Water body suspended sediment concentration monitoring method based on FUI water body classification

Abstract

A water body suspended sediment concentration monitoring method based on FUI water body classification comprises the steps that firstly, a target watershed Sentinel-2 L1C image from a certain year to a certain year is obtained from GEE, and after resampling is conducted till the resolution is 10 m, preprocessing such as 6S model atmospheric correction, proximity effect correction and quality control is conducted; 5 wave bands are used for calculating spectral angles to obtain FUI, and 14 is used as a threshold value to divide the water body into clear water and muddy water. And then selecting an optimal wave band combination according to the FUI classification to construct a semi-empirical inversion model, and finally calculating the suspended sediment concentration distribution by using the model and verifying the suspended sediment concentration distribution. The method solves the problem that the model generalization ability is low in the monitoring of the suspended sediment concentration of the target drainage basin, and the inversion precision is high.

Description

Technical Field

[0001] The present invention belongs to the technical field of water body suspended sediment concentration monitoring, and in particular relates to a water body suspended sediment concentration monitoring method based on FUI water body classification. Background Art

[0002] Monitoring suspended sediment concentrations in water bodies is crucial for studying ecosystems and hydrodynamics. Traditional methods rely on collecting water samples for laboratory analysis, which is highly accurate but time-consuming and labor-intensive. Furthermore, only 17 hydrological stations along the Yangtze River have monitoring capabilities, resulting in intermittent observations from space. Satellite remote sensing technology, owing to its multi-temporal and large-scale monitoring capabilities, has gradually become the mainstream method. Its inversion methods include empirical, analytical, and semi-analytical methods. Semi-analytical methods are particularly advantageous due to their clear physical mechanisms and strong generalization capabilities.

[0003] Optical classification of water bodies can improve model applicability. Traditional classification methods (such as single-band and band ratios) are susceptible to interference from atmospheric and observational conditions. The Forel-Ule index (FUI), which is related to turbidity, is resistant to interference from aerosols and observation geometry, and provides a stable classification threshold. Existing research on the Yangtze River basin has primarily focused on lakes and estuaries (such as Poyang Lake and the Yangtze Estuary), with limited research on the mainstream stream. This is due to the following challenges: complex hydrodynamics and a wide range of suspended sediment concentrations (20 to 2500 mg / L), which make single models inaccurate; narrow river channels, which are susceptible to interference from cross-river bank radiation; and the need for highly automated image processing (such as atmospheric correction parameter acquisition) for large-scale monitoring, which requires high sensor spatial resolution.

[0004] The present invention mainly solves the following technical problems: The inversion accuracy of suspended sediment concentration in the mainstream of the Yangtze River is insufficient: the suspended sediment concentration in the mainstream has a very large range. The traditional single model is easily saturated in the spectral signal in highly turbid water bodies (>50 mg / L). The sensitive band at low concentration (<50 mg / L) is different from that at high concentration, resulting in poor generalization ability. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a method for monitoring suspended sediment concentration in water bodies based on FUI water classification. The present invention uses FUI to classify water bodies into clear water and turbid water, and constructs inversion models for each (using the B4 / B3 bands for clear water and the B8A / B4 bands for turbid water). This solves the problem of spectral response differences in different concentration ranges and improves accuracy.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is: A method for monitoring suspended sediment concentration in water based on FUI water body classification, comprising the following steps: Image data processing: Acquire Sentinel-2 L1C images of the target basin from one year to another, resample to 10-meter resolution, apply 6S atmospheric correction to correct for proximity effects, remove interference using quality control indices, and evaluate the accuracy of Rrs using L2A products. Optical classification of water bodies: Based on the range of suspended sediment concentration in the Yangtze River, the spectral angles of the five bands of Sentinel-2 were calculated to obtain the FUI. Based on the relationship between FUI and measured concentration, the water bodies were classified into clear water and turbid water using an FUI of 14 as the threshold. Construct a suspended sediment concentration inversion model: align the measured suspended sediment concentration with the Rrs time and remove outliers; use a semi-empirical method to select the best band combination according to the FUI classification, construct an algorithm, and select the suspended sediment concentration inversion model with the highest fit; The suspended sediment concentration inversion model was used to monitor the suspended sediment concentration in water bodies: the constructed suspended sediment concentration inversion model was used to calculate the suspended sediment concentration distribution in water bodies based on the processed Sentinel-2 imagery. By comparing with field measurement data from the same period, the inversion accuracy and applicability of the model in different water types (clear water and turbid water) were evaluated, and the model's ability to monitor the dynamic changes in suspended sediment concentration in the mainstream of the Yangtze River was verified.

