A cyanobacterial bloom monitoring method and system

Abstract

The application provides a cyanobacterial bloom monitoring method and system, comprising: acquiring a target water body remote sensing image of a target water body region of a monitoring area; screening a chlorophyll a sensitive band of the target water body remote sensing image, obtaining a chlorophyll a inversion concentration based on a preset chlorophyll a concentration inversion model; calculating a normalized vegetation index according to a multi-band remote sensing reflectivity value of the target water body remote sensing image; constructing a cyanobacterial bloom grading threshold dynamic calibration model according to the chlorophyll a inversion concentration and the normalized vegetation index, and grading a cyanobacterial bloom risk of the target water body region; and generating a warning instruction corresponding to a risk level according to a cyanobacterial bloom risk grading result of the target water body region. The cyanobacterial bloom grading threshold can be dynamically adjusted, thereby eliminating the subjective bias of artificial empowerment, and effectively improving the real-time performance, accuracy and efficiency of cyanobacterial bloom monitoring and warning.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of geographic information, in particular to a cyanobacterial bloom monitoring method and system. BACKGROUND

[0002] Cyanobacterial bloom refers to the phenomenon of large-scale proliferation of cyanobacteria under suitable environmental conditions, which causes serious damage to aquatic ecosystems and even poses a threat to human health. With global climate change and increasing human activities, the frequency and intensity of cyanobacterial blooms have increased, making effective monitoring and early warning of cyanobacteria of great significance.

[0003] Currently, the monitoring and early warning of cyanobacterial blooms mainly rely on two technical paths: mathematical model method and analytic hierarchy process method. The mathematical model method simulates the evolution process of algal blooms through a multi-parameter coupled dynamic model (such as a three-dimensional response model of nutrients-light-water temperature), and then constructs an evaluation system through the analytic hierarchy process method. However, the determination of index weights in this method relies too much on expert experience and is easily influenced by subjective cognitive bias, resulting in significant differences in risk level classification of the same water body by different research teams, which leads to low efficiency and low accuracy in monitoring and early warning of cyanobacterial blooms. SUMMARY

[0004] The present application provides a cyanobacterial bloom monitoring method and system that can effectively improve the efficiency and accuracy of cyanobacterial bloom monitoring and early warning by utilizing the correlation between chlorophyll-a concentration and normalized difference vegetation index to dynamically adjust the grading threshold of cyanobacterial bloom.

[0005] A cyanobacterial bloom monitoring method, comprising:

[0006] obtaining a remote sensing image of a monitoring area and extracting multi-band remote sensing reflectance values of the remote sensing image;

[0007] extracting a target water body remote sensing image containing only the target water body area based on a preset water body classification model according to the multi-band remote sensing reflectance values of the remote sensing image;

[0008] conducting correlation analysis on the multi-band remote sensing reflectance values of the target water body remote sensing image and the chlorophyll-a concentration to screen the chlorophyll-a sensitive band of the target water body remote sensing image;

[0009] obtaining the chlorophyll-a inversion concentration based on a preset chlorophyll-a concentration inversion model according to the remote sensing reflectance values of the chlorophyll-a sensitive band;

[0010] analyzing the vegetation growth of the target water body based on a preset normalized difference vegetation index calculation formula according to the multi-band remote sensing reflectance values of the target water body remote sensing image to obtain the normalized difference vegetation index;

[0011] According to the chlorophyll-a inversion concentration and the normalized vegetation index, a blue-green algae bloom grading threshold dynamic calibration model is constructed, and a blue-green algae bloom risk grading of the target water body region is performed, including:

[0012] The target water body region is randomly collected to obtain a plurality of chlorophyll-a inversion concentrations and normalized vegetation indexes;

[0013] The normalized vegetation index and the chlorophyll-a inversion concentration are analyzed for correlation, and if the normalized vegetation index and the chlorophyll-a inversion concentration are linearly correlated, a linear model of the normalized vegetation index and the chlorophyll-a inversion concentration is established as the blue-green algae bloom grading threshold dynamic calibration model;

[0014] A preset chlorophyll-a concentration grading threshold is obtained, and the chlorophyll-a concentration grading threshold is input into the blue-green algae bloom grading threshold dynamic calibration model to obtain the normalized vegetation index grading threshold;

[0015] According to the chlorophyll-a inversion concentration and the chlorophyll-a concentration grading threshold, and the normalized vegetation index and the normalized vegetation index grading threshold, the target water body region is graded for blue-green algae bloom risk;

[0016] According to the blue-green algae bloom risk grading of the target water body region, a warning instruction is issued to a blue-green algae bloom early warning device.

[0017] The application also provides a blue-green algae bloom monitoring system, including:

[0018] A remote sensing image acquisition module is configured to acquire remote sensing images of a monitoring region and extract multi-band remote sensing reflectivity values of the remote sensing images;

[0019] A target water body remote sensing image extraction module is configured to extract target water body remote sensing images containing only target water body regions based on a preset water body classification model according to the multi-band remote sensing reflectivity values of the remote sensing images;

[0020] A sensitive band screening module is configured to screen chlorophyll-a sensitive bands of the target water body remote sensing images by performing correlation analysis on the multi-band remote sensing reflectivity values of the target water body remote sensing images and chlorophyll-a concentrations;

[0021] A chlorophyll-a concentration inversion module is configured to obtain chlorophyll-a inversion concentrations based on a preset chlorophyll-a concentration inversion model according to remote sensing reflectivity values of the chlorophyll-a sensitive bands;

[0022] The normalized vegetation index calculation module is configured to analyze the vegetation growth of the target water body based on a preset normalized vegetation index calculation formula according to the multi-band remote sensing reflectivity value of the target water body remote sensing image, and obtain a normalized vegetation index.

[0023] The target water body region grading module is configured to construct a cyanobacterial bloom grading threshold dynamic calibration model according to the chlorophyll-a inversion concentration and the normalized vegetation index, and grade the cyanobacterial bloom risk of the target water body region, including:

[0024] The target water body region is randomly collected to obtain a plurality of chlorophyll-a inversion concentrations and normalized vegetation indexes.

[0025] The correlation between the normalized vegetation index and the chlorophyll-a inversion concentration is analyzed. If the normalized vegetation index and the chlorophyll-a inversion concentration are linearly correlated, a linear model of the normalized vegetation index and the chlorophyll-a inversion concentration is established as the cyanobacterial bloom grading threshold dynamic calibration model.

[0026] A preset chlorophyll-a concentration grading threshold is obtained, and the chlorophyll-a concentration grading threshold is input into the cyanobacterial bloom grading threshold dynamic calibration model to obtain a normalized vegetation index grading threshold.

[0027] According to the chlorophyll-a inversion concentration and the chlorophyll-a concentration grading threshold, and the normalized vegetation index and the normalized vegetation index grading threshold, the cyanobacterial bloom risk of the target water body region is graded.

[0028] The warning instruction issuing module is configured to generate a warning instruction corresponding to the risk level according to the cyanobacterial bloom risk grading result of the target water body region, and transmit the warning instruction to a cyanobacterial bloom warning device.

