A three-dimensional aerosol tomography method based on a multi-angle imager

Abstract

The present application relates to the technical field of atmospheric remote sensing monitoring, and particularly relates to a three-dimensional aerosol tomography method based on a multi-angle imager, comprising: obtaining geometric parameters and kernel-driven model coefficients, constructing a surface BRDF parameter prior dataset based on space-time filtering and spectral conversion technology, and combining a semi-empirical kernel-driven model to estimate MISR surface reflectivity at each angle; obtaining MISR surface reflectivity estimation error at each angle, analyzing the estimation error to construct a surface reflectivity monthly dynamic error correction model, and optimizing the estimated MISR surface reflectivity at each angle; determining the AOD inversion result at each angle according to the optimized MISR surface reflectivity at each angle, combining a 6S radiation transfer model to construct a lookup table; constructing an aerosol tomography dataset; planning a tomography region and discretizing, establishing an aerosol tomography model, and iteratively solving and extracting aerosol three-dimensional distribution information. The three-dimensional aerosol tomography based on the multi-angle imager is realized, and the aerosol vertical distribution information is effectively extracted.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of atmospheric remote sensing monitoring, and particularly relates to a three-dimensional aerosol tomography method based on a multi-angle imager. BACKGROUND

[0002] With the acceleration of industrialization and urbanization, air pollution problems are prominent, aerosols, as an important part of the atmosphere, have a major impact on climate, environment and human health, and their global transmission with the movement of the atmospheric layer, accumulation, spread and diffusion in the atmosphere present three-dimensional dynamic changes, so accurate acquisition of aerosol three-dimensional distribution information is the key to studying the formation mechanism of air pollution, climate change and the like.

[0003] At present, the monitoring of land aerosols mainly includes three types: ground-based observation, aerial measurement and satellite remote sensing. Ground-based observation obtains high-precision aerosol parameter data at the site by deploying ground observation stations and using professional instruments to measure radiation attenuation, but it is limited to a specific location, the site distribution is uneven, and there are few remote areas, which cannot provide global data. Aerial measurement can flexibly select sites for monitoring by using aircraft or remote-controlled aircraft to carry professional instruments for aerosol vertical distribution detection, and can observe a certain height layer in a local area to provide data for a larger research area, and can obtain key data in areas difficult for humans to access by breaking through geographical limitations, but the operation and maintenance cost of the flight equipment is high, and professional personnel are needed, and the monitoring is limited to a specific time node and a preset height range, making it difficult to continuously monitor and provide global scale data at all times. Satellite remote sensing can continuously collect data on a global multi-scale and large area basis by relying on satellite sensors to receive solar radiation signals at the top of the earth's atmosphere, analyzing the intensity and polarization state difference of radiation signals in different wavelength spectral ranges, and inverting aerosol parameters, and can provide rich data basis for atmospheric aerosol research by normalizing monitoring in any area with observation conditions.

[0004] At present, satellite remote sensing AOD (Aerosol Optical Depth, i.e. aerosol optical thickness) inversion algorithms are gradually mature, and a large number of remote sensing AOD products provide core data support for aerosol horizontal transmission research. However, the exploration of the spatial differentiation characteristics and time evolution law of aerosol in the vertical direction needs to be deepened. In the existing aerosol vertical observation methods, traditional tethered airships and unmanned aerial vehicle detection are only suitable for short-time small-scale research, and the detection height is limited; although aerial flight can program trajectories to achieve local flexible measurement, it is difficult to support long-term sequential large-scale detection; ground-based laser radar can provide global aerosol vertical monitoring by laying monitoring points, but there are many disadvantages in China, such as few laser radar sites, high operation and maintenance cost, difficulty, and "point-to-area" problems; although air-based laser radar has the advantage of global observation, the field of view of the detection range is narrow, and it takes several weeks to observe a small part of the ground, making it difficult to obtain global coverage, and the spatial resolution is limited.

