Inland water transparency inversion method based on MAPE-DR and improved QAA

Abstract

The invention provides an inland water transparency inversion method based on MAPE-DR and improved QAA, and relates to the technical field of inland water transparency, and the method comprises the steps: obtaining hyperspectral remote sensing data, processing the hyperspectral remote sensing data based on the MAPE-DR, and obtaining an inland water transparency inversion result; an absorption coefficient and a backscattering coefficient are obtained based on an improved QAA algorithm according to the processed hyperspectral remote sensing data, a diffuse reflection attenuation coefficient is obtained through calculation based on the absorption coefficient and the backscattering coefficient, and the transparency of the water body is obtained through calculation of the diffuse reflection attenuation coefficient. The method can effectively improve the data quality, improves the Zsd inversion precision, has a wide application prospect, and provides a scientific basis for water environment monitoring and management.

Description

Technical Field

[0001] The invention relates to the technical field of inland water transparency, and in particular to an inland water transparency inversion method based on MAPE-DR and improved QAA. Background Art

[0002] Water transparency is an important water quality parameter in limnology and oceanography. It is closely related to the light field distribution under the water body and is of great significance to the diversity and productivity of aquatic organisms. Secchi disk depth (Zsd) is an important indicator of water transparency. It is usually measured by observing the disappearance depth of a white or black and white transparent disk in the water. Zsd can directly reflect the content of suspended particles, algae concentration and other pollutants in the water body, and is widely used in lake water quality monitoring. Changes in Zsd not only affect the health of lake and marine ecosystems, but also have important significance for ecological protection, water resource utilization, fishery production and tourism.

[0003] Traditional methods usually rely on a large amount of field sampling data to study the distribution characteristics of Zsd. However, this method not only requires long-term monitoring at multiple sampling points, but the field sampling data are often discontinuous in space, which makes it difficult to fully reflect the Zsd distribution and its dynamic changes of large-scale water bodies. In addition, traditional methods cannot accurately calculate the remote sensing reflectance of hyperspectral data. Therefore, it is necessary to design an inland water transparency inversion method based on MAPE-DR and improved QAA. Summary of the invention

[0004] In order to overcome the shortcomings of the prior art, the purpose of the present invention is to provide an inland water transparency inversion method based on MAPE-DR and improved QAA.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] The present invention provides an inland water transparency inversion method based on MAPE-DR and improved QAA, comprising:

[0007] Acquisition of hyperspectral remote sensing data;

[0008] Processing of hyperspectral remote sensing data based on MAPE-DR;

[0009] Based on the improved QAA algorithm, the absorption coefficient and backscattering coefficient are obtained according to the processed hyperspectral remote sensing data. The diffuse reflection attenuation coefficient is calculated based on the absorption coefficient and the backscattering coefficient. The water transparency is calculated through the diffuse reflection attenuation coefficient.

[0010] Preferably, the hyperspectral remote sensing data is ZY1-02D ​​hyperspectral data, which is obtained from the natural resources satellite remote sensing cloud service platform.

[0011] Preferably, the hyperspectral remote sensing data is processed based on MAPE-DR, specifically as follows:

[0012] Perform atmospheric correction on the hyperspectral data to obtain surface reflectance data, and convert the surface reflectance data into remote sensing reflectance based on remote sensing reflectance correction;

[0013] The remote sensing reflectance was denoised based on LF, HF, MF, SGF and HA denoising methods respectively;

[0014] For each band, the MAPE values ​​of different combination methods and ASD equivalent spectra were compared, the best method combination was selected as the optimal band, and all the optimal bands were stacked to form a reconstructed image.

[0015] Preferably, the absorption coefficient and the backscattering coefficient are obtained according to the processed hyperspectral remote sensing data based on the improved QAA algorithm, specifically:

[0016] Improvements are made based on QAAV6 and QAAL09. The specific forms are:

[0017] r rs =R rs / (0.52+1.7R rs )

[0018]

[0019] In the formula, r rs represents the remote sensing reflectivity below the water surface, R rs represents the remote sensing reflectivity above the water surface, u(λ) is r rs function, a(λ0) is the water absorption coefficient at the reference wavelength, b bp (λ0) is the water backscattering coefficient at the reference wavelength, a w represents the absorption coefficient of pure water, b bw is the backscattering coefficient of pure water, Y is an empirical parameter, a(λ) and b bp (λ) are the total absorption coefficient and backscattering coefficient of all wavelengths respectively;

[0020] The QAAA optimization steps are as follows: taking advantage of the ZY1-02D ​​hyperspectral spectrum, set λ0 to 713, 722, 730, and 739, respectively; fix λ2 to 748 nm, vary λ1 between 400 and 800 nm, and optimize the Y value.

[0021] Preferably, the diffuse reflection attenuation coefficient is calculated based on the absorption coefficient and the backscattering coefficient, and the water transparency is calculated by the diffuse reflection attenuation coefficient. The specific calculation form is:

[0022]

[0023] In the formula, K d (λ) diffuse reflection attenuation coefficient, Represents the minimum value of the diffuse reflection attenuation coefficient of the transparent window of water in the range of 410-665nm, is the remote sensing reflectance of the corresponding band, a(λ) and b bp (λ) are the total absorption coefficient and backscattering coefficient of water, b bw (λ) is the backscattering coefficient of pure water, θ s is the solar zenith angle, m0 is 0.005, m1 is 4.26, and m2 is 0.52.

