A method for extracting submerged vegetation and algal blooms in rivers and lakes based on sentinel-2 images

Abstract

The application provides a method for extracting submerged vegetation and algal blooms in a river-connected lake based on a Sentinel-2 image, and the method comprises the following steps: acquiring Sentinel-2 image data and performing image preprocessing; acquiring MNDWI, AVI, NDVI, BI, FAI and SABI index images respectively; determining the classification threshold value in the AVI image by using a linear mixing model; determining the classification threshold value in the FAI image and the SABI image by using a maximum gradient histogram method respectively; and taking an empirical threshold value for the remaining classification threshold values; constructing a decision tree classification model based on different index images and corresponding threshold values, identifying water bodies and non-water bodies first, and then performing pixel-by-pixel discrimination on the targets in the water bodies. The method fully considers the characteristics of the river-connected lake, can identify the spatial distribution of the submerged vegetation and the algal blooms in the river-connected lake on a large scale, and has the advantages of a clear classification process, a wide application range and high monitoring efficiency.

Description

Technical Field

[0001] This invention relates to the field of remote sensing application technology, and in particular to a method for extracting submerged vegetation and algal blooms from Tongjiang lakes based on Sentinel-2 imagery. Background Technology

[0002] Connecting lakes are an important type of water body formed by the interconnection of rivers and lakes, characterized by complex hydrological processes, frequent shoreline changes, and significant spatial and temporal variations in water transparency and suspended solids concentration. Under eutrophic conditions, the rise and fall of submerged vegetation and algal blooms in connecting lakes often have a significant impact on the structure and function of the lake ecosystem. Submerged vegetation is an important ecological factor for maintaining lake ecological balance, improving water quality, and stabilizing bottom sediments, while large-scale algal blooms can lead to water quality deterioration, decreased dissolved oxygen, and ecosystem degradation. Therefore, rapidly and accurately identifying the spatial distribution of submerged vegetation and algal blooms in connecting lakes is of great significance for lake ecological monitoring, eutrophication control, ecological restoration, and analysis of vegetation-algae conversion processes.

[0003] Currently, the main methods for extracting submerged vegetation and algal blooms in lakes include field surveys, paleolynological methods, and remote sensing monitoring. Although field surveys and paleolynological methods have high accuracy, they suffer from problems such as large workload, poor timeliness, and limited coverage. Remote sensing technology provides strong support for monitoring the succession of submerged vegetation and algae. In lakes connected to the Yangtze River, the challenges in identifying submerged vegetation and algal blooms are: (1) Lakes connected to the Yangtze River are significantly affected by factors such as water level fluctuations, sediment input, changes in the water-land transition zone, and exposure of shallow waters, which easily leads to spectral confusion between shallow waters, submerged vegetation, and algal blooms; (2) During the high-water season, when the water level rises, emergent / floating-leaved vegetation becomes difficult to distinguish from submerged vegetation after being submerged. Summary of the Invention

[0004] The purpose of this invention is to provide a method for extracting submerged vegetation and algal blooms in Tongjiang lakes based on Sentinel-2 imagery. By identifying the spatial distribution of submerged vegetation and algal blooms in Tongjiang lakes over a wide range, the method can improve the applicability and monitoring efficiency.

[0005] To achieve the above objectives, the present invention provides the following solution: A method for extracting submerged vegetation and algal blooms in Tongjiang lakes based on Sentinel-2 imagery includes the following steps: Sentinel-2 satellite surface reflectance image data of the study area were acquired and preprocessed to obtain preprocessed image data. The preprocessed image data includes: normalized difference water index image, aquatic vegetation index image, normalized vegetation index image, brightness image, phytoplankton index image and shoal-algal bloom discrimination index image. The first classification threshold is calculated based on the normalized differential water index image and the sample kernel density map; Based on aquatic vegetation index images, normalized vegetation index images, and brightness images, the second, third, and fourth classification thresholds are calculated according to a linear mixture model. Based on the phytoplankton index image and the shallow water-algal bloom discrimination index image, the fifth and sixth classification thresholds were calculated using the maximum gradient histogram method. A decision tree classification model is constructed based on the first, second, third, fourth, fifth, and sixth classification thresholds. The preprocessed image data is then conditionally discriminated using the decision tree classification model to obtain the classification results.

