A method for retrieving chlorophyll a concentration in surface water based on satellite remote sensing

Abstract

This invention presents a method for retrieving chlorophyll a concentration in surface water based on satellite remote sensing. The method acquires remote sensing images from a satellite that match the acquisition time of the water quality data. Based on the spectral information extracted from these images, a new grayscale image is generated. A grayscale co-occurrence matrix (HCEM) is calculated from this new image. The HCEM is then used to mine the texture features contained in the remote sensing images, and each texture feature is quantified. This results in a significantly improved model performance compared to models trained solely using spectral information. Furthermore, with the same amount of raw data, this invention achieves higher accuracy, which is beneficial for the precise measurement of surface water eutrophication.

Description

[Technical Field]

[0001] This invention demonstrates a method for retrieving chlorophyll a concentration in surface water based on satellite remote sensing, belonging to the field of quantitative environmental remote sensing technology. [Background Technology]

[0002] Climate change and human activities have significantly impacted the environmental quality of surface waters such as rivers, lakes, and reservoirs, with eutrophication being one of the major environmental problems. Eutrophication occurs when nutrients such as nitrogen and phosphorus accumulate in water bodies, leading to increased primary productivity and massive blooms of phytoplankton (algae). Eutrophication is known as algal blooms in freshwater and red tides in seawater. Lakes and reservoirs are key areas of concern regarding surface water eutrophication, and chlorophyll a is the most common pigment in their algal blooms, making it a core indicator for assessing the degree of eutrophication. Accurately determining chlorophyll a concentration is a crucial task in eutrophication research and management.

[0003] Remote sensing satellites, with their unique ability to acquire large-scale surface images, are widely used for water quality monitoring of large bodies of water such as lakes, reservoirs, and oceans. To track water quality changes and develop effective management strategies to control lake eutrophication, long-term lake image datasets are needed. Compared to ground-based measurement methods, remote sensing satellites can continuously observe chlorophyll a concentration in lake water, meeting the requirements for long-term dynamic monitoring. However, current water environment remote sensing focuses on using and analyzing the spectral information carried by the images, employing limited data characteristics. It only uses different combinations of spectral bands to invert the concentration of substances such as chlorophyll a, without deeply exploring the geometric texture features inherent in the images themselves. [Summary of the Invention]

[0004] The technical problem to be solved by the present invention is to provide a method for inverting the chlorophyll a concentration of surface water based on satellite remote sensing, which combines the texture information contained in the remote sensing image.

[0005] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0006] A method for retrieving chlorophyll a concentration in surface water based on satellite remote sensing, comprising a satellite remote sensor, and the steps of the inversion method are as follows:

[0007] S1. Acquire water quality data, select remote sensing images from satellite remote sensors whose transit time matches the acquisition time of the water quality data, and preprocess the remote sensing images to eliminate grayscale distortion in the remote sensing images.

[0008] S2. Based on the parameters of the satellite remote sensor and the spectral reflectance curve of chlorophyll a, extract the spectral reflectance values ​​of the pixels in the remote sensing image;

[0009] S3. Calculate and generate a new grayscale image based on the bands included in the inversion algorithm, compress the grayscale levels of the remote sensing image to determine the matrix order, select four movement directions and set the movement step size, matrix base window and matrix sliding window size, thereby calculating and forming a grayscale co-occurrence matrix.

[0010] S4. Select several texture features, calculate the values ​​of the texture features in each direction according to the gray-level co-occurrence matrix, and use the average value of the four directions as the texture feature index.

[0011] S5. Establish an inversion model based on the spectral reflectance value, the texture feature index, and the concentration of chlorophyll a in the water quality data.