[0029] Compared with the prior art, the present application determines the target water body region of the monitoring region based on the remote sensing image of the monitoring region, and obtains the chlorophyll-a inversion concentration and the normalized vegetation index of the target water body region according to the target water body remote sensing image of the target water body region. Secondly, by utilizing the correlation between the chlorophyll-a concentration and the normalized vegetation index, an unsupervised cyanobacterial bloom grading threshold dynamic calibration model is constructed, which can dynamically adjust the cyanobacterial bloom grading threshold, thereby eliminating the subjective bias of artificial empowerment and effectively improving the real-time performance, accuracy and efficiency of cyanobacterial bloom monitoring and warning.

[0030] In order to more clearly understand the present application, the specific embodiments of the present application will be described below in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0031] Figure 1A flow chart of a cyanobacterial bloom monitoring method of the present application;

[0032] Figure 2 A flow chart of a method for constructing a water body classification model in a cyanobacterial bloom monitoring method of the present application;

[0033] Figure 3 A flow chart of a method for evaluating the accuracy of a water body classification model in a cyanobacterial bloom monitoring method of the present application;

[0034] Figure 4 A flow chart of a method for screening a chlorophyll-a sensitive waveband in a cyanobacterial bloom monitoring method of the present application;

[0035] Figure 5 A flow chart of a method for constructing a chlorophyll-a concentration inversion model in a cyanobacterial bloom monitoring method of the present application;

[0036] Figure 6 A flow chart of a method for classifying the risk of cyanobacterial bloom in a target water body area in a cyanobacterial bloom monitoring method of the present application;

[0037] Figure 7 A field water reference map for a target water body area with a risk level of no risk in a cyanobacterial bloom monitoring method of the present application;

[0038] Figure 8 A field water reference map for a target water body area with a risk level of lower risk in a cyanobacterial bloom monitoring method of the present application;

[0039] Figure 9 A field water reference map for a target water body area with a risk level of low risk in a cyanobacterial bloom monitoring method of the present application;

[0040] Figure 10 A field water reference map for a target water body area with a risk level of medium risk in a cyanobacterial bloom monitoring method of the present application;

[0041] Figure 11 A field water reference map for a target water body area with a risk level of high risk in a cyanobacterial bloom monitoring method of the present application;

[0042] Figure 12 A schematic diagram of a cyanobacterial bloom monitoring system of the present application. DETAILED DESCRIPTION

[0043] With reference to the drawings, the technical solutions in the embodiments of the present application will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all the other embodiments obtained by a person of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0044] It should be understood that the schematic drawings are not drawn according to the actual proportions. The flowchart used in the present application shows the operations implemented according to some embodiments of the present application. It should be understood that the operations of the flowchart can be implemented in no order, the steps without logical context relationship can be reversed in order or implemented simultaneously. In addition, one or more other operations can be added to the flowchart or one or more operations can be removed from the flowchart by a person of ordinary skill in the art under the guidance of the content of the present application.

[0045] In the present application, the phrase "embodiment" means that the specific features, structures or characteristics described in combination with the embodiment can be contained in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily mean the same embodiment, nor is it an independent or alternative embodiment to other embodiments. A person of ordinary skill in the art understands explicitly and implicitly that the embodiments described herein can be combined with other embodiments.

[0046] Embodiment 1

[0047] Please refer to Figure 1 , Figure 1 The present application provides a flowchart of a cyanobacterial bloom monitoring method.

[0048] The present application provides a cyanobacterial bloom monitoring method, which specifically comprises the following steps:

[0049] S1: acquiring a remote sensing image of a monitoring area and extracting multi-band remote sensing reflectivity values of the remote sensing image;

[0050] S2: according to the multi-band remote sensing reflectivity values of the remote sensing image, extracting a target water body remote sensing image containing only a target water body area based on a preset water body classification model;

[0051] S3: filtering a chlorophyll a sensitive band of the target water body remote sensing image by performing correlation analysis on the multi-band remote sensing reflectivity values of the target water body remote sensing image and the chlorophyll a concentration;

[0052] S4: obtaining a chlorophyll a inversion concentration based on a preset chlorophyll a concentration inversion model according to the remote sensing reflectivity values of the chlorophyll a sensitive band;

[0053] S5: According to the multi-band remote sensing reflectivity value of the target water body remote sensing image, the vegetation growth condition of the target water body is analyzed based on a preset normalized difference vegetation index calculation formula, and a normalized difference vegetation index is obtained.

[0054] S6: According to the chlorophyll-a inversion concentration and the normalized difference vegetation index, a cyanobacterial bloom grading threshold dynamic calibration model is constructed, and the target water body region is graded according to the cyanobacterial bloom risk.

[0055] S7: According to the cyanobacterial bloom risk grading result of the target water body region, a warning instruction corresponding to the risk level is generated, and the warning instruction is transmitted to the cyanobacterial bloom warning device.

[0056] Compared with the prior art, the scheme determines the target water body region of the monitoring region based on the remote sensing image of the monitoring region, and obtains the chlorophyll-a inversion concentration and the normalized difference vegetation index of the target water body region according to the target water body remote sensing image of the target water body region. Secondly, by utilizing the correlation characteristics of chlorophyll-a concentration and normalized difference vegetation index to construct an unsupervised cyanobacterial bloom grading threshold dynamic calibration model, the dynamic adjustment of the cyanobacterial bloom grading threshold can be realized, so as to eliminate the subjective bias of artificial empowerment, and effectively improve the real-time performance, accuracy and efficiency of cyanobacterial bloom monitoring and warning.

[0057] The cyanobacterial bloom monitoring method of the present application can be executed by the following computer system, which comprises a cyanobacterial bloom monitoring database server, a data acquisition server and a cyanobacterial bloom monitoring server. The cyanobacterial bloom monitoring database server is used to store the remote sensing image of the monitoring region, the length, width and area of the target ecological corridor and other information, and the height and base area of each building in the buffer interface and other information, so as to construct the cyanobacterial bloom monitoring database.

[0058] The data acquisition server is used to acquire the remote sensing image and other information of the monitoring region from the cyanobacterial bloom monitoring database server, and send it to the cyanobacterial bloom monitoring server for processing.

[0059] The cyanobacterial bloom monitoring server executes the cyanobacterial bloom monitoring method of the present application, extracts the multi-band remote sensing reflectivity value of the remote sensing image of the monitoring area, extracts the target water body remote sensing image containing only the target water body area based on the water body classification model, obtains the chlorophyll a inversion concentration based on the chlorophyll a concentration inversion model through the chlorophyll a sensitive band in the target water body remote sensing image which is sensitive to the chlorophyll a concentration, calculates the normalized vegetation index through the multi-band remote sensing reflectivity value of the target water body remote sensing image, constructs a cyanobacterial bloom grading threshold dynamic calibration model according to the chlorophyll a inversion concentration and the normalized vegetation index, and grades the cyanobacterial bloom risk of the target water body area, generates a warning instruction corresponding to the risk level according to the cyanobacterial bloom risk grading result of the target water body area, and transmits the warning instruction to the cyanobacterial bloom warning device.

[0060] For step S1, in response to the calling instruction of the cyanobacterial bloom monitoring server to the cyanobacterial bloom monitoring database server, the remote sensing image of the monitoring area is obtained from the cyanobacterial bloom monitoring database server, the multi-band remote sensing reflectivity value of the remote sensing image is extracted, and the multi-band remote sensing reflectivity value is stored in the cyanobacterial bloom monitoring database server. The remote sensing image includes: unmanned aerial vehicle remote sensing image and satellite remote sensing image. In this embodiment, the unmanned aerial vehicle remote sensing image is a multispectral remote sensing image with a spatial resolution better than 10 m, including Blue band, Green band, Red band, Red Edge band and NIR band, which can be obtained from unmanned aerial vehicle platform or related scientific research institutions.

