Taihu lake cyanobacterial bloom extraction method based on sentinel No.2 image

Abstract

The invention relates to the technical field of environment monitoring, in particular to a Taihu lake cyanobacterial bloom extraction method based on sentinel No.2 images. The method comprises the following steps: S1, obtaining and preprocessing remote sensing data: correcting by using a Sen2Cor plug-in to obtain L2A-level data, and re-sampling waveband data to 10m through SNAP software; s2, water area contour extraction: according to the multispectral remote sensing data in the step S1, respectively calculating NDWI of each group of data, and according to a threshold value, completely excluding a land part to obtain a Taihu Lake water area range mask to be applied to subsequent cutting; and S3, contour extraction of cyanobacterial bloom: calculating the NDVI of each group of data, wherein the extraction threshold value of the NDVI is-0.0693. The Taihu Lake and cyanobacterial bloom images obtained according to the method have good consistency with field sampling data, and can be subsequently used for visual evaluation of the Taihu Lake cyanobacterial bloom.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of environmental monitoring, in particular to a method for extracting cyanobacterial blooms in Taihu Lake based on Sentinel-2 images. BACKGROUND

[0002] Sentinel-2 is a high-resolution multi-spectral imaging satellite developed by the European Space Agency (ESA), which has two satellites, 2A and 2B. Sentinel-2 multi-spectral images contain 13 spectral bands, with a width of about 290 kilometers and a spatial resolution of up to 10m. The complementary revisit period of the two satellites is 5 days, and in higher latitude areas it is only 3 days. In summary, Sentinel-2 has the advantages of high spatial resolution, short playback period, rich spectral bands, open source acquisition, etc.

[0003] Cyanobacteria have obvious differences in spectral reflectance from water bodies and other ground objects. Cyanobacteria contain photosynthetic pigments such as chlorophyll in their cells, which have specific absorption and reflection characteristics in the visible light band. In particular, there is a higher reflection peak in the green light band (500-600nm), which is caused by the composition of cyanobacterial pigments and cell structure. In the near-infrared band (700-1100nm), the reflectivity of cyanobacterial cells is significantly higher than that of normal water bodies due to structures such as bubbles in the cells, forming a distinct reflection platform. This unique spectral characteristic allows cyanobacteria to be distinguished from surrounding water bodies in remote sensing images.

[0004] Therefore, the prior art has used satellite remote sensing images for the monitoring of lake cyanobacterial blooms. For example, Chinese Patent Document CN114998724A proposes a cloud layer interference-resistant remote sensing monitoring method for lake cyanobacterial blooms based on a Landsat8 satellite, which includes the following steps: (1) constructing water body pixel, cyanobacterial bloom pixel, and cloud pixel sample sets; (2) calculating the normalization coefficients of the water body pixel, cyanobacterial bloom pixel, and cloud pixel; (3) constructing an index function CBI; (4) calculating a cyanobacterial bloom pixel threshold value; and (5) extracting cyanobacterial bloom pixels. This scheme solves the problems of cloud pixel misjudgment and time-consuming atmospheric correction by constructing an index function using a humidity component coefficient, and can improve monitoring accuracy and efficiency, achieving cloud layer interference-resistant remote sensing monitoring of cyanobacterial blooms. However, the Landsat8 satellite has a long playback cycle and cannot provide cyanobacterial bloom information in a timely manner during the outbreak period of cyanobacteria. Chinese Patent Document CN117115077A proposes a lake cyanobacterial bloom detection method based on a Sentinel2 satellite, which filters cloud-free satellite image pixels, uses a SWIR / NIR / Red band combination to construct an FUI pseudo-color index, performs histogram statistics on sample hue angles, establishes a threshold value for the hue angle a of aquatic plants and cyanobacterial blooms, and eliminates thin clouds, high-turbidity water bodies, and other phenomena that may cause "false positives" of cyanobacterial blooms, achieving rapid, accurate, and automated analysis of cyanobacterial blooms. The Sentinel2 satellite has high spatial resolution, a short playback cycle, and rich satellite spectral bands, which can effectively improve the accuracy and timeliness of cyanobacterial bloom identification. The method is used to monitor cyanobacterial blooms in Taihu Lake, and the monitoring results are consistent with those published by the Jiangsu Environmental Monitoring Center. SUMMARY

[0005] The present application aims to provide a method for extracting cyanobacterial blooms in Taihu Lake based on Sentinel-2 images, which uses a slope analysis method to determine the NDVI threshold value of cyanobacterial blooms in Taihu Lake, and the consistency of the extracted images based on the NDVI threshold value with the field sampling data is good.

