Aerosol complex refractive index self-adaptive adjustment method and system based on structure shaped like Chinese character'mi '

Abstract

The invention belongs to the technical field of atmospheric detection and remote sensing, particularly relates to an aerosol complex refractive index self-adaptive adjustment method and system based on a star-shaped structure, and provides a star-shaped structure function model and a comprehensive weight factor to fully consider the complexity of coastline space change. And when the aerosol lookup table is constructed, the imaginary part of the aerosol complex refractive index is adaptively adjusted according to the weight factor, so that the precision of optical thickness inversion of the atmospheric aerosol in the offshore region is improved.

Description

Technical Field

[0001] The present application belongs to the field of atmospheric detection and remote sensing technology, and specifically relates to an aerosol complex refractive index adaptive adjustment and system based on a crisscross structure. Background Art

[0002] The structure function method is a method that uses satellite image ambiguity to invert aerosol optical thickness. It assumes that the surface reflectivity is constant, ignores multiple scattering between the surface and the atmosphere, and is not restricted by the surface reflectivity. However, in existing technical literature, the structure function model used considers limited spatial directions when calculating the difference between pixels, and does not take into account the regional differences in the imaginary part of the aerosol complex refractive index. There is a lack of research on offshore areas, especially coastal areas. The limitations of these models, especially their application in coastal areas, have become a problem that needs to be solved in the field of satellite remote sensing inversion of atmospheric aerosols.

[0003] Compared with deserts and urban areas, the coastline is a curve with no regular distribution, and the surface reflectivity at the junction of land and sea varies significantly. The traditional pixel differences in several directions are difficult to cover the complexity of the coastline. In the satellite data grid, each pixel has eight adjacent pixels, and the lines between the center point of the pixel and the center points of the eight adjacent pixels form a cross. Taking the differences of the eight adjacent pixels into account in the structure function model can cover the complex changes of the coastline. In addition, aerosols in coastal areas are affected by both land-based anthropogenic sources and sea salt aerosols. The spatial and temporal distribution of aerosol optical properties varies significantly compared to inland areas, especially the imaginary part of the complex refractive index. However, in the existing technical literature, the regional differences in the imaginary part of the aerosol complex refractive index are rarely considered. When constructing the aerosol lookup table, the sun angle, satellite angle and aerosol optical parameters are set to several fixed values, and multiple linear interpolations are performed when inverting the aerosol. Although this method has higher accuracy and efficiency in inland areas, from the perspective of error transmission, multiple interpolations will bring errors into the inversion results of aerosol optical thickness. However, this method is not applicable in nearshore and coastal areas. Summary of the invention

[0004] In view of the shortcomings of the prior art, the present invention proposes an adaptive adjustment system for the complex refractive index of aerosols based on a crisscross structure, which takes into account the regional changes in the sun angle, satellite angle and aerosol optical parameters, and improves the inversion accuracy of atmospheric aerosol optical thickness in offshore areas. The technical solution is as follows: A method for adaptively adjusting the complex refractive index of an aerosol based on a pixilated structure comprises the following steps: S1. Obtain satellite images of the area to be inverted; S2. Preprocessing of L1 satellite remote sensing data in the area to be inverted; S3. Constructing a cross-shaped structure function model to calculate the apparent reflectivity structure function value of the image to be inverted and the surface reflectivity structure function value; S4. Calculate the weight factor for each pixel of the surface reflectance image, and calculate the window movement law to obtain the comprehensive weight factor after resampling; S5. Adaptively adjust the imaginary part of the aerosol complex refractive index based on the comprehensive weight factor to construct a lookup table for atmospheric aerosol optical thickness and atmospheric transmittance; S6. Substitute the ratio of the apparent reflectance to the structure function value of the surface reflectance into the fitted formula of atmospheric transmittance and optical thickness to realize the inversion of aerosol optical thickness using the structure function method.

[0005] Preferably, in step S1, the historical surface reflectance is used as a clear image, and the seasonal changes in the phenology of surface vegetation are taken into consideration. When downloading the surface reflectance products of the area to be inverted in the adjacent years, they need to be stored separately according to the seasons; the historical surface reflectance products are synthesized by minimum value by season to obtain surface reflectance images of the four seasons of spring, summer, autumn and winter.

