A surface water level extraction method based on SWOT satellite pixel cloud products

Abstract

The present invention discloses a method for extracting surface water levels based on SWOT satellite pixel cloud products, comprising: S1, extracting the extent of water bodies within a study area based on multi-source remote sensing images; S2, extracting the water surface elevation within the study area based on SWOT satellite pixel cloud product data; S3, performing outlier removal processing at different scales for the surface elevations of different types of inland water bodies to extract the corresponding surface water levels. The method provided by the present invention can ensure the scientific representativeness of the water level extraction results for inland water bodies. It also proposes two different processing flows for rivers and lakes, improving the temporal resolution and overall accuracy of the inverted water levels, and providing technical support for the application and promotion of SWOT satellite pixel cloud products in extracting water levels from inland water bodies.

Description

Technical Field

[0001] The present invention belongs to the technical field of satellite altimetry applications, and in particular relates to a surface water level extraction method based on SWOT satellite pixel cloud products. Background Art

[0002] Surface water level is an important indicator for studying the dynamic changes of surface water. Strengthening surface water level monitoring capabilities and accurately and comprehensively monitoring changes in river and lake water levels are of great significance to flood prevention and disaster reduction in river basins, water resource management and allocation, and river and lake ecological and environmental protection. Traditional surface water level measurement methods usually rely on ground-based measuring stations, cross-sectional water level gauges, or manual inspections. These methods have significant spatial limitations and high labor costs. At the same time, due to the uneven distribution of ground monitoring stations, their application in complex terrain and remote areas still faces many challenges. Relying solely on conventional hydrological observation methods cannot provide effective and continuous water level observations, making it difficult to meet the needs of dynamic monitoring.

[0003] With the continuous development of remote sensing technology, satellite altimetry has provided a new technical approach for water level monitoring. Satellite altimetry, a novel remote sensing technique, uses electromagnetic waves emitted to the ground and measures the round-trip propagation time of the echoes to accurately invert the surface elevations of oceans and land. It enables large-scale, high-precision, and periodic surface surveys, offering advantages unmatched by other observation techniques. It overcomes the limitations of traditional surface water level monitoring, provides wider spatial coverage, and fills the gap in water level monitoring in data-deficient areas. It has become a crucial technical tool for monitoring the elevation of oceans and inland water bodies worldwide.

[0004] The SWOT satellite, a groundbreaking space hydrological mission, is designed for dynamic monitoring of global water resources. Its Ka-band radar interferometer, using dual-antenna interferometric radar technology, can interferometrically measure water surface heights over 90% of Earth's surface over a 120 km wide swath. This enables high-resolution monitoring of global surface waters, achieving for the first time continuous two-dimensional observations of global surface water bodies. One of the SWOT satellite's core products is its high-resolution pixel cloud product (L2_HR_PIXC), which includes information such as the geolocation and elevation of water bodies. However, the SWOT satellite's pixel cloud data contains numerous outliers affected by noise, such as the mislabeling of "dark water" features caused by satellite attitude variations, phase unwrapping errors, and the influence of terrain and other geographical features. Effectively removing these outliers is crucial for accurate water level extraction. Therefore, there is an urgent need to develop high-precision, large-scale, and systematic surface water level extraction methods that utilize the SWOT satellite's high-resolution pixel cloud data, combined with multi-source remote sensing information and efficient data processing technologies. Summary of the Invention

[0005] In response to the above-mentioned deficiencies in the prior art, the present invention provides a surface water level extraction method based on SWOT satellite pixel cloud products. By using multi-source remote sensing data and SWOT satellite pixel cloud high-resolution data, and combining the distribution, classification attributes, and statistical filtering algorithms of SWOT satellite pixel cloud products, a surface water level extraction method based on SWOT satellite pixel cloud products is proposed, which realizes high-precision dynamic monitoring of surface water levels and provides important technical support for hydrological research and water resources management.

