Multi-point Inversion Method of Permanganate Index Based on Hyperspectral Remote Sensing and Water Quality Sequences

Abstract

The present invention discloses a method for multi-point inversion of permanganate index based on hyperspectral remote sensing and water quality sequence: obtaining hyperspectral remote sensing images and water quality time series data of points to be inverted; constructing a remote sensing regression inversion model and training it using a spectral information training set; constructing a Gaussian hidden Markov model and training it using grouped water quality time series data; inputting the spectral information in the test set into the remote sensing regression inversion model to obtain the inversion results of the points to be inverted; inputting the grouped water quality time series data in the test set into the Gaussian hidden Markov model, and inferring the current hidden state of the Gaussian hidden Markov model using the inversion results of the points to be inverted to obtain the final permanganate index prediction results of each point. This method combines a Gaussian hidden Markov model and a hyperspectral remote sensing regression model, utilizes multi-dimensional data, reduces the influence of sparse sampling points on accuracy, and simultaneously suppresses the interference of the phenomenon of different substances with the same spectrum, improving the accuracy of multi-point inversion of permanganate index.

Description

Technical Field

[0001] The present invention belongs to the field of water resource monitoring, and particularly relates to a multi-point inversion method for permanganate index based on hyperspectral remote sensing and water quality sequence. Background Art

[0002] Inland water bodies such as rivers and lakes play an important role in maintaining ecological balance. However, with the rapid development of social economy in recent years, the pollution and damage of water bodies have been continuously aggravated. Therefore, accurate and real-time monitoring of organic pollution at important points in the basin is the long-term foundation for ecological governance and restoration. The permanganate index (COD Mn ) refers to the amount of oxidant consumed when reducing substances in a unit water body are oxidized under specified conditions, and is an important indicator reflecting organic pollution in the basin. Monitoring it has quite important significance.

[0003] Traditional ground determination of COD Mn relies on laboratory sampling determination or automatic station determination, which has the problems of high consumption of human, material and financial resources and long determination period. With the continuous development of remote sensing technology, using satellite images for water quality inversion has gradually become the mainstream. However, the remote sensing inversion model of COD Mn is difficult to capture the correlation between COD Mn and spectral bands, and at the same time has relatively high requirements for the density of water quality sampling points, resulting in poor inversion accuracy and weak migration ability of the model.

[0004] In addition to the above two mainstream methods, the method of predicting based on water quality time series has also gradually developed in water quality monitoring technology, but its prediction accuracy is high while its ability to cope with sudden environmental changes is weak.

[0005] Currently, existing remote sensing inversion methods of COD Mn have relatively high requirements for spectral correlation and the distribution of water quality sampling points, resulting in difficulty for conventional remote sensing inversion models to adapt to scenarios with sparse sampling points. And simply relying on water quality time series for COD Mn prediction is difficult to capture sudden changes in water quality and cannot be applied in practice. Summary of the Invention

[0006] The present invention provides a multi-point inversion method for permanganate index based on hyperspectral remote sensing and water quality sequence, which combines the COD Mn water quality data and remote sensing spectral data of the point to be inverted for joint prediction, reduces the influence of sparse sampling points on accuracy and captures the change trend of COD Mn , and then improves the inversion accuracy of COD Mn . The multi-point inversion method for permanganate index includes:

[0007] Obtain the hyperspectral remote sensing image of the point to be inverted and perform preprocessing to obtain the preprocessed hyperspectral remote sensing image. Obtain the permanganate index data collected by the water quality automatic monitoring station within the range of the hyperspectral remote sensing image to obtain the permanganate index water quality time series data.

[0008] Extract the spectral information of the preprocessed hyperspectral remote sensing image and perform band alignment. Extract the permanganate index water quality time series data and perform time grouping to obtain the permanganate index water quality time series grouped data. Randomly sample the band-aligned spectral information and the permanganate index water quality time series grouped data at a ratio of 2:1 as the training set and the test set respectively.

[0009] Construct a remote sensing regression inversion model. The remote sensing regression inversion model includes sensitive band screening, inversion band combination screening, and determination of the final regression inversion model. Use the spectral information training set to train the remote sensing regression inversion model.