[0006] Preferably, in step S2, the preprocessing includes geometric correction, radiation calibration, and cloud detection; During geometric correction, the projection and coordinate system parameters should be set according to the location of the study area; Radiometric calibration uses the image's own correction coefficients to calculate and obtain the corresponding apparent reflectance; Cloud detection uses the threshold method to identify whether it is a cloud pixel by pixel. Cloud pixels are 1 and non-cloud pixels are 0. The cloud mask is stored as a binary image, and only 0-valued pixels are retained in subsequent processing.

[0007] Preferably, in step S3, assuming that the surface reflectivity remains unchanged and ignoring multiple scattering between the surface and the atmosphere, the ratio of the image structure function value of turbid weather to that of clear weather is the atmospheric transmittance, and thus the aerosol optical thickness is inverted according to the radiation transfer model. The cross-shaped structure function model is as follows: ; in, Indicates the minimum value; Indicates the row number, Indicates the number of columns, represents the pixel interval, Indicates the model calculation window size, Represents the image reflectivity.

[0008] Preferably, the cross-shaped structure function performs multiple pixel-by-pixel traversals within the calculation window. When Row, No. Starting from the column, calculate the distance between each pixel and the 8 directions The difference between pixels until the pixel spacing increases to , and then compute the mean of the multiple passes.

[0009] Preferably, in step S4: S41. Calculate the distance between land and sea based on the coastline position and pixel position ,use To express the weight of ocean contribution, based on The weights are set in sections: ; in: represents the ocean contribution of one unit, Indicates the number of ocean pixels. When the pixel is ocean, the distance between land and sea is is a positive value. When the pixel is land, the distance between land and sea is a negative number. For pixels on the coastline, their weight is proportional to the number of ocean pixels in the eight adjacent pixels. S42. Utilization To represent the weight of industrial life, the weight of industrial land is the largest, set , the weight of residential area is less than that of office building, so residential area is set as , the office building is set up ,in represents the contribution of a unit’s industrial life; S43. Obtain the comprehensive weight factor after resampling based on the movement law of the structure function method model window ,in , represents the proportion of ocean contribution, Indicates the proportion of industrial life, and the values ​​are as follows: ; When the land pixel is more than 20 km away from the coastline, the ocean contribution is ignored; when the ocean pixel is more than 20 km away from the coastline, the industrial life contribution is ignored.

[0010] Preferably, in step S5, after determining the calculation window size of the structure function model, the solar zenith angle, solar azimuth angle, satellite zenith angle and satellite azimuth angle are simultaneously resampled, and when constructing a lookup table using the 6S model, the resampled observation geometry value is directly input; Atmospheric mode: Use the custom mode to input the vertical profile data of atmospheric pressure, temperature, water vapor content and ozone content of the corresponding season according to the time of the image to be inverted; Aerosol mode: Select Junge spectrum, you need to input the Junge spectrum index and complex refractive index, and use the comprehensive weight factor obtained in step S4 to adaptively change the imaginary part of the complex refractive index. The final form of the comprehensive weight factor is , the imaginary part after adaptive adjustment is ,in, represents the ocean contribution coefficient, represents the industrial life contribution coefficient, represents the imaginary part of sea salt aerosol particles, Represents the imaginary part of continental aerosol particles.

[0011] Preferably, in step S5, regarding the sensor code, 6S ​​comes with a configured sensor band. For sensor bands that are not covered, the corresponding spectral response function needs to be converted and added to the program; 6S simulates the transmission process of solar electromagnetic radiation between 0.25μm and 4.0μm, and divides the electromagnetic wave into intervals of 2.5nm to obtain 1501 wavelength values , so when adding a band, you only need to add the starting value of the wavelength of the band to be added. and According to the nearest neighbor interpolation method, Determine the corresponding wavelength value in and ,in and Indicates the position of the interpolated wavelength starting value in 1501 elements, and is obtained by bilinear interpolation arrive Spectral response values ​​between , set the corresponding spectral values ​​of the remaining positions to 0; after given the observation geometry, atmospheric model, aerosol model and sensor code, a lookup table containing atmospheric transmittance and 550nm aerosol optical thickness is constructed.

