Multi-modal remote sensing image land surface stable water body extraction method integrating priori knowledge

Abstract

The invention provides a multi-modal remote sensing image land surface stable water body extraction method integrated with prior knowledge, and the method comprises the steps: obtaining Sentinel-1 SAR remote sensing data and Sentinel-2 optical remote sensing data, and carrying out the preprocessing of the Sentinel-1 SAR remote sensing data and Sentinel-2 optical remote sensing data, so as to obtain a monthly time sequence SAR remote sensing image covering the whole domain, and a monthly median optical remote sensing image covering the whole domain; calculating an SDWI water body index and removing an abnormal value; acquiring a plurality of stable water body samples in the whole domain; extracting an SDWI pixel value of a stable water body sample coverage area, and determining an SDWI water body segmentation threshold value; calculating monthly water body frequency based on an SDWI water body segmentation threshold value; calculating and determining a monthly initial stable water body based on the monthly water body frequency; calculating an optimization index based on the monthly median optical remote sensing image and removing a non-water body part in the monthly initial stable water body; calculating an MNDWI water body index based on the monthly median remote sensing image; and merging and integrating the MNDWI water body index and the monthly initial stable water body without the non-water body part, and supplementing the water body leakage part to obtain the final monthly stable water body.

Description

Technical Field

[0001] The present application relates to the technical field of water body identification, and in particular to a method for extracting stable water bodies from land surfaces using multimodal remote sensing images that integrates prior knowledge. Background Art

[0002] Water is an important natural resource that is indispensable for human production and life and the sustainable development of nature. Among them, land surface water, as an important freshwater resource, often exists in the form of natural or artificial rivers, lakes, reservoirs and ponds. The spatiotemporal distribution information of land surface water bodies is helpful for the dynamic management of water resources, water ecological governance and restoration, and emergency monitoring of mountain torrent disasters. It has important scientific value in the fields of climate change, hydrological simulation, and water environment governance. Therefore, it is of great significance to efficiently and accurately extract wide-area land surface water bodies.

[0003] The rapid development of multi-source satellite remote sensing has laid a solid foundation for the acquisition of land surface water information. The method based on remote sensing water index is widely used in water system mapping. It is easy to calculate and uses the spectral characteristics of water body reflection, absorption and scattering of solar radiation to construct an index through band combination. Combined with threshold segmentation, it can quickly obtain small area surface water information.

[0004] However, when using the remote sensing water index method to obtain land surface water information on a wide scale, the land surface water information obtained is inaccurate. Summary of the invention

[0005] The embodiment of the present application provides a method for extracting stable water bodies on the land surface from multimodal remote sensing images that integrates prior knowledge to accurately obtain stable water body information on the land surface over a wide area.

[0006] In a first aspect, the present application provides a method for extracting stable water bodies from land surfaces using multimodal remote sensing images that integrates prior knowledge, including:

[0007] Acquire Sentinel-1 SAR remote sensing data and Sentinel-2 optical remote sensing data;

[0008] Preprocessing the Sentinel-1SAR remote sensing data to obtain monthly time-series SAR remote sensing images covering the entire region;

[0009] Preprocessing the Sentinel-2 optical remote sensing data to obtain a monthly median optical remote sensing image covering the entire region;

[0010] Calculate the SDWI water index based on the monthly time-series SAR remote sensing images and remove outliers;

[0011] Using GSW water products as prior information, global stable water samples are extracted;

[0012] Based on the SDWI water body index and the stable water body samples, the SDWI pixel values ​​of the area covered by the stable water body samples are extracted, and the SDWI water body segmentation threshold is determined;

[0013] Monthly water body frequency calculation based on SDWI water body segmentation threshold calculation;

[0014] Calculate and determine the monthly initial stable water body based on the monthly water body frequency;

[0015] The optimization index is calculated based on the monthly median optical remote sensing image and the non-water part of the monthly initial stable water body is removed;

[0016] Based on the monthly median remote sensing image, the MNDWI water index is calculated;

[0017] The MNDWI water index is integrated with the monthly initial stable water body after removing the non-water body part, and the missed water body part is supplemented to obtain the final monthly stable water body.