[0024] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0025] The present invention provides an inland water transparency inversion method based on MAPE-DR and improved QAA, the method comprising: acquiring hyperspectral remote sensing data, processing the hyperspectral remote sensing data based on MAPE-DR, obtaining absorption coefficient and backscattering coefficient according to the processed hyperspectral remote sensing data based on the improved QAA algorithm, calculating diffuse reflection attenuation coefficient based on the absorption coefficient and the backscattering coefficient, and obtaining water transparency through diffuse reflection attenuation coefficient calculation. The present invention can effectively improve data quality, improve Zsd inversion accuracy, has broad application prospects, and provides a scientific basis for water environment monitoring and management. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0027] Figure 1 This is a schematic diagram of the study area and sampling point locations;

[0028] Figure 2 Schematic diagram of different atmospheric correction effects;

[0029] Figure 3a This is a schematic diagram of the first type of remote sensing reflectivity accuracy assessment;

[0030] Figure 3b This is a schematic diagram of the second type of remote sensing reflectivity accuracy assessment;

[0031] Figure 3c This is a schematic diagram of the third type of remote sensing reflectivity accuracy assessment;

[0032] Figure 4 Schematic diagram for the performance evaluation of the denoising algorithm;

[0033] Figure 5 This is a schematic diagram of band accuracy evaluation of the MAPE-DR method;

[0034] Figure 6 This is a schematic diagram of point accuracy evaluation of the MAPE-DR method;

[0035] Figure 7 Schematic diagram of the accuracy evaluation of different versions of QAA algorithms;

[0036] Figure 8 This is a schematic diagram of the QAA722 accuracy assessment of the ZY1-02 image data;

[0037] Fig. 9 This is a schematic diagram of the spatial distribution of Zsd of Dushan Lake water body;

[0038] Fig.10 This is a schematic diagram of the spatial distribution of Zsd of Weishan Lake water body;

[0039] Fig.11 A flow chart of a method provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0040] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0041] The purpose of the present invention is to provide an inland water transparency inversion method based on MAPE-DR and improved QAA, which can effectively improve data quality and Zsd inversion accuracy, has broad application prospects, and provides a scientific basis for water environment monitoring and management.

[0042] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0043] The present invention takes Nansihu Lake in Shandong Province as the research background and describes the method recorded in the present invention in detail;

[0044] First, let’s introduce the South Four Lakes: The South Four Lakes (37°27′-35°20′N, 116°34′-117°21′E) are located in the southwest of Shandong Province, China. Figure 1 As shown, it is the largest freshwater lake in Shandong Province. The total area of ​​the South Four Lakes is about 1266km 2It is formed by Nanyang Lake, Dushan Lake, Zhaoyang Lake and Weishan Lake connected from northwest to southeast. It is 125km long from north to south and 6-25km wide from east to west. The average water depth is 1.5m, the maximum water depth during the flood season is 2.5-3m, and the maximum flood control storage capacity is 5.4 billion m3. It is an important regulating lake in the East Route of the South-to-North Water Diversion Project and an important biodiversity protection area.

[0045] The present invention conducted water transparency and spectrum measurements in the South Four Lakes on March 31 and April 1, 2023. The sampling points were set in Dushan Lake and Weishan Lake, where the waters of the South Four Lakes are relatively open. Figure 1 The transparency of water bodies was measured using a Secchi disk, and a total of 82 transparency samples were collected. The transparency data statistics are shown in Figure 1 As shown in the middle bar graph, the maximum transparency of the 82 sample points is 1.10m, the minimum is 0.16m, and the average value is about 0.50m. According to the water spectrum measurement specifications formulated by NASA, the "surface measurement method" is used to measure the water spectrum. The measuring instrument is the Fieldspec4 high-resolution spectroradiometer (spectral range 350-2500nm, interval 1nm) produced by ASD Company of the United States. A total of 46 water spectra synchronized with transparency are collected. During the sampling period, there was no wind and no clouds, and the weather was good. Ten spectra were measured for each content and the average was taken to ensure the measurement accuracy. The water surface remote sensing reflectivity is calculated by substituting into the following formula:

[0046]

[0047] In the formula, R rs (λ) is the remote sensing reflectivity of the water surface, r is the sky light reflectivity (Fresnel reflectivity) of the water-air interface, and according to existing research, the r value of the present invention is set to 0.025, ρ p is the reflectivity of a standard white board, π is 3.14, E s is the whiteboard mean, and the measured R on the water surface is calculated rs (λ), the measured R rs (λ) is converted to satellite equivalent band:

[0048]

[0049] In the formula, RSR i (λ) represents the spectral response function (SRF) of the satellite, λ max and λ min is the wavelength range of SRF in each band, R rs (λ) is the measured water surface remote sensing reflectivity, R rs_ (λ i) is the equivalent band remote sensing reflectance of the measured data on the satellite. The SRF of ZY1-02D ​​is assumed to be a Gaussian function, which is often used for SRF simulation of hyperspectral satellites. The specific method is as follows:

[0050]

[0051] In the formula, λ i and Δλ i They represent the central wavelength and full-width at half-maximum (FWHM) of the ith band respectively. The FWHM of the AHSI sensor of the ZY1-02D ​​satellite in the visible near-infrared and short-wave infrared bands are 8.67nm and 16.26nm respectively.

