Water bloom prediction method and device based on meteorological and remote sensing data

Abstract

The present invention discloses a method and device for predicting water bloom based on meteorological and remote sensing data, and belongs to the field of satellite remote sensing technology. The present invention calculates the difference between the baseline of the near infrared band and the red light band and the short-wave infrared band for each pixel, as well as the meteorological data corresponding to the same moment, combines the binary graph of water bloom results, and uses a statistical method to calculate a set of data corresponding to each pixel to obtain a regression model for water bloom prediction, and determines whether water bloom occurs according to the water bloom threshold screening, selects the threshold corresponding to the occurrence of water bloom through data analysis, and then carries out water bloom prediction for each pixel, and finally obtains the prediction results of the predicted occurrence location and the size of the occurrence area of ​​water bloom. The present invention improves the accuracy of water bloom prediction, and at the same time has universality, and is simple and easy to operate.

Description

Technical Field

[0001] This invention relates to the field of satellite remote sensing technology, and in particular to a method and apparatus for predicting algal blooms based on meteorological and remote sensing data. Background Technology

[0002] Against the backdrop of eutrophication of lakes and global warming caused by human activities, cyanobacteria are overgrown in a wide range of inland waters worldwide, leading to frequent cyanobacterial blooms and further deteriorating water quality. Monitoring cyanobacterial blooms utilizes both ground-based and remote sensing methods to monitor the bloom process. Cyanobacterial blooms change rapidly in the short term; therefore, predicting blooms and their locations simultaneously provides technical support for bloom control. Bloom prediction models consist of two parts: accurate identification of bloom-driving factors (determining model inputs) and the establishment of the prediction model. The response relationship of blooms to driving factors is a crucial step in predicting the probability of bloom occurrence. Bloom prediction is mainly divided into three types: data-driven, process-driven, and fusion model.

[0003] The complex mechanisms of cyanobacterial blooms lead to their inherent uncertainties, making bloom prediction challenging. Current research in China focuses on three main categories: data-driven models, process-driven models, and fusion models. Data-driven models require high-frequency and spatially dense data inputs, making data monitoring costly and collection inconvenient for large water areas. Process-driven models need to comprehensively consider meteorological, hydrological, water quality, sediment, and biological factors. Given the complexity and randomness of cyanobacterial bloom processes, these models require numerous input parameters and are difficult to construct. Fusion models require deep learning models combining multi-source data, making them complex and difficult to implement. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a method and apparatus for predicting algal blooms based on meteorological and remote sensing data, which improves the accuracy of algal bloom prediction while being universally applicable and easy to operate.

[0005] The technical solution provided by this invention is as follows:

[0006] A method for predicting algal blooms based on meteorological and remote sensing data, the method comprising:

[0007] S1: Acquire remote sensing images and meteorological data corresponding to the time points when the remote sensing images pass over the area;

[0008] The remote sensing images include near-infrared band, red band and short-wave infrared band;

[0009] S2: Obtain atmospheric reflectance data results using remote sensing images, and obtain the reflectance value of each pixel;

[0010] S3: For each pixel, calculate the difference value F between the near-infrared band and the baselines of the red band and short-wave infrared band;

[0011] F = R rc，NIR -R′ rc，NIR

[0012] R′ rc,NIR =R rc，RED +(R rc,SWIR -R rc，RED )*(λ NIR -λ RED ) / (λ SWIR -λ RED )

[0013] Among them, R rc,RED 、R rc,NIR and R rc,SWIR λ represents the reflectance values ​​for the red band, near-infrared band, and short-wave infrared band, respectively. RED , λ NIR and λ SWIR These represent the center wavelengths of the red light band, near-infrared band, and short-wave infrared band, respectively.

[0014] S4: Compare the difference value F of each pixel with the set threshold to obtain the binary map of algal bloom;

[0015] S5: Resample the meteorological data into a grid with the same resolution as the remote sensing image, and obtain the meteorological data for each grid.