[0061] Before extracting the multi-band remote sensing reflectivity value of the unmanned aerial vehicle remote sensing image, the unmanned aerial vehicle remote sensing image can be preprocessed, such as format conversion and distortion correction. After radiometric calibration and atmospheric correction of the unmanned aerial vehicle remote sensing image, the multi-band remote sensing reflectivity value of the unmanned aerial vehicle remote sensing image is extracted; after radiometric calibration and atmospheric correction of the unmanned aerial vehicle remote sensing image, orthographic correction, image color uniformity and image inlaying can also be performed. Among them, format conversion refers to converting the original unmanned aerial vehicle sensor data such as RAW format into a general raster format; distortion correction is used to eliminate the optical distortion of unmanned aerial vehicle lens and ensure the geometric fidelity of unmanned aerial vehicle remote sensing image; radiometric calibration converts the original digital (Digital Number, DN) value of the original unmanned aerial vehicle remote sensing image into absolute radiance or apparent reflectivity; atmospheric correction removes the influence of atmospheric scattering and absorption, and converts apparent reflectivity into remote sensing reflectivity value; orthographic correction eliminates the geometric deformation caused by the terrain undulation of the monitoring area to generate an orthographic image; image color uniformity adjusts the radiation consistency between multiple unmanned aerial vehicle remote sensing images, such as color balance, to reduce the color difference in the splicing area; image inlaying splices multiple orthographic images to cover the complete area.

[0062] The satellite remote sensing image is a Sentinel-2A remote sensing image, the time resolution of which is 5 days, and the spatial resolution is 10 m, 20 m or 60 m. The Sentinel-2A remote sensing image includes Band2, Band3, Band4, Band5, Band6, Band7, Band8 and Band10, etc. and can be obtained from the official website of the European Space Agency.

[0063] The satellite remote sensing image can be sequentially subjected to radiation calibration, geometric precise correction, orthorectification, image registration and atmospheric correction, so as to complete the extraction of the remote sensing reflectance value of the satellite remote sensing image. The radiation calibration is to convert the original digital value recorded by the sensor into radiance or apparent reflectance. The purpose of the geometric precise correction is to eliminate the geometric distortion caused by the sensor attitude, platform motion or terrain undulation, and to ensure the accuracy of the spatial position of the image. The orthorectification is to correct the displacement caused by the terrain such as projection difference in combination with the digital elevation model of the monitoring area, and to ensure the orthographic projection characteristics of the image. The image registration is to align the multi-temporal, multi-sensor or different band remote sensing images to a unified coordinate system, and to ensure the consistency of the data in space.

[0064] The atmospheric correction is to remove the influence of atmospheric scattering and absorption, and to convert the apparent reflectance into the real ground reflectance.

[0065] For step S2, in response to the calling instruction of the cyanobacterial bloom monitoring server to the cyanobacterial bloom monitoring database server, the multi-band remote sensing reflectance value of the remote sensing image is obtained from the cyanobacterial bloom monitoring database server, the target water body remote sensing image containing only the target water body region is extracted based on the preset water body classification model, and the target water body remote sensing image is stored in the cyanobacterial bloom monitoring database server.

[0066] In this embodiment, please refer to Figure 2 , Figure 2 The flow chart of the method for constructing the water body classification model in the cyanobacterial bloom monitoring method of the application is shown in the figure. The water body classification model is constructed, including:

[0067] S21: obtaining the historical remote sensing image of the monitoring area, wherein the historical remote sensing image includes the multi-band remote sensing reflectance value corresponding to the water body and the non-water body;

[0068] S22: labeling the water body and the non-water body of the historical remote sensing image to obtain the standard historical remote sensing image;

[0069] S23: taking the multi-band remote sensing reflectance value of the historical remote sensing image and the standard historical remote sensing image as the second sample data, and dividing the second sample data into the second training sample data set and the second test sample data set according to the ratio of 8:2;

[0070] S24: inputting the multi-band remote sensing reflectivity values of the second training sample data set as input features and the standard historical remote sensing image of the second training sample data set as a target variable, and constructing the water body classification model based on a machine learning algorithm;

[0071] S25: inputting the multi-band remote sensing reflectivity values of the second test sample data set into the water body classification model to obtain a water body classification test result corresponding to the second test sample data set;

[0072] S26: performing precision evaluation on the water body classification model by counting the number of correct water body and non-water body classifications in the water body classification test result, to obtain a water body classification model precision evaluation result;

[0073] S27: when the water body classification model precision evaluation result is lower than a preset water body classification model precision threshold, adjusting the hyperparameters of the water body classification model by using a second optimization algorithm.

[0074] For step S21, the historical remote sensing image of the monitoring area refers to a UAV remote sensing image or a satellite remote sensing image of the monitoring area acquired by remote sensing technology at a certain time point or time period in the past. The historical remote sensing image includes multi-band remote sensing reflectivity values corresponding to water bodies and non-water bodies. The types of water bodies include general water bloom, aquatic vegetation and normal water body, etc. The general water bloom refers to surface scum or film formed due to the large-scale reproduction of algae in the water body. The aquatic vegetation refers to plants growing in water. The normal water body refers to a water body that has not been significantly polluted or abnormally affected. The non-water body refers to a part of the earth's surface other than water bodies, including land, vegetation, buildings, etc. The multi-band remote sensing reflectivity values corresponding to the water bodies and the non-water bodies refer to specific reflectivity values that the water bodies and the non-water bodies have under different bands. These values constitute the spectral characteristics of the water bodies and the non-water bodies in the historical remote sensing image, and can be used to classify water bodies and non-water bodies.

[0075] For step S22, ArcGIS or QGIS or other software can be used to judge water bodies based on artificial visual interpretation, such as color difference and texture characteristics, etc. to realize the labeling of water bodies and non-water bodies in the historical remote sensing image, specifically, water bodies are labeled as 1 and non-water bodies are labeled as 0. Of course, in other embodiments, the labels of water bodies and non-water bodies in the historical remote sensing image can also be labeled by calculating spectral indices by calling the multi-band remote sensing reflectivity values corresponding to the water bodies and the non-water bodies, thereby generating the standard historical remote sensing image, such as normalized difference water index NDWI and improved normalized difference water index MNDWI.

[0076] For step S23-24, the second training sample dataset is used to build the water body classification model, so that the water body classification model learns the spectral features of water body and non-water body. The second test sample dataset is used to evaluate the performance of the water body classification model. The machine learning algorithm includes support vector machine (SVM) and random forest.

[0077] By taking the multi-band remote sensing reflectance values of the second training sample dataset as an input feature to form a multi-dimensional feature vector, and taking the standard historical remote sensing image of the second training sample dataset as a target variable to provide a supervision signal, a machine learning algorithm such as support vector machine (SVM) or random forest is used to establish a mapping relationship between the input feature and the target variable. By iteratively adjusting model hyperparameters such as penalty parameters and kernel function parameters in support vector machine (SVM) or the number and depth of trees in random forest, the difference between the predicted label and the true label is minimized to build the water body classification model. The predicted label refers to the water body classification training result of the second training sample dataset output by the model during the training process, and the true label refers to the standard historical remote sensing image of the second training sample dataset.