[0007] Preferably, the specific steps of image data processing are as follows: S1.1. Data acquisition and preprocessing: Acquire Sentinel-2 L1C-level TOA images of the target basin from one year to another from the GEE platform, select available data with cloud coverage less than 40%, resample to 10-meter spatial resolution, and obtain a total of N images; S1.2. Atmospheric Correction: Perform atmospheric correction on TOA images based on the 6S model. Input parameters include image metadata (sensor and solar geometry parameters), MODIS aerosol product (AOD555), NCEP-NCAR water vapor data, TOMS / OMI column ozone concentration, and SRTM altitude data. Surface reflectance and Rrs are calculated using the formula. Rrs represents remote sensing reflectance. S1.3, Proximity Effect Correction: To address the narrow width of rivers, the Sentinel-2 technical documentation method was used, with a 1 km convolution kernel window set to calculate the pixel average Rrs. Combined with the diffuse / direct transmittance ratio output by the 6S model, the cross-radiation interference of riverbank objects was corrected using a formula. S1.4. Quality Control: Calculate the SGI index using the red band (B3) and shortwave infrared band (B12), eliminating interfering pixels such as solar flares and whitecaps with an SGI ≥ -0.5; evaluate the accuracy of the inverted Rrs by comparing it with the L2A standard product of the same period; Preferably, the specific steps for performing optical classification of water bodies are: S2.1. Spectral angle calculation: Select the five bands of Sentinel-2 (443, 490, 560, 665, and 705 nm), calculate the spectral angle (α), and match the 21-level Forel-Ule index (FUI) lookup table to obtain the FUI value. S2.2. Threshold determination and classification: Group the measured suspended sediment concentration and FUI values ​​into 10 mg / L intervals, draw a correlation line graph, and determine FUI = 14 as the classification threshold: FUI < 14 is classified as clear water (average SSC ≈ 35 mg / L), and FUI ≥ 14 is classified as turbid water (average SSC ≈ 289 mg / L). Preferably, the specific method for constructing the suspended sediment concentration inversion model is: S3.1. Data alignment and screening: Match the field-measured SSC with the Sentinel-2 inverted Rrs in a 12-hour time window, and remove abnormal spectral data with Rrs median values ​​< 0 in the near-infrared band; S3.2. Band combination optimization: Analyze the response characteristics of SSC to the spectrum. For low concentrations (<50 mg / L), use the red / green band combination (B4 / B3). For high concentrations (≥50 mg / L), use the near-infrared / red band combination (B8A / B4). Determine the optimal band ratio through regression analysis. S3.3. Model construction and fitting: Based on the semi-empirical method, linear, exponential, and logarithmic algorithms were constructed for each band combination. The coefficient of determination (r²) was used as an indicator, and the model with the highest fit (r²=0.54 for clear water and r²=0.57 for turbid water) was selected as the final inversion model. Preferably, the specific steps of monitoring the suspended sediment concentration in water using the suspended sediment concentration inversion model are: S4.1. Concentration inversion calculation: Apply the constructed model to the preprocessed Sentinel-2 image, batch calculate the Rrs band ratio, and obtain the spatial distribution of water body SSC through inversion of the model formula; S4.2. Accuracy Assessment and Verification: Compare field sampling data from the same period, calculate the inversion error (e.g., RMSE, MAE) for different water types (clear water / muddy water), analyze the model's ability to monitor dynamic changes in SSC (20-2500 mg / L), and verify its applicability in the mainstream of the Yangtze River; Preferably, in S1.2, the altitude data is obtained from the digital elevation data provided by the Shuttle Radar Topography Mission (SRTM); the atmospheric correction parameters are output after calculation by the 6S model, and then the TOA image is rasterized to obtain , the calculation formula is as follows:

[0008] in, is the pixel radiance of TOA image; 、 and is the correction coefficient calculated by the 6S model; is the surface reflectivity; is the downward direct irradiance of solar radiation, is the diffuse irradiance caused by aerosol and molecular scattering, both calculated by the 6S model; the sky spectral reflectance of the air-water interface It was set to 0.0245 in this study and typically ranges from 0.022 (in calm weather) to 0.025 (in wind speeds up to 5 m / s); Preferably, in S1.3, the proximity effect refers to the cross-radiation effect caused by bright objects at the boundary of the water body, which depends on the height distribution of aerosols above the objects and affects the entrance pupil radiance received by the sensor. This interference is particularly obvious in urban areas. In remote sensing monitoring of rivers, because the rivers are narrow and mostly flow through urban areas, the proximity effect interference they receive cannot be ignored. This method adopts the Sentinel-2 technical document method to weaken the proximity effect by reducing the contrast between the pixels in the convolution kernel and the background pixels. The steps include: setting the convolution kernel window radius to 1 km, calculating the average remote sensing reflectance of each pixel in the convolution kernel ( ), correct the proximity effect according to equations (5) and (6):

[0009] Where N is the number of pixels in the convolution kernel; is the diffuse transmittance from the sensor to the ground surface ( ) and direct transmittance ( ) ratio, calculated by the 6S model; After correction for proximity effect ; Preferably, in S1.4, a Quality Control Index (QCI) is proposed, which uses red light and short-wave infrared bands to remove interference such as solar flares and whitecaps. The calculation formula is as follows:

[0010] Among them, B12 and B3 represent the TOA radiation of the twelfth and third bands, respectively; pixels with SGI less than -0.5 are considered to be free of sunlight reflection pollution; In order to analyze the accuracy of satellite-derived Rrs, this paper uses the simultaneously captured L2A product to evaluate the Rrs consistency between the method in this study and the L2A standard product.

[0011] A water body suspended sediment concentration monitoring system based on FUI water body classification adopts the water body suspended sediment concentration monitoring method based on FUI water body classification. A computer device comprising: one or more processors; One or more execution programs are stored in the processor; When one or more execution programs are executed by the one or more processors, a method for monitoring suspended sediment concentration in water based on FUI water body classification is implemented.

[0012] The present invention can achieve the following beneficial effects: This invention overcomes the difficulties of satellite-based observations of river suspended sediment concentration, such as narrow river widths and a wide range of suspended sediment concentrations, by constructing an inversion model. The inversion model achieved leading accuracy in experiments to invert suspended sediment content in water bodies, with consistent and good inversion performance across different Yangtze River basins, and was able to effectively reflect the fluctuations in suspended sediment concentration along the Yangtze River mainstream.

[0013] This paper uses FUI to divide water bodies into two categories: clear water and turbid water. It selects the optimal band combination based on the spectral response characteristics of different types of water bodies to construct an inversion model. This effectively solves the problem of insufficient generalization ability of a single model caused by the large span of suspended sediment concentration in the mainstream of the Yangtze River, and enhances the model's adaptability to complex water environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The present invention will be further described below with reference to the accompanying drawings and examples: Figure 1 Flow chart of the method of the present invention; Figure 2 This is a flow chart of Sentinel-2 image preprocessing in the present invention; Figure 3 This is a graph showing the relationship between the FUI calculated by the present invention and the measured suspended sediment concentration; Figure 4 This is a box plot of suspended sediment concentration corresponding to different optical water body types in the measured data set of the present invention. DETAILED DESCRIPTION