[0016] S6: Take the difference value F of each pixel and the meteorological data of the corresponding raster as a data group, and obtain the data group corresponding to each pixel;

[0017] S7: Using the probability of algal bloom in each pixel as the dependent variable and the data set of each pixel as the independent variable, a regression model is established through a linear regression equation.

[0018] The regression model is expressed as: p = α + β1x1 + β2x2 + β3x3 + ... + β m x m

[0019] Where p is the probability of a pixel undergoing algal bloom, x1, x2, x3, ..., x m For each data point in the data set of this pixel, m is the total number of data points in the data set; α, β1, β2, β3, ..., β m The parameters to be determined;

[0020] S8: Based on the known probability of algal blooms occurring in each pixel and its data set, solve for the parameters of the regression model;

[0021] S9: Predict algal blooms based on the solved regression model.

[0022] Furthermore, S4 includes:

[0023] S41: Compare the difference value F of each pixel with the set threshold, and identify pixels with a difference value F greater than the threshold as algal bloom areas, otherwise identify them as non-algal bloom areas.

[0024] S42: Assign a value of 1 to each cell in the algal bloom region and a value of 0 to each cell in the non-algal bloom region to obtain the algal bloom binary image.

[0025] Furthermore, S8 includes:

[0026] S81: Use the value of each pixel on the binary map of algal bloom as the probability of algal bloom occurring in each pixel;

[0027] S82: Based on the probability of algal blooms occurring in each pixel and its data set, the parameters of the estimated regression model are solved using the maximum likelihood method.

[0028] Furthermore, S7 also includes:

[0029] Perform a logit transformation on p in the regression model, using logit(p) as the dependent variable, and transform the regression model as follows:

[0030] logit(p)=ln(p / (1-p))=α+β1x1+β2x2+β3x3+……+β m x m .

[0031] Furthermore, S9 includes:

[0032] S91: Acquire remote sensing images and meteorological data of the area to be predicted, and calculate the data set for each pixel to be predicted;

[0033] S92: Use the data set of each pixel to be predicted as the input of the solved regression model, and output the probability of algal bloom in the pixel to be predicted;

[0034] S93: Compare the probability of algal bloom in the cell to be predicted with the set algal bloom threshold. If the probability is greater than the algal bloom threshold, mark the cell to be predicted as an algal bloom area; otherwise, mark it as a non-algal bloom area.

[0035] S94: Determine the location of algal bloom based on the predicted algal bloom area, and calculate the predicted algal bloom area based on the number of pixels in the predicted algal bloom area and the spatial resolution of the pixels.

[0036] An algal bloom prediction device based on meteorological and remote sensing data, the device comprising:

[0037] The data acquisition module is used to acquire remote sensing images and meteorological data corresponding to the time points when the remote sensing images pass over the area;

[0038] The remote sensing images include near-infrared band, red band and short-wave infrared band;

[0039] The remote sensing data processing module is used to obtain atmospheric reflectance data results from remote sensing images and to obtain the reflectance value of each pixel.

[0040] The band difference calculation module is used to calculate the difference value F between the near-infrared band and the baselines of the red band and the short-wave infrared band for each pixel;

[0041] F = R rc，NIR -R′ rc，NIR

[0042] R′ rc,NIR =R rc，RED +(R rc,SWIR -R rc，RED )*(λ NIR -λ RED ) / (λ SWIR -λ RED )

[0043] Among them, R rc,RED 、R rc,NIR and R rc,SWIR λ represents the reflectance values ​​for the red band, near-infrared band, and short-wave infrared band, respectively. RED , λ NIR and λ SWIR These represent the center wavelengths of the red light band, near-infrared band, and short-wave infrared band, respectively.

[0044] The binary image calculation module is used to compare the difference value F of each pixel with the set threshold to obtain the algal bloom binary image.