[0078] For step S26, please also refer to Figure 3 , Figure 3 A method flowchart for evaluating the accuracy of a water body classification model in a cyanobacterial bloom monitoring method. The accuracy of the water body classification model is evaluated by counting the number of water body classification correct points and non-water body classification correct points in the water body classification test results, and the water body classification model accuracy evaluation result is obtained, including:

[0079] S261: Randomly generating a plurality of test points in the water body classification test results;

[0080] S262: Counting the number of water body classification correct points and the number of non-water body classification correct points in the water body classification test results;

[0081] S263: According to the total number of test points, the number of water body classification correct points and the number of non-water body classification correct points, the water body classification model accuracy evaluation result is obtained based on the following formula:

[0082]

[0083] In the formula, OA is the water body classification model accuracy evaluation result, is the number of water body classification correct points, is the number of non-water body classification correct points, and N is the total number of test points.

[0084] For steps S261-S264, the water body classification test results can be read using geographic information system software such as ArcGIS, and the test points can be randomly generated using tools such as creating random points. In this embodiment, the total number of test points can be set to 200, and to avoid too dense random points, the minimum distance between two random test points can also be set.

[0085] The judgment rule can be that a person skilled in the art, based on the inventive concept of the present application, combines the common knowledge in the art, judges each test point in the water body classification test results to be water body or non-water body by artificial judgment, and calculates the number of test points with correct water body classification and the number of test points with correct non-water body classification.

[0086] In other embodiments, the water body classification model can also be evaluated for accuracy by calculating coefficients such as Kappa and F1-Score, to obtain the water body classification model accuracy evaluation result.

[0087] For step S27, the water body classification model accuracy threshold is preferably set to 90%, and the second optimization algorithm can be an iterative algorithm such as grid search, Bayesian optimization, genetic algorithm, gradient boosting, etc. Of course, the water body classification model accuracy threshold can be adaptively modified according to actual needs.

[0088] For step S3, in response to the calling instruction of the cyanobacterial bloom monitoring server to the cyanobacterial bloom monitoring database server, the multi-band remote sensing reflectance values of the target water body remote sensing image are obtained from the cyanobacterial bloom monitoring database server, the chlorophyll a sensitive band of the target water body remote sensing image is screened, and the remote sensing reflectance values of the chlorophyll a sensitive band sensitive to the chlorophyll a concentration are stored in the cyanobacterial bloom monitoring database server.

[0089] In one embodiment, please refer to Figure 4 , Figure 4 is a flowchart of a method for screening a chlorophyll a sensitive band in a cyanobacterial bloom monitoring method of the present application. The chlorophyll a sensitive band of the target water body remote sensing image is screened by performing correlation analysis on the multi-band remote sensing reflectance values of the target water body remote sensing image and the chlorophyll a concentration, and includes:

[0090] S31: Obtain first sample data of the target water body region at different historical monitoring times, wherein the first sample data includes historical chlorophyll a monitoring concentration and historical target water body remote sensing image;

[0091] S32: Calculate the Pearson correlation coefficient between the remote sensing reflectance of each band in the historical target water body remote sensing image and the historical chlorophyll a monitoring concentration according to the following formula:

[0092]

[0093] in which, denotes the Pearson correlation coefficient of the jth waveband, denotes the remote sensing reflectance of the jth waveband in the historical target water remote sensing image of the ith first sample data, denotes the average value of the remote sensing reflectance of the jth waveband, denotes the historical monitoring concentration of chlorophyll a of the ith first sample data, denotes the average value of the historical monitoring concentration of chlorophyll a, is the number of the first sample data;

[0094] S33: According to the Pearson correlation coefficient of the remote sensing reflectance of each waveband and the historical monitoring concentration of chlorophyll a, a corresponding waveband is selected as the chlorophyll a sensitive waveband based on a preset sensitive waveband screening standard.

[0095] For steps S31-S33, the first sample data of the target water area at different historical monitoring times refers to the historical monitoring concentration of chlorophyll a and the historical target water remote sensing image obtained at the same time point or time period, and the historical monitoring concentration of chlorophyll a and the historical target water remote sensing image strictly correspond in spatial position (the same geographic coordinate point) and time dimension (the same monitoring time), which can be obtained from local environmental monitoring agencies or academic research institutions, etc.

[0096] The sensitive waveband screening standard is specifically that according to the Pearson correlation coefficient of the remote sensing reflectance of each waveband and the historical monitoring concentration of chlorophyll a, each waveband is sorted in descending order, and the top five wavebands are selected as the chlorophyll a sensitive waveband.

[0097] In other embodiments, a correlation threshold value can also be preset according to actual needs or statistical standards, and a waveband with a Pearson correlation coefficient of the remote sensing reflectance of each waveband and the historical monitoring concentration of chlorophyll a greater than the preset correlation threshold value is selected as the chlorophyll a sensitive waveband. For example, in the research of Thousand Island Lake, it is found through Pearson analysis that the (B4+B2) / B3 waveband combination is significantly correlated with the concentration of chlorophyll a, with a determination coefficient r²=0.7366, that is, the Pearson correlation coefficient r≈0.858, and if the correlation threshold value is set to 0.8 (corresponding to r²≥0.64), the waveband combination meets the screening standard.

[0098] For step S4, in response to the call instruction of the cyanobacterial bloom monitoring server to the cyanobacterial bloom monitoring database server, the remote sensing reflectivity value of the chlorophyll a concentration sensitive chlorophyll a sensitive wave band is obtained from the cyanobacterial bloom monitoring database server, the chlorophyll a concentration inversion model is obtained based on the preset chlorophyll a concentration inversion model, and the chlorophyll a concentration inversion model is stored in the cyanobacterial bloom monitoring database server.

[0099] In the embodiment, please refer to Figure 5 , Figure 5 A method flow chart for constructing a chlorophyll a concentration inversion model in a cyanobacterial bloom monitoring method of the application. The chlorophyll a concentration inversion model is constructed, comprising:

[0100] S41: dividing the first sample data into a first training sample data set and a first test sample data set according to a ratio of 8:2;

[0101] S42: taking the chlorophyll a historical monitoring concentration and the remote sensing reflectivity value of the sensitive wave band in the historical target water body remote sensing image of the first training sample data set as input features, and constructing the chlorophyll a concentration inversion model based on a machine learning algorithm;

[0102] S43: taking the remote sensing reflectivity of the sensitive wave band in the historical target water body remote sensing image of the first test sample data set as an input feature, and obtaining the chlorophyll a inversion concentration test result corresponding to the first test sample data set based on the chlorophyll a concentration inversion model;

[0103] S44: performing accuracy evaluation on the chlorophyll a concentration inversion model according to the chlorophyll a inversion concentration test result and the chlorophyll a historical monitoring concentration of the first test sample data set, and obtaining a chlorophyll a concentration inversion model accuracy evaluation result.

[0104] For step S41, the first training sample data set is used to construct the chlorophyll a concentration inversion model, so that the chlorophyll a concentration inversion model learns the relationship between the remote sensing reflectivity value of the sensitive wave band in the historical target water body remote sensing image and the chlorophyll a concentration. The first test sample data set is used to evaluate the performance of the chlorophyll a concentration inversion model.