[0012] The beneficial effects of using the present invention are:

[0013] In this invention, remote sensing images matching the acquisition time of the water quality data are acquired from satellite remote sensors. These images are then preprocessed to eliminate grayscale distortion caused by other factors, improving the accuracy of the invention. Furthermore, based on the extracted spectral information from the remote sensing images, a new grayscale image is generated. A grayscale co-occurrence matrix is ​​calculated based on this new grayscale image. The texture features contained in the remote sensing images are mined through this grayscale co-occurrence matrix, and each texture feature is quantified. This results in a significant performance improvement in the model established by this invention compared to a model trained solely on spectral information. Additionally, with the same amount of raw data, this invention achieves higher accuracy, which is beneficial for the precise measurement of surface water eutrophication.

[0014] Preferably, the transit time in S1 is ±1 day of the water quality data acquisition time.

[0015] Preferably, the spectral reflectance value of a pixel in the remote sensing image in step S2 is the median value of several adjacent pixels.

[0016] Preferably, the four directions selected in S3 are 0°, 45°, 90° and 135°.

[0017] Preferably, the texture feature indicators include mean, variance, homogeneity, entropy, energy, correlation, and autocorrelation, and the calculation formulas for the texture feature indicators are as follows:

[0018] The mean is M.

[0019]

[0020] The variance is V.

[0021] V = ∑ i ∑j p(i, j)*(iM) 2 ;

[0022] The homogeneity is H.

[0023]

[0024] The entropy is S.

[0025] S=-∑ i ∑ j p(i,j)*log(p(i,j));

[0026] The energy is E.

[0027] E = ∑ i ∑ j p(i, j) 2 ;

[0028] The correlation is C.

[0029]

[0030] The autocorrelation is A.

[0031] A = ∑ i ∑ j p(i,j)*i*j;

[0032] Where i, j are the values ​​represented by the indices in the gray-level co-occurrence matrix, p(i, j) is the i-th and j-th values ​​in the gray-level co-occurrence matrix, and d is the width of the sliding window of the matrix.

[0033] Other features and advantages of the present invention will be disclosed in detail in the following detailed description and accompanying drawings. [Attached Image Description]

[0034] The invention will be further described below with reference to the accompanying drawings:

[0035] Figure 1 This is a flowchart of the inversion method described in this embodiment of the invention;

[0036] Figure 2 This is a multispectral satellite image from an embodiment of the present invention;

[0037] Figure 3 This is a texture feature index map in an embodiment of the present invention;

[0038] Figure 4 This is a comparison chart of the inversion results of the two models in this embodiment of the invention;

[0039] Figure 5This is a comparison diagram of the inversion fitting relationship between the two models in an embodiment of the present invention.

Detailed Implementation Methods

[0040] The technical solutions of the embodiments of the present invention will be explained and described below with reference to the accompanying drawings. However, the following embodiments are only preferred embodiments of the present invention and not all of them. Other embodiments obtained by those skilled in the art based on the embodiments in the implementation methods without creative effort are all within the protection scope of the present invention.

[0041] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "clockwise," and "counterclockwise," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0042] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise expressly defined.

[0043] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0044] Example 1:

[0045] like Figures 1 to 5As shown in the figure, this embodiment demonstrates a method for retrieving chlorophyll a concentration in surface water based on satellite remote sensing. The steps of the inversion method are as follows: First, acquire the data monitored at the water quality monitoring station. Based on the location information of the monitoring station, select a remote sensing image from the satellite remote sensor whose transit time matches the acquisition time of the water quality data. To improve accuracy, in this embodiment, the transit time is ±1 day of the water quality data acquisition time. Second, to ensure the integrity of the remote sensing image, images with low cloud cover in the area or where the monitoring station and its surroundings are not obscured by clouds should be selected. In this embodiment, Landsat 5, Landsat 7, and Landsat 8 images from 2000 to 2019 that met the time requirements and were cloud-free were selected, totaling 15 remote sensing images. These images were then preprocessed with radiometric correction, atmospheric correction, and image cropping to eliminate grayscale distortion. In this embodiment, due to a malfunction of the Landsat-7ETM+ airborne scan line corrector (SLC) on May 31, 2003, data striping was lost in subsequently acquired images, severely impacting the usability of Landsat ETM remote sensing images. Therefore, destriping was necessary before preprocessing.