[0018] In a feasible implementation, the preprocessing of the Sentinel-1SAR remote sensing data to obtain monthly time-series SAR remote sensing images covering the entire region includes:

[0019] The Sentinel-1SAR remote sensing data is subjected to terrain correction, Refined Lee filtering, splicing and cropping to form monthly time-series SAR images covering the entire region.

[0020] In a feasible implementation, the monthly time-series SAR image is a ground-range multi-view image in an interferometric wide-band mode (, the polarization modes are VV polarization and VH polarization, and the image spatial resolution is 10 m.

[0021] In a feasible implementation, the preprocessing of the Sentinel-2 optical remote sensing data to obtain a monthly median optical remote sensing image covering the entire region includes:

[0022] The Sentinel-2 optical remote sensing data is processed in the cloud using the QA60 band, and a monthly median optical remote sensing image covering the entire area is formed through stitching, monthly median synthesis and cropping.

[0023] In a feasible implementation, the calculation formula of the SDWI water body segmentation threshold is:

[0024] SDWI=ln(10·VV·VH)-8;

[0025] Among them, SDWI represents the SDWI water body segmentation threshold, VV represents the VV polarization band, and VH represents the VH polarization band.

[0026] In a feasible implementation, the calculation formula of the SDWI water body segmentation threshold is:

[0027] T=M SDWI -2S SDWI ;

[0028] Where, T represents the SDWI water body segmentation threshold, M SDWI represents the mean value of the SDWI water index in the corresponding area of ​​the stable water sample point, S SDWI Represents the standard deviation of the SDWI water index in the corresponding area of ​​the stable water sample point.

[0029] In a feasible implementation, the monthly water body frequency calculation based on the SDWI water body segmentation threshold includes:

[0030] When the SDWI water index of a pixel is greater than the SDWI water segmentation threshold, the pixel is marked as a water body. Based on this, the regional monthly time series water body mask is obtained in turn. On this basis, the monthly water body frequency value is calculated, that is, the frequency of each pixel being identified as a water body in the monthly time series. The specific calculation formula is:

[0031]

[0032] Where F represents the frequency of pixels being identified as water bodies in the monthly time series; N water Indicates the number of times the pixel is identified as a water body, N valid Represents the valid observation number of the pixel.

[0033] In a feasible implementation, pixels with a monthly water body frequency greater than 0.3 are determined as monthly initial stable water bodies;

[0034] Based on the monthly median optical remote sensing image, the optimization index is calculated and the non-water body part of the monthly initial stable water body is removed. When the optimization index value is greater than 0, it indicates a non-water body. It is used to remove pixels in the low scattering area of ​​the monthly initial stable water body that are mistakenly classified as water bodies.

[0035] The optimization index calculation formula is:

[0036]

[0037] Where OI represents the optimization index result, ρ SWIR1 represents the short-wave infrared 1 band reflectivity, ρ SWIR2 represents the shortwave infrared 2 band reflectivity, ρ R represents the reflectivity of red light band, ρ G Represents the reflectivity of the green light band.

[0038] In a feasible implementation method, the MNDWI water index is calculated based on the monthly median remote sensing image. When the MNDWI water index value is greater than 0, it is a water body.

[0039] The calculation formula of MNDWI water index is:

[0040]

[0041] Where, MNDWI represents the MNDWI water index, ρ G represents the reflectivity of green light band, ρ SWIR1 Indicates the shortwave infrared 1 band reflectivity.