[0010] Construct a Gaussian hidden Markov model. The Gaussian hidden Markov model includes three hidden states and optimal step size selection. Use the permanganate index water quality time series grouped data training set to input into the Gaussian hidden Markov model for training.

[0011] Input the spectral information in the test set into the pre-trained remote sensing regression inversion model to obtain the inversion result of the point to be inverted; input the permanganate index water quality time series grouped data in the test set into the pre-trained Gaussian hidden Markov model to obtain the prediction results under three different hidden states.

[0012] Use the inversion result of the point to be inverted to infer the current hidden state of the Gaussian hidden Markov model to obtain the final permanganate index prediction result for each point. Specifically: verify the accuracy of the final regression inversion model to obtain the mean error as the threshold, and use the previous day's data in the permanganate index water quality time series data as the reference value; sort the prediction results of the Gaussian hidden Markov model from small to large to obtain the first hidden state, the second hidden state, and the third hidden state; judge whether the regression inversion result fluctuates compared with the previous day's data. If there is no fluctuation or the fluctuation range up and down does not exceed the threshold, select the second hidden state as the final permanganate index prediction result; if the downward fluctuation range exceeds the threshold, select the first hidden state as the final permanganate index prediction result; if the upward fluctuation range exceeds the threshold, select the third hidden state as the final permanganate index prediction result.

[0013] Further, the preprocessing is to perform radiometric calibration, atmospheric correction, and orthorectification on the hyperspectral remote sensing image. The calculation formula for radiometric calibration is: L = Gain·DN + Bias, where L is the pixel radiance, with the unit of W / (cm 2 ·μm·sr), DN is the pixel brightness value of the remote sensing image, Gain is the radiometric calibration gain, and Bias is the radiometric calibration offset, both with the unit of W / (cm 2 ·μm·sr); the calculation formula for atmospheric correction is: where C is the pixel radiance after atmospheric correction, ρ is the pixel reflectance, ρ e is the average reflectance of the surrounding area of the pixel, A and B are calculated from the actual atmospheric and geometric conditions, S is the atmospheric spherical albedo, and C a is the path radiance of the atmosphere.

[0014] Further, obtain the permanganate index data collected by the water quality automatic monitoring stations within the range of the hyperspectral remote sensing image. Specifically, obtain the permanganate index data collected in the 14 days before the shooting date of the hyperspectral remote sensing image, and group the permanganate index water quality time series data in units of N days to obtain the grouped data of the permanganate index water quality time series, where 7 ≤ N ≤ 14.

[0015] Further, the processing method for band alignment is as follows: for the hyperspectral remote sensing images taken by different hyperspectral satellites, select the image taken by one of the hyperspectral satellites as the reference image, and the others as the images to be registered; when the difference between the central bands of each band of the image to be registered and each band of the reference image is within the specified range, keep the band; when the difference exceeds the specified range, take the mean of the band of the image to be registered and the previous or next band as the new central band, so that the difference of the new central band is within the specified range, and take the mean of the radiance values of the previous and next bands as the radiance value of the new central band.

[0016] Further, for the sensitive band screening and inversion band combination screening in the remote sensing regression inversion model, the Pearson correlation coefficient method is used to select the band or band combination with the highest correlation as the calculation formula from spectral information to the independent variable of remote sensing regression inversion. The formula for the Pearson correlation coefficient is: where r is the Pearson correlation coefficient, X i is the independent variable of remote sensing regression inversion, Y i is the measured permanganate index water quality data on the day when the hyperspectral remote sensing image corresponding to X i was taken, n is the number of samples, is the sample mean of X i and Y i , and S X and SY is the sample standard deviation; among the final regression inversion models in the remote sensing regression inversion model, after the linear regression model, logarithmic regression model, polynomial regression model, and power regression model are established, R 2 is used for accuracy verification, and the regression inversion model with the highest coefficient of determination is selected. Among them, where R 2 is the coefficient of determination, SSR is the regression sum of squares, and SST is the total sum of squares. is the inverted water quality corresponding to Y i .