[0052] Fig.11 A flow chart of a method provided by an embodiment of the present invention, such as Fig.11 As shown, the present invention provides an inland water transparency inversion method based on MAPE-DR and improved QAA, comprising:

[0053] Step 100: Acquire hyperspectral remote sensing data;

[0054] Step 200: Processing hyperspectral remote sensing data based on MAPE-DR;

[0055] Step 300: Based on the improved QAA algorithm, the absorption coefficient and the backscattering coefficient are obtained according to the processed hyperspectral remote sensing data, the diffuse reflection attenuation coefficient is calculated based on the absorption coefficient and the backscattering coefficient, and the water body transparency is calculated through the diffuse reflection attenuation coefficient.

[0056] In step 100, the hyperspectral remote sensing data is ZY1-02D ​​hyperspectral data, which is obtained from the natural resources satellite remote sensing cloud service platform. The ZY1-02D ​​data closest to the sampling time is selected, among which Dushan Lake is on March 27, 2023, and Weishan Lake is on March 30, 2023. All images are cloudless and of good quality. Radiometric calibration, orthorectified image cropping and water body extraction are performed in ENVI software. For water body extraction, the normalized difference water index (NDWI) is first calculated, and then the improved histogram bimodal method (MHBM) is used to segment the water area from other land cover types.

[0057] In step 200, the hyperspectral remote sensing data is processed based on MAPE-DR, specifically:

[0058] Step 201: Perform atmospheric correction processing on the hyperspectral data to obtain surface reflectance data, and convert the surface reflectance data into remote sensing reflectance based on remote sensing reflectance correction, specifically:

[0059] The atmospheric correction processing part is introduced: Atmospheric correction is used to remove the interference of atmospheric absorption and scattering on water body signals, so as to effectively invert water quality parameters. The present invention briefly introduces several atmospheric correction methods:

[0060] 1. Based on the radiation transfer model, Flaash accurately corrects the atmospheric effects by inputting scene parameters (such as sensor height, sun angle, aerosol type, etc.) and handles complex atmospheric phenomena such as water vapor absorption and aerosol scattering;

[0061] 2. QUAC is based on scene statistics, assuming that the mean spectral reflectance is constant, and does not require external atmospheric parameters to achieve rapid correction;

[0062] 3. The 6S model is used to simulate satellite atmospheric transmission within the solar spectrum and is suitable for multispectral and hyperspectral data;

[0063] 4. Modtran provides a variety of standard atmospheric models for correction by simulating the transmission process of electromagnetic radiation in the atmosphere;

[0064] Introduction to remote sensing reflectance correction: Hyperspectral data after atmospheric correction is essentially surface reflectance data. In water color remote sensing, remote sensing reflectance is often used to estimate various water quality parameters. The difference in water vapor and land-air radiation transmission processes will lead to a great difference between the two reflectances. In view of this, the present invention refers to a remote sensing reflectance correction method developed by Wang et al. to convert the surface reflectance into remote sensing reflectance. The specific method is as follows:

[0065]

[0066] In the formula, Represents the corrected remote sensing reflectance, R rs (λ) is the surface reflectance after processing of each data source, min(R swir ) represents the minimum value of the short-wave infrared band, and π is 3.14;

[0067] In the present invention, the remote sensing reflectance correction is performed using the mean value of the visible light band minus the short-wave infrared band (1530-1630nm);

[0068] The remote sensing reflectivity is denoised based on the LF, HF, MF, SGF and HA denoising methods, respectively, as follows:

[0069] Denoising algorithms are used to eliminate noise in spectral data and enhance signal quality. Low-pass filtering (LF) and High-pass filtering (HF) allow low-frequency and high-frequency signals to pass through respectively, and weaken signals that are not within the set frequency range. Median filtering (MF) is a nonlinear filtering technology based on sorting statistics. It sorts pixels through a sliding template and takes the median to suppress noise. Savitzky-Golay filtering (SGF) is based on the local polynomial least squares method, smoothes the spectral curve and calculates derivatives of each order. It is suitable for removing spectral glitches caused by mechanical vibration. Harmonic analysis (HA) decomposes the spectral curve into multiple sine and cosine waves, and reconstructs it to effectively remove high-frequency noise and achieve spectral smoothing.

