometric correction: adjusting the geometric distortion of the image to ensure that the converted image correctly represents the surface position in the geographic coordinate system. The deformation results in the direction of the radar line of sight are converted into deformation results in the direction perpendicular to the ground. Geocoding is an important step in converting SAR images from a slant range coordinate system to a geographic coordinate system, which enables the deformation results to be more intuitively used for geographic analysis and integration with other geographic data.

[0069] S3, obtains Landsat series remote sensing images of the target area based on optical satellite remote sensing technology.

[0070] Specifically, the Landsat series of satellites is a series of Earth observation satellites jointly operated by the National Aeronautics and Space Administration (NASA) and the United States Geological Survey (USGS). Since the launch of Landsat 1 in 1972, several generations of satellites have been launched, including Landsat 2, 3, 4, 5, 7, and 8. These satellites carry a variety of sensors capable of acquiring optical images of the Earth's surface. These sensors primarily include the Multispectral Scanner (MSS), the Thematic Mapper (TM), the Enhanced Thematic Mapper (ETM+), and the Operational Land Imager (OLI). For example, the OLI sensor aboard the Landsat 8 satellite has nine bands, ranging in wavelength from visible light to shortwave infrared, providing rich spectral information on land features with a spatial resolution of 30 meters (for most bands), enabling it to clearly distinguish between different land feature types such as farmland, forests, and cities. Landsat series remote sensing images of the target area can be obtained through relevant channels.

[0071] S4. Establish a wetland classification system and an object feature set of the wetland classification system for the target area based on the Landsat series remote sensing images. The object feature set includes spectral features, texture features, geometric features, and surface deformation features of the target area.

[0072] According to one embodiment of the present invention, step S4 includes:

[0073] S41, preprocessing of Landsat series remote sensing images.

[0074] According to one embodiment of the present invention, step S41 includes: performing radiometric calibration on the Landsat series remote sensing images to eliminate errors in the sensor itself; performing FLAASH atmospheric correction on the Landsat series remote sensing images to eliminate the effects of atmospheric refraction and scattering on the reflection of surface landscape elements; using the WGS84 coordinate system, performing orthorectification on the Landsat series remote sensing images and resampling the remote sensing images into orthophoto images; and using the vector file of the target area to crop the orthorectified Landsat series remote sensing images to remove invalid data and improve processing efficiency.

[0075] S42, performing multi-scale segmentation on the pre-processed Landsat series remote sensing images according to preset segmentation conditions to generate a segmented object layer.

[0076] S43, establishing a wetland classification system for the target area based on the segmented object layer, wherein the wetland classification system includes natural wetlands and artificial wetlands, wherein natural wetlands include: rivers, lakes, and mudflats; artificial wetlands include: reservoirs, ponds, coal mining subsidence wetlands, and urban water surfaces.

[0077] S44, establishing object feature sets corresponding to each subset in natural wetlands and artificial wetlands respectively.

[0078] S5: Input the object feature set into an object-oriented decision tree classification model corresponding to the target area to identify coal mining subsidence wetlands in the target area. Coal mining subsidence wetlands are areas where the surface deformation rate is within a preset deformation rate range. The preset deformation rate range can be calibrated based on actual conditions and can be between 5 mm / year and -20 mm / year.

[0079] According to one embodiment of the present invention, in step S5, an object-oriented decision tree classification model corresponding to the target area is constructed, including: identifying and extracting the wetland distribution of the target area through AWEInsh spectral index extraction; distinguishing vegetation and non-vegetation from wetlands through NDMI and NDVI indices; distinguishing water bodies and ponds from non-vegetation based on brightness and mNDWI index; distinguishing linear water bodies and non-linear water bodies from water bodies through geometric features, wherein the geometric features include aspect ratio and shape index, and linear water bodies are rivers; distinguishing artificial water bodies and natural water bodies from non-linear water bodies through artificial trace features, wherein artificial water bodies include reservoirs, coal mining subsidence wetlands, and urban water surfaces, and natural water bodies include lakes; combining the surface deformation rate, identifying the area in the artificial water body where the surface deformation rate is within the preset deformation rate range as a coal mining subsidence wetland. Among them, the object-oriented decision tree classification model is as follows: Figure 2 shown.

[0080] According to one embodiment of the present invention, step S5 includes: inputting the object feature set into the object-oriented decision tree classification model corresponding to the target area, and the object-oriented decision tree classification model executes the decision tree algorithm to identify the object feature set to identify the coal mining subsidence wetland in the target area.

[0081] Throughout this specification, reference to terms such as "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" means that a specific feature, structure, material, or characteristic described in conjunction with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations 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 any one or more embodiments or examples.

[0082] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present invention, "plurality" means at least two, such as two, three, etc., unless otherwise specifically defined.