[0042] The embodiment of the present application provides a method for extracting stable water bodies from land surfaces using multimodal remote sensing images that integrate prior knowledge, including: acquiring Sentinel-1SAR remote sensing data and Sentinel-2 optical remote sensing data; preprocessing the Sentinel-1SAR remote sensing data to obtain monthly time-series SAR remote sensing images covering the entire domain; preprocessing the Sentinel-2 optical remote sensing data to obtain monthly median optical remote sensing images covering the entire domain; calculating the SDWI water body index based on the monthly time-series SAR remote sensing images and removing outliers; acquiring multiple stable water body samples from the entire domain; and SDWI water body index and stable water body samples, extract the SDWI pixel values ​​of the area covered by the stable water body samples, and determine the SDWI water body segmentation threshold; calculate the monthly water body frequency based on the SDWI water body segmentation threshold; determine the monthly initial stable water body based on the monthly water body frequency calculation; calculate the optimization index based on the monthly median optical remote sensing image and remove the non-water body part of the monthly initial stable water body; calculate the MNDWI water body index based on the monthly median remote sensing image; merge and integrate the MNDWI water body index with the monthly initial stable water body with the non-water body part removed, and make up for the missed water body part to obtain the final monthly stable water body.

[0043] This method combines active remote sensing and passive remote sensing, and can overcome the problems of interference from cloudy and rainy weather and insufficient effective ground observations, and greatly weakens the phenomena of imperfect water body edges and small patches, omission of urban water bodies, and misclassification of water bodies in low-scattering areas in the results; secondly, this method uses the sample points of existing remote sensing water body products as prior information to automatically and quickly obtain the remote sensing water body index segmentation threshold, reducing the uncertainty of threshold segmentation; in addition, this method also comprehensively applies multi-source remote sensing water body indices, and improves the robustness of land surface water body information extraction through time series frequency judgment, partition integration, and result optimization integration. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The drawings described herein are used to provide further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present application and do not constitute improper limitations on the present invention.

[0045] In the attached picture:

[0046] Figure 1 It is a flow chart of a method for extracting stable water bodies from land surfaces using multimodal remote sensing images with integrated prior knowledge provided in one embodiment of the present application;

[0047] Figure 2 It is a schematic diagram of the monthly stable water extraction results of the demonstration area provided in one embodiment of the present application;

[0048] Figure 3 It is a partial schematic diagram of the water extraction results of the cultivation area in the demonstration area provided by an embodiment of the present application;

[0049] Figure 4 It is a partial schematic diagram of the water body extraction results of the built-up area of ​​the demonstration area provided in one embodiment of the present application;

[0050] Figure 5 It is a partial schematic diagram of the water body extraction results of the lake area in the demonstration area provided by one embodiment of the present application. DETAILED DESCRIPTION

[0051] In order to enable those skilled in the art to better understand the technical solutions in the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work should fall within the scope of protection of the present application.

[0052] In the description of the embodiments of the present application, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present application, "plurality" means at least two, for example, two, three, etc., unless otherwise clearly and specifically defined.

[0053] In this application, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements, unless otherwise clearly defined. For ordinary technicians in this field, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0054] In the present application, unless otherwise clearly specified and limited, a first feature being "above" or "below" a second feature may mean that the first and second features are in direct contact, or the first and second features are in indirect contact through an intermediate medium. Moreover, a first feature being "above", "above" or "above" a second feature may mean that the first feature is directly above or obliquely above the second feature, or simply means that the first feature is higher in level than the second feature. A first feature being "below", "below" or "below" a second feature may mean that the first feature is directly below or obliquely below the second feature, or simply means that the first feature is lower in level than the second feature.

[0055] Water is an important natural resource that is indispensable for human production and life and the sustainable development of nature. Among them, land surface water, as an important freshwater resource, often exists in the form of natural or artificial rivers, lakes, reservoirs and ponds. The spatiotemporal distribution information of land surface water bodies is helpful for the dynamic management of water resources, water ecological governance and restoration, and emergency monitoring of mountain torrent disasters. It has important scientific value in the fields of climate change, hydrological simulation, and water environment governance. Therefore, it is of great significance to efficiently and accurately extract wide-area land surface water bodies.

[0056] The rapid development of multi-source satellite remote sensing has laid a solid foundation for the acquisition of land surface water information. The method based on remote sensing water index is widely used in water system mapping. It is easy to calculate and uses the spectral characteristics of water body reflection, absorption and scattering of solar radiation to construct an index through band combination. Combined with threshold segmentation, it can quickly obtain small area surface water information.