[0012] Preferably, in step S6, an exponential function is used for fitting, and the fitting formula is as follows: , ; in represents the atmospheric transmittance, represents the atmospheric aerosol optical thickness at 550nm, and are the fitting coefficients, where is a negative number; during the fitting process, the ratio of the structure function value of the apparent reflectance to the surface reflectance is , and is the coefficient determined by fitting, and the aerosol optical thickness can be obtained.

[0013] An aerosol complex refractive index adaptive adjustment system based on a pixilated structure, comprising a data acquisition unit, a data processing unit and an output unit; The data acquisition unit is used to obtain satellite images of the area to be inverted; The data processing unit: processes the data, constructs a cross-shaped structure function model, calculates the apparent reflectivity structure function value of the image to be inverted and the surface reflectivity structure function value; calculates the weight factor for each pixel of the surface reflectivity image, and obtains the comprehensive weight factor after resampling according to the calculation window movement law of the structure function model; according to the comprehensive weight factor, adaptively adjusts the imaginary part of the aerosol complex refractive index, and constructs a lookup table of atmospheric aerosol optical thickness and atmospheric transmittance; brings the ratio of the structure function value of the apparent reflectivity to the surface reflectivity into the fitted formula of atmospheric transmittance and optical thickness, and realizes the inversion of aerosol optical thickness by the structure function method; Output unit: Output the results visually.

[0014] Compared with the prior art, the present invention has the following beneficial effects: 1. More detailed spatial structure: The cross-shaped structure model is adopted, and the pixel differences in eight directions are considered at the same time, and the spatial structure covers a wide range; 2. Considering the mixing of multi-source aerosol particles: Since the offshore area is affected by both the ocean and the land, a comprehensive weight factor is used to simultaneously consider the multi-source aerosol mixing effect brought by the ocean and industrial life, and the aerosol complex refractive index can be adaptively adjusted; 3. Accurate observation geometry: This application generates a targeted atmospheric aerosol lookup table for each pixel. After determining the migration law of the structure function model, it also completes the resampling of the solar zenith angle and azimuth angle, the satellite zenith angle and azimuth angle. Compared with traditional technologies, it reduces the error transmission of aerosol optical thickness. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a technical flow chart for inverting atmospheric aerosol content using the cross-shaped structure function model; Figure 2 It is a comparison diagram of the structure function model, (A) is a single-direction model, (B) is a three-direction model, (C) is a city model, and (D) is a cross-shaped model; Figure 3 It is a comparison chart of the ratios of the four structure function models; Figure 4 This is a comparison chart of the inversion accuracy of 550nm atmospheric aerosol optical thickness. (A) is a single-direction model, (B) is a three-direction model, (C) is a city model, and (D) is a cross-shaped model. DETAILED DESCRIPTION

[0016] The technical solution of the present application is described in detail below through specific embodiments and drawings. It should be understood that the embodiments of the present application and the specific features in the embodiments are detailed descriptions of the technical solution of the present application, rather than limitations on the technical solution of the present application, and the specific technical features may be combined with each other.

[0017] A method for adaptively adjusting the complex refractive index of an aerosol based on a pixilated structure comprises the following steps: S1. Obtain satellite images of the area to be inverted; The historical surface reflectance is used as a clear image, and the seasonal changes in surface vegetation phenology are taken into consideration. When downloading the surface reflectance products of the area to be inverted in the adjacent years, they need to be stored separately according to the seasons; the historical surface reflectance products are synthesized at the minimum value by season to obtain the surface reflectance images of the four seasons of spring, summer, autumn and winter.