[0045] The meteorological data processing module is used to resample the meteorological data into a grid with the same resolution as the remote sensing image, and to acquire the meteorological data of each grid.

[0046] The image data group acquisition module is used to collect the difference value F of each pixel and the meteorological data of its corresponding location raster as a data group, and to acquire the data group corresponding to each pixel.

[0047] The model building module is used to build a regression model using the probability of algal bloom in each pixel as the dependent variable and the data set of each pixel as the independent variable, through a linear regression equation.

[0048] The regression model is expressed as: p = α + β1x1 + β2x2 + β3x3 + ... + β m xm

[0049] Where p is the probability of a pixel undergoing algal bloom, x1, x2, x3, ..., x m For each data point in the data set of this pixel, m is the total number of data points in the data set; α, β1, β2, β3, ..., β m The parameters to be determined;

[0050] The model parameter solving module is used to solve the parameters of the regression model based on the known probability of algal blooms in each pixel and its data set;

[0051] The prediction module is used to predict algal blooms based on the solved regression model.

[0052] Furthermore, the binary image calculation module includes:

[0053] The comparison unit is used to compare the difference value F of each pixel with a set threshold, and to identify pixels with a difference value F greater than the threshold as algal bloom areas, and otherwise as non-algal bloom areas.

[0054] The assignment unit is used to assign a value of 1 to each cell in the algal bloom region and a value of 0 to each cell in the non-algal bloom region, thus obtaining the algal bloom binary image.

[0055] Furthermore, the model parameter solving module includes:

[0056] The probability determination unit uses the value of each pixel in the binary map of algal bloom as the probability of algal bloom occurring in each pixel.

[0057] The solution unit is used to solve for the parameters of the estimated regression model by using the maximum likelihood method based on the probability of algal blooms occurring in each pixel and its data set.

[0058] Furthermore, the model building module is also used for:

[0059] Perform a logit transformation on p in the regression model, using logit(p) as the dependent variable, and transform the regression model as follows:

[0060] logit(p)=ln(p / (1-p))=α+β1x1+β2x2+β3x3+……+β m x m .

[0061] Furthermore, the prediction module includes:

[0062] The prediction data acquisition unit is used to acquire remote sensing images and meteorological data of the area to be predicted, and to calculate the data set for each pixel to be predicted.

[0063] The prediction probability calculation unit is used to take the data set of each pixel to be predicted as the input of the solved regression model and output the probability of algal bloom in the pixel to be predicted.

[0064] The threshold comparison unit is used to compare the probability of algal bloom in the pixel to be predicted with the set algal bloom threshold. If the probability is greater than the algal bloom threshold, the pixel to be predicted is marked as an algal bloom region; otherwise, it is marked as a non-algal bloom region.

[0065] The area calculation unit is used to determine the location of algal bloom based on the predicted algal bloom area, and to calculate the predicted algal bloom area based on the number of pixels in the predicted algal bloom area and the spatial resolution of the pixels.

[0066] The present invention has the following beneficial effects:

[0067] This invention calculates the difference between the near-infrared band and the baselines of the red and short-wave infrared bands for each pixel, along with the corresponding meteorological data at the same time. Combining this with a binary map of algal bloom results, a regression model for algal bloom prediction is calculated using statistical methods from a set of data corresponding to each pixel. An algal bloom threshold is used to determine whether an algal bloom has occurred. Through data analysis, a threshold corresponding to algal bloom occurrence is selected, and algal bloom prediction is then performed for each pixel, ultimately yielding predictions of the location and area of ​​the algal bloom. The difference between the near-infrared band and the baselines of the red and short-wave infrared bands can represent the actual algal bloom situation, which helps improve the accuracy of algal bloom prediction. Furthermore, the method of this invention is universal, simple, and easy to operate, requiring no complex water quality or biological process data. It aims to provide new ideas for predicting cyanobacterial blooms in lakes, offer references for algal bloom control and management decisions, and provide suggestions for water environment protection measures in lake basins. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 This is a flowchart of the rapid monitoring method for vegetation anomalies along the ecological protection red line according to the present invention;