[0006] In order to achieve the above-mentioned object of the invention, the technical solution adopted by the present invention is: a surface water level extraction method based on SWOT satellite pixel cloud product, comprising the following steps:

[0007] S1. Extract the scope of water bodies in the study area based on multi-source remote sensing images;

[0008] S2. Extract the water surface elevation in the study area based on the SWOT satellite pixel cloud product data;

[0009] S3. For the water surface elevations of different types of inland water bodies, outliers of different scales are eliminated to extract the corresponding surface water levels.

[0010] Furthermore, the step S1 includes the following sub-steps:

[0011] S11. Obtain Sentinel-1 and Sentinel-2 remote sensing images of the study area and the study period respectively;

[0012] S12. Obtain the boundaries of inland water bodies within the study area, determine the boundary buffer zone, and clip the Sentinel-1 and Sentinel-2 remote sensing images within the study area based on the boundary buffer zone; the inland water bodies include rivers and lakes;

[0013] S13, calculating the water body index and performing threshold segmentation on the cropped Sentinel-1 or Sentinel-2 remote sensing image to obtain the corresponding water body range;

[0014] S14. Binarize the extracted water body range and export the processed raster data as a surface vector file for clipping pixel cloud data.

[0015] Furthermore, in step S13, for the Sentinel-1 remote sensing image, speckle noise is removed by Refined Lee filtering, and the Otsu method is used to automatically select a suitable threshold value to extract the water range in the study area based on the dual-polarization water index (SDWI);

[0016] For Sentinel-2 remote sensing images, we screened out those with cloud cover less than 20% during the study period, and used the Otsu method to automatically select a suitable threshold based on the automatic water index AWEIsh to extract the water range within the study area.

[0017] In step S14, when the extracted water body range is binarized, the Sentinel-2 remote sensing image synchronized with the SWOT satellite transit time is preferentially selected to extract the water body range. When the Sentinel-2 remote sensing image is interfered by cloud cover and the water body range cannot be effectively extracted, the water body range extraction result of the Sentinel-1 remote sensing image is used.

[0018] Furthermore, in step S13, the dual-polarization water index SDWI is:

[0019] SDWI=ln(10×VV×VH)-8

[0020] Where VV / VH represents two different polarization modes;

[0021] In step S14, the automatic water index AWEIsh is:

[0022] AWEIsh=Blue+2.5×Green-1.5×(NIR+SWIR1)-0.25×SWIR2

[0023] Wherein, Blue represents the blue band; Green represents the green band; NIR represents the near-infrared band; SWIR1 / 2 represents the short-wave infrared band.

[0024] Furthermore, step S2 includes the following steps:

[0025] S21. Obtain SWOT satellite pixel cloud product data within the study area and study period;

[0026] S22. Save specific fields in the SWOT satellite pixel cloud product data as a CSV file;

[0027] S23, cutting the CSV file according to the surface vector file of the water body range in the study area, and extracting the CSV file within the water body range in the study area;

[0028] S24. For the CSV file within the water body range of the study area, calculate the water surface elevation of the study area based on specific fields.

[0029] Furthermore, in step S24, the water surface elevation WSE of the study area is:

[0030] WSE=height-geoid-solid_earth_tide-load_tide-pole_tide

[0031] Where height represents the water surface height, that is, the height of the water surface in the study area relative to the reference ellipsoid; geoid represents the height difference of the geoid relative to the reference ellipsoid; solid_earth_tide represents the solid earth tide height; load_tide represents the load tide height; pole_tide represents the pole tide height.

[0032] Furthermore, step S3 includes the following sub-steps:

[0033] S31. Preliminary elimination of outliers in the water surface elevation within the study area;

[0034] S32, on the basis of preliminary elimination of outliers;

[0035] For pixel cloud data of inland water bodies such as rivers, multiple statistical filtering methods are used to perform statistical analysis by calculating the distance between the sampling point and the neighborhood point set, identify and remove outliers, and extract the corresponding surface water level;

[0036] For the pixel cloud data of inland water bodies such as lakes, pixel cloud data quality screening, open water pixel cloud data screening and multiple statistical filtering are carried out in sequence. Statistical analysis is performed by calculating the distance between the sampling point and the neighborhood point set, outliers are identified and removed, and the corresponding surface water level is extracted.