[0070] For each band, the MAPE values ​​of different combination methods and ASD equivalent spectra are compared, the best method combination is selected as the optimal band, and all the optimal bands are stacked to form a reconstructed image, specifically:

[0071] In the process of processing inland water hyperspectral remote sensing data, how to maximize the accuracy of remote sensing reflectance and reduce errors is still a challenge. To overcome this problem, the present invention proposes a data processing method (MAPE-DR) based on MAPE optimization and data reconstruction. Using MAPE-DR to process inland water hyperspectral data can ensure that the error between each band and the ASD equivalent spectrum is minimized. For each band, the MAPE values ​​of different algorithm combinations and the ASD equivalent spectrum are compared, the best algorithm combination is selected as the optimal band, and all the optimal bands are stacked to form a reconstructed image. In short, the hyperspectral remote sensing data is divided into multiple bands, and for each band, atmospheric correction is performed based on four algorithms to obtain four results. These four results are denoised based on five algorithms, and a total of twenty results are obtained. The MAPE values ​​of different algorithm combinations and the ASD equivalent spectrum are compared, the best algorithm combination is selected as the optimal band, and all the optimal bands are stacked to form a reconstructed image.

[0072] In step 300, the absorption coefficient and the backscattering coefficient are obtained according to the processed hyperspectral remote sensing data based on the improved QAA algorithm, specifically:

[0073] Improvements are made based on QAAV6 and QAAL09. The specific forms are:

[0074] r rs =R rs / (0.52+1.7R rs )

[0075]

[0076] In the formula, r rs represents the remote sensing reflectivity below the water surface, R rs represents the remote sensing reflectivity above the water surface, u(λ) is r rs function, a(λ0) is the water absorption coefficient at the reference wavelength, b bp (λ0) is the water backscattering coefficient at the reference wavelength, a w represents the absorption coefficient of pure water, b bw is the backscattering coefficient of pure water, Y is an empirical parameter, a(λ) and b bp (λ) are the total absorption coefficient and backscattering coefficient of all wavelengths respectively;

[0077] The QAAA optimization steps are as follows: taking advantage of the ZY1-02D ​​hyperspectral spectrum, set λ0 to 713, 722, 730, and 739, respectively; fix λ2 to 748 nm, vary λ1 between 400 and 800 nm, and optimize the Y value.

[0078] The diffuse reflection attenuation coefficient is calculated based on the absorption coefficient and the backscattering coefficient, and the water transparency is calculated by the diffuse reflection attenuation coefficient. The specific calculation form is:

[0079]

[0080] In the formula, K d (λ) diffuse reflection attenuation coefficient, Represents the minimum value of the diffuse reflection attenuation coefficient of the transparent window of water in the range of 410-665nm, is the remote sensing reflectance of the corresponding band, a(λ) and b bp (λ) are the total absorption coefficient and backscattering coefficient of water, b bw (λ) is the backscattering coefficient of pure water, θ s is the solar zenith angle, m0 is 0.005, m1 is 4.26, and m2 is 0.52.

[0081] Combining the two steps in the above step 300, the origin of step 300 is described in detail:

[0082] The semi-analytical model based on the radiation transfer model and underwater visibility theory is robust in inverting water transparency. The core is to calculate the diffuse reflection attenuation coefficient through the absorption coefficient and backscattering coefficient obtained by QAA, and then calculate the water transparency through the diffuse reflection attenuation coefficient. The specific formula is as follows:

[0083]

[0084] In the formula, Represents the diffuse reflection attenuation coefficient of the transparent window of water in the range of 410-665nm, is the remote sensing reflectance of the corresponding band, a(λ) and b bp (λ) are the total absorption coefficient and backscattering coefficient of water, b bw (λ) is the backscattering coefficient of pure water. According to the results, this method does not need to adjust the parameters. Therefore, it is particularly important to obtain the applicable water absorption coefficient and backscattering coefficient through the QAA algorithm. The commonly used QAA versions are QAAv5, QAAv6, QAAL09 and QAAM14. The calculation details of QAAV5 are shown in Table 1.

[0085] Table 1 QAAV5 algorithm calculation steps

[0086]

[0087]

[0088] Table 1 rs represents the remote sensing reflectivity below the water surface, R rs represents the remote sensing reflectivity above the water surface, u(λ) is r rs function, a(λ0) is the water absorption coefficient at the reference wavelength, b bp (λ0) is the water backscattering coefficient at the reference wavelength, a w represents the absorption coefficient of pure water, b bw is the backscattering coefficient of pure water, Y is an empirical parameter, a(λ) and b bp (λ) are the total absorption coefficient and backscattering coefficient of all wavelengths respectively. QAAV6 is the judgment of Rrs in Step 2 of QAAV5. If Rrs(670)>0.0015sr -1 , then the reference wavelength is changed to 670nm, and its absorption coefficient is calculated according to the following formula:

[0089]

[0090] QAAL09 moves the reference wavelength to 710nm and ignores the image of the backscattering coefficient of pure water. The reference wavelength backscattering coefficient and the empirical parameter Y are calculated as follows:

[0091]

[0092] QAAM14 mainly improves the calculation of the absorption coefficient of the reference wavelength. The specific calculation method is shown in the following formula:

[0093]

[0094] The present invention proposes a Zsd inversion model suitable for the South Four Lakes based on QAAL09 and the idea of ​​iterative optimization. The specific optimization is divided into the following two steps:

[0095] 1. Taking advantage of the high spectral resolution of ZY1-02D, moving the reference wavelength to a longer band can improve the accuracy of QAA in inland waters, and setting λ0 to 713, 722, 730, and 739 respectively;

[0096] 2. Fix λ2 to 748, change λ1 between 400-800nm, and optimize the Y value.