[0069] Figure 2 This is a schematic diagram of the rapid monitoring device for abnormal vegetation along the ecological protection red line according to the present invention. DETAILED DESCRIPTION

[0070] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0071] Example 1:

[0072] This invention provides a method for predicting algal blooms based on meteorological and remote sensing data, such as... Figure 1 As shown, the method includes:

[0073] S1: Acquire remote sensing images covering lakes and other bodies of water, as well as meteorological data corresponding to the time points when the remote sensing images pass over the area; the remote sensing images are required to include near-infrared (NIR), red (RED), and short-wave infrared (SWIR) bands.

[0074] In some cases, it is also necessary to remove the non-water areas by separating the land and water or by vector cropping, and only retain the remote sensing images of the water areas.

[0075] To improve the accuracy of algal bloom prediction, long-term time-series analysis of cyanobacterial blooms in lakes can be conducted to obtain multiple long-term remote sensing images, providing a data foundation for improving model accuracy.

[0076] S2: Obtain atmospheric reflectance data results using remote sensing images, and obtain the reflectance value of each pixel.

[0077] In this invention, atmospheric reflectance data can be obtained by using preprocessing methods such as atmospheric correction; or, if atmospheric reflectance data already exists, it can be obtained directly.

[0078] S3: For each pixel, calculate the difference value F between its near-infrared band and the baselines of the red band and short-wave infrared band;

[0079] F = R rc，NIR -R′ rc，NIR

[0080] R′ rc,NIR =R rc，RED +(R rc,SWIR -R rc，RED )*(λ NIR -λ RED ) / (λ SWIR -λ RED )

[0081] Among them, R rc,RED 、R rc,NIR and R rc,SWIR λ represents the reflectance values ​​for the red band, near-infrared band, and short-wave infrared band, respectively. RED , λ NIR and λ SWIR These represent the center wavelengths of the red light band, near-infrared band, and short-wave infrared band, respectively.

[0082] S4: Compare the difference value F of each pixel with the set threshold to obtain the binary map of algal bloom.

[0083] Specifically, this step includes:

[0084] S41: Compare the difference value F of each pixel with the set threshold. Pixels with a difference value F greater than the threshold are identified as algal bloom areas, and others are identified as non-algal bloom areas.

[0085] S42: Assign each pixel in the algal bloom region a value of 1 and each pixel in the non-algal bloom region a value of 0 to obtain the algal bloom binary image and obtain the corresponding 0 / 1 values ​​for each pixel.

[0086] S5: Resample the meteorological data into rasters with the same resolution as the remote sensing image, and obtain the meteorological data for each raster.

[0087] S6: Collect the difference value F of each pixel and the meteorological data of the corresponding raster as a data group, and obtain the data group corresponding to each pixel.

[0088] For example, a set of data for a single cell can be represented as x1, x2, x3, ..., x m , where m is the total number of data in the data set.

[0089] S7: Using the probability of algal bloom occurring in each pixel as the dependent variable and the data set of each pixel as the independent variable, a regression model is established by fitting the variables through a linear regression equation.

[0090] The regression model is expressed as: p = α + β1x1 + β2x2 + β3x3 + ... + β m x m

[0091] Where p is the dependent variable, i.e., the probability of a pixel experiencing algal bloom. The dependent variable is defined as having a value of 1 when algal bloom occurs, and 0 otherwise. Thus, the probability of algal bloom is p (y = 1), and 0 ≤ p ≤ 1. x1, x2, x3, ..., x m For each data point in the data set of this pixel, m is the total number of data points in the data set; α, β1, β2, β3, ..., β m The parameter to be determined.

[0092] S8: Based on the known probability of algal blooms occurring in each pixel and its data set, solve for the parameters of the regression model.