[0006] To achieve the above-mentioned purpose, the technical solution of the present application is as follows: The method for extracting cyanobacterial blooms in Taihu Lake based on Sentinel-2 images includes the following steps: S1, obtaining and preprocessing remote sensing data: obtaining remote sensing images of Sentinel-2 and multispectral remote sensing data corresponding to the remote sensing images, and correcting the data using a Sen2Cor plug-in to obtain L2A level data, and then resampling the band data to 10m using SNAP software; S2, contour extraction of water area: calculating the NDWI of each group of data according to the multispectral remote sensing data in step S1, and completely excluding the land part according to the threshold value to obtain a mask of the water area range of Taihu Lake for application in subsequent cropping, wherein, or ; NDWI is a normalized water index, R Green is a green band reflectance, R NIR is a near-infrared band reflectance, R Band3 is a Sentinel 2 third band reflectance, R Band5 is a Sentinel 2 fifth band reflectance, R S3, profile extraction of cyanobacterial bloom: according to the multispectral remote sensing data in step S1, NDVI of each group of data is calculated respectively, and the extraction threshold of NDVI is-0.0693, or ; NDVI is a normalized vegetation index, R Green is a red band reflectance, R NIR is a near-infrared band reflectance, R Band8 is a Sentinel 2 eighth band reflectance, R Band4 is a Sentinel 2 fourth band reflectance.

[0007] As an improvement, in the step S1, the remote sensing image without cloud obstruction is selected.

[0008] As a further improvement, the multispectral remote sensing data is greater than 40 groups.

[0009] As a still further improvement, in the step S3, the extraction threshold of NDVI is obtained by the following method: S41, slope analysis is performed on the NDVI value by using the slope tool in the Arcgis software, and the natural breakpoint classification method is adopted to divide the slope into two categories of high slope and low slope; S42, the slope map is reclassified, and 1 and 0 values are assigned to the high slope and the low slope respectively, to obtain the NDVI value at the junction of the cyanobacterial bloom area and the non-cyanobacterial bloom area; S43, statistical analysis is performed on the NDVI value, the mean minus twice standard deviation method is used to obtain the cyanobacterial pixel extraction threshold, and the extraction threshold of all remote sensing images is averaged.

[0010] The method has the beneficial effect that the images of the Taihu Lake and the cyanobacterial bloom obtained by the method are consistent with the in-situ sampling data, and can be used for subsequent intuitive evaluation of the cyanobacterial bloom in the Taihu Lake. BRIEF DESCRIPTION OF DRAWINGS

[0011] Figure 1 is the extracted profile of the Taihu Lake water area; Figure 2 is the extraction effect of the cyanobacterial bloom in the Taihu Lake on a certain day. DETAILED DESCRIPTION

[0012] The application will be further described in connection with specific embodiments. It should be understood that these embodiments are only used to illustrate the application and not used to limit the scope of the application. Furthermore, it should be understood that those skilled in the art can make various modifications or changes to the application after reading the content of the application, and these equivalent forms also fall within the scope defined by the appended claims. Embodiments

[0013] In this study, Sentinel-2 multispectral remote sensing data was downloaded from the European Space Agency website (https: / / dataspace.copernicus.eu / ), with a time range of 2021-2023. L1C level data was processed using the Sen2Cor plug-in for atmospheric correction and other operations to obtain L2A level data, and then the band data was resampled to 10m using SNAP software. In addition, there should be 3-4 remote sensing images per month in the Taihu Basin in theory, but in actual research, due to local climate conditions, there are a large number of cloud clusters over the Taihu Lake in some months, making it impossible to analyze the remote sensing images, resulting in fewer available remote sensing images, which may cause incomplete analysis or poor results. However, this phenomenon is unavoidable in the selection of remote sensing images, so these images are still discarded.

[0014] Before extracting the cyanobacterial bloom in the Taihu Lake area, in order to prevent non-lake area features from interfering with the extraction results, the lake area to be studied needs to be accurately extracted in advance. A total of 44 images with no cloud cover in the lake area were selected, NDWI was calculated for each image, and the appropriate threshold was selected to completely exclude the land part.

[0015] In this study, the normalized water index (NDWI) was calculated to extract the water area.

[0016] NDWI (Normalized Difference Water Index, Normalized Difference Water Index), that is, the normalized difference processing of specific bands of remote sensing images. Through this processing method, the water information in the image can be highlighted significantly, so as to more accurately identify and extract the water area. Its calculation formula is: Or , where R Green is the green band reflectance, R NIR is the near-infrared band reflectance, R Band3 is the third band reflectance of Sentinel-2, and R Band5 is the fifth band reflectance of Sentinel-2.

[0017] After calculation, the Taihu Lake water area mask is obtained and applied to subsequent cropping. The water extraction effect is shown in Figure 1 .

[0018] NDVI (Normalized Difference Vegetation Index) is a remote sensing index that combines red and near-infrared bands and has high sensitivity to vegetation information, which can be used for lake cyanobacterial bloom monitoring. NDVI is calculated based on multispectral data, and the calculation formula is: or . Wherein, RGreen is the reflectivity of the red band, RNIR is the reflectivity of the near-infrared band, RBand8 is the reflectivity of the 8th band of Sentinel-2, and RBand4 is the reflectivity of the 4th band of Sentinel-2. The basic information of each band of Sentinel-2 is common knowledge, and the specific information is shown in Table 1:

[0019]