[0097] Combined with specific data, based on the experimental results obtained by the present invention, this is explained:

[0098] 1. Accuracy assessment of atmospheric correction and remote sensing reflectivity correction

[0099] The ZY1-02D ​​data were processed using the Flaash, QUAC, 6S, and Modtran algorithms to obtain the surface reflectance data. The results are as follows: Figure 2 As shown;

[0100] The present invention uses the band within the range of 400-900nm in the subsequent transparency inversion, that is, the 2nd to 59th bands of ZY1-02D. Flaash will have negative values ​​in some bands. This is because the algorithm over- or under-corrects the atmospheric effect in these bands. The spectral characteristics of QUAC are more obvious at 404-687nm (Band2-35), and the reflectivity is 0.025-0.08sr -1 The reflectance of the bands after 696nm (Band 36) is close to 0, indicating that the algorithm has a poor effect in these bands. The reflectance of the 6S algorithm is abnormal in the first few bands, but the spectral characteristics of the water body are also obvious, with a reflectance of 0.02-0.08sr -1 The reflectivity of different points in the Modtran algorithm varies greatly in the short wavelength band, and the overall reflectivity is between 0.02 and 0.08 sr -1 The red dotted line in the figure represents the mean reflectivity of 1530-1630nm. It can be found that the red dotted lines of the four algorithms are basically below the visible light and near-infrared bands (404-894nm). Therefore, we use the visible light and near-infrared bands minus the mean reflectivity of 1530-1630nm, and then divide it by π to convert the surface reflectivity into remote sensing reflectivity. The results are as follows Figure 3a-3c As shown;

[0101] Figure 3a It shows that the ZY1-02D ​​data after remote sensing reflectance correction is more consistent with the ASD equivalent spectrum. Figure 3cThe MRE and RMSE before and after correction are shown. Flaash has negative values ​​in multiple bands, resulting in poor correction effect. However, after 500nm, the MRE before correction is 64%-94%, and the RMSE is 0.02-0.04sr-1; the MRE after correction is 22%-73%, and the RMSE is 0.002-0.005, and the average MRE is increased by about 37%. QUAC has a large number of zero values ​​in the reflectivity of the band after 696nm. Only the correction effect in the range of 404-687nm is evaluated. The MRE before correction is 50%-82%, and the RMSE is 0.01-0.04; the MRE after correction is 12%-61%, and the RMSE is 0.002-0.006sr-1, the average MRE increased by about 41%, the MRE before 6S algorithm correction was 71%-94%, the RMSE was 0.02-0.05sr-1, the MRE after correction was 11%-76%, the RMSE was 0.002-0.008sr-1, the MRE in the range of 404-730nm was less than 50%, of which the MRE in 490-662nm was less than 20%, and the average MRE increased by 45%. There were a few outliers after Modtran correction, but the overall correction effect was significant, the MRE in 499-722nm was less than 50%, and the MRE in 524-585nm was less than 30%;

[0102] Ten bands and two ratios of ZY1-02D ​​were selected, and the accuracy of different atmospheric correction algorithms in each band was systematically evaluated. The results are as follows: Figure 3b As shown in the figure, the MAPE of the 6S algorithm at 447nm is 37%, and the Modtran has the worst accuracy (71%); the 6S algorithm performs best at 490nm (MAPE=26%), and its accuracy is improved by 16%, 6% and 14% respectively compared with Flaash, QUAC and Modtran; at 550nm, the accuracy of QUAC and 6S is similar (12% and 13%), and the accuracy of Flaash is the lowest (23%); the MAPE of 6S at 620nm is 15%, indicating that in the range of 447-620nm, the 6S algorithm has the best atmospheric correction effect on the water body of the South Four Lakes, but in the range of 670-748nm, the accuracy of 6S decreases, and the accuracy of Flaash increases, and the longer the wavelength, the greater the gap between the two;

[0103] QUAC has moderate effect at short wavelengths, but its accuracy in long-wavelength bands deteriorates. Modtran’s accuracy drops to 151% at 739nm and 174% at 748nm. For two ratios (490 / 560 and 709 / 665), the ratio form can improve the accuracy of a single band. For example, the MAPE of QUAC at 490nm is 32%, but the MAPE of 490 / 550 drops to 17%. The MAPE of Modtran at 705nm is 36%, while the MAPE of 705 / 670 is only 3%, which effectively offsets the common errors of different bands and reduces the impact of atmospheric scattering and absorption.