[0105] For step S42, the machine learning model includes support vector machine (SVM), random forest, XGBoost, etc. By constructing a multi-dimensional feature vector as an input feature by taking the historical monitoring concentration of chlorophyll a of the first training sample data set and the remote sensing reflectance value of the sensitive band in the historical target water body remote sensing image, and iteratively adjusting the model hyperparameters, such as the penalty parameter and kernel function parameter in the vector machine SVM or the number and depth of trees in the random forest, based on machine learning algorithms such as vector machine SVM, random forest or XGBoos, the chlorophyll a concentration inversion model is constructed.

[0106] For step S44, according to the chlorophyll a inversion concentration test result and the historical monitoring concentration of chlorophyll a of the first test sample data set, the accuracy of the chlorophyll a concentration inversion model is evaluated to obtain the chlorophyll a concentration inversion model accuracy evaluation result, including:

[0107] According to the chlorophyll a inversion concentration test result and the historical monitoring concentration of chlorophyll a of the first test sample data set, based on the determination coefficient calculation formula, the chlorophyll a concentration inversion model accuracy evaluation result is obtained, and the determination coefficient calculation formula is:

[0108]

[0109] In the formula, is the chlorophyll a concentration inversion model accuracy evaluation result, is the predicted chlorophyll a concentration, is the historical monitoring concentration of chlorophyll a of the first training sample data set, is the average value of the historical monitoring concentration of chlorophyll a of the first training sample data set, and n is the size of the first sample data.

[0110] Or according to the chlorophyll a inversion concentration test result and the historical monitoring concentration of chlorophyll a of the first test sample data set, based on the root mean square error calculation formula, the chlorophyll a concentration inversion model accuracy evaluation result is obtained, and the root mean square error calculation formula is:

[0111]

[0112] In the formula, is the chlorophyll a concentration inversion model accuracy evaluation result, is the chlorophyll a inversion concentration test result, is the historical monitoring concentration of chlorophyll a of the first test sample data set, and n is the size of the first sample data.

[0113] When the chlorophyll a concentration inversion model precision threshold is set to 0.7, if the chlorophyll a concentration inversion model precision evaluation result is lower than 0.7, the hyperparameters of the chlorophyll a concentration inversion model are adjusted by the first optimization algorithm, such as the penalty parameter and the kernel function parameter in the vector machine SVM or the number and depth of trees in the random forest.

[0114] When the chlorophyll a concentration inversion model precision threshold is set to 0.1, if the chlorophyll a concentration inversion model precision evaluation result is higher than 0.1, the hyperparameters of the chlorophyll a concentration inversion model are adjusted by the first optimization algorithm, such as the penalty parameter and the kernel function parameter in the vector machine SVM or the number and depth of trees in the random forest.

[0115] The first optimization algorithm can select iterative algorithms such as grid search, Bayesian optimization, and genetic algorithm. Of course, in other embodiments, the chlorophyll a concentration inversion model can also be evaluated for precision by using functions such as mean square error loss function and mean absolute error loss function. According to the performance requirements of the chlorophyll a concentration inversion model and the actual selection of the formula for evaluating the precision of the chlorophyll a concentration inversion model, the chlorophyll a concentration inversion model precision threshold can be adaptively modified.

[0116] For step S5, in response to the calling instruction of the cyanobacterial bloom monitoring server to the cyanobacterial bloom monitoring database server, the multi-band remote sensing reflectivity value of the target water body remote sensing image is obtained from the cyanobacterial bloom monitoring database server, the vegetation growth of the target water body is analyzed based on the preset normalized difference vegetation index calculation formula, the normalized difference vegetation index is obtained, and the normalized difference vegetation index is stored in the cyanobacterial bloom monitoring database server.

[0117] The multi-band remote sensing reflectivity value of the target water body remote sensing image includes the remote sensing reflectivity value of the near-infrared band and the red light band. The normalized difference vegetation index is obtained based on the preset normalized difference vegetation index calculation formula according to the multi-band remote sensing reflectivity value of the target water body remote sensing image, including:

[0118] The normalized difference vegetation index is obtained according to the remote sensing reflectivity value of the near-infrared band and the red light band according to the normalized difference vegetation index calculation formula, and the normalized difference vegetation index calculation formula is:

[0119]

[0120] wherein, the normalized vegetation index, the remote sensing reflectance value of the near-infrared band, the remote sensing reflectance value of the red light band.

[0121] For step S6, in response to the calling instruction of the cyanobacterial bloom monitoring server to the cyanobacterial bloom monitoring database server, the chlorophyll a inversion concentration and the normalized vegetation index are obtained from the cyanobacterial bloom monitoring database server, a cyanobacterial bloom grading threshold dynamic calibration model is constructed, and the target water body region is graded for cyanobacterial bloom risk.

[0122] In this embodiment, please refer to Figure 6 , Figure 6 is a method flow chart for grading a target water body region for cyanobacterial bloom risk in a cyanobacterial bloom monitoring method according to the present application. The cyanobacterial bloom grading threshold dynamic calibration model is constructed according to the chlorophyll a inversion concentration and the normalized vegetation index, and the target water body region is graded for cyanobacterial bloom risk, which comprises:

[0123] S61: randomly collecting the target water body region to obtain a plurality of chlorophyll a inversion concentrations and normalized vegetation indexes;

[0124] S62: performing correlation analysis on the normalized vegetation index and the chlorophyll a inversion concentration, and if the normalized vegetation index and the chlorophyll a inversion concentration are linearly correlated, a linear model of the normalized vegetation index and the chlorophyll a inversion concentration is established as the cyanobacterial bloom grading threshold dynamic calibration model;

[0125] S63: obtaining a preset chlorophyll a concentration grading threshold, and inputting the chlorophyll a concentration grading threshold into the cyanobacterial bloom grading threshold dynamic calibration model to obtain a normalized vegetation index grading threshold;

[0126] S64: grading the target water body region for cyanobacterial bloom risk according to the chlorophyll a inversion concentration and the chlorophyll a concentration grading threshold, and the normalized vegetation index and the normalized vegetation index grading threshold.

[0127] For step S61, a random sampling is performed on the target water body area to obtain a plurality of chlorophyll-a inversion concentrations and normalized vegetation indexes, which can be achieved by using GIS software such as ArcGIS or QGIS. Specifically, the target water body remote sensing image of the target water body area is imported into ArcGIS or QGIS; the target water body remote sensing image is used as a background to set a sampling area, and a random point generation tool is used to randomly generate sampling points in the sampling area; the remote sensing reflectance values at the corresponding positions on the target water body remote sensing image are extracted according to the coordinates of the generated sampling points, and the chlorophyll-a inversion concentration and the normalized vegetation index of each sampling point are obtained based on the chlorophyll-a concentration inversion model and the normalized vegetation index calculation formula.

[0128] For step S62, the Pearson correlation coefficient of the normalized vegetation index and the chlorophyll-a inversion concentration is calculated to perform correlation analysis. When the Pearson correlation coefficient of the normalized vegetation index and the chlorophyll-a inversion concentration is greater than 0, it indicates that the normalized vegetation index and the chlorophyll-a inversion concentration are positively correlated.

[0129] Of course, the normalized vegetation index and the chlorophyll-a inversion concentration can also be subjected to a significance test, for example, the significance level of the normalized vegetation index and the chlorophyll-a inversion concentration is calculated by t-test or F-test. When the significance level of the normalized vegetation index and the chlorophyll-a inversion concentration is less than a preset significance level threshold, it is considered that the linear relationship between the normalized vegetation index and the chlorophyll-a inversion concentration is significant. The significance level threshold can be set to 0.05 or 0.01, and in other embodiments, the significance level threshold can be adaptively adjusted according to actual needs.