[0020] In view of the fact that the spectral characteristics of cyanobacterial bloom pixels and non-cyanobacterial bloom pixels differ significantly at their boundaries, the NDVI slope can be used to determine the cyanobacterial extraction threshold. The slope tool in Arcgis software is used to analyze the slope of the NDVI value, and the natural breakpoint classification method is used to classify the slope into high slope and low slope. The slope map is reclassified, and the high slope and low slope are assigned values of 1 and 0, respectively, to obtain the NDVI value at the boundary between the cyanobacterial bloom area and the non-cyanobacterial bloom area. In this part of the data, the determined cyanobacterial bloom pixels are excluded with the condition of NDVI>0.2, and the non-cyanobacterial bloom pixels such as water are excluded with the condition of NDVI<0. The remaining pixels are screened, and the NDVI values of the remaining pixels are statistically analyzed. The mean minus twice the standard deviation method is used to obtain the cyanobacterial pixel extraction threshold for a single image, and the extraction thresholds for all images are averaged to obtain the NDVI extraction uniform threshold based on Sentinel-2 remote sensing images, which is -0.0693. The cyanobacterial bloom extraction threshold for a single image is shown in Table 2:

[0021] Date NDVI threshold Date NDVI threshold Date NDVI threshold 2021-01-08 -0.0789 2022-03-04 -0.0614 2023-01-18 -0.0660 2021-01-13 -0.0754 2022-03-09 -0.0695 2023-01-28 -0.0759 2021-01-18 -0.0750 2022-03-24 -0.0706 2023-03-04 -0.0664 2021-02-07 -0.0654 2022-03-29 -0.0742 2023-03-09 -0.0642 2021-05-03 -0.0645 2022-04-03 -0.0706 2023-03-14 -0.0687 2021-06-22 -0.0677 2022-04-08 -0.0654 2023-04-08 -0.0739 2021-10-05 -0.0699 2022-05-03 -0.0576 2023-05-23 -0.0698 2021-11-09 -0.0648 2022-07-27 -0.0679 2023-05-28 -0.0712 2021-11-14 -0.0661 2022-08-11 -0.0576 2023-10-15 -0.0692 2021-11-24 -0.0671 2022-10-10 -0.0693 2023-10-30 -0.0743 2021-12-04 -0.0647 2022-12-14 -0.0639 2023-11-14 -0.0660 2021-12-19 -0.0651 2022-12-19 -0.0750 2023-11-19 -0.0696 2021-12-29 -0.0774 2022-12-24 -0.0788 2023-11-24 -0.0660 2022-01-03 -0.0634 2023-01-03 -0.0718 2023-12-24 -0.0784 2022-02-27 -0.0714 2023-01-08 -0.0795

[0022] Taking the image of June 22, 2021 as an example, the extraction effect is shown in Figure 2 . Compared with the visual interpretation of Taihu water, the cyanobacterial bloom extraction effect is good and basically conforms to the distribution. By querying the real-time water sample values in the Taihu Basin, the consistency between the cyanobacterial particle concentration and chlorophyll concentration data and the results obtained by the method is good.

Claims

1. A method for extracting cyanobacterial blooms in Taihu Lake based on Sentinel-2 images, characterized in that, The method comprises the following steps: S1, obtaining and preprocessing remote sensing data: obtaining remote sensing images of Sentinel-2 and corresponding multispectral remote sensing data, correcting the remote sensing images by using a Sen2Cor plug-in to obtain L2A level data, and resampling band data to 10 m by using SNAP software; S2, water area contour extraction: according to the multispectral remote sensing data in step S1, the NDWI of each group of data is calculated respectively, and the land part is completely excluded according to the threshold value to obtain a mask of the water area range of Taihu Lake to be applied to subsequent cropping, wherein, or ; NDWI is a normalized water index, R Green is a green band reflectance, R NIR is a near-infrared band reflectance, R Band3 is a third band reflectance of Sentinel 2, and R Band5 is a fifth band reflectance of Sentinel 2. S3, Contour extraction of cyanobacterial bloom: According to the multispectral remote sensing data in step S1, the NDVI of each group of data is calculated respectively, and the extraction threshold of NDVI is -0.0693, or ; NDVI is the normalized vegetation index, R Green is the red band reflectivity, R NIR is the near-infrared band reflectivity, R Band8 is the reflectivity of the 8th band of Sentinel 2, R Band4 is the reflectivity of the 4th band of Sentinel 2. 2.The method for extracting Taihu cyanobacterial bloom based on Sentinel-2 image according to claim 1, characterized in that, In the step S1, remote sensing images without cloud obstruction are selected. 3.The method of claim 2, wherein, The multispectral remote sensing data are greater than 40 groups. 4.The method of claim 1, wherein, In the step S3, the extraction threshold of NDVI is obtained by the following method: S41, slope analysis is performed on the NDVI value by using a slope tool in Arcgis software, and the slope is divided into high slope and low slope by using a natural breakpoint classification method; S42, the slope map is reclassified, and 1 and 0 values are respectively assigned to the high slope and the low slope to obtain the NDVI value at the junction of the cyanobacterial bloom region and the non-cyanobacterial bloom region; S43, statistical analysis is performed on the NDVI value, the mean value minus double standard deviation method is used to obtain the cyanobacterial pixel extraction threshold, and average processing is performed on the extraction threshold of all remote sensing images.

Images