[0057] However, the method of using remote sensing water index to obtain land surface water information on a wide scale is inaccurate. When the monitoring area becomes larger, the climate conditions are relatively complex and easily disturbed by cloudy and rainy weather, especially in the rainy season in the south, where weather conditions change frequently in a short period of time. The optical remote sensing imaging conditions and revisit cycle seriously restrict the number of effective ground observations, resulting in large-scale water extraction based on optical remote sensing being difficult to meet the needs of current monitoring, and the water body edge and small patch information obtained when the scene is complex is incomplete. Synthetic aperture radar (SAR) satellite images have all-day and all-weather observation capabilities and are free from the interference of cloudy and rainy weather, but due to the strong echo diffusion effect of buildings, it is easy to cause water bodies in built-up areas to be missed, and there is a phenomenon of water body misclassification in low-scattering areas such as smooth surfaces and bare soil. Large-scale regional geographical scenes are complex, landscape types are diverse, water body morphology and water quality environment are changeable, and the temporal and spatial information presented in the image is quite different. The water body index of a single source or a single type is challenged in terms of versatility and universality, and the uncertainty of threshold segmentation in terms of effectiveness increases.

[0058] In order to solve the above problems, an embodiment of the present application provides a method and system for extracting stable water bodies from land surfaces in multimodal remote sensing images that integrates prior knowledge. The solution provided by the embodiment of the present application will be described in detail below in conjunction with the accompanying drawings in the specification.

[0059] Figure 1 It is a flowchart of a method for extracting stable water bodies from land surfaces using multimodal remote sensing images that integrates prior knowledge, provided in one embodiment of the present application.

[0060] Reference Figure 1 As shown, the embodiment of the present application provides a method for extracting stable water bodies from land surfaces using multimodal remote sensing images that integrates prior knowledge, including:

[0061] S100: Acquire Sentinel-1 SAR remote sensing data and Sentinel-2 optical remote sensing data.

[0062] Specifically, according to the entire regional scope, Sentinel-1 SAR remote sensing data and Sentinel-2 optical remote sensing data of the designated area are obtained. Sentinel-1 SAR remote sensing data are SAR remote sensing data collected by the Sentinel-1 satellite, and Sentinel-2 optical remote sensing data are optical remote sensing data collected by the Sentinel-2 satellite. It should be noted that Sentinel-1 and Sentinel-2 are two sentinel satellites. Sentinel-1 is an earth observation satellite series under the Copernicus program of the European Space Agency (ESA). The main task of this satellite series is to provide all-weather and all-day radar imaging services for land and ocean observations. The Sentinel-1 satellite is equipped with a C-band synthetic aperture radar (SAR), which can penetrate clouds and darkness to obtain high-resolution radar images; Sentinel-2 is a high-resolution multispectral imaging satellite series under the Copernicus program of the European Space Agency. The main task of this satellite series is land monitoring, providing images of vegetation, soil and water cover, inland waterways and coastal areas, etc., and can also be used for emergency rescue services. The Sentinel-2 satellite is equipped with a multispectral imager (MSI) that can provide image data in 13 spectral bands, covering the visible, near-infrared and short-wave infrared regions, with spatial resolutions of 10 meters, 20 meters and 60 meters respectively, all of which are known to those skilled in the art.

[0063] S200: Preprocess the Sentinel-1SAR remote sensing data to obtain monthly time-series SAR remote sensing images covering the entire area.

[0064] Specifically, the Sentinel-1SAR remote sensing data is terrain corrected, Refined Lee filtered, spliced ​​and cropped to form monthly time-series SAR images covering the entire area. The monthly time-series SAR images are multi-view images of ground distance in the interferometric wide-band mode, with VV polarization and VH polarization, and the image spatial resolution is 10m. Refined Lee filtering is an adaptive filtering algorithm for speckle noise in synthetic aperture radar (SAR) images. It is based on the improvement of the classic Lee filter and can better retain edge and detail information while suppressing noise.

[0065] S300: Preprocess the Sentinel-2 optical remote sensing data to obtain the monthly median optical remote sensing image covering the entire domain.