[0017] Furthermore, the Gaussian hidden Markov model includes:

[0018] Construct a Gaussian hidden Markov model for describing the hidden state of water quality, and initialize the parameters of the model, including the state transition matrix, initial probability matrix, mean matrix, and covariance matrix;

[0019] Use the grouped data of the permanganate index water quality time series to construct the Gaussian hidden Markov observation sequence;

[0020] Set the number of water quality hidden states to 3, and use the Baum-Welch algorithm to estimate the parameters of the Gaussian hidden Markov model to obtain the trained state transition matrix, initial probability matrix, mean matrix, and covariance matrix;

[0021] Calculate the weighted average of the expected observed values of each state in the next stage through the state transition matrix and the mean matrix, and use it as the predicted value of the hidden state.

[0022] Preferably, use test samples to verify the accuracy of the permanganate remote sensing inversion results.

[0023] Preferably, the mean error is taken from the mean absolute error of the final regression inversion model during the training process, and its formula is where MAE is the mean absolute error, Y i is the measured permanganate index water quality data, is the inverted water quality corresponding to Y i .

[0024] Preferably, the hyperspectral remote sensing images are taken from the Ziyuan-1 02D hyperspectral camera, Ziyuan-1 02E hyperspectral camera, and GF-5 02 satellite hyperspectral camera.

[0025] The multi-point inversion method of permanganate index based on hyperspectral remote sensing and water quality sequence provided by the present invention combines hyperspectral remote sensing images and measured water quality time series data, uses multi-dimensional data, reduces the influence of sparse sampling points on accuracy, and simultaneously suppresses the interference of the phenomenon of different substances with the same spectrum, thereby improving the accuracy of multi-point inversion of permanganate index. Description of the Drawings

[0026] Figure 1 It is a flow chart of the multi-point inversion method of permanganate index based on hyperspectral remote sensing and water quality sequence.

[0027] Figure 2 It is a schematic diagram of the model principle of the multi-point inversion method of permanganate index.

[0028] Figure 3 It is a comparison chart of the inversion errors of each model in the embodiment. Detailed Embodiment

[0029] The embodiment of the present invention provides a multi-point inversion method of permanganate index based on hyperspectral remote sensing and water quality sequence. As shown in the attached Figure 1 and attached Figure 2 shown, the multi-point inversion method of permanganate index includes:

[0030] First step, obtain the hyperspectral remote sensing image of the point to be inverted and perform preprocessing to obtain the preprocessed hyperspectral remote sensing image, and obtain the permanganate index data collected by the water quality automatic monitoring station within the range of the hyperspectral remote sensing image to obtain the permanganate index water quality time series data.

[0031] In this step, the hyperspectral remote sensing image is taken from the hyperspectral camera of Ziyuan-1 02D, the hyperspectral camera of Ziyuan-1 02E, and the hyperspectral camera of GF-5 02 satellite, and the permanganate index water quality time series data is taken from the permanganate index data collected by the water quality automatic monitoring station set within the range of the remote sensing image in the 14 days before the shooting date of the hyperspectral remote sensing image.

[0032] The preprocessing is to perform radiometric calibration, atmospheric correction, and orthorectification on the hyperspectral remote sensing image. The calculation formula for radiometric calibration is: L = Gain·DN + Bias, where L is the pixel radiance, with the unit of W / (cm 2 ·μm·sr); DN is the pixel brightness value of the remote sensing image, Gain is the radiometric calibration gain, and Bias is the radiometric calibration offset, both with the unit of W / (cm 2 ·μm·sr); the calculation formula for atmospheric correction is: where C is the pixel radiance after atmospheric correction, ρ is the pixel reflectivity, ρ eis the average reflectance of the surrounding area of the pixel, and A and B are calculated from actual atmospheric and geometric conditions; S is the atmospheric spherical albedo, and Ca is the atmospheric path radiance.

[0033] In the second step, extract the spectral information of the preprocessed hyperspectral remote sensing image and perform band alignment, extract the time series data of the permanganate index water quality and perform time grouping to obtain the grouped data of the time series of the permanganate index water quality. Randomly sample the band-aligned spectral information and the grouped data of the time series of the permanganate index water quality at a ratio of 2:1 as the training set and the test set respectively.