[0130] The expression of the cyanobacterial bloom grading threshold dynamic calibration model is:

[0131]

[0132] In the formula, is the chlorophyll-a inversion concentration, is the normalized vegetation index, a is a regression coefficient, indicating the expected change in chlorophyll-a concentration when NDVI increases by 1 unit, is the intercept, indicating the estimated value of chlorophyll-a concentration when NDVI is 0.

[0133] The regression coefficient a of the cyanobacterial bloom grading threshold dynamic calibration model can be obtained by the least squares method, and the specific formula is:

[0134]

[0135] In the formula, a is a regression coefficient of the cyanobacterial bloom grading threshold dynamic calibration model, is a normalized vegetation index of the i th sampling point, is an average value of the normalized vegetation index, is a chlorophyll-a inversion concentration of the i th sampling point, is an average value of the chlorophyll-a inversion concentration, and m is a number of the sampling points.

[0136] The intercept b of the cyanobacterial bloom grading threshold dynamic calibration model can be obtained by the following formula:

[0137]

[0138] In the formula, b is an intercept of the cyanobacterial bloom grading threshold dynamic calibration model, is an average value of the normalized vegetation index, is an average value of the chlorophyll-a inversion concentration, and a is a regression coefficient of the cyanobacterial bloom grading threshold dynamic calibration model.

[0139] After the cyanobacterial bloom grading threshold dynamic calibration model is obtained, the model can be tested by residual analysis or R value calculation to ensure the fitting effect and prediction ability of the model.

[0140] For step S63, the chlorophyll-a concentration grading threshold is a value obtained by dividing the content of chlorophyll-a in the water body into different levels or ranges, which is used to evaluate the eutrophication state of the water body, the occurrence risk of cyanobacterial bloom and the water quality. The normalized vegetation index grading threshold is a value obtained by dividing the normalized vegetation index in the water body into different levels or ranges, which is used to evaluate the vegetation coverage of the water body. The chlorophyll-a concentration grading threshold can be obtained by the water quality monitoring standards published by the environmental protection department, such as the surface water environmental quality standard. It can be seen that by using the correlation characteristics of chlorophyll-a concentration and normalized vegetation index, the cyanobacterial bloom grading threshold dynamic calibration model can be adjusted in real time, and the normalized vegetation index grading threshold can be dynamically adjusted, so as to eliminate the subjective bias of artificial weighting and effectively improve the real-time, accuracy and efficiency of cyanobacterial bloom monitoring and early warning.

[0141] For step S64, in the embodiment, the target water body area is graded according to the risk of cyanobacterial bloom, which can be divided into no risk (level I), lower risk (level II), low risk (level III), medium risk (level IV) and high risk (level V). The chlorophyll-a concentration grading threshold and the normalized vegetation index grading threshold are shown in the following table 1.

[0142] Table 1:

[0143]

[0144] When the concentration of the concentration of chlorophyll a inversion is in the interval of 0~10ug / L and the normalized vegetation index value of the target water area is in the interval of [-1,-0.07), it is determined that the target water area has no risk of cyanobacterial bloom, at this time the water color of the target water area is normal, there is no algae aggregation on the water surface, and basically no algae particles can be identified in the water, as shown in Figure 7 .

[0145] When the concentration of the concentration of chlorophyll a inversion is in the interval of 10~15ug / L and the normalized vegetation index value of the target water area is in the interval of [-0.07,-0.05), it is determined that the target water area is at low risk, at this time the water color of the target water area is light yellow-green, and there is no algae aggregation on the water surface, as shown in Figure 8 .

[0146] When the concentration of the concentration of chlorophyll a inversion is in the interval of 15~50ug / L and the normalized vegetation index value of the target water area is in the interval of [-0.05,0.07), it is determined that the target water area is at low risk, at this time the water color of the target water area is yellow-green, and a small amount of algae particles can be identified in the water, as shown in Figure 9 .

[0147] When the concentration of the concentration of chlorophyll a inversion is in the interval of 50~100ug / L and the normalized vegetation index value of the target water area is in the interval of [0.07,0.28), it is determined that the target water area is at medium risk, at this time the water color of the target water area is yellow-green, and there is sporadic algae aggregation on the water surface, or the algae in the water is obviously visible, as shown in Figure 10 .

[0148] When the concentration of the concentration of chlorophyll a inversion is greater than 100ug / L and the normalized vegetation index value of the target water area is greater than 0.28, it is determined that the target water area is at high risk, at this time the water color of the target water area is yellow-green, and there is algae aggregation on the water surface, or the algae in the water is obviously visible, as shown in Figure 11 .

[0149] When the concentration of the concentration of chlorophyll a inversion is inconsistent with the normalized vegetation index judgment standard, the higher level is selected. For example, when the concentration of the concentration of chlorophyll a inversion is in the interval of 15~50ug / L (low risk), but the normalized vegetation index is in the interval of [0.07,0.28) (medium risk), the risk level of the normalized vegetation index is taken as the evaluation result, that is, the target water area is at medium risk.

[0150] Of course, in other embodiments, the chlorophyll-a concentration grading threshold value can also be adaptively modified according to the actual water quality monitoring standard. At the same time, the normalized difference vegetation index grading threshold value can also be adaptively modified according to the chlorophyll-a concentration grading threshold value and the cyanobacterial bloom grading threshold value dynamic calibration model.

[0151] For step S7, in response to the calling instruction of the cyanobacterial bloom monitoring server to the cyanobacterial bloom monitoring database server, a warning instruction corresponding to the risk level is generated according to the cyanobacterial bloom risk grading result of the target water body area, and the warning instruction is transmitted to the cyanobacterial bloom warning device.

[0152] In the embodiment, the warning instruction includes the generation time, the geographic coordinates of the target water body area, the risk level, the key parameters, and the warning suggestion, etc. The risk level includes the no risk (level I), the lower risk (level II), the low risk (level III), the medium risk (level IV), and the high risk (level V). The key parameters include the chlorophyll-a inversion concentration and the normalized difference vegetation index, and the chlorophyll-a concentration grading threshold value and the normalized difference vegetation index grading threshold value.

[0153] When the cyanobacterial bloom risk grading result of the target water body area is no risk, the warning suggestion can be to maintain regular monitoring, record water quality data, and there is no need to start emergency response. When the cyanobacterial bloom risk grading result of the target water body area is lower risk, the warning suggestion can be to strengthen the patrol frequency and submit a monitoring report to the management department. When the cyanobacterial bloom risk grading result of the target water body area is low risk, the warning suggestion can be to start warning monitoring (once a day), limit the surrounding agricultural / industrial pollution, and publish water quality safety tips to the public. When the cyanobacterial bloom risk grading result of the target water body area is medium risk, the warning suggestion can be to start emergency monitoring (once an hour), close part of the water area, deploy salvage equipment, and prepare for algae removal operations. When the cyanobacterial bloom risk grading result of the target water body area is high risk, the warning suggestion can be to issue a red alert, evacuate surrounding personnel, start cross-departmental linkage (environmental protection, water conservancy, emergency), and implement emergency treatment.