[0037] Furthermore, in step S31, outliers in the water surface elevation are preliminarily eliminated by the interquartile range method, and the formula is:

[0038] IQR=Q3-Q1

[0039] WSE 异 <Q1-1.5×IQR&WSE 异 >Q3+1.5×IQR

[0040] Where, IQR represents the interquartile range, which reflects the degree of dispersion of the middle 50% of the water surface elevation data, Q3 represents the third quartile, Q1 represents the first quartile, and WSE represents the 异 Indicates outliers in water surface elevation.

[0041] Furthermore, in step S32, the formula for statistical filtering is:

[0042]

[0043]

[0044]

[0045] T=μ+α×σ

[0046] Where, d i represents the average neighborhood distance; k represents the nearest neighbor point; p i Represents each point in the point cloud; ||p i -p ij || represents point p i The Euclidean distance between the point and its j-th nearest neighbor; μ represents the global mean; σ represents the global standard deviation; T represents the threshold; α represents the scale factor; and n represents the total number of point clouds.

[0047] The beneficial effects of the present invention are:

[0048] (1) The surface water level extraction method based on the SWOT satellite pixel cloud product proposed in this paper can effectively eliminate the outliers and noise data of the pixel cloud product, greatly improve the extraction accuracy of the surface water level, and provide an efficient and accurate surface water level extraction solution.

[0049] (2) The present invention designs treatment strategies that are suitable for different inland water bodies such as rivers and lakes, which significantly improves the applicability of the method to different inland water bodies, helps to accurately grasp the dynamic changes of inland water bodies, and provides an important basis for the scientific formulation of water resource management strategies, optimization of scheduling plans, and flood prevention and disaster reduction. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 The figure shows a flow chart of a surface water level extraction method based on SWOT satellite pixel cloud products provided by an embodiment of the present invention.

[0051] Figure 2 Shown is a comparison diagram of the longitude distribution of river pixel clouds and the filtering effect in the section from Yichang Hydrological Station to Zhicheng Hydrological Station in the middle reaches of the Yangtze River provided by an embodiment of the present invention.

[0052] Figure 3 Shown is a comparison diagram of the longitude distribution and filtering effect of the Taihu Lake pixel cloud provided by an embodiment of the present invention.

[0053] Figure 4 Shown is a comparison diagram of pixel cloud elevation distribution before and after river statistical filtering in the section from Yichang Hydrological Station to Zhicheng Hydrological Station in the middle reaches of the Yangtze River provided by an embodiment of the present invention.

[0054] Figure 5 Shown is a comparison diagram of pixel cloud elevation distribution before and after statistical filtering of Taihu Lake provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0055] The specific embodiments of the present invention are described below to facilitate understanding of the present invention by those skilled in the art. However, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the appended claims, these changes are obvious, and all inventions and creations utilizing the concepts of the present invention are protected.

[0056] The embodiment of the present invention provides a surface water level extraction method based on SWOT satellite pixel cloud products, such as Figure 1 As shown, the following steps are included:

[0057] S1. Extract the scope of water bodies in the study area based on multi-source remote sensing images;

[0058] S2. Extract the water surface elevation in the study area based on the SWOT satellite pixel cloud product data;

[0059] S3. For the water surface elevations of different types of inland water bodies, outliers of different scales are eliminated to extract the corresponding surface water levels.

[0060] Step S1 of the embodiment of the present invention includes the following sub-steps:

[0061] S11. Obtain Sentinel-1 and Sentinel-2 remote sensing images of the study area and the study period respectively;

[0062] S12. Obtain the boundaries of inland water bodies within the study area, determine the boundary buffer zone, and clip the Sentinel-1 and Sentinel-2 remote sensing images within the study area based on the boundary buffer zone; the inland water bodies include rivers and lakes;

[0063] S13, calculating the water body index and performing threshold segmentation on the cropped Sentinel-1 or Sentinel-2 remote sensing image to obtain the corresponding water body range;

[0064] S14. Binarize the extracted water body range and export the processed raster data as a surface vector file for clipping pixel cloud data.