[0015] The preferred solution is Figures 1 to 4 As shown in FIG, a method for monitoring suspended sediment concentration in water based on FUI water body classification is described, and the specific steps are as follows: Step 1: Sentinel-2 image data processing. The image data used in this method are Sentinel-2 Level-1C (L1C) top of atmosphere reflectance (TOA) images with cloud coverage less than 40% from 2016 to 2021 obtained from Google Earth Engine (GEE). This method uses all available MSI L1C images in the Yangtze River Basin and resamples them to 10-meter resolution. A total of 488 images are used. The image processing steps of this method are as follows: Figure 2 shown.

[0016] This method uses the 6S (Second simulation of the satellite signal in the solarspectrum) model to perform atmospheric correction on TOA images. The aerosol types are predefined by the model, including continental, combustion, urban, and oceanic types. The input parameters required for correction come from the image metadata and other auxiliary data. Among them, the geometric parameters such as the zenith angle and altitude angle of the sensor and the sun are obtained from the image attribute file. The aerosol optical depth at 555 nm (AOD555) is obtained from the synchronously observed MODIS MCD19A2 daily average aerosol product [1] or the MODIS MOD08 atmospheric monthly scale product [2]. The water vapor content is obtained from the National Centers for Environmental Prediction-National Center for Atmospheric Research (NCEP-NCAR) global reanalysis product [3]. The column ozone concentration is obtained from the TOMS / OMI product provided by NASA's Goddard Earth Science Data and Information Service (GES DISC) [4]. The altitude data is obtained from the digital elevation data provided by the Shuttle Radar Topography Mission (SRTM). The atmospheric correction parameters are calculated by the 6S model (Second simulation of the satellite signal in the solar spectrum), and then the TOA image is rasterized to obtain , the calculation formula is as follows:

[0017] in, is the pixel radiance of TOA image; 、 and is the correction coefficient calculated by the 6S model; is the surface reflectivity; is the downward direct irradiance of solar radiation, is the diffuse irradiance caused by aerosol and molecular scattering, both calculated by the 6S model; the sky spectral reflectance of the air-water interface It was set to 0.0245 in this study and typically ranges from 0.022 (in calm weather) to 0.025 (in winds up to 5 m / s).

[0018] The proximity effect refers to the cross-radiation effect caused by bright objects at the boundary of the water body. It depends on the height distribution of aerosols above the objects and affects the entrance pupil radiance received by the sensor. This interference is particularly obvious in urban areas. In remote sensing monitoring of rivers, because the river width is narrow and most of it flows through the city, the interference from the proximity effect cannot be ignored. This method adopts the Sentinel-2 technical document method to weaken the proximity effect by reducing the contrast between the pixels in the convolution kernel and the background pixels. The steps include: setting the convolution kernel window radius to 1 km, calculating the average remote sensing reflectance of each pixel in the convolution kernel ( ), correct the proximity effect according to equations (5) and (6):

[0019] Where N is the number of pixels in the convolution kernel; is the diffuse transmittance from the sensor to the ground surface ( ) and direct transmittance ( ) ratio, calculated by the 6S model; After correction for proximity effect .

[0020] White caps, shadows of ships and hulls caused by environmental factors on the water surface will have a certain impact on the inversion of water quality parameters. In addition, when sunlight is incident on the water surface and is reflected by the mirror and received by the sensor, it appears as a bright flare area on the image, also known as a transient anomaly. This type of optical signal is unrelated to the concentration of photoactive substances in the water body itself and needs to be removed in the preprocessing step. Therefore, this method proposes a quality control index (QCI), which uses red light and short-wave infrared bands to remove interference such as solar flares and whitecaps. The calculation formula is as follows:

[0021] Where B12 and B3 represent the TOA radiation of the twelfth and third bands, respectively. Pixels with an SGI less than -0.5 are considered to be free of sunlight reflection contamination.