[0154] The cyanobacterial bloom monitoring server can push the warning instruction to the cyanobacterial bloom warning device through a wireless network. The cyanobacterial bloom warning device can be a water quality monitoring buoy, a shore display screen, a mobile terminal, etc. Of course, the cyanobacterial bloom monitoring server and the warning device system can also be interfaced.

[0155] In other embodiments, the warning instruction can be adaptively modified according to the actual monitoring needs.

[0156] Embodiment 2

[0157] Please refer toFigure 12 , Figure 12 FIG. 1 is a schematic diagram of a cyanobacterial bloom monitoring system according to the present application.

[0158] The present application also provides a cyanobacterial bloom monitoring system, comprising:

[0159] a remote sensing image acquisition module 1 configured to acquire remote sensing images of a monitoring area and extract multi-band remote sensing reflectance values of the remote sensing images;

[0160] a target water body remote sensing image extraction module 2 configured to extract target water body remote sensing images containing only target water body regions based on a preset water body classification model according to the multi-band remote sensing reflectance values of the remote sensing images;

[0161] a sensitive band screening module 3 configured to screen chlorophyll a sensitive bands of the target water body remote sensing images by performing correlation analysis on the multi-band remote sensing reflectance values of the target water body remote sensing images and chlorophyll a concentrations;

[0162] a chlorophyll a concentration inversion module 4 configured to obtain chlorophyll a inversion concentrations based on a preset chlorophyll a concentration inversion model according to the remote sensing reflectance values of the chlorophyll a sensitive bands;

[0163] a normalized difference vegetation index calculation module 5 configured to analyze vegetation growth conditions of the target water bodies based on a preset normalized difference vegetation index calculation formula according to the multi-band remote sensing reflectance values of the target water body remote sensing images to obtain normalized difference vegetation indexes;

[0164] a target water body region grading module 6 configured to construct a cyanobacterial bloom grading threshold dynamic calibration model according to the chlorophyll a inversion concentrations and the normalized difference vegetation indexes, and grade cyanobacterial bloom risks of the target water body regions;

[0165] a warning instruction issuing module 7 configured to generate warning instructions corresponding to risk levels according to cyanobacterial bloom risk grading results of the target water body regions, and transmit the warning instructions to cyanobacterial bloom warning devices.

[0166] It should be noted that the data obtained by the cyanobacterial bloom monitoring system provided by the present application when implementing the cyanobacterial bloom monitoring method are saved in the storage of the system in a one-to-one correspondence, and the data required for calculation can be directly obtained from the storage when relevant calculations are needed.

[0167] It also needs to be explained that the above embodiment provides a blue-green algae water bloom monitoring system in the implementation of a blue-green algae water bloom monitoring method, only the above-mentioned functional module division is used as an example, and in actual application, the above-mentioned functions can be distributed by different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the blue-green algae water bloom monitoring system provided by the above embodiment and the blue-green algae water bloom monitoring method of embodiment 1 belong to the same concept, and the implementation process is described in detail in the method embodiment, which will not be repeated here.

[0168] Based on the same inventive concept, the present application also provides an electronic device, which can be a server, a desktop computing device or a mobile computing device (for example, a laptop computer, a handheld computing device, a tablet computer, a netbook, etc.). The device includes one or more processors and a memory, wherein the processor is used to execute a program to implement the blue-green algae water bloom monitoring method; the memory is used to store a computer program executable by the processor.

[0169] The present application is not limited to the above-mentioned embodiments, and if various modifications or changes of the present application do not deviate from the spirit and scope of the present application, the present application also intends to include these modifications and changes, provided that these modifications and changes belong to the scope of the claims and equivalent technology of the present application.

Claims

1. A method for monitoring cyanobacterial blooms, characterized in that, Includes the following steps: Acquire remote sensing images of the monitored area and extract multi-band remote sensing reflectance values ​​from the remote sensing images; Based on the multi-band remote sensing reflectance values ​​of the remote sensing images, and based on a preset water body classification model, target water body remote sensing images containing only the target water body area are extracted. By performing correlation analysis between the multi-band remote sensing reflectance values ​​of the target water body remote sensing image and the chlorophyll a concentration, the chlorophyll a sensitive bands of the target water body remote sensing image were screened. Based on the remote sensing reflectance value of the chlorophyll a sensitive band, and using a preset chlorophyll a concentration inversion model, the chlorophyll a inversion concentration is obtained. Based on the multi-band remote sensing reflectance values ​​of the target water body's remote sensing image, and using a preset normalized vegetation index calculation formula, the vegetation growth of the target water body is analyzed to obtain the normalized vegetation index. Based on the chlorophyll a inversion concentration and the normalized vegetation index, a dynamic calibration model for cyanobacterial bloom classification thresholds is constructed, and cyanobacterial bloom risk classification is performed on the target water body area, including: Randomly collect data from the target water body area to obtain several chlorophyll a inversion concentrations and normalized vegetation indices. A correlation analysis was performed on the normalized vegetation index and the chlorophyll a inversion concentration. If the normalized vegetation index and the chlorophyll a inversion concentration are linearly correlated, a linear model of the normalized vegetation index and the chlorophyll a inversion concentration is established as the dynamic calibration model for the cyanobacterial bloom classification threshold. Obtain a preset chlorophyll a concentration classification threshold, and input the chlorophyll a concentration classification threshold into the dynamic calibration model of the cyanobacterial bloom classification threshold to obtain the normalized vegetation index classification threshold. Based on the chlorophyll a inversion concentration and chlorophyll a concentration classification threshold, as well as the normalized vegetation index and normalized vegetation index classification threshold, the target water body area is classified for cyanobacterial bloom risk. Based on the risk classification results of cyanobacterial blooms in the target water area, an early warning instruction corresponding to the risk level is generated and transmitted to the cyanobacterial bloom early warning device.

2. The method for monitoring cyanobacterial blooms according to claim 1, characterized in that, The process of screening chlorophyll-a sensitive bands in the remote sensing images of the target water body by performing correlation analysis between multi-band remote sensing reflectance values ​​and chlorophyll a concentration includes: Acquire first sample data of the target water body area under different historical monitoring times, wherein the first sample data includes historical monitoring concentration of chlorophyll a and historical remote sensing images of the target water body; The Pearson correlation coefficient between the remote sensing reflectance of each band in the historical target water body remote sensing image and the historical monitoring concentration of chlorophyll a is calculated according to the following formula: In the formula, This represents the Pearson correlation coefficient for the j-th band. Let represent the remote sensing reflectance of the j-th band in the historical remote sensing image of the target water body for the i-th sample data. This represents the average remote sensing reflectance of the j-th band. This represents the historical monitoring concentration of chlorophyll a in the i-th data set of the first sample. This represents the average historical monitoring concentration of chlorophyll a. The number of the first sample data; Based on the Pearson correlation coefficient between the remote sensing reflectance of each band and the historical monitoring concentration of chlorophyll a, and based on the preset sensitive band screening criteria, the corresponding bands are selected as the sensitive bands for chlorophyll a.