[0065] In step S13 of this embodiment, for the Sentinel-1 remote sensing image, speckle noise is removed by Refined Lee filtering, and the Otsu method is used to automatically select a suitable threshold value based on the dual-polarization water index (SDWI) to extract the water range within the study area.

[0066] For Sentinel-2 remote sensing images, we screened out those with cloud cover less than 20% during the study period, and used the Otsu method to automatically select a suitable threshold based on the automatic water index AWEIsh to extract the water range within the study area.

[0067] Among them, the dual polarization water index SDWI is:

[0068] SDWI=ln(10×VV×VH)-8

[0069] Where VV / VH represents two different polarization modes;

[0070] The automatic water index AWEIsh is:

[0071] AWEIsh=Blue+2.5×Green-1.5×(NIR+SWIR1)-0.25×SWIR2

[0072] Wherein, Blue represents the blue band; Green represents the green band; NIR represents the near-infrared band; SWIR1 / 2 represents the short-wave infrared band.

[0073] In step S14 of this embodiment, when the extracted water body range is binarized, the Sentinel-2 remote sensing image synchronized with the SWOT satellite transit time is preferentially selected to extract the water body range. When the Sentinel-2 remote sensing image is interfered with by cloud cover and the water body range cannot be effectively extracted, the water body range extraction result of the Sentinel-1 remote sensing image is used.

[0074] Specifically, when determining the extent of water bodies, the high precision of optical data and the all-weather capability of radar data are used to compensate for the limitations of a single data source, ensure the temporal continuity of the extracted water body extent, and improve the spatiotemporal consistency of matching the SWOT satellite transit time.

[0075] Step S2 of the embodiment of the present invention includes the following steps:

[0076] S21. Obtain SWOT satellite pixel cloud product data within the study area and study period;

[0077] S22. Save specific fields in the SWOT satellite pixel cloud product data as a CSV file;

[0078] S23, cutting the CSV file according to the surface vector file of the water body range in the study area, and extracting the CSV file within the water body range in the study area;

[0079] S24. For the CSV file within the water body of the study area, calculate the water surface elevation of the study area according to the specific fields;

[0080] Among them, the water surface elevation WSE of the study area is:

[0081] WSE=height-geoid-solid_earth_tide-load_tide-pole_tide

[0082] Where height represents the water surface height, that is, the height of the water surface in the study area relative to the reference ellipsoid; geoid represents the height difference of the geoid relative to the reference ellipsoid; solid_earth_tide represents the solid earth tide height; load_tide represents the load tide height; pole_tide represents the pole tide height.

[0083] In this embodiment, the SWOT satellite pixel cloud product is the SWOT satellite L2_HR_PIXC pixel cloud product.

[0084] Step S3 of the embodiment of the present invention includes the following sub-steps:

[0085] S31. Preliminary elimination of outliers in the water surface elevation within the study area;

[0086] Among them, the interquartile range method is used to preliminarily eliminate outliers in the water surface elevation. The formula is:

[0087] IQR=Q3-Q1

[0088] WSE 异 <Q1-1.5×IQR&WSE 异 >Q3+1.5×IQR

[0089] Where, IQR represents the interquartile range, which reflects the degree of dispersion of the middle 50% of the water surface elevation data, Q3 represents the third quartile, Q1 represents the first quartile, and WSE represents the 异 Indicates outliers in water surface elevation.

[0090] S32, on the basis of preliminary elimination of outliers;

[0091] For pixel cloud data of inland water bodies such as rivers, multiple statistical filtering methods are used to perform statistical analysis by calculating the distance between the sampling point and the neighborhood point set, identify and remove outliers, and extract the corresponding surface water level;

[0092] For pixel cloud data of inland lakes, we perform pixel cloud data quality screening, open water pixel cloud data screening, and multiple statistical filtering. We perform statistical analysis by calculating the distance between the sampling point and the neighboring point set, identify and remove outliers, and extract the corresponding surface water level.