[0006] Optionally, preprocessing includes: boundary clipping, declouding masking, and reflectivity scaling.

[0007] Optionally, the first classification threshold is used to distinguish between water areas and non-water areas, and the formula for calculating the first classification threshold is: ;in, For green band reflectivity, This refers to the reflectivity in the near-infrared band.

[0008] Optionally, the second and third classification thresholds are used to distinguish between aquatic vegetation areas and non-aquatic vegetation areas in water bodies; the fourth classification threshold is used to distinguish between submerged vegetation areas; the formula for calculating the second classification threshold is: ,in, The coefficients are linear. and These are the average AVI values ​​of pixels in pure water and pixels in areas with dense aquatic vegetation, respectively; the formula for calculating the third classification threshold is: ,in, For near-infrared reflectivity, The formula for calculating the fourth classification threshold is: ,in, This refers to the reflectivity in the green band.

[0009] Optionally, based on the phytoplankton index image and the shoal-bloom discrimination index image, the fifth and sixth classification thresholds are calculated using the maximum gradient histogram method, including: Different gradient images are generated based on the phytoplankton index image and the shoal-bloom discrimination index image; the gradient of the gradient image is the difference between the pixel and its neighboring pixel in a 3×3 window; Based on the sample histogram distribution, sparse or dense regions in the gradient image are removed to obtain the intermediate boundary region pixels. A histogram is generated based on the gradient image of the pixels in the intermediate boundary region. The average value of all pixels corresponding to the mode of the maximum gradient in the histogram in the phytoplankton index image or the shoal-algal bloom discrimination index image is used as the fifth classification threshold or the sixth classification threshold. The fifth classification threshold is used to distinguish between shoal algal bloom areas and open water areas in non-aquatic vegetation areas, and the sixth classification threshold is used to distinguish between shoals and algal blooms in shoal algal bloom areas.

[0010] Optionally, a decision tree classification model is constructed based on a first classification threshold, a second classification threshold, a third classification threshold, a fourth classification threshold, a fifth classification threshold, and a sixth classification threshold. The preprocessed image data is then conditionally judged using the decision tree classification model to obtain classification results, including: The study area was divided into water bodies and non-water bodies based on the first classification threshold. Water bodies are classified into aquatic vegetation and non-aquatic vegetation based on the second and third classification thresholds. Aquatic vegetation is divided into submerged vegetation and shallow water vegetation based on the fourth classification threshold. Based on the fifth classification threshold, non-aquatic vegetation is divided into shallow water algal blooms and open water. Based on the sixth classification threshold, shallow algal blooms are divided into algal blooms and shallow waters.

[0011] Optionally, a decision tree classification model is constructed based on the first, second, third, fourth, fifth, and sixth classification thresholds. The preprocessed image data is then conditionally judged using the decision tree classification model to obtain the classification results. The method also includes calculating the overall classification accuracy and Kappa coefficient of the classification results. The formula for calculating the overall classification accuracy is: ,in, The total number of samples, The total number of correctly classified samples; the formula for calculating the Kappa coefficient is: ,in, The total number of categories, For the i-th type of diagonal element in the confusion matrix, and The first i Line number i The total number of pixels in the column, where N is the total number of samples.