[0066] Specifically, the QA60 band is used to perform cloud processing on the Sentinel-2 optical remote sensing data, and through stitching, monthly median synthesis and cropping, a monthly median optical remote sensing image covering the entire area is formed.

[0067] The optical remote sensing image is an L2A surface reflectance product. The image used consists of 10 spectral bands, including blue band, green band, red band, vegetation red edge 1 band, vegetation red edge 2 band, vegetation red edge 3 band, near infrared band, narrow edge near infrared band, shortwave infrared 1 band and shortwave infrared 2 band. Among them, the spatial resolution of the blue band, green band, red band and near infrared band is 10m; the spatial resolution of vegetation red edge 1 band, vegetation red edge 2 band, vegetation red edge 3 band, narrow edge near infrared band, shortwave infrared 1 band and shortwave infrared 2 band is 20m. The spatial resolution of the bands with a spatial resolution of 20m needs to be resampled to 10m through nearest neighbor interpolation.

[0068] S400: Calculate the SDWI water index based on the monthly time-series SAR remote sensing images and remove outliers.

[0069] The SDWI water index is calculated based on the monthly time-series SAR remote sensing images, and the abnormally large and small values ​​in the calculation results are set to 0. The calculation formula of the SDWI water index is:

[0070] SDWI=ln(10·VV·VH)-8

[0071] Among them, SDWI represents SDWI water index, VV represents VV polarization band, and VH represents VH polarization band.

[0072] S500: Using GSW water products as prior information, extract global stable water samples.

[0073] Based on the GSW (Global Surface Water) remote sensing water product, the permanent water body information is obtained, and several permanent water body sample points are randomly selected within the coverage area as stable water body samples using stratified random sampling. The GSW is the global water body time series coverage obtained by the Joint Research Center of the European Commission under the Copernicus program using Landsat data. This data product provides information on the spatial distribution, changes, seasonality and maximum range of water bodies.

[0074] S600: Based on the SDWI water body index and the stable water body samples, the SDWI pixel values ​​of the area covered by the stable water body samples are extracted, and the SDWI water body segmentation threshold is determined.

[0075] Based on the SDWI water index obtained in S400 and the stable water samples obtained in S500, the SDWI pixel values ​​of the area covered by the stable water samples are extracted, and the segmentation threshold of the SDWI is determined by a statistical method.

[0076] The calculation formula of SDWI water body segmentation threshold is:

[0077] T=MSDWI -2S SDWI

[0078] Where, T represents the SDWI water body segmentation threshold, M SDWI represents the mean value of the SDWI water index in the corresponding area of ​​the stable water sample point, S SDWI Represents the standard deviation of the SDWI water index in the corresponding area of ​​the stable water sample point.

[0079] S700: Calculate monthly water body frequency based on SDWI water body segmentation threshold.

[0080] Based on the SDWI water body segmentation threshold obtained in S700, when the SDWI water body index of a pixel is greater than the SDWI water body segmentation threshold, the pixel is marked as a water body. Based on this, the regional monthly time series water body mask is obtained in turn, and the monthly water body frequency value is calculated on this basis, that is, the frequency of each pixel being identified as a water body in the monthly time series. The specific calculation formula is:

[0081]

[0082] Where F represents the frequency of pixels being identified as water bodies in the monthly time series; N water Indicates the number of times the pixel is identified as a water body, N valid Represents the valid observation number of the pixel.

[0083] S800: Determine the monthly initial stable water body based on the monthly water body frequency calculation.

[0084] Based on the monthly water body frequency obtained in step S700, pixels with a monthly water body frequency greater than 0.3 are determined as stable water bodies, and the monthly initial stable water body of the region is obtained accordingly.

[0085] S900: Calculate the optimization index based on the monthly median optical remote sensing image and remove the non-water part of the monthly initial stable water body.

[0086] Based on the monthly median optical remote sensing image, the optimization index is calculated and the non-water body part of the monthly initial stable water body is removed. When the optimization index value is greater than 0, it indicates a non-water body. It is used to remove pixels in the low scattering area of ​​the monthly initial stable water body that are mistakenly classified as water bodies.