[0093] In this embodiment, the formula for performing statistical filtering is:

[0094]

[0095]

[0096]

[0097] T=μ+α×σ

[0098] Where, d i represents the average neighborhood distance; k represents the nearest neighbor point; p i Represents each point in the point cloud; ||p i -p ij || represents point p i The Euclidean distance between the point and its j-th nearest neighbor; μ represents the global mean; σ represents the global standard deviation; T represents the threshold; α represents the scale factor; and n represents the total number of point clouds.

[0099] When extracting surface water levels from pixel cloud data for inland rivers, extensive experiments revealed that selecting only data with a quality flag of zero for pixel clouds located above rivers can result in empty pixel clouds. Furthermore, even with a high quality flag value, the water surface elevation at some pixels remains similar to that obtained when the quality flag is zero. Therefore, to retain as much valid pixel cloud data as possible, we employed a method that, after initially removing outliers, employed multiple statistical filtering methods to identify and remove outliers. Figure 2 The figure shows the longitudinal distribution of river pixel clouds and the comparison of filtering effects in the section from Yichang Hydrological Station to Zhicheng Hydrological Station in the middle reaches of the Yangtze River provided by an embodiment of the present invention; Figure 4 Shown is a comparison diagram of pixel cloud elevation distribution before and after river statistical filtering in the section from Yichang Hydrological Station to Zhicheng Hydrological Station in the middle reaches of the Yangtze River provided by an embodiment of the present invention.

[0100] When extracting surface water levels from pixel cloud data of inland lakes, we first screen pixel cloud data with reliable quality and select pixel cloud data with a quality indicator attribute (geolocation_qual) of zero. Next, we further screen pixel cloud data that are automatically classified as open water (class = 4). Finally, we further adopt a multiple statistical filtering method to calculate the distance between the sampling point and the neighboring point set for statistical analysis to identify and remove outliers.

[0101] Specifically, the geolocation quality flag attribute (geolocation_qual) of the SWOT satellite pixel cloud product is used to assess the quality of the geolocation data (including altitude, longitude, and latitude). A value of zero indicates reliable data quality, while larger values ​​indicate lower reliability. Pixel cloud products are classified into seven categories: land (class = 1), land near water (class = 2), water near land (class = 3), open water (class = 4), dark water (class = 5), low-coherence water near land (class = 6), and open low-coherence water (class = 7). Different categories represent different pixel cloud classifications. Therefore, only pixel cloud data with a quality flag of zero and classified as open water were selected. Figure 3 The figure shows the distribution of Taihu Lake pixel cloud in the longitude direction and the comparison of filtering effects provided by an embodiment of the present invention. Figure 5 Shown is a comparison diagram of pixel cloud elevation distribution before and after statistical filtering of Taihu Lake provided by an embodiment of the present invention.

[0102] Specific embodiments are used in the present invention to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core ideas. At the same time, for those skilled in the art, according to the ideas of the present invention, there may be changes in the specific implementation methods and application scopes. In summary, the contents of this specification should not be understood as limiting the present invention.

[0103] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and it should be understood that the scope of protection of the present invention is not limited to such specific descriptions and embodiments. Those skilled in the art can make various other specific variations and combinations based on the technical teachings disclosed in the present invention without departing from the essence of the present invention, and such variations and combinations are still within the scope of protection of the present invention.

Claims

1. A surface water level extraction method based on SWOT satellite pixel cloud products, characterized in that: The following steps are involved: S1. Extract the scope of water bodies in the study area based on multi-source remote sensing images; S2. Extract the water surface elevation in the study area based on the SWOT satellite pixel cloud product data; S3. Perform outlier removal processing of different scales for the water surface elevations of different types of inland water bodies to extract the corresponding surface water levels, including the following steps: S31. Preliminary elimination of outliers in the water surface elevation within the study area; S32, on the basis of preliminary elimination of outliers; For pixel cloud data of inland water bodies such as rivers, multiple statistical filtering methods are used to perform statistical analysis by calculating the distance between the sampling point and the neighborhood point set, identify and remove outliers, and extract the corresponding surface water level; For the pixel cloud data of inland water bodies such as lakes, pixel cloud data quality screening, open water pixel cloud data screening and multiple statistical filtering are carried out in sequence. Statistical analysis is performed by calculating the distance between the sampling point and the neighborhood point set, outliers are identified and removed, and the corresponding surface water level is extracted.