[0018] S2. Preprocessing of L1 satellite remote sensing data in the area to be inverted; Preprocessing mainly includes geometric correction, radiometric calibration, and cloud detection. Whether the L1 data is geometrically corrected depends on the instructions for use of the satellite sensor data. For example, the L1B data of MODIS (MODerate-resolution Imaging Spectroradiometer) is not geometrically corrected, while the L1C data of Sentinel-2 and the L1B data of VIIRS (Visible Infrared Imaging Radiometer Suite) are geometrically corrected. During geometric correction, the projection and coordinate system parameters should be set according to the location of the study area. Albers conformal projection and WGS-84 coordinate system are usually selected in China. Radiometric calibration can be calculated using the correction coefficients of the image itself to obtain the corresponding apparent reflectance. Cloud detection is mainly done by using the threshold method to identify whether it is a cloud pixel by pixel. Cloud pixels are 1 and non-cloud pixels are 0. The cloud mask is stored as a binary image, and only 0-value pixels are retained in subsequent processing.

[0019] S3. Construct a cross-shaped structure function model to calculate the apparent reflectivity structure function value of the image to be inverted and the surface reflectivity structure function value.

[0020] In order to invert aerosols over high reflectivity surfaces, Tanre proposed a structure function method that uses the ambiguity of satellite images in turbid weather to invert aerosol optical thickness, which is not restricted by the surface reflectivity. The structure function method requires satellite images of clear weather as a reference, and calculates the structure function value after removing the atmospheric effect. Assuming that the surface reflectivity remains unchanged and ignoring multiple scattering between the surface and the atmosphere, the ratio of the image structure function value of turbid weather to that of clear weather is the atmospheric transmittance, and then the aerosol optical thickness is inverted according to the radiation transfer model, as shown in formula (1): (1); in, is the structure function value of the surface reflectance obtained by atmospheric correction of the clear weather image, is the structure function value of the apparent reflectivity of the image to be inverted, , , and are the solar zenith angle and the satellite zenith angle, respectively.

[0021] According to formula (1), the core of the structure function method lies in the calculation of the structure function value. There are three commonly used structure function models, as shown in formulas (2), (3), and (4): (2); (3); (4); in, Indicates the row number, Indicates the number of columns, represents the pixel interval, Indicates the model calculation window size, represents the image reflectivity, Represents the maximum value of the pixel interval, Indicates the minimum value of the pixel interval.

[0022] Formulas (2), (3), and (4) correspond to the one-way model, three-way model, and city model, respectively. The one-way model calculates the pixel-by-pixel distance in the window and its row direction interval. The three-way model is to calculate the pixel difference in the window pixel by pixel and its spacing in the row direction, column direction and diagonal direction The city model calculates the pixel difference of each pixel in the window with a certain distance. The coastline is an irregular curve with complex spatial angle changes. In order to fully cover the structural characteristics of the coastline, this application proposes a cross-shaped structure model. The calculation formula is as follows: (5); in, express The minimum value in .

[0023] Compared with other structural models, the cross-shaped structure proposed in this patent will perform multiple pixel-by-pixel traversals within the calculation window. When Row, No. Starting from the column, calculate the distance between each pixel and the 8 directions The difference between pixels until the pixel spacing increases to , and then compute the mean of the multiple passes.

[0024] S4. Calculate the weight factor for each pixel of the surface reflectance image, and calculate the window movement law to obtain the comprehensive weight factor after resampling.

[0025] The comprehensive weight factor proposed in this application takes into account both the contribution of the ocean and the impact of industrial life.

[0026] First, the professional geographic information system software ARCGIS is used to calculate the distance between land and sea based on the coastline position and pixel position. This application uses To express the weight of ocean contribution, based on The weights are set in sections: ; The unit is KM. represents the ocean contribution of one unit, Indicates the number of ocean pixels. When the pixel is an ocean, the distance between land and sea is is a positive value, and when the pixel is land, the distance between land and sea is a negative number. For pixels on the coastline, their weight is proportional to the number of ocean pixels in the eight adjacent pixels.

[0027] Secondly, the industrial land classification in the land use data is used to determine factories, residential areas, and office buildings. To represent the weight of industrial life, the weight of industrial land is the largest, set , the weight of residential area is less than that of office building, so residential area is set as , the office building is set up ,in Represents the industrial life contribution of a unit.