[0012] According to specific embodiments provided by the present invention, the following technical effects are disclosed: The method for extracting submerged vegetation and algal blooms in Tongjiang lakes based on Sentinel-2 imagery provided by the present invention includes: acquiring Sentinel-2 satellite surface reflectance image data of the study area and performing preprocessing to obtain preprocessed image data; the preprocessed image data includes: normalized difference water index image, aquatic vegetation index image, normalized vegetation index image, brightness image, phytoplankton index image, and shoal-algal bloom discrimination index image; based on the normalized difference water index image and sample kernel density... The first classification threshold is calculated using the image. Based on the aquatic vegetation index image, normalized difference vegetation index image, and brightness image, the second, third, and fourth classification thresholds are calculated using a linear mixture model. Based on the phytoplankton index image and the shoal-algal bloom discrimination index image, the fifth and sixth classification thresholds are calculated using the maximum gradient histogram method. A decision tree classification model is constructed based on the first, second, third, fourth, fifth, and sixth classification thresholds, and the preprocessed image data is conditionally discriminated using the decision tree classification model to obtain the classification results. This method fully considers the characteristics of connected lakes and can identify submerged vegetation and algal blooms in real time, over a large area, and with high accuracy. It has important practical significance for reconstructing the grass-algae conversion process and for lake management. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 This is a flowchart of the method for extracting submerged vegetation and algal blooms from Tongjiang Lakes according to the present invention; Figure 2 This is a schematic diagram illustrating the working principle of the extraction process of submerged vegetation and algal blooms from Tongjiang Lake according to an embodiment of the present invention. Figure 3 These are MNDWI kernel density maps of various types of land features in embodiments of the present invention; Figure 4 This is a histogram of MNDWI frequencies in water bodies according to an embodiment of the present invention. Figure 5 This is an AVI frequency histogram of submerged vegetation according to an embodiment of the present invention. Figure 6 This is a kernel density diagram of submerged vegetation and shallow waters according to an embodiment of the present invention; Figure 7 This is a remote sensing extraction result image of submerged vegetation and algal blooms in Poyang Lake according to an embodiment of the present invention. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0017] This invention uses Poyang Lake, the largest typical lake connected to the Yangtze River in a certain country, as an example. The data on submerged vegetation and algal blooms in this example are derived from previous research and field surveys. Algal blooms in Poyang Lake are mainly distributed in areas with relatively calm water and high nutrient levels, often forming aggregations in local bays, nearshore waters, and water stagnation zones. Submerged vegetation is mostly distributed in shallower, more transparent, and more stable saucer-shaped lake areas. Figure 1 and Figure 2 As shown, this embodiment provides a method for extracting submerged vegetation and algal blooms in Tongjiang lakes based on Sentinel-2 imagery, including the following steps: Step 100: Acquire Sentinel-2 satellite surface reflectance image data of the study area and perform preprocessing to obtain preprocessed image data; the preprocessed image data includes: normalized difference water index image, aquatic vegetation index image, normalized vegetation index image, brightness image, phytoplankton index image and shoal-algal bloom discrimination index image. Step 200: Calculate the first classification threshold based on the normalized differential water index image and the sample kernel density map; Step 300: Based on the aquatic vegetation index image, normalized vegetation index image, and brightness image, calculate the second classification threshold, the third classification threshold, and the fourth classification threshold according to the linear mixture model; Step 400: Based on the phytoplankton index image and the shoal-bloom discrimination index image, calculate the fifth and sixth classification thresholds using the maximum gradient histogram method; Step 500: Construct a decision tree classification model based on the first, second, third, fourth, fifth, and sixth classification thresholds, and use the decision tree classification model to perform conditional discrimination on the preprocessed image data to obtain the classification results.

[0018] In the specific implementation process, step 100 downloads the Sentinel-2 satellite surface reflectance image data (COPERNICUS / S2_SR_HARMONIZED) from the Google Earth Engine cloud platform on October 23, 2023, which is synchronized with the sample point acquisition time and covers the Poyang Lake area. Then, using the uploaded Poyang Lake vector boundary cropping image, the cropping image is de-clouded and masked through the QA60 band. Finally, reflectance scaling is performed with a scaling factor of 0.0001 to achieve image data preprocessing.

[0019] In the specific implementation process, step 200 acquires the improved normalized difference water index (MNDWI) image from October 23, 2023, and, in conjunction with previous research and sample kernel density maps, determines the classification threshold T. MNDWI And it is used to distinguish between water areas and non-water areas. For example... Figure 3 As shown, the formula for calculating MNDWI is: ;in, For green band reflectivity, For near-infrared reflectance, in this embodiment, T MNDWI The preferred value is 0.2.