[0005] CALIOP (Cloud-Aerosol Lidar with Orthogonal Polarization) is the first detector to provide information on the vertical distribution of aerosols globally, and it is currently the most mature observation method capable of providing high-resolution global aerosol vertical distribution data. Its profile data, for the first time, revealed the characteristics of the global aerosol vertical distribution and the analytical representation of aerosol radiation effects in the vertical direction, providing a three-dimensional structure of aerosol distribution globally. However, its field of view is narrow, covering only 0.2% of the global area in one week of operation (approximately 16 days), and its horizontal resolution is only 5 km, making it unable to provide information on the vertical structure of aerosols over large areas. Furthermore, to date, very little information on the vertical structure of atmospheric aerosols has been obtained from passive remote sensing measurements. Summary of the Invention

[0006] To address the challenges of acquiring large-scale aerosol vertical distribution data and the inability to effectively utilize existing passive satellite sensor observation data, this invention aims to provide a three-dimensional aerosol tomography method based on a multi-angle imager. The specific technical solution adopted is as follows:

[0007] Geometric parameters and kernel-driven model coefficients were obtained. Based on spatiotemporal filtering and spectral conversion techniques, a prior dataset of surface BRDF parameters was constructed. The surface reflectance of MISR at various angles was estimated by combining a semi-empirical kernel-driven model.

[0008] The estimation error of surface reflectance at various angles of MISR was obtained. The estimation error was analyzed to construct a monthly dynamic error correction model for surface reflectance and optimize the estimated surface reflectance of MISR at various angles.

[0009] Based on the optimized MISR surface reflectance at each angle, a lookup table was constructed in conjunction with the 6S radiative transfer model to determine the AOD inversion results at each angle.

[0010] An aerosol tomography dataset was constructed by combining geometric parameters and AOD inversion results from various angles;

[0011] The aerosol chromatography region is planned and discretized, an aerosol chromatography model is established, and the three-dimensional distribution information of aerosols is extracted by iterative solution.

[0012] Preferably, the geometric parameters include solar zenith angle, satellite zenith angle, solar azimuth angle, and satellite azimuth angle.

[0013] Preferably, a prior dataset of surface BRDF parameters is constructed, specifically as follows:

[0014] MCD43A1 data was acquired and processed using a penalized least squares estimation spatiotemporal filtering algorithm based on discrete cosine transform to obtain a prior dataset of surface BRDF parameters.

[0015] Preferably, the estimation error of surface reflectance at each angle of the MISR is obtained, the estimation error is analyzed to construct a monthly dynamic error correction model for surface reflectance, and the estimated surface reflectance of the MISR at each angle is optimized, including:

[0016] Acquire apparent reflectance data and AERONET AOD data from the MISR sensor, perform atmospheric correction based on the 6S radiative transfer model, back-calculate the atmospheric-corrected reflectance data of MISR, and compare the estimation errors of surface reflectance of MISR at various angles.

[0017] Similarly, atmospheric correction is performed on the MODIS data to obtain the corresponding MODIS atmospheric corrected reflectance;

[0018] The MODIS atmospheric corrected reflectance was compared and analyzed with the estimated MISR surface reflectance at various angles. A monthly dynamic error correction model for surface reflectance was constructed using regression analysis.

[0019] The surface reflectance of MISR at various angles was optimized and estimated using a monthly dynamic error correction model for surface reflectance.

[0020] Preferably, a lookup table is constructed to determine the AOD inversion results from each angle, including:

[0021] The atmospheric radiative transfer process was simulated using the 6S radiative transfer model. Geometric parameters, atmospheric model parameters, aerosol model parameters, sensor spectral characteristic parameters, and surface reflectivity parameters were set. The 6S radiative transfer model was run, and lookup tables were constructed under different atmospheric conditions.

[0022] When performing AOD inversion at various angles, the corresponding parameters are matched from the lookup table, and the interpolation algorithm is used to obtain the AOD inversion results at each angle.

[0023] Preferably, a quality control standard is preset, and outliers and large error data in the AOD inversion results are removed according to the quality control standard.

[0024] Preferably, the chromatography region is planned and discretized, an aerosol chromatography model is established, and the three-dimensional distribution information of aerosols is extracted through iterative calculation, including:

[0025] The area to be chromatography is defined as the chromatography region, which is then divided into multiple voxel grids and discretized into several layers.