[0028] Finally, the comprehensive weight factor after resampling is obtained according to the movement law of the structure function method model window. ,in , represents the proportion of ocean contribution, Indicates the proportion of industrial life, and the values ​​are as follows: ; When the land pixel is more than 20 km away from the coastline, the ocean contribution is ignored; when the ocean pixel is more than 20 km away from the coastline, the industrial life contribution is ignored.

[0029] S5. Based on the comprehensive weight factor, the imaginary part of the aerosol complex refractive index is adaptively adjusted to construct a lookup table for atmospheric aerosol optical thickness and atmospheric transmittance.

[0030] The 6S (Second Simulation of the Satellite Signal in the Solar Spectrum) radiation transfer model is used to construct a lookup table. In addition to the angle information of the sun and the satellite, the atmospheric mode, aerosol mode, aerosol content, and sensor code need to be given. The present invention uses curve fitting to invert the atmospheric aerosol optical thickness, which has high requirements for the lookup table generated by the radiation transfer model. Usually, the lookup table is constructed by considering only a number of solar zenith angles, solar azimuth angles, satellite zenith angles, and satellite azimuth angles, and does not consider the changes in the observation geometry in the region. The whole domain defaults to a value. When this angle does not exist in the lookup table, interpolation is performed. However, this process will bring errors into the subsequent atmospheric aerosol optical thickness inversion. Therefore, after determining the calculation window size of the structure function model, the present invention simultaneously resamples the solar zenith angle, solar azimuth angle, satellite zenith angle, and satellite azimuth angle. When the 6S model is used to construct the lookup table, the resampled observation geometry value is directly input.

[0031] Atmospheric mode uses a custom mode and inputs the vertical profile data of atmospheric pressure, temperature, water vapor content and ozone content of the corresponding season according to the time of the image to be inverted.

[0032] Aerosol mode, select Junge spectrum, you need to input the Junge spectrum index and complex refractive index, where the real part of the complex refractive index does not change much. This application mainly uses the comprehensive weight factor obtained in step S4 to adaptively change the imaginary part of the complex refractive index. The final form of the comprehensive weight factor is , represents the ocean contribution coefficient, Represents the contribution coefficient of industrial life, and further obtains the imaginary part after adaptive adjustment ,in, represents the ocean contribution coefficient, represents the industrial life contribution coefficient, represents the imaginary part of sea salt aerosol particles, Represents the imaginary part of continental aerosol particles.

[0033] For the aerosol content module, this application uses aerosol optical depth. Since there is no situation where the aerosol content is zero, the parameters of aerosol optical depth are set to: 0.001, 0.05, 0.1~1.0 (interval of 0.1), 1.0~2.0 (interval of 0.2), and 2.5.

[0034] Regarding sensor codes, 6S comes with 59 configured sensor bands. For sensor bands that are not covered, the corresponding spectral response function needs to be converted and added to the program. 6S can simulate the transmission process of solar electromagnetic radiation between 0.25μm and 4.0μm, and divide the electromagnetic waves into 2.5nm intervals to obtain 1501 wavelength values. , so when adding a band, you only need to add the starting value of the wavelength of the band to be added ( and ) According to the nearest neighbor interpolation method, Determine the corresponding wavelength value in and ,in and Indicates the position of the interpolated wavelength starting value in 1501 elements, and is obtained using bilinear interpolation arrive Spectral response values ​​between , and set the corresponding values ​​of the spectrum at the remaining positions to 0. Given the observation geometry, atmospheric model, aerosol model and sensor code, a lookup table containing atmospheric transmittance and 550nm aerosol optical thickness is constructed.

[0035] S6. Substitute the ratio of the apparent reflectance to the structure function value of the surface reflectance into the fitted formula of atmospheric transmittance and optical thickness to realize the inversion of aerosol optical thickness using the structure function method.

[0036] Through the simulation of 6S radiation transmission model, it is found that there is a strong exponential relationship between atmospheric transmittance and aerosol optical thickness. Therefore, this application uses exponential function for fitting. The fitting formula is as follows: ,in Indicates atmospheric transmittance (values ​​range from 0 to 1, indicating a negative index). represents the atmospheric aerosol optical thickness at 550nm, and are the fitting coefficients, where is a negative number. During the fitting process, the ratio of the structure function value of the apparent reflectance to the surface reflectance is , and is the coefficient determined by fitting, and the aerosol optical thickness can be obtained .