[0093] This step can be implemented in the following ways:

[0094] S81: Use the value of each pixel in the binary map of algal bloom as the probability of algal bloom occurring in each pixel.

[0095] In other words, the dependent variable takes the value of 1 when algal bloom occurs, and 0 otherwise. Since the binary map of algal bloom is known, the probability of algal bloom occurring in each pixel is known.

[0096] S82: Based on the probability of algal blooms occurring in each pixel and its data set, the parameters of the estimated regression model are solved using the maximum likelihood method.

[0097] When solving the problem, since p is a binary variable that can only take the values ​​0 or 1, the linear regression model encounters difficulties: the right side of the regression model equation has continuous values ​​while the left side can only take the values ​​[0,1], which cannot be correlated. Therefore, this invention performs a logit transformation on p, using logit(p) as the dependent variable to establish the regression model.

[0098] logit(p)=ln(p / (1-p))=α+β1x1+β2x2+β3x3+……+β m x m

[0099] S9: Predict algal blooms based on the solved regression model.

[0100] Specifically, this step includes:

[0101] S91: Acquire remote sensing images and meteorological data of the area to be predicted, and calculate the data set for each pixel to be predicted.

[0102] The specific calculation method is similar to that of S1 to S3, S5 and S6 mentioned above, and will not be repeated here.

[0103] S92: Use the data set of each pixel to be predicted as input to the solved regression model, and output the probability of algal bloom in the pixel to be predicted.

[0104] S93: Compare the probability of algal bloom in the cell to be predicted with the set algal bloom threshold. If the probability is greater than the algal bloom threshold, mark the cell to be predicted as an algal bloom region; otherwise, mark it as a non-algal bloom region.

[0105] For example, the algal bloom threshold can be set based on historical algal bloom data.

[0106] S94: Determine the location of algal bloom based on the predicted algal bloom area, and calculate the predicted algal bloom area based on the number of pixels in the predicted algal bloom area and the spatial resolution of the pixels.

[0107] This invention calculates the difference between the near-infrared band and the baselines of the red and short-wave infrared bands for each pixel, along with the corresponding meteorological data at the same time. Combining this with the binary map of algal bloom results, a regression model for algal bloom prediction is calculated using statistical methods from a set of data corresponding to each pixel. Based on the algal bloom threshold, it is determined whether algal bloom has occurred. Through data analysis, the threshold corresponding to the occurrence of algal bloom is selected, and then algal bloom prediction is carried out for each pixel. Finally, the prediction results of the location and size of the algal bloom are obtained.

[0108] This invention, based on routine monitoring of cyanobacterial blooms in lakes, introduces pixel-by-pixel differences between the near-infrared, red, and short-wave infrared baselines, along with meteorological data. These differences represent the actual bloom conditions, improving the accuracy of bloom prediction. Furthermore, the method is universally applicable, simple, and easy to operate, requiring no complex water quality or biological process data. It aims to provide new insights for predicting cyanobacterial blooms in lakes, offer references for bloom control and management decisions, and provide suggestions for water environment protection measures in lake basins.

[0109] Example 2:

[0110] This invention provides an algal bloom prediction device based on meteorological and remote sensing data, such as... Figure 2 As shown, the device includes:

[0111] Data acquisition module 1 is used to acquire remote sensing images and meteorological data corresponding to the time points when the remote sensing images pass over the area.

[0112] The remote sensing images include near-infrared, red, and short-wave infrared bands.

[0113] Remote sensing data processing module 2 is used to obtain atmospheric reflectance data results from remote sensing images and obtain the reflectance value of each pixel.

[0114] Band difference calculation module 3 is used to calculate the difference value F between the near-infrared band and the baselines of the red band and short-wave infrared band for each pixel.