[0020] In the specific implementation process, step 300 acquires the aquatic vegetation index (AVI), normalized difference vegetation index (NDVI), and brightness (BI) images from October 23, 2023, to determine the classification threshold T. AVI T NDVI And T BI T AVI and T NDVI Used to distinguish between aquatic vegetation areas and non-aquatic vegetation areas in aquatic areas, T BI Used to remove shallow areas and obtain submerged vegetation zones. Specifically, the AVI index is calculated using the following formula: Wherein, "Wetness" is the humidity index, obtained by weighting the spectrum across different wavelengths. The formula for calculating the humidity index is: ; in, For blue band reflectivity, For green band reflectivity, For red band reflectivity, For near-infrared reflectivity, For shortwave infrared band 1 reflectivity, T represents the reflectivity of the shortwave infrared band 2. AVI The calculation formula is obtained from the linear mixture model: ; Among them, T AVIThe threshold for AVI, and These are the average AVI values ​​of pixels in pure water and pixels in areas with dense aquatic vegetation, respectively. The linear coefficient is preferably 70% in this embodiment. For example... Figure 4 As shown, in this embodiment, the MNDWI image of the scene is acquired, and the MNDWI value corresponding to the highest peak in the MNDWI histogram is used as the threshold for extracting pure water, which is calculated to be 0.935. All pixels with MNDWI values ​​greater than 0.754 are extracted as pure water pixels, and AVI(w) is calculated to be -0.0669. Figure 5 As shown, pixels in areas with dense aquatic vegetation growth are segmented through visual interpretation. Based on the histogram calculation, AVI(v) = -0.0085 is obtained, and T is thus calculated. AVI =-0.0494.

[0021] The formula for calculating NDVI is: .like Figure 6 As shown, to further exclude shallow water areas and obtain pixels of pure submerged vegetation, the brightness BI is calculated using the following formula: Based on the kernel density maps of shallow and submerged vegetation pixels and previous studies, this embodiment uses T... BI =7.5%.

[0022] It should be noted that BI was introduced into the vegetation sample candidate region to perform secondary separation of submerged vegetation and shoals at the brightness level, overcoming the shortcomings of existing methods that rely solely on vegetation indices and are difficult to separate highly reflective shoals.

[0023] In the specific implementation process, step 400 calculates the phytoplankton index (FAI) and the shallow water-bloom discriminant index (SABI) based on image data, and determines T using the maximum gradient histogram. FAI T SABI Specifically, the formulas for calculating FAI and SABI are as follows: ; ; ; in, The center wavelength of the near-infrared band is 835.1 micrometers in this embodiment; The center wavelength of the red band is 664.5 micrometers in this embodiment; The center wavelength of the shortwave infrared band 1 is 1613.7 micrometers in this embodiment.

[0024] Next, the maximum gradient histogram method is used to calculate and determine T. FAI T SABI ,include: 1) Generate the gradient image of the exponential image and define the gradient of a pixel as the difference between it and its neighboring pixels in a 3×3 window; 2) Based on the histogram distribution of the samples, remove overly sparse or dense areas, and retain only the pixels in the intermediate boundary areas; 3) Generate a histogram from the gradient images of the remaining pixels (i.e., pixels in the intermediate boundary region), determine the mode of its maximum gradient, and use the average value of all pixels corresponding to this gradient mode in the exponential image as the classification threshold. In this embodiment, T is preferred. FAI =0.02, T SABI =0.0167.

[0025] It should be noted that SABI, as a novel indicator proposed in this invention, is introduced into the high FAI candidate region and used to further distinguish between real algal blooms and high-reflectivity shoals, thus overcoming the problem of shoal misjudgment caused by the direct identification of algal blooms by a single FAI threshold in existing methods.

[0026] like Figure 7 As shown, in the specific implementation process, step 500 is based on T MNDWI T AVI T NDVI T BI T FAI And T SABI Determine the discrimination criteria and build a decision tree classification model. The model construction conditions include: Condition 1: The study area is divided into water bodies and non-water bodies based on the MNDWI index. If the MNDWI of the study area pixels is greater than 0.2, it is classified as a water body; otherwise, it is classified as a non-water body. Condition 2: For water bodies, classification is performed using the AVI and NDVI indices. When AVI > -0.0494 and NDVI > -0.02, the body is classified as aquatic vegetation; otherwise, it is classified as non-aquatic vegetation. Condition 3: For aquatic vegetation, the BI index is used for classification. When BI < 7.5%, it is classified as submerged vegetation; otherwise, it is classified as shallow water. Condition 4: For non-aquatic vegetation, the FAI index is used for classification. When FAI > 0.02, it is classified as shallow water algal bloom; otherwise, it is classified as open water. Condition 5: For algal blooms in shallow waters, the SABI index is used for classification. When SABI > 0.0167, it is classified as an algal bloom; otherwise, it is classified as a shallow water.