[0037] In order to explain the implementation process of the present invention in detail, the atmospheric aerosol content in the coastal area of ​​Bohai Bay monitored by VIIRS data is taken as an example to demonstrate the implementation process of inverting the atmospheric aerosol in the offshore area using the cross-shaped structure function model. Figure 1 As shown, the implementation steps are as follows: Step 1. Based on the date of the data to be inverted, obtain the VNP09 surface reflectance product of the corresponding season in the previous two years of the study area, and perform minimum value synthesis to obtain the surface reflectance image of the study area; Step 2. Obtain VNP03 data of the same time as VNP02 data, and use ENVI software to complete the geometric correction of VNP02 data; use IDL to read the radiation calibration parameters of VNP02 to obtain the apparent reflectance, and use the threshold method to remove cloud pixels to obtain the apparent reflectance image of the study area, the sun and the sensor observation angle; Step 3. Use the cross-shaped structure function model shown in formula (5) to calculate the structure function values ​​of the surface reflectance image and the apparent reflectance image respectively; Step 4. Calculate the weight factor pixel by pixel for the surface reflectance image based on land use, UAV and high-resolution satellite data, and resample the weight factor according to the calculation window movement law of the structure function model; Step 5. According to the comprehensive weight factor, the imaginary part of the aerosol complex refractive index is adaptively adjusted to construct a lookup table for the 550nm atmospheric aerosol optical thickness and the total atmospheric transmittance.

[0038] According to formula (1), the ratio of the apparent reflectance to the structure function value of the surface reflectance can be understood as the atmospheric transmittance of the image to be inverted. The atmospheric aerosol optical thickness at 550 nm corresponding to the atmospheric transmittance can be determined by using the curve fitting method.

[0039] In order to elaborate on the advantages of the cross-shaped structure function model, the cross-shaped structure function model is compared with three traditional structure function models, such as Figure 2 As shown. Among them, Figure 2 (A) is a single-direction model. Figure 2 The middle (B) is a three-direction model. Figure 2 The middle (C) is the city model. Figure 2 The middle (D) is the cross-shaped model proposed by the present invention. Obviously, the cross-shaped model takes the spatial structure into more detailed consideration.

[0040] according to Figure 1 The algorithm flow shown in the figure inverts the atmospheric aerosol optical thickness in the coastal area of ​​Bohai Bay on July 23, 2024. The structure function ratios of the four models are as follows: Figure 3 As shown, the fitting coefficient and The inversion results are compared with the aerosol products of MODIS, and the distribution of pixels with absolute inversion errors within 10%, 20% and 30% are plotted, as shown in Figure 2. Figure 4 shown.

[0041] This application does not take into account the directionality of ocean reflectivity. When calculating the structural function ratio of apparent reflectivity to surface reflectivity, there will be a situation where the ratio is greater than or equal to 1, which contradicts the transmittance being less than 1. This application counts the proportion of pixels with a ratio less than 1. The four models account for 53.56%, 50.02%, 50.63%, and 54.52% respectively. Obviously, the cross-shaped pixels have the highest proportion.

[0042] Figure 4 In the figure, blue indicates that the absolute value of the error is less than 10%, green indicates that the absolute value of the error is less than 20%, and red indicates that the absolute value of the error is less than 30%. It is not difficult to find that the inversion accuracy of the cross-shaped model proposed in this application in nearshore land and ocean areas is relatively high, and the results are more stable compared with the other three models.

[0043] An aerosol complex refractive index adaptive adjustment system based on a pixilated structure, comprising a data acquisition unit, a data processing unit and an output unit; The data acquisition unit is used to obtain satellite images of the area to be inverted; The data processing unit processes the data and constructs a cross-shaped structure function model to calculate the apparent reflectivity structure function value of the image to be inverted and the surface reflectivity structure function value; calculates the weight factor for each pixel of the surface reflectivity image, and obtains the comprehensive weight factor after resampling according to the calculation window movement law of the structure function model; according to the comprehensive weight factor, the imaginary part of the aerosol complex refractive index is adaptively adjusted to construct a lookup table of atmospheric aerosol optical thickness and atmospheric transmittance; the ratio of the structure function value of the apparent reflectivity to the surface reflectivity is brought into the fitted formula of atmospheric transmittance and optical thickness to realize the inversion of aerosol optical thickness by the structure function method; Output unit: Output the results visually.