[0022] In order to analyze the accuracy of satellite-derived Rrs, this paper uses the simultaneously captured L2A product to evaluate the Rrs consistency between the method in this study and the L2A standard product.

[0023] Step 2: Water Optical Classification. Water optical classification can improve the model's representation of suspended sediment concentrations, especially across a wide range of suspended sediment concentrations. The visible spectrum is sensitive to low to moderate suspended sediment concentrations (approximately <50 mg / L) but becomes saturated in highly turbid waters (approximately >50 mg / L). Suspended sediment concentrations in the Yangtze River range from 20 to 2500 mg / L. Therefore, this method uses the FUI to classify water bodies into two categories (clear and turbid), and then constructs an inversion model for each category. Five Sentinel-2 bands (with central wavelengths of 443, 490, 560, 665, and 705 nm) are used to calculate the spectral angle (α) in color space. The FUI is then obtained using a 21-level FUI lookup table.

[0024] This paper determined a classification threshold based on the relationship between the FUI and measured suspended sediment concentration. The FUI was calculated from Sentinel-2 spectra of satellite / ground-based data pairs. The data pairs were reordered by suspended sediment concentration and grouped into intervals of 10 mg / L. The average suspended sediment concentration and corresponding average FUI were calculated for each group and then plotted. When the suspended sediment concentration was low, the FUI increased with increasing suspended sediment concentration. When the FUI exceeded 14, the index no longer increased monotonically with increasing suspended sediment concentration, but instead fluctuated between 15 and 17. Therefore, an FUI of 14 can be used as the classification threshold to classify natural water bodies into clear and turbid water categories. The average suspended sediment concentrations for the two water types were 35.34 ± 21.55 mg / L (clear water) and 288.97 ± 208.21 mg / L (turbid water), respectively.

[0025] Step 3: Model Construction. To develop the suspended sediment concentration inversion model, this method temporally aligns field-measured suspended sediment concentrations with Sentinel-2-derived Rrs within a 12-hour window. Rrs spectra with values ​​less than zero in the near-infrared (VIS-NIR: B1-B8A) band are considered outliers and removed.

[0026] This method uses a semi-empirical approach to construct a suspended sediment concentration inversion model. Analysis of the spectral response of suspended sediment concentration revealed that the red and green bands are most sensitive to suspended sediment concentration at low suspended sediment concentrations (<50 mg / L). As suspended sediment concentration increases, the spectral signal in the VIS band approaches saturation, and the sensitive band shifts to the NIR spectral region. Previous studies have demonstrated that band-switching methods using the VIS–NIR band for suspended sediment concentration inversion can achieve good results. Therefore, this method uses the FUI to classify the data into two water categories. Through trial and error, the best performing algorithm for each category is selected: B4 (665 nm) / B3 (560 nm) for clear water (r² = 0.54), and B8A (865 nm) / B4 (665 nm) for turbid water (r² = 0.57). Using the selected exponents, various algorithms are constructed, including linear, exponential, and logarithmic formulas. The model with the best fit is selected as the final model for each water category.

[0027] The suspended sediment concentration inversion model was used to monitor the suspended sediment concentration in water bodies: the constructed suspended sediment concentration inversion model was used to calculate the suspended sediment concentration distribution in water bodies based on the processed Sentinel-2 imagery. By comparing the model with field measurement data from the same period, the inversion accuracy and applicability of the model in different water body types were evaluated, and the model's ability to monitor the dynamic changes in suspended sediment concentration in the mainstream of the Yangtze River was verified.

[0028] Concentration inversion calculation: The constructed model is applied to the preprocessed Sentinel-2 image, the Rrs band ratio is calculated in batches, and the spatial distribution of water body SSC is obtained through inversion of the model formula; Accuracy assessment and verification: Comparison with field sampling data from the same period, calculation of inversion errors (such as RMSE and MAE) for different water types (clear water / muddy water), analysis of the model's ability to monitor dynamic changes in SSC (20-2500 mg / L), and verification of its applicability to the mainstream of the Yangtze River.