[0027] It should be noted that by introducing a dual judgment mechanism of NDVI and AVI, misjudgment of mixed pixels by a single index is avoided, and the accuracy of submerged vegetation candidate area extraction is improved. Furthermore, BI and the newly proposed SABI are used to achieve secondary separation of "submerged vegetation-shoals" and "algal bloom-shoals" respectively. Compared with the traditional AVI index, which has insufficient judgment power, is easily affected by suspended sediment, and has the problem that the AVI of water areas may be greater than that of submerged vegetation, this method fully considers the characteristics of water level fluctuations, sediment input, changes in the water-land transition zone, and exposed shoals in connected lakes. This effectively overcomes the problem of misjudgment of mixed pixels and high-reflectivity shoals in single index classification, significantly improving the accuracy and anti-interference ability of submerged vegetation and algal bloom identification in connected lakes.

[0028] Furthermore, this embodiment also plots the classification results from step 500 in ArcGIS, and calculates the overall classification accuracy (OA) and Kappa coefficient based on the measured sample points and classification results. The calculation formulas are as follows: ; ; in, The total number of samples, The total number of samples that were correctly classified. The total number of categories, For the i-th type of diagonal element in the confusion matrix, and The first i Line number i The total number of pixels in the column, N is the total number of samples. The classification results are evaluated by the overall classification accuracy and Kappa coefficient, as shown in Table 1. The validation set is derived from balanced stratified random sampling, combined with Google high-resolution maps and field surveys, totaling 240 sample points.

[0029] Table 1. Accuracy Evaluation of Automatic Classification and Mapping Results of Submerged Vegetation and Algal Blooms in Tongjiang Lakes

[0030] It should be noted that, through point-to-point verification of the classification results and sample points, it can be seen that the method of the present invention has high identification accuracy for land cover types such as submerged vegetation and algal blooms, with an overall classification accuracy of 90.4% and a Kappa coefficient of 0.872, which can meet the needs of ecological environment monitoring and classification mapping of lakes connected to the Yangtze River.

[0031] The beneficial effects of this invention are as follows: 1) This invention relies on the Google Earth Engine cloud platform to carry out image preprocessing, index calculation and classification mapping. It has the advantages of high efficiency, wide range and strong repeatability. It is suitable for large-scale and long-term lake ecological monitoring and can provide effective technical support for algal bloom early warning, aquatic vegetation dynamic analysis and lake ecological restoration management. 2) By comprehensively utilizing multiple remote sensing indices such as MNDWI, AVI, NDVI, BI, FAI, and SABI, a multi-indicator collaborative classification system for complex water environments in lakes connected to the Yangtze River was constructed, which effectively improved the ability to distinguish between submerged vegetation, algal blooms, open water areas, and shallow waters. 3) The key classification thresholds were determined by using the linear mixture model and the maximum gradient histogram method respectively, which enabled the automatic extraction of thresholds for targets such as vegetation-non-vegetation and algal bloom-shoal, reducing the subjectivity of traditional manual experience-based threshold selection and improving the stability and reliability of classification results. 4) A decision tree classification model was constructed based on multiple indices and their thresholds. The model is classified step by step according to the process of water body identification, vegetation discrimination, shoal removal and algal bloom extraction. The classification logic is clear and the degree of automation is high. It can realize the rapid and accurate extraction and mapping of target features in lakes and rivers.

[0032] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0033] Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this invention. Furthermore, those skilled in the art will recognize that, based on the ideas of this invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. A method for extracting submerged vegetation and algal blooms in Tongjiang lakes based on Sentinel-2 imagery, characterized in that, Includes the following steps: Sentinel-2 satellite surface reflectance image data of the study area was acquired and preprocessed to obtain preprocessed image data; the preprocessed image data includes: normalized difference water index image, aquatic vegetation index image, normalized vegetation index image, brightness image, phytoplankton index image and shoal-algal bloom discrimination index image. The first classification threshold is calculated based on the normalized differential water index image and the sample kernel density map; Based on the aquatic vegetation index image, the normalized vegetation index image, and the brightness image, the second classification threshold, the third classification threshold, and the fourth classification threshold are calculated according to the linear mixture model. Based on the phytoplankton index image and the shallow water-algal bloom discrimination index image, the fifth classification threshold and the sixth classification threshold are calculated according to the maximum gradient histogram method. A decision tree classification model is constructed based on the first classification threshold, the second classification threshold, the third classification threshold, the fourth classification threshold, the fifth classification threshold, and the sixth classification threshold. The preprocessed image data is then conditionally judged using the decision tree classification model to obtain the classification result.