[0044] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A method for adaptively adjusting the complex refractive index of aerosol based on a 'Piece-shaped' structure, characterized in that: The following steps are involved: S1. Obtain satellite images of the area to be inverted; S2. Preprocessing of L1 satellite remote sensing data in the area to be inverted; S3. Constructing a cross-shaped structure function model to calculate the apparent reflectivity structure function value of the image to be inverted and the surface reflectivity structure function value; S4. Calculate the weight factor for each pixel of the surface reflectance image, and calculate the window movement law to obtain the comprehensive weight factor after resampling; S5. Adaptively adjust the imaginary part of the aerosol complex refractive index based on the comprehensive weight factor to construct a lookup table for atmospheric aerosol optical thickness and atmospheric transmittance; S6. Substitute the ratio of the apparent reflectance to the structure function value of the surface reflectance into the fitted formula of atmospheric transmittance and optical thickness to realize the inversion of aerosol optical thickness using the structure function method.

2. The method for adaptively adjusting the aerosol complex refractive index based on the cross-shaped structure according to claim 1, characterized in that: In step S1, the historical surface reflectance is used as a clear image, and the seasonal changes in the surface vegetation phenology are taken into consideration. When downloading the surface reflectance products of the area to be inverted in the adjacent years, they need to be stored separately according to the seasons; the historical surface reflectance products are synthesized at the minimum value by season to obtain the surface reflectance images of the four seasons of spring, summer, autumn and winter.

3. The method for adaptively adjusting the aerosol complex refractive index based on the cross-shaped structure according to claim 1, characterized in that: In step S2, preprocessing includes geometric correction, radiometric calibration, and cloud detection; During geometric correction, the projection and coordinate system parameters should be set according to the location of the study area; Radiometric calibration uses the image's own correction coefficients to calculate and obtain the corresponding apparent reflectance; Cloud detection uses the threshold method to identify whether it is a cloud pixel by pixel. Cloud pixels are 1 and non-cloud pixels are 0. The cloud mask is stored as a binary image, and only 0-valued pixels are retained in subsequent processing.

4. The method for adaptively adjusting the aerosol complex refractive index based on the cross-shaped structure according to claim 1, characterized in that: In step S3, assuming that the surface reflectivity remains unchanged and ignoring multiple scattering between the surface and the atmosphere, the ratio of the image structure function value of turbid weather to that of clear weather is the atmospheric transmittance, and the aerosol optical thickness is inverted according to the radiation transfer model. The cross-shaped structure function model is as follows: ； in, Indicates the minimum value; Indicates the row number, Indicates the number of columns, represents the pixel interval, Indicates the model calculation window size, Represents the image reflectivity.

5. The method for adaptively adjusting the aerosol complex refractive index based on the cross-shaped structure according to claim 4, characterized in that: The cross-shaped structure function performs multiple pixel-by-pixel traversals within the calculation window. When Row, No. Starting from the column, calculate the distance between each pixel and the 8 directions The difference between pixels until the pixel spacing increases to , and then compute the mean of the multiple passes.

6. The method for adaptively adjusting the aerosol complex refractive index based on the cross-shaped structure according to claim 1, characterized in that: In step S4: S41. Calculate the distance between land and sea based on the coastline position and pixel position ,use To express the weight of ocean contribution, based on The weights are set in sections: ； in: represents the ocean contribution of one unit, Indicates the number of ocean pixels. When the pixel is ocean, the distance between land and sea is is a positive value. When the pixel is land, the distance between land and sea is a negative number. For pixels on the coastline, their weight is proportional to the number of ocean pixels in the eight adjacent pixels. S42. Utilization To represent the weight of industrial life, the weight of industrial land is the largest, set , the weight of residential area is less than that of office building, so residential area is set as , the office building is set up ,in represents the contribution of a unit’s industrial life; S43. Obtain the comprehensive weight factor after resampling based on the movement law of the structure function method model window ,in , represents the proportion of ocean contribution, Indicates the proportion of industrial life, and the values ​​are as follows: ； When the land pixel is more than 20 km away from the coastline, the ocean contribution is ignored; when the ocean pixel is more than 20 km away from the coastline, the industrial life contribution is ignored.