[0034] In this step, the processing method for band alignment is as follows: Process the hyperspectral remote sensing images taken by different hyperspectral satellites respectively, select the image taken by one hyperspectral satellite as the reference image, and the rest as the images to be registered; When the difference between the central bands of each band of the image to be registered and the central bands of each band of the reference image is within the specified range, retain the band; When the difference exceeds the specified range, take the average value of the current band and the previous or next band of the image to be registered as the new central band, so that the difference of the new central band is within the specified range, and take the average value of the radiance values of the previous and next bands as the radiance value of the new central band. The processing method for time grouping is as follows: According to the shooting date of the hyperspectral remote sensing image, extract the permanganate index data within a specific time period before this shooting time from the water quality time series dataset to form grouped data, and group the time series data of the permanganate index water quality in units of N days, where 7 ≤ N ≤ 14. In this embodiment, select the hyperspectral image taken by Ziyuan-1 02E satellite as the reference image, and the comparison results of the band alignment part are shown in Tables 1 and 2.

[0035] Table 1 Central wavelengths of each band before band alignment

[0036]

[0037]

[0038] Table 2 Central wavelengths of each band after band alignment

[0039] Band Name ZY-1 02E ZY-1 02E ZY-1 02E b1 395.44 395.86 395.60 b2 404.03 404.45 404.15 b3 412.61 413.04 412.70 b4 421.20 421.63 421.25 b5 429.79 430.23 429.80 b6 438.37 438.58 438.36 b7 446.96 447.10 446.91 b8 455.54 455.79 455.46 b9 464.13 464.37 464.01 b10 472.71 473.21 472.56 … … … …

[0040] After obtaining the above band alignment results, the band-aligned spectral information and the time series data of the permanganate index water quality can also be randomly sampled at a ratio of 2:1 as the training set and the test set respectively, and the accuracy of the hyperspectral remote sensing inversion result can be verified using the test set.

[0041] In the third step, a remote sensing regression inversion model is constructed. The remote sensing regression inversion model includes sensitive band screening, inversion band combination screening, and determination of the final regression inversion model. The spectral information training set is used to input into the remote sensing regression inversion model for training.

[0042] For the sensitive band screening and inversion band combination screening in the remote sensing regression inversion model, the Pearson correlation coefficient method is used to select the band or band combination with the highest correlation as the calculation formula from the spectral band information to the independent variable of the remote sensing regression inversion. The formula for the Pearson correlation coefficient is: where r is the Pearson correlation coefficient, X i is the independent variable of the remote sensing regression inversion, Y i is the measured permanganate index water quality data corresponding to the day when the hyperspectral remote sensing image was taken, n is the number of samples, i is the sample mean of X and Y i and S i are the sample standard deviations; for the final regression inversion model in the remote sensing regression inversion model, after the linear regression model, logarithmic regression model, polynomial regression model, and power regression model are established, R X is used for accuracy verification, and the regression inversion model with the highest coefficient of determination is selected. Among them, Y where R 2 is the coefficient of determination, SSR is the regression sum of squares, SST is the total sum of squares, is the inverted water quality corresponding to the measured water quality Y 2 . In this embodiment, the band and band combination with the highest correlation is b22, and the final regression inversion model is a polynomial model. is the inverted water quality corresponding to the measured water quality Y i . In this embodiment, the band and band combination with the highest correlation is b22, and the final regression inversion model is a polynomial model.

[0043] In the fourth step, a Gaussian hidden Markov model is constructed. The Gaussian hidden Markov model includes three hidden states and optimal step size selection. The permanganate index water quality time series data training set is used to train the Gaussian hidden Markov model.

[0044] In this step, the Gaussian hidden Markov model includes:

[0045] Construct a Gaussian hidden Markov model for describing the hidden state of water quality, and initialize the parameters of the model, including the state transition matrix, initial probability matrix, mean matrix, and covariance matrix;

[0046] Use the permanganate index water quality time series grouped data to construct the Gaussian hidden Markov observation sequence;

[0047] Set the number of water quality hidden states to 3, and use the Baum-Welch algorithm to estimate the parameters of the Gaussian hidden Markov model to obtain the trained state transition matrix, initial probability matrix, mean matrix, and covariance matrix;

[0048] Calculate the weighted average of the expected observation values of each state in the next stage through the state transition matrix and the mean matrix, and use it as the predicted value of the hidden state.