2. The method for extracting submerged vegetation and algal blooms in Tongjiang lakes based on Sentinel-2 imagery according to claim 1, characterized in that, The preprocessing includes: boundary clipping, declouding masking, and reflectivity scaling.

3. The method for extracting submerged vegetation and algal blooms in Tongjiang lakes based on Sentinel-2 imagery according to claim 1, characterized in that, The first classification threshold is used to distinguish between water areas and non-water areas, and the formula for calculating the first classification threshold is: ;in, For green band reflectivity, This refers to the reflectivity in the near-infrared band.

4. The method for extracting submerged vegetation and algal blooms in Tongjiang lakes based on Sentinel-2 imagery according to claim 1, characterized in that, The second and third classification thresholds are used to distinguish between aquatic vegetation areas and non-aquatic vegetation areas in water bodies; the fourth classification threshold is used to distinguish between submerged vegetation areas; the calculation formula for the second classification threshold is: ,in, The coefficients are linear. and These are the average AVI values ​​of pixels in pure water and pixels in areas with dense aquatic vegetation, respectively; the formula for calculating the third classification threshold is: ,in, For near-infrared reflectivity, The formula for calculating the fourth classification threshold is: ,in, This refers to the reflectivity in the green band.

5. The method for extracting submerged vegetation and algal blooms in Tongjiang lakes based on Sentinel-2 imagery according to claim 1, characterized in that, Based on the phytoplankton index image and the shallow water-algal bloom discrimination index image, the fifth and sixth classification thresholds are calculated using the maximum gradient histogram method, including: Different gradient images are generated based on the phytoplankton index image and the shallow water-algal bloom discrimination index image, respectively; the gradient of the gradient image is the difference between the pixel and its neighboring pixel in a 3×3 window; Based on the sample histogram distribution, sparse or dense regions in the gradient image are removed to obtain intermediate boundary region pixels; A histogram is generated based on the gradient image of the pixels in the intermediate boundary region, and the average value of all pixels corresponding to the mode of the maximum gradient in the histogram in the phytoplankton index image or the shoal-algal bloom discrimination index image is used as the fifth classification threshold or the sixth classification threshold. The fifth classification threshold is used to distinguish between shoal algal bloom areas and open water areas in non-aquatic vegetation areas, and the sixth classification threshold is used to distinguish between shoals and algal blooms in shoal algal bloom areas.

6. The method for extracting submerged vegetation and algal blooms in Tongjiang lakes based on Sentinel-2 imagery according to claim 1, characterized in that, A decision tree classification model is constructed based on the first classification threshold, the second classification threshold, the third classification threshold, the fourth classification threshold, the fifth classification threshold, and the sixth classification threshold. The preprocessed image data is then conditionally judged using the decision tree classification model to obtain classification results, including: The study area is divided into water bodies and non-water bodies based on the first classification threshold. The water body is divided into aquatic vegetation and non-aquatic vegetation according to the second classification threshold and the third classification threshold. The aquatic vegetation is divided into submerged vegetation and shallow water vegetation according to the fourth classification threshold. Based on the fifth classification threshold, the non-aquatic vegetation is divided into shallow algal blooms and open waters; The shallow algal blooms are classified into algal blooms and shallow waters based on the sixth classification threshold.

7. The method for extracting submerged vegetation and algal blooms in Tongjiang lakes based on Sentinel-2 imagery according to claim 1, characterized in that, A decision tree classification model is constructed based on the first, second, third, fourth, fifth, and sixth classification thresholds. The preprocessed image data is then conditionally classified using this model to obtain classification results. The method further includes calculating the overall classification accuracy and Kappa coefficient of the classification results. The formula for calculating the overall classification accuracy is: ,in, The total number of samples, The total number of correctly classified samples; the formula for calculating the Kappa coefficient is: ,in, The total number of categories, For the i-th type of diagonal element in the confusion matrix, and The first i Line 1 i The total number of pixels in the column, where N is the total number of samples.

Images