[0029] The above embodiments are merely preferred technical solutions of the present invention and should not be construed as limiting the present invention. The scope of protection of the present invention shall be the technical solutions set forth in the claims, including equivalent alternatives to the technical features of the technical solutions set forth in the claims. In other words, equivalent alternatives and improvements within this scope are also within the scope of protection of the present invention.

Claims

1. A method for monitoring suspended sediment concentration in water based on FUI water body classification, characterized in that The following steps are involved: Image data processing: Acquire Sentinel-2 L1C images of the target basin from one year to another, resample to 10-meter resolution, apply 6S atmospheric correction to correct for proximity effects, remove interference using quality control indices, and evaluate the accuracy of Rrs using L2A products. Optical classification of water bodies: Based on the range of suspended sediment concentration in the Yangtze River, the spectral angles of the five bands of Sentinel-2 were calculated to obtain the FUI. Based on the relationship between FUI and measured concentration, the water bodies were classified into clear water and turbid water using an FUI of 14 as the threshold. Construct a suspended sediment concentration inversion model: align the measured suspended sediment concentration with the Rrs time and remove outliers; use a semi-empirical method to select the best band combination according to the FUI classification, construct an algorithm, and select the suspended sediment concentration inversion model with the highest fit; The suspended sediment concentration inversion model was used to monitor the suspended sediment concentration in water bodies: the constructed suspended sediment concentration inversion model was used to calculate the suspended sediment concentration distribution in water bodies based on the processed Sentinel-2 imagery. By comparing the model with field measurement data from the same period, the inversion accuracy and applicability of the model in different water body types were evaluated, and the model's ability to monitor the dynamic changes in suspended sediment concentration in the mainstream of the Yangtze River was verified.

2. The method for monitoring suspended sediment concentration in water based on FUI water classification according to claim 1, characterized in that: The specific steps for image data processing are as follows: S1.

1. Data acquisition and preprocessing: Acquire Sentinel-2 L1C-level TOA images of the target basin from one year to another from the GEE platform, select available data with cloud coverage less than 40%, resample to 10-meter spatial resolution, and obtain a total of N images; S1.

2. Atmospheric Correction: Perform atmospheric correction on TOA images based on the 6S model. Input parameters include image metadata, MODIS aerosol products, NCEP-NCAR water vapor data, TOMS / OMI column ozone concentration, and SRTM altitude data. Surface reflectance and Rrs are calculated using the formula. Rrs represents remote sensing reflectance. S1.3, Proximity Effect Correction: To address the narrow width of the river, the Sentinel-2 technical documentation method was used, with a 1km convolution kernel window set to calculate the pixel average Rrs. Combined with the diffuse / direct transmittance ratio output by the 6S model, the cross-radiation interference of riverbank objects was corrected using a formula. S1.

4. Quality control: Calculate the SGI index using the red and shortwave infrared bands, and remove interfering pixels such as solar flares and whitecaps with SGI ≥ -0.5; evaluate the accuracy of the inverted Rrs by comparing it with the L2A standard product of the same period.

3. The method for monitoring suspended sediment concentration in water based on FUI water classification according to claim 1, characterized in that: The specific steps for optical classification of water bodies are: S2.

1. Spectral angle calculation: Select the five bands of Sentinel-2 (443, 490, 560, 665, and 705 nm), calculate the spectral angle α, and match it with the 21-level Forel-Ule index lookup table to obtain the FUI value; S2.

2. Threshold determination and classification: Group the measured suspended sediment concentration and FUI value into 10 mg / L intervals, draw a correlation line graph, and determine FUI=14 as the classification threshold: FUI<14 is classified as clear water, and FUI≥14 is classified as turbid water.