[0087] The optimization index calculation formula is:

[0088]

[0089] Where OI represents the optimization index result, ρ SWIR1 represents the short-wave infrared 1 band reflectivity, ρ SWIR2 represents the shortwave infrared 2 band reflectivity, ρ R represents the reflectivity of red light band, ρG Represents the reflectivity of the green light band.

[0090] S1000: Calculate the MNDWI water index based on the monthly median remote sensing image.

[0091] Specifically, the MNDWI water index is calculated based on the monthly median remote sensing image. When the MNDWI water index value is greater than 0, it means that the part is a water body.

[0092] The calculation formula of MNDWI water index is:

[0093]

[0094] Where, MNDWI represents the MNDWI water index, ρ G represents the reflectivity of green light band, ρ SWIR1 Indicates the shortwave infrared 1 band reflectivity.

[0095] S1100: The MNDWI water index is integrated with the monthly initial stable water body after removing the non-water body part, and the missed water body part is supplemented to obtain the final monthly stable water body.

[0096] Figure 2 It is a schematic diagram of the monthly stable water extraction results of the demonstration area provided in one embodiment of the present application; Figure 3 It is a partial schematic diagram of the water extraction results of the cultivation area in the demonstration area provided by an embodiment of the present application; Figure 4 It is a partial schematic diagram of the water body extraction results of the built-up area of ​​the demonstration area provided in one embodiment of the present application; Figure 5 This is a partial schematic diagram of the water extraction results of the lake area in the demonstration area provided by an embodiment of the present application. Figures 2 to 5 As shown, after comparison, it can be seen that this method can accurately obtain information about stable water bodies.

[0097] This method combines active remote sensing and passive remote sensing, and can overcome the problems of interference from cloudy and rainy weather and insufficient effective ground observations, and greatly weakens the phenomena of imperfect water body edges and small patches, omission of urban water bodies, and misclassification of water bodies in low-scattering areas in the results; secondly, this method uses the sample points of existing remote sensing water body products as prior information to automatically and quickly obtain the remote sensing water body index segmentation threshold, reducing the uncertainty of threshold segmentation; in addition, this method also comprehensively applies multi-source remote sensing water body indices, and improves the robustness of land surface water body information extraction through time series frequency judgment, partition integration, and result optimization integration.

[0098] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0099] It is easy to understand that those skilled in the art can combine, split, reorganize, etc. the embodiments of the present application to obtain other embodiments based on the several embodiments provided in the present application, and these embodiments do not exceed the protection scope of the present application.

[0100] The above specific implementation methods further explain in detail the purpose, technical solutions and beneficial effects of the embodiments of the present application. It should be understood that the above are only specific implementation methods of the embodiments of the present application and are not used to limit the protection scope of the embodiments of the present application. Any modifications, equivalent substitutions, improvements, etc. made on the basis of the technical solutions of the embodiments of the present application should be included in the protection scope of the embodiments of the present application.

Claims

1. A method for extracting stable water bodies from multimodal remote sensing images based on prior knowledge, characterized in that: include: Acquire Sentinel-1 SAR remote sensing data and Sentinel-2 optical remote sensing data; Preprocessing the Sentinel-1SAR remote sensing data to obtain monthly time-series SAR remote sensing images covering the entire region; Preprocessing the Sentinel-2 optical remote sensing data to obtain a monthly median optical remote sensing image covering the entire region; Calculate the SDWI water index based on the monthly time-series SAR remote sensing images and remove outliers; Using GSW water products as prior information, global stable water samples are extracted; Based on the SDWI water body index and the stable water body samples, the SDWI pixel values ​​of the area covered by the stable water body samples are extracted, and the SDWI water body segmentation threshold is determined; Monthly water body frequency calculation based on SDWI water body segmentation threshold calculation; Calculate and determine the monthly initial stable water body based on the monthly water body frequency; The optimization index is calculated based on the monthly median optical remote sensing image and the non-water part of the monthly initial stable water body is removed; Based on the monthly median remote sensing image, the MNDWI water index is calculated; The MNDWI water index is integrated with the monthly initial stable water body after removing the non-water body part, and the missed water body part is supplemented to obtain the final monthly stable water body.