[0115] F = R rc，NIR -R′ rc，NIR

[0116] R′ rc,NIR =R rc，RED +(R rc，SWIR -R rc，RED )*(λ NIR -λ RED ) / (λ SWIR -λ RED )

[0117] Among them, R rc,RED 、R rc,NIR and R rc,SWIR λ represents the reflectance values ​​for the red band, near-infrared band, and short-wave infrared band, respectively. RED , λ NIR and λ SWIR These represent the center wavelengths of the red light band, near-infrared band, and short-wave infrared band, respectively.

[0118] Binary image calculation module 4 is used to compare the difference value F of each pixel with the set threshold to obtain the algal bloom binary image.

[0119] Meteorological data processing module 5 is used to resample meteorological data into rasters with the same resolution as remote sensing images, and to acquire meteorological data for each raster.

[0120] The image data group acquisition module 6 is used to collect the difference value F of each pixel and the meteorological data of its corresponding location raster as a data group, and to acquire the data group corresponding to each pixel.

[0121] Model building module 7 is used to build a regression model using the probability of algal bloom in each pixel as the dependent variable and the data set of each pixel as the independent variable, through a linear regression equation.

[0122] The regression model is expressed as: p = α + β1x1 + β2x2 + β3x3 + ... + β m x m

[0123] Where p is the probability of a cell undergoing algal bloom, x1, x2, x3, ..., x m For each data point in the data set of this pixel, m represents the total number of data points in the data set; α, β1, β2, β3, ..., β m These are the parameters to be determined.

[0124] Model parameter solving module 8 is used to solve the parameters of the regression model based on the known probability of algal blooms in each pixel and its data set.

[0125] Prediction module 9 is used to predict algal blooms based on the solved regression model.

[0126] As an example, the binary graph computation module includes:

[0127] The comparison unit is used to compare the difference value F of each pixel with a set threshold. Pixels with a difference value F greater than the threshold are identified as algal bloom areas, and those without are identified as non-algal bloom areas.

[0128] The assignment unit is used to assign a value of 1 to each cell in the algal bloom region and a value of 0 to each cell in the non-algal bloom region, thus obtaining the algal bloom binary image.

[0129] The aforementioned model parameter solution module includes:

[0130] The probability determination unit uses the value of each pixel in the binary map of algal bloom as the probability of algal bloom occurring in each pixel.

[0131] The solution unit is used to solve for the parameters of the estimated regression model by using the maximum likelihood method based on the probability of algal blooms occurring in each pixel and its data set.

[0132] Furthermore, the model building module is also used for:

[0133] Perform a logit transformation on p in the regression model, using logit(p) as the dependent variable, and transform the regression model as follows:

[0134] logit(p)=ln(p / (1-p))=α+β1x1+β2x2+β3x3+……+β m x m .

[0135] As one implementation method, the prediction module includes:

[0136] The prediction data acquisition unit is used to acquire remote sensing images and meteorological data of the area to be predicted, and to calculate the data set for each pixel to be predicted.

[0137] The prediction probability calculation unit is used to take the data set of each pixel to be predicted as the input of the solved regression model and output the probability of algal bloom in the pixel to be predicted.

[0138] The threshold comparison unit is used to compare the probability of algal bloom in the pixel to be predicted with the set algal bloom threshold. If the probability is greater than the algal bloom threshold, the pixel to be predicted is marked as an algal bloom region; otherwise, it is marked as a non-algal bloom region.

[0139] The area calculation unit is used to determine the location of algal bloom based on the predicted algal bloom area, and to calculate the predicted algal bloom area based on the number of pixels in the predicted algal bloom area and the spatial resolution of the pixels.

[0140] The apparatus provided in this embodiment of the invention has the same implementation principle and technical effects as the aforementioned method embodiment. For the sake of brevity, any parts not mentioned in the apparatus embodiment can be referred to the corresponding content in the aforementioned method embodiment 1. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the apparatus and units described above can all be referred to the corresponding processes in the above method embodiments, and will not be repeated here.