2. The surface water level extraction method based on SWOT satellite pixel cloud products according to claim 1 is characterized in that: The step S1 includes the following sub-steps: S11. Obtain Sentinel-1 and Sentinel-2 remote sensing images of the study area and the study period respectively; S12. Obtain the boundaries of inland water bodies within the study area, determine the boundary buffer zone, and clip the Sentinel-1 and Sentinel-2 remote sensing images within the study area based on the boundary buffer zone; the inland water bodies include rivers and lakes; S13, calculating the water body index and performing threshold segmentation on the cropped Sentinel-1 or Sentinel-2 remote sensing image to obtain the corresponding water body range; S14. Binarize the extracted water body range and export the processed raster data as a surface vector file for clipping pixel cloud data.

3. The surface water level extraction method based on SWOT satellite pixel cloud products according to claim 2 is characterized in that: In step S13, for the Sentinel-1 remote sensing image, speckle noise is removed by Refined Lee filtering, and the Otsu method is used to automatically select a suitable threshold value based on the dual-polarization water index (SDWI) to extract the water range in the study area; For Sentinel-2 remote sensing images, we screened out those with cloud cover less than 20% during the study period, and used the Otsu method to automatically select a suitable threshold based on the automatic water index AWEIsh to extract the water range within the study area. In step S14, when the extracted water body range is binarized, the Sentinel-2 remote sensing image synchronized with the SWOT satellite transit time is preferentially selected to extract the water body range. When the Sentinel-2 remote sensing image is interfered by cloud cover and the water body range cannot be effectively extracted, the water body range extraction result of the Sentinel-1 remote sensing image is used.

4. The surface water level extraction method based on SWOT satellite pixel cloud products according to claim 2 is characterized in that: In step S13, the dual polarized water index for: Where, VV / VH Indicates two different polarization modes; In step S14, the automatic water index AWEIsh for: Where, Blue represents the blue band; Green Indicates the green band; NIR represents the near-infrared band; SWIR1 / 2 Indicates the shortwave infrared band.

5. The surface water level extraction method based on SWOT satellite pixel cloud products according to claim 1 is characterized in that: The step S2 comprises the following steps: S21. Obtain SWOT satellite pixel cloud product data within the study area and study period; S22. Save specific fields in the SWOT satellite pixel cloud product data as a CSV file; S23, cutting the CSV file according to the surface vector file of the water body range in the study area, and extracting the CSV file within the water body range in the study area; S24. For the CSV file within the water body range of the study area, calculate the water surface elevation of the study area based on specific fields.

6. The surface water level extraction method based on SWOT satellite pixel cloud products according to claim 5 is characterized in that: In step S24, the water surface elevation of the research area WSE for: Where, height represents the water surface height, that is, the height of the water surface in the study area relative to the reference ellipsoid; geoid It represents the height difference of the geoid relative to the reference ellipsoid; solid_earth_tide represents the tidal height of the solid Earth; load_tide represents the load tidal height; pole_tide Indicates the height of the high tide.

7. The surface water level extraction method based on SWOT satellite pixel cloud products according to claim 1 is characterized in that: In step S31, outliers in the water surface elevation are initially eliminated by the interquartile range method, and the formula is: Where, It represents the interquartile range, reflecting the degree of dispersion of the middle 50% of the data in the water surface elevation. Q 3 represents the third quartile, Q 1 represents the first quartile, WSE 异 Indicates outliers in water surface elevation.

8. The surface water level extraction method based on SWOT satellite pixel cloud products according to claim 1 is characterized in that: In step S32, the formula for statistical filtering is: Where, represents the average neighborhood distance; represents the nearest neighbor point; Represents each point in the point cloud; Indicates a point With its j The Euclidean distance between the nearest neighbors; represents the global mean; represents the global standard deviation; Indicates the threshold value; represents the proportionality coefficient; n Indicates the total number of point clouds.

Images