2. The method for extracting stable water bodies from multimodal remote sensing images based on prior knowledge according to claim 1 is characterized in that: The preprocessing of the Sentinel-1SAR remote sensing data to obtain monthly time-series SAR remote sensing images covering the entire region includes: The Sentinel-1SAR remote sensing data is subjected to terrain correction, Refined Lee filtering, splicing and cropping to form monthly time-series SAR images covering the entire region.

3. The method for extracting stable water bodies from multimodal remote sensing images based on prior knowledge according to claim 2 is characterized in that: The monthly time-series SAR images are ground-range multi-view images in an interferometric wide-band mode, with VV polarization and VH polarization, and an image spatial resolution of 10 m.

4. The method for extracting stable water bodies from multimodal remote sensing images based on prior knowledge according to claim 2 is characterized in that: The preprocessing of the Sentinel-2 optical remote sensing data to obtain a monthly median optical remote sensing image covering the entire region includes: The Sentinel-2 optical remote sensing data is processed in the cloud using the QA60 band, and a monthly median optical remote sensing image covering the entire area is formed through stitching, monthly median synthesis and cropping.

5. The method for extracting stable water bodies from multimodal remote sensing images with integrated prior knowledge according to claim 1, characterized in that: The calculation formula of the SDWI water index is: SDWI=ln(10·VV·VH)-8 Among them, SDWI represents the SDWI water body segmentation threshold, VV represents the VV polarization band, and VH represents the VH polarization band.

6. The method for extracting stable water bodies from multimodal remote sensing images based on prior knowledge according to claim 1, characterized in that: The calculation formula of the SDWI water body segmentation threshold is: T=M SDWI -2S SDWI Where T represents the SDWI water body segmentation threshold, M SDWI represents the mean value of the SDWI water index in the area corresponding to the stable water sample point, S SDWI Represents the standard deviation of the SDWI water index in the corresponding area of ​​the stable water sample point.

7. The method for extracting stable water bodies from multimodal remote sensing images based on prior knowledge according to claim 1, characterized in that: The monthly water body frequency calculation based on the SDWI water body segmentation threshold includes: When the SDWI water index of a pixel is greater than the SDWI water segmentation threshold, the pixel is marked as a water body. Based on this, the regional monthly time series water body mask is obtained in turn. On this basis, the monthly water body frequency value is calculated, that is, the frequency of each pixel being identified as a water body in the monthly time series. The specific calculation formula is: Where F represents the frequency of pixels being identified as water bodies in the monthly time series; N water Indicates the number of times the pixel is identified as a water body; N valid Represents the valid observation number of the pixel.

8. The method for extracting stable water bodies from multimodal remote sensing images with integrated prior knowledge according to claim 1, characterized in that: Pixels with a monthly water body frequency greater than 0.3 were identified as monthly initial stable water bodies; Based on the monthly median optical remote sensing image, the optimization index is calculated and the non-water body part of the monthly initial stable water body is removed. When the optimization index value is greater than 0, it indicates a non-water body. It is used to remove pixels in the low scattering area of ​​the monthly initial stable water body that are mistakenly classified as water bodies. The optimization index calculation formula is: Where OI represents the optimization index result, ρ SWIR1 represents the short-wave infrared 1 band reflectivity, ρ SWIR2 represents the shortwave infrared 2 band reflectivity, ρ R represents the reflectivity of red light band, ρ G Represents the reflectivity of the green light band.

9. The method for extracting stable water bodies from multimodal remote sensing images with integrated prior knowledge according to claim 1, characterized in that: Based on the monthly median remote sensing image, the MNDWI water index is calculated. When the MNDWI water index value is greater than 0, it is a water body. The calculation formula of MNDWI water index is: Where, MNDWI represents the MNDWI water index, ρ G represents the reflectivity of green light band, ρ SWIR1 Indicates the shortwave infrared 1 band reflectivity.

Images