3. The method for monitoring cyanobacterial blooms according to claim 2, characterized in that, Constructing the chlorophyll a concentration inversion model includes: The first sample data is divided into a first training sample dataset and a first test sample dataset in an 8:2 ratio; The historical monitoring concentration of chlorophyll a inversion model is constructed based on machine learning algorithms, using the historical monitoring concentration of chlorophyll a in the first training sample dataset and the remote sensing reflectance values ​​of sensitive bands in the historical remote sensing images of the target water body as input features. Using the remote sensing reflectance of sensitive bands in the historical target water remote sensing images of the first test sample dataset as input features, and based on the chlorophyll a concentration inversion model, the chlorophyll a inversion concentration test results corresponding to the first test sample dataset are obtained. Based on the chlorophyll a inversion concentration test results and the historical monitoring concentration of chlorophyll a in the first test sample dataset, the accuracy of the chlorophyll a concentration inversion model is evaluated, and the accuracy evaluation result of the chlorophyll a concentration inversion model is obtained.

4. The method for monitoring cyanobacterial blooms according to claim 3, characterized in that, The accuracy evaluation of the chlorophyll a concentration inversion model is performed based on the chlorophyll a inversion concentration test results and the historical monitoring concentration of chlorophyll a in the first test sample dataset, resulting in an accuracy evaluation result for the chlorophyll a concentration inversion model, including: Based on the chlorophyll a concentration inversion test results and the historical chlorophyll a concentration monitoring data of the first test sample dataset, the accuracy evaluation result of the chlorophyll a concentration inversion model is obtained based on the coefficient of determination calculation formula. The coefficient of determination calculation formula is as follows: In the formula, This is the accuracy evaluation result of the chlorophyll a concentration inversion model. The results of the chlorophyll a inversion concentration test are as follows. The historical monitoring concentration of chlorophyll a in the first test sample dataset. is the average historical monitoring concentration of chlorophyll a in the first test sample dataset, and n is the size of the first sample data; Alternatively, based on the chlorophyll a concentration inversion test results and the historical chlorophyll a concentration of the first test sample dataset, the accuracy evaluation result of the chlorophyll a concentration inversion model can be obtained using the root mean square error (RMSE) calculation formula. The RMSE calculation formula is as follows: In the formula, This is the accuracy evaluation result of the chlorophyll a concentration inversion model. The results of the chlorophyll a inversion concentration test are as follows. is the historical monitoring concentration of chlorophyll a in the first test sample dataset, and n is the size of the first sample data.

5. The method for monitoring cyanobacterial blooms according to claim 1, characterized in that, Constructing the water body classification model includes: Acquire historical remote sensing images of the monitored area, wherein the historical remote sensing images include multi-band remote sensing reflectance values ​​corresponding to water bodies and non-water bodies; The historical remote sensing images are labeled with water bodies and non-water bodies to obtain standard historical remote sensing images; The multi-band remote sensing reflectance values ​​of the historical remote sensing images and the standard historical remote sensing images are used as the second sample data, and the second sample data is divided into the second training sample dataset and the second test sample dataset in an 8:2 ratio. The multi-band remote sensing reflectance values ​​of the second training sample dataset are used as input features, and the standard historical remote sensing images of the second training sample dataset are used as target variables. The water body classification model is constructed based on machine learning algorithms. The multi-band remote sensing reflectance values ​​of the second test sample dataset are input into the water body classification model to obtain the water body classification test results corresponding to the second test sample dataset. By statistically analyzing the number of correctly classified water bodies and non-water bodies in the water body classification test results, the accuracy of the water body classification model is evaluated, and the accuracy evaluation result of the water body classification model is obtained. When the accuracy evaluation result of the water body classification model is lower than the preset accuracy threshold of the water body classification model, the hyperparameters of the water body classification model are adjusted using the second optimization algorithm.

6. The method for monitoring cyanobacterial blooms according to claim 5, characterized in that, The accuracy of the water body classification model is evaluated by statistically analyzing the number of correctly classified water bodies and non-water bodies in the water body classification test results, resulting in an accuracy evaluation result for the water body classification model, including: Several test points are randomly generated from the water body classification test results; Count the number of test points that correctly classified the water body in the water body classification test results and the number of test points that did not correctly classify the water body. Based on the total number of test points, the number of test points correctly classified as water bodies, and the number of test points incorrectly classified as water bodies, the accuracy evaluation result of the water body classification model is obtained using the following formula: In the formula, OA represents the accuracy evaluation result of the water body classification model. The number of test points for correctly classifying water bodies. N represents the number of test points that correctly classified as non-water bodies, and N is the total number of test points.

7. The method for monitoring cyanobacterial blooms according to claim 1, characterized in that, The multi-band remote sensing reflectance values ​​of the target water body remote sensing image include: remote sensing reflectance values ​​of the near-infrared band and the red light band; The step involves analyzing the vegetation growth of the target water body based on the multi-band remote sensing reflectance values ​​of the remote sensing image and a preset normalized vegetation index (NVI) calculation formula, to obtain the NVI, including: Based on the remote sensing reflectance values ​​in the near-infrared and red bands, the normalized vegetation index (NWRI) is obtained according to the aforementioned formula. The NWRI calculation formula is as follows: in, The normalized vegetation index is... The remote sensing reflectance value for the near-infrared band is [value missing]. The value is the remote sensing reflectance value for the red light band.

8. A cyanobacterial bloom monitoring system, characterized in that, include: Remote sensing image acquisition module: used to acquire remote sensing images of the monitored area and extract multi-band remote sensing reflectance values ​​of the remote sensing images; Target water body remote sensing image extraction module: used to extract target water body remote sensing images containing only the target water body area based on the multi-band remote sensing reflectance values ​​of the remote sensing images and a preset water body classification model. Sensitive band screening module: used to screen the chlorophyll a sensitive bands of the remote sensing image of the target water body by performing correlation analysis between the multi-band remote sensing reflectance values ​​and chlorophyll a concentration. Chlorophyll a concentration inversion module: used to obtain the chlorophyll a inversion concentration based on the remote sensing reflectance value of the chlorophyll a sensitive band and a preset chlorophyll a concentration inversion model; Normalized Difference Vegetation Index (NDVI) Calculation Module: This module is used to analyze the vegetation growth of the target water body based on the multi-band remote sensing reflectance values ​​of the remote sensing image of the target water body and a preset NDVI calculation formula, and to obtain the NDVI. Target water body area classification module: used to construct a dynamic calibration model for cyanobacterial bloom classification thresholds based on the chlorophyll a inversion concentration and the normalized vegetation index, and to classify the cyanobacterial bloom risk of the target water body area, including: Randomly collect data from the target water body area to obtain several chlorophyll a inversion concentrations and normalized vegetation indices. A correlation analysis was performed on the normalized vegetation index and the chlorophyll a inversion concentration. If the normalized vegetation index and the chlorophyll a inversion concentration are linearly correlated, a linear model of the normalized vegetation index and the chlorophyll a inversion concentration is established as the dynamic calibration model for the cyanobacterial bloom classification threshold. Obtain a preset chlorophyll a concentration classification threshold, and input the chlorophyll a concentration classification threshold into the dynamic calibration model of the cyanobacterial bloom classification threshold to obtain the normalized vegetation index classification threshold. Based on the chlorophyll a inversion concentration and chlorophyll a concentration classification threshold, as well as the normalized vegetation index and normalized vegetation index classification threshold, the target water body area is classified for cyanobacterial bloom risk. Early warning instruction issuing module: used to generate an early warning instruction corresponding to the risk level based on the risk classification results of cyanobacterial blooms in the target water area, and transmit the early warning instruction to the cyanobacterial bloom early warning device.

9. An electronic device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, when the processor executes the computer program, it implements a method for monitoring cyanobacterial blooms as described in any one of claims 1 to 7.

Images