[0049] The Gaussian hidden Markov model contains three hidden states. The parameter estimation uses the Baum-Welch algorithm and does not use the state sequence inference algorithm. After the permanganate index grouped data is input into the Gaussian hidden Markov model, three prediction results corresponding to the three hidden states are generated. In this embodiment, the minimum step size for training the Gaussian hidden Markov model is 7, and the maximum step size is 14. The accuracies of different step sizes obtained using the training set are shown in Table 3. Finally, step size 10 is selected as the step size for the subsequent Gaussian hidden Markov model.

[0050] Table 3 Changes in model accuracy under different step sizes

[0051]

[0052]

[0053] Step 5: Input the spectral information in the test set into the pre-trained remote sensing regression inversion model to obtain the inversion results of the points to be inverted; input the permanganate index water quality time series grouped data in the test set into the pre-trained Gaussian hidden Markov model to obtain the prediction results under three different hidden states.

[0054] In this embodiment, the pre-trained remote sensing regression inversion model refers to a polynomial regression model fitted using the spectral band information training set, and the pre-trained Gaussian hidden Markov model refers to determining the model step size as 10 using the permanganate index training set grouped data.

[0055] Step 6: Infer the current hidden state of the Gaussian hidden Markov model using the inversion results of the points to be inverted to obtain the final permanganate index prediction results for each point.

[0056] The steps for obtaining the permanganate index prediction result using the inversion result are as follows: First, verify the accuracy of the final regression inversion model to obtain the mean error as the threshold, and use the data of the previous day in the permanganate index water quality time series data as the reference value; Second, sort the three prediction results generated by the Gaussian hidden Markov model from smallest to largest, corresponding to the first, second, and third hidden states respectively; Third, determine whether the regression inversion result fluctuates compared with the data of the previous day. If there is no fluctuation or the range of up and down fluctuations does not exceed the threshold, select the second hidden state as the final permanganate index prediction result; if the downward fluctuation range exceeds the threshold, select the first hidden state as the final permanganate index prediction result; if the upward fluctuation range exceeds the threshold, select the third hidden state as the final permanganate index prediction result.

[0057] In this embodiment, the mean error is taken from the mean absolute error of the final regression inversion model during the training process, and its formula is where MAE is the mean absolute error.

[0058] After obtaining the final permanganate index prediction results at each point, the accuracy of the permanganate remote sensing inversion results can also be verified using the test samples. The results of the accuracy verification are shown in Table 4, and the comparison of the inversion errors of each model is attached Figure 3 as shown, where Figure 3 (a) is the inversion error of the polynomial inversion model, Figure 3 (b) is the inversion error of the Gaussian hidden Markov model, Figure 3 (c) is the inversion error of the multi-point inversion method for permanganate index based on hyperspectral remote sensing and water quality sequence.

[0059] Table 4 Results of accuracy verification

[0060]

[0061] The above multi-point inversion method for permanganate index by combining Gaussian hidden Markov and hyperspectral remote sensing regression is constructed as follows: 1) Set the mean absolute error of the regression inversion model as the threshold, and use the data of the permanganate index grouped training set with a length of 7 to 14 of the predicted value of the regression inversion model relative to the water quality data of the previous day as the input to the Gaussian hidden Markov model with a hidden state of 3 to obtain the optimal step length for predicting the permanganate index of the Gaussian hidden Markov model in the study area. 2) After fitting the regression inversion model using the training set, record the mean absolute error of the regression inversion model, and then input the test set to obtain the predicted value of the regression inversion model. 3) Use the water quality data of the previous day of the target prediction date in the permanganate index grouped data as the reference value, and intervene in the predicted value of the Gaussian hidden Markov model according to the fluctuation range of the data to obtain the final prediction results at each point.

[0062] The length of the permanganate index grouped training set used above is from 7 to 14, enabling the Gaussian hidden Markov model to obtain sufficient historical permanganate index change information while retaining the model's sensitivity to short-term changes. The number of hidden states of the Gaussian hidden Markov model is set to 3, corresponding to the increasing, stable, and decreasing trends of the permanganate index respectively. By recording the mean absolute error of the regression inversion model and combining it with the previous day's data, the fluctuation range of the permanganate index can be obtained in a regression inversion model with relatively low accuracy, thereby capturing the fluctuation trend of the permanganate index on the predicted day. Using the above-mentioned fluctuation trend to select the predicted value of the hidden state of the corresponding Gaussian hidden Markov model as the prediction result makes up for the deficiency that the Gaussian hidden Markov model responds slowly to environmental changes, and further improves the model prediction accuracy.