7. The method for adaptively adjusting the aerosol complex refractive index based on the cross-shaped structure according to claim 1, characterized in that: In step S5, after determining the calculation window size of the structure function model, the solar zenith angle, solar azimuth angle, satellite zenith angle and satellite azimuth angle are simultaneously resampled, and when constructing a lookup table using the 6S model, the resampled observation geometry value is directly input; Atmospheric mode: Use the custom mode to input the vertical profile data of atmospheric pressure, temperature, water vapor content and ozone content of the corresponding season according to the time of the image to be inverted; Aerosol mode: Select Junge spectrum, you need to input the Junge spectrum index and complex refractive index, and use the comprehensive weight factor obtained in step S4 to adaptively change the imaginary part of the complex refractive index. The final form of the comprehensive weight factor is , the imaginary part after adaptive adjustment is ,in, represents the ocean contribution coefficient, represents the industrial life contribution coefficient, represents the imaginary part of sea salt aerosol particles, Represents the imaginary part of continental aerosol particles.

8. The method for adaptively adjusting the aerosol complex refractive index based on the cross-shaped structure according to claim 7, characterized in that: In step S5, regarding the sensor code, 6S ​​comes with configured sensor bands. For sensor bands that are not covered, the corresponding spectral response function needs to be converted and added to the program; 6S simulates the transmission process of solar electromagnetic radiation between 0.25μm and 4.0μm, and divides the electromagnetic waves into intervals of 2.5nm to obtain 1501 wavelength values , so when adding a band, you only need to add the starting value of the wavelength of the band to be added. and According to the nearest neighbor interpolation method, Determine the corresponding wavelength value in and ,in and Indicates the position of the interpolated wavelength starting value in 1501 elements, and is obtained by bilinear interpolation arrive Spectral response values ​​between , set the corresponding spectral values ​​of the remaining positions to 0; after given the observation geometry, atmospheric model, aerosol model and sensor code, a lookup table containing atmospheric transmittance and 550nm aerosol optical thickness is constructed.

9. The method for adaptively adjusting the aerosol complex refractive index based on the cross-shaped structure according to claim 1, characterized in that: In step S6, an exponential function is used for fitting, and the fitting formula is as follows: ， ； in represents the atmospheric transmittance, represents the atmospheric aerosol optical thickness at 550nm, and are the fitting coefficients, where is a negative number; during the fitting process, the ratio of the structure function value of the apparent reflectance to the surface reflectance is , and is the coefficient determined by fitting, and the aerosol optical thickness can be obtained.

10. An aerosol complex refractive index adaptive adjustment system based on a 'P' structure, using the aerosol complex refractive index adaptive adjustment method based on a 'P' structure according to any one of claims 1 to 9, characterized in that: It includes a data acquisition unit, a data processing unit and an output unit; The data acquisition unit is used to obtain satellite images of the area to be inverted; The data processing unit processes the data and constructs a cross-shaped structure function model to calculate the apparent reflectivity structure function value of the image to be inverted and the surface reflectivity structure function value; calculates the weight factor for each pixel of the surface reflectivity image, and obtains the comprehensive weight factor after resampling according to the calculation window movement law of the structure function model; according to the comprehensive weight factor, the imaginary part of the aerosol complex refractive index is adaptively adjusted to construct a lookup table of atmospheric aerosol optical thickness and atmospheric transmittance; the ratio of the structure function value of the apparent reflectivity to the surface reflectivity is brought into the fitted formula of atmospheric transmittance and optical thickness to realize the inversion of aerosol optical thickness by the structure function method; Output unit: Output the results visually.

Images