[0046] Then, based on the parameters of the satellite remote sensor and the spectral reflectance curve of chlorophyll a, the spectral reflectance values ​​of the pixels in the remote sensing image are extracted. The spectral reflectance value of the pixels in the remote sensing image is the median value of several adjacent pixels. In this embodiment, the extraction function built into the software ENVI is used to import the vector file of the water quality monitoring station into the software ENVI, and extract the median value of the corresponding point within a 3x3 range in the remote sensing image raster data as the spectral reflectance value of the monitoring station. At the same time, the data of the stations obscured by cloud cover are deleted. Finally, a total of 119 spectral emissivity data matching the on-site measurement data of the monitoring station are obtained, as shown in the sample data column in Table 1.

[0047]

[0048]

[0049] Table 1 - Samples of spectral data extracted from Landsat remote sensing satellite and measured data

[0050] Among them, B2 is the blue band, B3 is the green band, B4 is the red band, B5 is the near-infrared band, B6 is the shortwave infrared 1, and B7 is the shortwave infrared 2.

[0051] Then, a new grayscale image is calculated and generated based on the bands included in the inversion algorithm. The grayscale levels of the remote sensing image are compressed to determine the matrix order. Four directions are selected and the step size, matrix base window size, and matrix sliding window size are set. The grayscale co-occurrence matrix of the grayscale image is calculated using Python programming language and third-party libraries such as GDAL and NumPy. In this embodiment, a near-infrared / red band combination method is used to calculate the new grayscale image. The size of the grayscale co-occurrence matrix is ​​set to 32×32. The matrix base window is centered on the current pixel and has a size of 7×7. The size of the matrix sliding window is the same as that of the base window. The four selected moving directions are 0°, 45°, 90°, and 135°, and the moving step size is set to 2 pixels wide.

[0052] In this embodiment, seven texture features were selected, namely mean, variance, homogeneity, entropy, energy, correlation, and autocorrelation. The texture feature values ​​in each direction were calculated based on the gray-level co-occurrence matrix, and the average value of the four directions was used as the texture feature index. The calculation formula for the texture feature index is as follows:

[0053] The mean is M.

[0054]

[0055] The variance is V.

[0056] V = ∑ i ∑ j p(i, j)*(iM) 2 ;

[0057] The homogeneity is H.

[0058]

[0059] The entropy is S.

[0060] S=-∑ i ∑ j p(i,j)*log(p(i,j));

[0061] The energy is E.

[0062] E = ∑ i ∑ j P(i, j) 2 ;

[0063] The correlation is C.

[0064]

[0065] The autocorrelation is A.

[0066] A = ∑ i ∑ j p(i,j)*i*j;

[0067] Where i, j are the values ​​represented by the indices in the gray-level co-occurrence matrix, p(i,j) is the i-th and j-th values ​​in the gray-level co-occurrence matrix, and d is the width of the matrix sliding window.

[0068] Some texture index data in this embodiment are shown in Table 2:

[0069]

[0070] Table 2 - Sample of some texture index data

[0071] Finally, an inversion model was established based on the spectral reflectance value, the texture feature index, and the concentration of chlorophyll a in the water quality data. According to the spectral characteristics of chlorophyll a, this embodiment selected the ratio of the near-infrared band to the red band as the spectral feature independent variable data, and performed principal component analysis on all bands. Since only the eigenvalue of the first principal component is greater than 1, the first principal component was selected and added as an independent variable. The ratio of the near-infrared band to the red band, the first principal component score of the spectral information of all bands, and the seven texture feature indexes were used as independent variables, and the chlorophyll a concentration was used as the dependent variable for multiple linear stepwise regression. The independent variables that finally entered the model were: B5 / B4, the first principal component score, the mean, homogeneity, and autocorrelation.