[0063] In this example, the hyperspectral remote sensing data are the hyperspectral remote sensing images taken by the Ziyuan-1 02D hyperspectral camera, Ziyuan-1 02E hyperspectral camera, and GF-5 02 satellite hyperspectral camera in the study area of this example during the period from December 2022 to December 2023. The permanganate index data are from the daily water quality data collected by the automatic water quality monitoring stations in the study area of this example during the period from December 2022 to December 2023. Considering the characteristics of traditional remote sensing water quality inversion and water quality prediction, in order to quantitatively evaluate the performance of the method of the present invention, the following accuracy evaluation indicators are used: mean absolute error (MAE), mean relative error (MAPE), coefficient of determination (R 2 ).

[0064] Those of ordinary skill in the art should understand that the discussions of the above embodiments are only exemplary and are not intended to imply that the protection scope of the present application is limited to these examples; under the concept of the present application, the technical features in the above embodiments can also be combined, the steps can be implemented in any order, and there are many other variations in different aspects of one or more embodiments of the present application as described above, and they are not provided in detail for the sake of brevity.

[0065] One or more embodiments of the present application are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the present application. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of one or more embodiments of the present application shall be included within the protection scope of the present application.

Claims

1. A method for multi-point inversion of permanganate index based on hyperspectral remote sensing and water quality sequence, characterized in that The method includes: Obtaining a hyperspectral remote sensing image of the point to be inverted and performing preprocessing to obtain a preprocessed hyperspectral remote sensing image, obtaining permanganate index data collected by a water quality automatic monitoring station within the range of the hyperspectral remote sensing image, and obtaining permanganate index water quality time series data; Extracting the spectral information of the preprocessed hyperspectral remote sensing image and performing band alignment, extracting the permanganate index water quality time series data and performing time grouping to obtain permanganate index water quality time series grouped data, and randomly sampling the band-aligned spectral information and the permanganate index water quality time series grouped data into a training set and a test set at a ratio of 2:1 respectively; Constructing a remote sensing regression inversion model, the remote sensing regression inversion model includes sensitive band screening, inversion band combination screening and final regression inversion model determination, and using the spectral information training set to train the remote sensing regression inversion model; Constructing a Gaussian hidden Markov model, the Gaussian hidden Markov model includes three hidden states and optimal step size selection, and using the permanganate index water quality time series grouped data training set to input the Gaussian hidden Markov model for training; Inputting the spectral information in the test set into the pre-trained remote sensing regression inversion model to obtain an inversion result of the point to be inverted; inputting the permanganate index water quality time series grouped data in the test set into the pre-trained Gaussian hidden Markov model to obtain prediction results under three different hidden states; Inferring the current hidden state of the Gaussian hidden Markov model by using the inversion result of the point to be inverted to obtain the final permanganate index prediction result for each point, specifically: verifying the accuracy of the final regression inversion model to obtain the mean error as a threshold, and using the previous day's data in the permanganate index water quality time series data as a reference value; sorting the prediction results of the Gaussian hidden Markov model from smallest to largest to obtain the first hidden state, the second hidden state, and the third hidden state; judging whether the regression inversion result fluctuates compared with the previous day's data. If there is no fluctuation or the upper and lower fluctuation ranges do not exceed the threshold, then select the second hidden state as the final permanganate index prediction result; if the downward fluctuation range exceeds the threshold, then select the first hidden state as the final permanganate index prediction result; if the upward fluctuation range exceeds the threshold, then select the third hidden state as the final permanganate index prediction result.