4. The method for monitoring suspended sediment concentration in water based on FUI water classification according to claim 1, characterized in that: The specific method for constructing the suspended sediment concentration inversion model is as follows: S3.

1. Match the field-measured SSC with the Sentinel-2 inverted Rrs in a 12-hour time window and remove abnormal spectral data with Rrs median < 0 in the near-infrared band; S3.

2. Analyze the spectral response characteristics of SSC. Select a red / green wavelength combination for low concentrations and a near-infrared / red wavelength combination for high concentrations. Determine the optimal wavelength ratio through regression analysis. S3.

3. Based on the semi-empirical method, linear, exponential, and logarithmic algorithms are constructed for each band combination. The coefficient of determination r² is used as the indicator, and the model with the highest fit is selected as the final suspended sediment concentration inversion model.

5. The method for monitoring suspended sediment concentration in water based on FUI water classification according to claim 1, characterized in that: The specific steps for monitoring the suspended sediment concentration in water bodies using the suspended sediment concentration inversion model are as follows: S4.

1. Apply the constructed model to the preprocessed Sentinel-2 imagery, batch calculate the Rrs band ratios, and obtain the spatial distribution of water body SSC through inversion of the model formula; S4.

2. Compare the field sampling data from the same period, calculate the inversion errors for different water body types, analyze the model's ability to monitor the dynamic changes of SSC, and verify its applicability in the mainstream of the Yangtze River.

6. The method for monitoring suspended sediment concentration in water based on FUI water classification according to claim 2, characterized in that: In S1.2, the altitude data is obtained from the digital elevation data provided by the Space Shuttle Radar Topography Mission; the atmospheric correction parameters are output after calculation by the 6S model, and then the TOA image is rasterized to obtain , the calculation formula is as follows: in, is the pixel radiance of TOA image; 、 and is the correction coefficient calculated by the 6S model; is the surface reflectivity; is the downward direct irradiance of solar radiation, is the diffuse irradiance caused by aerosol and molecular scattering, both calculated by the 6S model; the sky spectral reflectance of the air-water interface Set to 0.0245.

7. The method for monitoring suspended sediment concentration in water based on FUI water classification according to claim 2, characterized in that: In S1.3, the Sentinel-2 technical documentation method is used to weaken the proximity effect by reducing the contrast between the pixels in the convolution kernel and the background pixels. The steps include: Set the convolution kernel window radius to 1 km and calculate the average remote sensing reflectance of each pixel within the convolution kernel. , correct the proximity effect according to equations (5) and (6): Where N is the number of pixels in the convolution kernel; is the diffuse transmittance from the sensor to the ground surface Direct transmittance The ratio is calculated by the 6S model; After correction for proximity effect .

8. The method for monitoring suspended sediment concentration in water based on FUI water classification according to claim 1, characterized in that: In S1.4, the SGI index is calculated using the red and shortwave infrared bands, and interfering pixels such as solar flares and whitecaps with an SGI ≥ -0.5 are removed. The steps are as follows: The quality control index is used to remove solar flares and whitecaps using red light and shortwave infrared bands. The calculation formula is as follows: Among them, B12 and B3 represent the TOA radiation of the twelfth and third bands, respectively; pixels with SGI less than -0.5 are considered to be free from sunlight reflection contamination; in order to analyze the accuracy of satellite-retrieved Rrs, the L2A product captured at the same time is used to evaluate the Rrs consistency with the L2A standard product.

9. A water suspended sediment concentration monitoring system based on FUI water body classification, characterized by: A method for monitoring suspended sediment concentration in water based on FUI water body classification according to any one of claims 1 to 8 is adopted.

10. A computer device, characterized in that: include: one or more processors; One or more execution programs are stored in the processor; When one or more execution programs are executed by the one or more processors, a method for monitoring suspended sediment concentration in water based on FUI water body classification as described in any one of claims 1 to 8 is implemented.

Images