[0083] In the present invention, unless otherwise specified or limited, the terms "installed," "connected," "connect," "fixed," etc. should be understood in a broad sense. For example, they can refer to fixed connection, detachable connection, or integration; mechanical connection, electrical connection; direct connection, or indirect connection through an intermediate medium; internal communication between two components, or interaction between two components, unless otherwise specified. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0084] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A method for identifying coal mining subsidence wetlands, characterized in that: The method comprises: S1, acquires SLC image data of the target area in the wide-band mode of Sentinel-1 interferometry; S2, processing the SLC image data using SBAS-InSAR technology to identify surface deformation characteristics of the target area, wherein the surface deformation characteristics include subsidence rate; S3, acquiring a Landsat series of remote sensing images of the target area based on optical satellite remote sensing technology; S4, establishing a wetland classification system for the target area and an object feature set of the wetland classification system based on Landsat series remote sensing images, wherein the object feature set includes spectral features, texture features, geometric features, and surface deformation features of the target area; S5, inputting the object feature set into the object-oriented decision tree classification model corresponding to the target area to identify the coal mining subsidence wetland in the target area, wherein the coal mining subsidence wetland is an area where the surface deformation rate is within a preset deformation rate range.

2. The method for identifying coal mining subsidence wetlands according to claim 1, characterized in that: Step S2 includes: S21, preprocessing the SLC image data according to the precise track data, the digital elevation model, and the vector file of the target area corresponding to the SLC image data to obtain an image set of the target area; S22, selecting a super master image from the image set, and performing interferometric image pair pairing on the image set according to a spatial baseline threshold and a temporal baseline threshold, so as to generate temporally consecutive interferometric image pair combinations within the image set, wherein each interferometric image pair consists of two SAR images; S23, performing interference pattern generation, interference pattern flattening, adaptive filtering, coherence calculation, and phase unwrapping on the interference image pair combination to generate a phase map of the target area; S24, performing SBAS inversion processing on the phase map of the target area to obtain the surface deformation characteristics of the target area in the SAR slant range coordinate system, wherein the SBAS inversion includes a first inversion and a second inversion, wherein the first inversion calculates the residual terrain phase and deformation rate through a linear model, removes the residual terrain phase from the interferometric phase, and re-performs phase unwrapping and filtering processing; the second inversion calculates the displacement in the time series; S25, converting the surface deformation characteristics of the target area in the SAR slant range coordinate system into the geographic coordinate system WGS1984 for geocoding processing, and converting the subsidence results in the radar line of sight direction into subsidence results in the direction perpendicular to the ground, and calculating the subsidence rate based on the subsidence results in the direction perpendicular to the ground.

3. The method for identifying coal mining subsidence wetlands according to claim 2, characterized in that: Step S21 includes: Positioning the SLC image data according to the precise orbit data to reduce orbit error; removing terrain phase and performing terrain correction on the SLC image data using a digital elevation model; The SLC image data is cropped using the vector file of the target area, and a VV polarization mode is set to obtain an image set of the target area.

4. The method for identifying coal mining subsidence wetlands according to claim 1, characterized in that: Step S4 includes: S41, preprocessing of Landsat series remote sensing images; S42, performing multi-scale segmentation on the pre-processed Landsat series remote sensing images according to preset segmentation conditions to generate a segmented object layer; S43, establishing a wetland classification system for the target area based on the segmented object layer, wherein the wetland classification system includes natural wetlands and artificial wetlands, wherein natural wetlands include rivers, lakes, and mudflats; and artificial wetlands include reservoirs, ponds, coal mining subsidence wetlands, and urban water surfaces; S44, establishing object feature sets corresponding to each subset in natural wetlands and artificial wetlands respectively.

5. The method for identifying coal mining subsidence wetlands according to claim 4, characterized in that: Step S41 includes: Radiometric calibration of Landsat series remote sensing images to eliminate sensor errors; Perform FLAASH atmospheric correction on Landsat remote sensing images to eliminate the effects of atmospheric refraction and scattering on the reflection of surface landscape elements; Use the WGS84 coordinate system to perform orthorectification on the Landsat series remote sensing images and resample the remote sensing images into orthophoto images; The orthorectified Landsat series remote sensing images are cropped using the vector file of the target area to remove invalid data and improve processing efficiency.

6. The method for identifying coal mining subsidence wetlands according to claim 4, characterized in that: In step S5, an object-oriented decision tree classification model corresponding to the target area is constructed, including: Identify and extract the wetland distribution in the target area through AWEInsh spectral index extraction; Distinguish vegetation and non-vegetation from wetlands using NDMI and NDVI indices; Distinguish water bodies and ponds from non-vegetated areas based on brightness and mNDWI index; Distinguishing linear water bodies from nonlinear water bodies by geometric features, wherein the geometric features include aspect ratio and shape index, and linear water bodies are rivers; Distinguish artificial water bodies from natural water bodies by using artifact features from nonlinear water bodies. Artificial water bodies include reservoirs, coal mining subsidence wetlands, and urban water surfaces, while natural water bodies include lakes. Combined with the surface deformation rate, areas in artificial water bodies where the surface deformation rate is within the preset deformation rate range are identified as coal mining subsidence wetlands.

7. The method for identifying coal mining subsidence wetlands according to claim 6, characterized in that: Step S5 includes: The object feature set is input into an object-oriented decision tree classification model corresponding to the target area, and the object-oriented decision tree classification model executes a decision tree algorithm to identify the object feature set to identify the coal mining subsidence wetland in the target area.

Images