[0141] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit it. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the scope of the technology disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. All should be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for predicting water bloom based on meteorological and remote sensing data, characterized in that: The method comprises: S1: Obtain remote sensing images and meteorological data corresponding to the time points when the remote sensing images pass through; Wherein, the remote sensing image includes near infrared band, red light band and short-wave infrared band; S2: Use remote sensing images to obtain the reflectance data of the bottom layer of the atmosphere and obtain the reflectance value of each pixel; S3: For each pixel, calculate the difference value F between the near infrared band and the baselines of the red light band and the short-wave infrared band; F=R rc,NIR -R ′ rc,NIR R ′ rc,NIR =R rc,RED +(R rc,SWIR -R rc,RED )*(λ NIR -l RED ) / (λ SWIR -l RED ) Among them, R rc,RED , R rc,NIR and R rc,SWIR Respectively represent the reflectivity values ​​of the red light band, near infrared band and short-wave infrared band, λ RED ,λ NIR and λ SWIR Represent the central wavelengths of the red light band, near infrared band and short-wave infrared band respectively; S4: Compare the difference value F of each pixel with the set threshold value to obtain a binary map of water bloom; S5: resampling the meteorological data into a grid having the same resolution as the remote sensing image, and obtaining meteorological data of each grid; S6: taking the difference value F of each pixel and the meteorological data of the corresponding position grid as a data group, and obtaining the data group corresponding to each pixel; S7: Taking the probability of water bloom in each pixel as the dependent variable and the data set of each pixel as the independent variable, a regression model is established through a linear regression equation; The expression of the regression model is: p = α + β1x1 + β2x2 + β3x3 + ... + β m x m Among them, p is the probability of water bloom in the pixel, x1,x2,x3,…,x m is each data in the data set of the pixel, m is the total number of data in the data set; α, β1, β2, β3, …, β m is the parameter to be requested; S8: according to the known probability of algal bloom in each pixel and its data set, solve the parameters of the regression model; S9: Predict algal blooms based on the solved regression model; The S9 includes: S91: Acquire remote sensing images and meteorological data of the area to be predicted, and calculate a data group for each pixel to be predicted; S92: taking the data group of each pixel to be predicted as the input of the solved regression model, and outputting the probability of occurrence of water bloom in the pixel to be predicted; S93: comparing the probability of water bloom occurring in the pixel to be predicted with the set water bloom threshold, if the probability is greater than the water bloom threshold, marking the pixel to be predicted as a predicted water bloom area, otherwise marking it as a predicted non-water bloom area; S94: determining the location of the water bloom according to the predicted water bloom area, and calculating the predicted water bloom area according to the number of pixels in the predicted water bloom area and the spatial resolution of the pixels.

2. The water bloom prediction method based on meteorological and remote sensing data according to claim 1, characterized in that: The S4 includes: S41: Compare the difference value F of each pixel with the set threshold value, and regard the pixel whose difference value F is greater than the threshold value as the water bloom area, otherwise as the non-water bloom area; S42: Assign a value of 1 to each pixel in the water bloom area, and assign a value of 0 to each pixel in the non-water bloom area, to obtain a water bloom binary map.

3. The water bloom prediction method based on meteorological and remote sensing data according to claim 2, characterized in that: The S8 includes: S81: taking the value of each pixel on the algae bloom binary map as the probability of algae bloom occurring in each pixel; S82: Based on the probability of algal bloom occurring in each pixel and its data set, the parameters of the estimated regression model are solved by the maximum likelihood method.