[0104] 2. Denoising algorithm accuracy evaluation

[0105] The atmospheric correction results of Flaash, QUAC, 6S and Modtran were denoised by LF, HF, MF, SGF and HA respectively, and the SNR of all data was calculated using the local variance method. The results are shown in Figure 4 As shown;

[0106] Figure 4The results show that the effects of denoising algorithms in different atmospheric correction algorithms are not the same. The SNR of Flaash original data has a relatively stable upward trend between wavelengths of 400-500nm and 680-750nm, and the SNR reaches a peak at 748nm. With the increase of wavelength, the signal quality after denoising by the Flaash algorithm improves. The effects of MF, HA and SGF are relatively similar, especially between wavelengths of 500-700nm. The SNR curves of these methods are basically higher than the original data, indicating that they have a certain enhancement effect on the signal. HF has the worst effect. In the entire band, the SNR is almost maintained at a very low level, indicating that HF ​​weakens the signal strength while removing low-frequency noise. LF shows a more unique trend, and the SNR remains at a high level, indicating that LF retains the effective signal while removing high-frequency noise. The SNR of QUAC is between 400-600nm. It is better than the Flaash algorithm and has no negative values, but the SNR at 600-700nm decreases slightly and then increases suddenly. The effects of different denoising algorithms on QUA are significantly different. Overall, LF has the best denoising effect, HF has the worst, and MF is better than SGF. The SNR curve of 6S shows large fluctuations between 400-900nm and reaches a peak near 740nm, but the decline after the peak is more significant than that of Flaash and QUAC. LF and MF show good denoising effects, indicating that these two methods have significant advantages in signal enhancement. SGF performs relatively smoothly, but its SNR is slightly lower than MF and LF, and it almost overlaps with the original data in some bands. Modtran's SNR performance is similar to Flaash. LF and MF significantly improve the SNR of the original data. HA and SGF have advantages in some bands, and they overlap with the original data as a whole. HF has poor effect.

[0107] 3. Reconstructed data accuracy assessment

[0108] From the above conclusions, it can be seen that different atmospheric correction algorithms on ZY1-02D ​​data have different effects on different bands, and the denoising algorithm can significantly improve the signal-to-noise ratio of hyperspectral data. In order to improve the correction accuracy of hyperspectral images in inland waters, 10 bands required for the subsequent QAA algorithm are selected, and the MAPE-DR method proposed in this invention is used to process ZY1-02D ​​hyperspectral data. The accuracy of the reconstructed image is compared with that of the traditional method. Figure 5 As shown;

[0109] Figure 5It can be seen that the MAPE of the reconstructed data is at the global minimum. The MAPE of the reconstructed data at 447nm (Band7) can be reduced by 35% at most. The QUAC effect is best at 490nm (Band12). The accuracy of the reconstructed data at 490nm can be improved by 3%-19%, and the accuracy of the reconstructed data at 620nm (Band27) is improved by about 8%. At 670-722nm (Band 27-39), QUAC, 6S and Modtran are not as good as Flaash. In these bands, the MAPE of the reconstructed data is between 23% and 39%, indicating that the red light band of the water body of the South Four Lakes is more suitable for Flaash processing. At 739nm (Band 41), QUAC and Flaash are not much different, but at 748nm (Band 42), the accuracy of the reconstructed data can be improved by 25%, 13%, 76% and 104% respectively compared with other methods. The above results show that this method is suitable for ZY1-02D ​​data and has good transferability. In the bands where some algorithms perform poorly, the advantages of other algorithms can be used to make up for the shortcomings. The reconstruction method significantly improves the matching accuracy of remote sensing data with the true value. In addition, 4 points were randomly selected in the upper and lower lakes of the South Four Lakes, and the MAPE and RMSE of the reconstructed data and other methods were evaluated based on the point positions. The results are as follows: Figure 6 As shown;

[0110] Figure 6 The MAPE radar chart in the results shows that the MAPE of the reconstructed data is the lowest among all algorithms, especially at Point3 and Point8. The MAPE of the reconstructed data of Point3 is 28%, while the MAPEs of other methods are 46%, 70%, 98% and 73% respectively; the MAPE of the reconstructed data of Point8 is 48%, while the MAPEs of other methods are as high as 92%, 63%, 136% and 166%. Although the MAPE of Flaash at Point2 is close to the reconstructed data, the overall accuracy is still not as stable as the reconstruction method, which is also due to the further improvement of data accuracy by the denoising algorithm. The similar trend is also obvious in the RMSE radar chart. Flaash performs close to the reconstructed data at some measurement points, but the overall RMSE is still higher than the reconstructed data. QUAC and Modtran perform relatively poorly. The RMSE of the reconstructed data at Point4 is the lowest, and 6S is close to the reconstructed data, but QUAC and Modtran have higher RMSE at this point. Modtran has the highest error value at multiple points. The point-based accuracy evaluation further illustrates the accuracy and stability of MAPE-DR in the processing of inland water hyperspectral remote sensing data.

[0111] The present invention also provides a detailed process of using the field data of Nanshan Lake to perform the Zsd inversion model to verify the feasibility of the method of the present invention. According to the above description, the Zsd inversion model of the South Four Lakes is first developed using the in-situ data, and the QAAV5, QAAV6, QA-L09 and QAA-M14 algorithms are respectively constructed. By comparison, it is found that QAA-L09 is more suitable for the South Four Lakes in Shandong Province. In the process of QAA, there are some empirical steps, but the present invention does not have the measured absorption coefficient and backscattering coefficient data. Therefore, referring to the research of Feng et al., the measured Zsd data is used to correct the algorithm results. The results show that the power function is used to correct the QAA results in the South Four Lakes with higher accuracy, and the correction forms of different algorithms are shown in the following formulas:

[0112]