2. The permanganate index multi-point inversion method according to claim 1, characterized in that The preprocessing is to perform radiometric calibration, atmospheric correction, and orthorectification on the hyperspectral remote sensing image. The calculation formula for radiometric calibration is: L = Gain·DN + Bias, where L is the pixel radiance, with the unit of W / (cm 2 ·μm·sr), DN is the pixel brightness value of the remote sensing image, Gain is the radiometric calibration gain, and Bias is the radiometric calibration offset, both with the unit of W / (cm 2 ·μm·sr); the calculation formula for atmospheric correction is: C = where C is the pixel radiance after atmospheric correction, ρ is the pixel reflectance, ρ e is the average reflectance of the surrounding area of the pixel, A and B are calculated from the actual atmospheric and geometric conditions, S is the atmospheric spherical albedo, and C a is the atmospheric path radiance.

3. The permanganate index multi-point inversion method according to claim 1, characterized in that, Obtaining permanganate index data collected by a water quality automatic monitoring station within the range of the hyperspectral remote sensing image, specifically obtaining permanganate index data collected in the 14 days before the shooting date of the hyperspectral remote sensing image, and grouping the permanganate index water quality time series data in units of N days to obtain permanganate index water quality time series grouped data, where 7 ≤ N ≤ 14.

4. The permanganate index multi-point inversion method according to claim 1, characterized in that, The processing method for band alignment is as follows: for hyperspectral remote sensing images taken by different hyperspectral satellites, select the image taken by one hyperspectral satellite as the reference image, and the rest as the images to be registered; when the difference between the central bands of each band of the images to be registered and the reference image is within the specified range, retain the band; when the difference exceeds the specified range, take the average of the current band and the previous or next band of the image to be registered as the new central band, so that the difference of the new central band is within the specified range, and take the average of the radiance values of the previous and next bands as the radiance value of the new central band.

5. The permanganate index multi-point inversion method according to claim 1, characterized in that, For the sensitive band screening and the inversion band combination screening in the remote sensing regression inversion model, the Pearson correlation coefficient method is used to select the band or band combination with the highest correlation as the calculation formula from spectral information to the independent variable of remote sensing regression inversion. The formula for the Pearson correlation coefficient is: where r is the Pearson correlation coefficient, X i is the independent variable of remote sensing regression inversion, and Y i is the measured permanganate index water quality data corresponding to the hyperspectral remote sensing image on the day of shooting, n is the number of samples, i is the sample mean of X and Y i respectively, and S i and S X are the sample standard deviations; for the final regression inversion model in the remote sensing regression inversion model, after the linear regression model, logarithmic regression model, polynomial regression model, and power regression model are established, R Y is used for accuracy verification, and the regression inversion model with the highest coefficient of determination is selected. Among them, 2 where R is the coefficient of determination, SSR is the regression sum of squares, SST is the total sum of squares, 2 and is the inverted water quality corresponding to Y i .

6. The permanganate index multi-point inversion method according to claim 1, characterized in that, The Gaussian Hidden Markov Model includes: Construct a Gaussian Hidden Markov Model for describing the hidden state of water quality, and initialize the parameters of the model, including the state transition matrix, the initial probability matrix, the mean matrix and the covariance matrix; Use the grouped data of the permanganate index water quality time series to construct the Gaussian Hidden Markov observation sequence; Set the number of water quality hidden states to 3, and use the Baum-Welch algorithm to estimate the parameters of the Gaussian Hidden Markov Model to obtain the trained state transition matrix, initial probability matrix, mean matrix and covariance matrix; Calculate the weighted average of the expected observation values of each state in the next stage through the state transition matrix and the mean matrix, and use it as the predicted value of the hidden state.

7. The permanganate index multi-point inversion method according to any one of claims 1-6, characterized in that, Use test samples to verify the accuracy of the hyperspectral remote sensing inversion results of permanganate.

8. The permanganate index multi-point inversion method according to any one of claims 1-6, characterized in that, The mean error is taken from the mean absolute error of the final regression inversion model during the training process, and its formula is where MAE is the mean absolute error, and Y i is the measured permanganate index water quality data, is the inversion water quality corresponding to Y i ​ 9. The permanganate index multi-point inversion method according to any one of claims 1-6, characterized in that, The hyperspectral remote sensing images are taken from the Hyperion camera on Ziyuan-1 02D satellite, the Hyperion camera on Ziyuan-1 02E satellite and the hyperspectral camera on GF-5 02 satellite.

Images