4. The water bloom prediction method based on meteorological and remote sensing data according to claim 3, characterized in that: The S7 further comprises: Perform logit transformation on p of the regression model, take logit(p) as the dependent variable, and transform the regression model into: logit(p)=ln(p / (1-p))=α+β1x1+β2x2+β3x3+……+β m x m 。 5. A water bloom prediction device based on meteorological and remote sensing data, characterized in that: The device comprises: A data acquisition module is used to acquire remote sensing images and meteorological data corresponding to the time points at which the remote sensing images pass through; Wherein, the remote sensing image includes near infrared band, red light band and short-wave infrared band; The remote sensing data processing module is used to obtain the reflectivity data of the bottom layer of the atmosphere using remote sensing images and obtain the reflectivity value of each pixel; The band difference calculation module is used to calculate the difference value F between the near infrared band and the baselines of the red light band and the short-wave infrared band for each pixel; F=R rc,NIR -R ′ rc,NIR R ′ rc,NIR =R rc,RED +(R rc,SWIR -R rc,RED )*(λ NIR -l RED ) / (λ SWIR -l RED ) Among them, R rc,RED , R rc,NIR and R rc,SWIR Respectively represent the reflectivity values ​​of the red light band, near infrared band and short-wave infrared band, λ RED ,λ NIR and λ SWIR Represent the central wavelengths of the red light band, near infrared band and short-wave infrared band respectively; A binary image calculation module is used to compare the difference value F of each pixel with the set threshold value to obtain a binary image of water bloom; A meteorological data processing module, used for resampling the meteorological data into a grid with the same resolution as the remote sensing image, and obtaining meteorological data for each grid; A pixel data group acquisition module is used to take the difference value F of each pixel and the meteorological data of the corresponding position grid as a data group, and acquire the data group corresponding to each pixel; A model building module is used to build a regression model through a linear regression equation using the probability of water bloom occurring in each pixel as a dependent variable and the data group of each pixel as an independent variable; The expression of the regression model is: p = α + β1x1 + β2x2 + β3x3 + ... + β m x m Among them, p is the probability of water bloom in the pixel, x1,x2,x3,…,x m is each data in the data set of the pixel, m is the total number of data in the data set; α, β1, β2, β3, …, β m is the parameter to be requested; The model parameter solving module is used to solve the parameters of the regression model according to the known probability of water bloom occurrence in each pixel and its data group; A prediction module, used to predict algal blooms based on the solved regression model; The prediction module comprises: A prediction data acquisition unit is used to acquire remote sensing images and meteorological data of the area to be predicted, and calculate a data group for each pixel to be predicted; A prediction probability calculation unit is used to use the data group of each pixel to be predicted as the input of the solved regression model and output the probability of water bloom occurring in the pixel to be predicted; A threshold comparison unit is used to compare the probability of water bloom occurring in the pixel to be predicted with the set water bloom threshold. If the probability is greater than the water bloom threshold, the pixel to be predicted is marked as a predicted water bloom area, otherwise it is marked as a predicted non-water bloom area; The area calculation unit is used to determine the location of the water bloom according to the predicted water bloom area, and calculate the predicted water bloom area according to the number of pixels in the predicted water bloom area and the spatial resolution of the pixels.

6. The water bloom prediction device based on meteorological and remote sensing data according to claim 5, characterized in that: The binary image calculation module includes: A comparison unit, used for comparing the difference value F of each pixel with a set threshold value, and taking the pixel whose difference value F is greater than the threshold value as an algae bloom area, otherwise as a non-algae bloom area; The assignment unit is used to assign a value of 1 to each pixel in the water bloom area and a value of 0 to each pixel in the non-water bloom area, thereby obtaining a water bloom binary map.

7. The water bloom prediction device based on meteorological and remote sensing data according to claim 6, characterized in that: The model parameter solving module includes: The probability determination unit uses the value of each pixel on the water bloom binary map as the probability of water bloom occurring in each pixel; The solving unit is used to solve and estimate the parameters of the regression model through the maximum likelihood method according to the probability of water bloom occurring in each pixel and its data group.

8. The water bloom prediction device based on meteorological and remote sensing data according to claim 7, characterized in that: The model building module is also used to: Perform logit transformation on p of the regression model, take logit(p) as the dependent variable, and transform the regression model into: logit(p)=ln(p / (1-p))=α+β1x1+β2x2+β3x3+……+β m x m 。

Images