[0113]

[0114] In the formula, Represents the results of different algorithms after the precise measurement of Zsd data is corrected by the power function, and x is the original result of the Zsd algorithm;

[0115] Comparison of the accuracy of different versions of QAA algorithms Figure 7 As shown in Figure 2, the Zsd inverted by QAA722 is more consistent with the field measurement value, and its scattered point distribution is more in line with the 1:1 straight line. The algorithm can also produce a higher R 2 and lower RMSE (0.8539 and 0.0544), the errors between the Zsd estimated by QAAV5 and QAAM14 and the measured values ​​are large, and the overestimation is serious in the medium concentration range (0.4-0.6m). When the Zsd range is 0.6-0.8m, the underestimation phenomenon of QAAV5 is more serious. QAAL09 and QAAV6 also have good inversion effects. 2 The accuracy of QAAL09 in the low and medium concentration Zsd range (0.2-0.5 m) is close to that of QAA722, but its overall accuracy is still significantly lower than that of QAA722. Compared with QAAV5, QAAV6, QAAL09 and QAAM14, QAA722 has a lower accuracy in the inversion of Zsd in the South Four Lakes. 2 The accuracy can be improved by 37%, 7%, 6% and 12% respectively, and the RMSE can be reduced by 38%, 15%, 14% and 22% respectively. In summary, the QAA722 proposed in this study has effectively improved the existing QAA algorithm, significantly improved the inversion accuracy and model robustness of Zsd, and can provide a more adaptable solution for the high-precision demand of Zsd inversion in the water bodies of the South Four Lakes.

[0116] The present invention provides image inversion result evaluation, specifically:

[0117] The image spectral features of all sample points were extracted from Flaash, QUAC, 6S, Modtran and reconstructed data to construct the QAA722 algorithm. Since the reflectance of QUAC in the band after 696nm is almost 0, the correction effect is not ideal, which will lead to a large number of negative values ​​when using QAA. Therefore, the performance of QAA722 on the other four images is evaluated only by MAPE, RMSE, AURE and MAE. The results are shown in the figure. Figure 8 As shown;

[0118] Figure 8 The results show that there are obvious differences in prediction accuracy among different algorithms. The QAA722 algorithm based on reconstructed data performs best in all evaluation indicators, with an RMSE of only 0.1774, which is about 37%, 42% and 37% lower than Flaash, 6S and Modtran respectively. The MAPE of QAA722 based on reconstructed data is 0.2328, which is 36%, 44% and 41% higher than other algorithms respectively, showing a strong relative accuracy advantage. AURE and MAE also show similar trends, with reconstructed data about 55.1% and 48.4% lower than 6S, about 46% and 44% lower than Flaash, and about 51% and 44% lower than Modtran.

[0119] In summary, the reconstructed data has significant advantages in the inversion of the water body Zsd of the South Four Lakes, which can improve the inversion accuracy and reduce errors. 6S performs the worst, especially in MAPE and AURE. The inversion accuracy of Flaash and Modtran can be improved by about 10% compared with 6S. The QAA722 algorithm based on reconstructed data is the best choice for the inversion of the water body Zsd of the South Four Lakes in this analysis, with strong prediction accuracy and stability. For future research, especially in the inversion of water bodies with complex water quality and lighting conditions, it is recommended to give priority to the QAA722 algorithm. The QAA algorithm proposed in this study is applied to the reconstructed data to obtain the spatial distribution map of the water body of the South Four Lakes. The image inversion results of Dushan Lake are shown in Figure 2. Fig. 9 As shown, the results of Weishan Lake are as follows Fig.10 As shown;

[0120] Fig. 9It can be seen that the Zsd of the water body of Dushan Lake shows obvious spatial distribution differences. The central area of ​​the lake is dominated by higher Zsd, and the Zsd values ​​in the southern area are mostly concentrated above 0.9m, reflecting a relatively clear water condition. In the area near the edge of the lake, Zsd decreases significantly, especially near areas with frequent human activities (such as the northwest and eastern coast), the transparency value drops to below 0.4m. The center of the lake is far away from the direct impact of human activities, such as agricultural runoff, domestic sewage and industrial wastewater discharge, so the degree of water pollution is relatively low and the water body Zsd is higher. In addition, the water depth in the central area of ​​the lake is larger, and the suspended matter sedimentation rate is faster, which also helps to increase the Zsd level. The edge of the lake is greatly affected by human activities. Agricultural irrigation, aquaculture activities and domestic sewage are directly discharged into the lake, which aggravates the input of nutrients and leads to eutrophication of the water body. At the same time, the water depth at the edge of the lake is shallow, and the sediment disturbance and algae growth also have an adverse effect on Zsd.