[0026] An aerosol chromatography model was established, and the multiplicative algebraic reconstruction method was used to iteratively solve the aerosol chromatography model to obtain the aerosol concentration distribution in each voxel block.

[0027] The three-dimensional distribution information of aerosols in the chromatography region is reconstructed based on the aerosol concentration distribution of each grid.

[0028] Preferably, during the iterative solution process, the aerosol concentration distribution is corrected in each iteration, wherein the ratio of the aerosol concentration obtained through iteration to the aerosol concentration in the aerosol chromatography dataset is corrected.

[0029] The present invention has the following beneficial effects:

[0030] This invention estimates the surface reflectance at various angles using MISR (Multi-angle Imaging Spectroradiometer) based on spatiotemporal filtering and spectral conversion techniques, obtains the estimation error, and constructs a monthly dynamic error correction model for surface reflectance by analyzing the estimation error to optimize the estimated surface reflectance at various angles of MISR, thereby improving the accuracy of surface reflectance data. Then, it constructs a lookup table in conjunction with the 6S radiative transfer model to determine the AOD inversion results at each angle, thus realizing AOD inversion at each angle and ensuring the accuracy of AOD inversion. Finally, it constructs an aerosol tomography dataset for subsequent tomography and discretization analysis to extract aerosol three-dimensional distribution information. The whole system couples multi-angle AOD inversion algorithms and tomography techniques, effectively utilizing observation data from existing passive satellite sensors without relying on a large amount of active remote sensing data or ground observation data, reducing data acquisition costs. It has important theoretical value and practical guiding significance for deepening the study of the vertical distribution structure of aerosol optical properties and revealing the formation mechanism of pollution events.

[0031] Therefore, effectively utilizing existing passive satellite sensor observation data to extract aerosol vertical observation information and deepen the research on the vertical distribution structure of aerosol optical properties is of great scientific significance for a more comprehensive study of the spatiotemporal variations of aerosols and the formation mechanism of pollution events, and for better revealing the climate and environmental effects of aerosols. Attached Figure Description

[0032] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 This is a flowchart illustrating an embodiment of a three-dimensional aerosol chromatography method based on a multi-angle imager provided by the present invention.

[0034] Figure 2 This is a flowchart illustrating the steps of a three-dimensional aerosol chromatography method based on a multi-angle imager, as provided in one embodiment of the present invention. Detailed Implementation

[0035] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a three-dimensional aerosol chromatography method based on a multi-angle imager proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0036] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0037] The following description, in conjunction with the accompanying drawings, details a specific scheme for a three-dimensional aerosol chromatography method based on a multi-angle imager provided by the present invention.

[0038] Please combine Figure 1 and Figure 2 It illustrates an implementation flowchart and step flowchart of a three-dimensional aerosol chromatography method based on a multi-angle imager provided in the first embodiment of the present invention, the method comprising:

[0039] Step S1: Obtain geometric parameters and kernel-driven model coefficients. Based on spatiotemporal filtering and spectral conversion techniques, construct a prior dataset of surface BRDF parameters and estimate the surface reflectance of MISR at various angles using a semi-empirical kernel-driven model.

[0040] Step S2: Obtain the estimation error of surface reflectance at each angle of MISR, analyze the estimation error to construct a monthly dynamic error correction model for surface reflectance, and optimize the estimated surface reflectance of MISR at each angle;

[0041] Step S3: Based on the optimized MISR surface reflectance at each angle, construct a lookup table using the 6S radiative transfer model to determine the AOD inversion results at each angle;

[0042] Step S4: Construct an aerosol tomography dataset by combining geometric parameters and AOD inversion results from various angles;

[0043] Step S5: Plan and discretize the chromatography region, establish an aerosol chromatography model, and iteratively solve to extract the three-dimensional distribution information of aerosols.

[0044] To better illustrate, the Multi-Angle Imager (MISR) can simultaneously capture images of the Earth's surface from multiple angles, acquiring data from four tilt angles and nine different perspectives in the vertical direction. It is equipped with multiple spectral channels for each perspective, covering the visible to near-infrared bands, enabling analysis of surface reflectivity. The extracted three-dimensional aerosol distribution information includes aerosol concentration variations at different altitudes, times, and spaces, revealing the distribution characteristics of aerosols in different layers such as the troposphere and stratosphere. This provides information on the vertical structure of atmospheric aerosols, allowing for a more accurate assessment of their optical properties and radiative effects.