[0072] Finally, the chlorophyll a concentration inversion model equation that combines spectral and texture information in this embodiment is:

[0073] y = 0.1931x1 - 0.0028x2 + 1.29 * 10 -2 x3 + 2.263 * 10 -3 x4+2.299*10 -5 x5-0.1272

[0074] Where x1 is the ratio of near-infrared band to red band, x2 is the score of the first principal component, x3 is the mean, x4 is the homogeneity, and x5 is the autocorrelation. Since the goodness of fit of the inversion model is usually measured using the coefficient of determination R... 2 The coefficient of determination, or R, is the ratio of the regression sum of squares to the total sum of squares in linear regression. Its value ranges from 0 to 1; a higher value indicates a better fit. The coefficient of determination R for the model was obtained based on the data in this embodiment. 2 It is 0.7534.

[0075] The traditional chlorophyll a concentration inversion model equation based on spectral information is as follows:

[0076] y = 0.1879x1 - 0.0024x2 - 0.0443

[0077] Where x1 is the ratio of the near-infrared band to the red band, and x2 is the score of the first principal component of the spectrum. Based on the spectral data extracted in this embodiment, the above model is used to invert the chlorophyll a concentration, and the coefficient of determination R of the model is obtained. 2 It is 0.7192.

[0078] Based on the parameter chlorophyll a concentration inversion results of the above two models, for example... Figure 4 As shown, the fitting relationship between the simulated concentration and the measured concentration for the two models is as follows: Figure 5 As shown, the chlorophyll a concentration inversion model combining spectral and texture information has higher fitting accuracy and is more reliable than the chlorophyll a concentration inversion model combining only spectral information. In addition, with the same amount of original data, this embodiment can achieve higher accuracy, which is beneficial for the accurate measurement of surface water eutrophication.

[0079] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Those skilled in the art should understand that the present invention includes, but is not limited to, the contents described in the accompanying drawings and the specific embodiments above. Any modifications that do not depart from the functional and structural principles of the present invention will be included within the scope of the claims.

Claims

1. A method for retrieving chlorophyll a concentration in surface water based on satellite remote sensing, comprising a satellite remote sensor, characterized in that: The inversion method involves the following steps: S1. Acquire water quality data, select remote sensing images from satellite remote sensors whose transit time matches the acquisition time of the water quality data, and preprocess the remote sensing images to eliminate grayscale distortion in the remote sensing images. S2. Based on the parameters of the satellite remote sensor and the spectral reflectance curve of chlorophyll a, extract the spectral reflectance values ​​of the pixels in the remote sensing image; S3. Calculate and generate a new grayscale image based on the bands included in the inversion algorithm, compress the grayscale levels of the remote sensing image to determine the matrix order, select four movement directions and set the movement step size, matrix base window and matrix sliding window size, thereby calculating and forming a grayscale co-occurrence matrix. S4. Select several texture features, calculate the values ​​of the texture features in each movement direction according to the gray-level co-occurrence matrix, and use the average value of the four movement directions as the texture feature index. S5. Establish an inversion model based on the spectral reflectance value, the texture feature index, and the concentration of chlorophyll a in the water quality data; The equation for the chlorophyll a concentration inversion model is: y=0.1931x1-0.0028x2+1.29×10 -2 x3+2.263×10 -3 x4+2.299×10 -5 x5-0.1272; Where x1 is the ratio of near-infrared band to red band, x2 is the score of the first principal component, x3 is the mean, x4 is the homogeneity, and x5 is the autocorrelation.

2. The method for retrieving chlorophyll a concentration in surface water based on satellite remote sensing according to claim 1, characterized in that: The transit time in S1 is ±1 day of the water quality data acquisition time.

3. The method for retrieving chlorophyll a concentration in surface water based on satellite remote sensing according to claim 1, characterized in that: The spectral reflectance value of a pixel in the remote sensing image in S2 is the median value of several adjacent pixels.

4. The method for retrieving chlorophyll a concentration in surface water based on satellite remote sensing according to claim 1, characterized in that: The four moving directions selected in S3 are 0°, 45°, 90° and 135°.

Images