[0121] Fig.10 The results show that the water Zsd distribution of Weishan Lake is more complex and the overall level is lower. Zsd is mainly concentrated between 0.1-0.5 meters. The Zsd in the central area of ​​the lake is generally lower than that of Dushan Lake, and most areas are concentrated between 0.35-0.55 meters. The Zsd along the coast is mainly concentrated below 0.3 meters, especially in the southeastern coastal area. From the perspective of natural conditions, Weishan Lake has a larger area, weaker hydrodynamic conditions, and it is difficult for suspended matter to settle, which aggravates the turbidity of the overall water body. From the perspective of human activities, the population along the coast of Weishan Lake is dense, farmland and industry are widely distributed, and agricultural irrigation drainage, domestic sewage and industrial wastewater are the main sources of pollution. In particular, agricultural activities in the northwest and urban life emissions in the southeast have a particularly prominent impact on the water quality of the lake. The input of a large amount of nutrients leads to eutrophication of the lake, and the large-scale reproduction of algae further reduces Zsd. In addition, aquaculture activities along the coast of Weishan Lake and local sand mining operations will also aggravate the deterioration of the lake's Zsd.

[0122] The present invention is summarized. Taking the South Four Lakes in Shandong Province as an example, the present invention uses the MAPE-DR method to process the ZY1-02D ​​hyperspectral remote sensing data. The results show that the MAPE of MAPE-DR at 447nm in ZY1-02 can be reduced by 35% at most, the accuracy at 490nm can be improved by 3%-19%, the accuracy at 620nm can be improved by about 8%, and the accuracy at 748nm can be improved by 25%, 13%, 76% and 104% respectively compared with other methods. MAPE-DR is based on the AS The MAPE of the D equivalent spectrum is calculated by selecting the optimal algorithm combination in each band. The reconstructed data has a higher matching degree with the real spectral data, and it has both robustness in band and spatial positioning evaluation. Based on the existing QAA algorithm, the idea of ​​iterative optimization is used to move the reference wavelength to a longer band, and the empirical parameter Y value is improved. A QAA722 algorithm suitable for Zsd inversion of the South Four Lakes is developed. Compared with QAAV5, QAAV6, QAAL09 and QAAM14, QAA722 has a higher performance in Zsd inversion of the South Four Lakes. 2 The data can be improved by 37%, 7%, 6% and 12% respectively, and the RMSE can be reduced by 38%, 15%, 14% and 22% respectively. In the Zsd inversion of inland water bodies, the present invention can effectively improve the data quality and the Zsd inversion accuracy, has a wide range of application prospects, and provides a scientific basis for water environment monitoring and management.

[0123] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0124] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for inland water transparency inversion based on MAPE-DR and improved QAA, characterized in that: include: Acquisition of hyperspectral remote sensing data; Processing of hyperspectral remote sensing data based on MAPE-DR; Based on the improved QAA algorithm, the absorption coefficient and backscattering coefficient are obtained according to the processed hyperspectral remote sensing data. The diffuse reflection attenuation coefficient is calculated based on the absorption coefficient and the backscattering coefficient. The water transparency is calculated through the diffuse reflection attenuation coefficient.

2. The method according to claim 1, characterized in that The hyperspectral remote sensing data is ZY1-02D ​​hyperspectral data, which is obtained from the natural resources satellite remote sensing cloud service platform.

3. The method according to claim 2, characterized in that The hyperspectral remote sensing data is processed based on MAPE-DR, specifically: Perform atmospheric correction on the hyperspectral data to obtain surface reflectance data, and convert the surface reflectance data into remote sensing reflectance based on remote sensing reflectance correction; The remote sensing reflectance was denoised based on LF, HF, MF, SGF and HA denoising methods respectively; For each band, the MAPE values ​​of different combination methods and ASD equivalent spectra were compared, the best method combination was selected as the optimal band, and all the optimal bands were stacked to form a reconstructed image.

4. The method according to claim 3, characterized in that Based on the improved QAA algorithm, the absorption coefficient and backscattering coefficient are obtained according to the processed hyperspectral remote sensing data, specifically: Improvements are made based on QAAV6 and QAAL09. The specific forms are: In the formula, r rs represents the remote sensing reflectivity below the water surface, R rs represents the remote sensing reflectivity above the water surface, u(λ) is r rs function, a(λ0) is the water absorption coefficient at the reference wavelength, b bp (λ0) is the water backscattering coefficient at the reference wavelength, a w represents the absorption coefficient of pure water, b bw is the backscattering coefficient of pure water, Y is an empirical parameter, a(λ) and b bp (λ) are the total absorption coefficient and backscattering coefficient of all wavelengths respectively; The QAAA optimization steps are as follows: taking advantage of the ZY1-02D ​​hyperspectral spectrum, set λ0 to 713, 722, 730, and 739, respectively; fix λ2 to 748 nm, vary λ1 between 400 and 800 nm, and optimize the Y value.

5. The method according to claim 4, characterized in that The diffuse reflection attenuation coefficient is calculated based on the absorption coefficient and the backscattering coefficient, and the water transparency is calculated by the diffuse reflection attenuation coefficient. The specific calculation form is: In the formula, K d (λ) diffuse reflection attenuation coefficient, Represents the minimum value of the diffuse reflection attenuation coefficient of the transparent window of water in the range of 410-665nm, is the remote sensing reflectance of the corresponding band, a(λ) and b bp (λ) are the total absorption coefficient and backscattering coefficient of water, b bw (λ) is the backscattering coefficient of pure water, θ s is the solar zenith angle, m0 is 0.005, m1 is 4.26, and m2 is 0.52.

Images