[0045] Furthermore, the geometric parameters include solar zenith angle, satellite zenith angle, solar azimuth angle, and satellite azimuth angle.

[0046] Preferably, the relevant data for geometric parameters comes from NASA’s official MISR and MODIS (Moderate Resolution Imaging Spectrophotometer) data. MISR acquires images of the same area from multiple perspectives, providing comprehensive geometric information; MODIS is known for its high spatial resolution and wide band coverage.

[0047] It can be explained that the geometric parameters are used in conjunction with the prior dataset of BRDF (Bidirectional Reflectance Distribution Function) parameters to estimate the corresponding surface reflectance parameters. Here, the observation point refers to the ground pixel within the observation view of MISR and MODIS; the solar zenith angle is the angle between the sunlight and the vertical direction of the observation point; the satellite zenith angle is the angle between the satellite sensor and the vertical direction of the observation point; the solar azimuth angle is the angle between the projection direction of the sunlight on the horizontal plane and the reference direction (usually north); and the satellite azimuth angle is the angle between the projection direction of the satellite sensor on the horizontal plane and the reference direction.

[0048] As an optional implementation, when constructing the aerosol-level dataset, the geometric parameters also include pixel latitude and longitude data and DEM data. Pixel latitude and longitude are the specific geographical coordinates of each pixel on the Earth's surface, expressed in longitude and latitude. DEM (Digital Elevation Model) data is used to reflect the height information of surface pixels, that is, to reflect the altitude relative to sea level through height values.

[0049] Further, in step S1, a priori dataset of surface BRDF parameters is constructed, specifically as follows:

[0050] MCD43A1 data was acquired and processed using a penalized least squares estimation spatiotemporal filtering algorithm based on discrete cosine transform to obtain a prior dataset of surface BRDF parameters.

[0051] To clarify, the MCD43A1 product refers to the BRDF product generated by MODIS, which contains the reflection characteristics of the Earth's surface to solar radiation. This product provides detailed information about the Earth's surface reflection characteristics through satellite observations, including reflectance, albedo, and related quality assessment parameters in different bands. MCD43A1 data, on the other hand, records the processed and corrected reflectance and albedo values, usually stored in raster format, and contains long-term series information for analyzing surface changes and long-term trends.

[0052] It can be noted that due to the influence of atmospheric conditions, cloud cover, and noise during data processing, the kernel-driven model coefficients in the MCD43A1 product have certain uncertainties and some missing values, making the data variations in different times and spaces complex and difficult to predict, thus increasing the difficulty of data processing. Therefore, a penalized least squares estimation spatiotemporal filtering algorithm based on discrete cosine transform is adopted. This algorithm uses spatiotemporal filtering technology to effectively balance data fidelity by calculating the sum of squared residuals and a penalty term reflecting the roughness of smooth data. Furthermore, a Bisquare weighting function is used to iteratively allocate weights, assigning higher weights to high-quality data points in the MCD43A1 data. This ensures that the algorithm remains robust at outliers and missing values, effectively filtering noise and maintaining the authenticity and accuracy of the data.

[0053] Then, a prior dataset of surface BRDF parameters is constructed. Preferably, in this embodiment, a prior dataset of surface BRDF shape indicator factors with an 8-day interval and a resolution of 500 m is constructed.

[0054] To better illustrate, the semi-empirical kernel-driven model is a nuclear physics model that combines theoretical calculations and experimental data. It improves prediction accuracy by introducing some empirical parameters to correct the results of purely theoretical calculations and by combining sparse reflectivity and thick-layer scattering effects to accurately simulate the light reflection behavior of complex surfaces.

[0055] Specifically, in step S1, the semi-empirical BRDF model driven by the RTLSR (Ross Thick-Li Sparse R) kernel shows that:

[0056]

[0057] in, Represents surface reflectance. Indicates the zenith angle of the sun. Indicates the zenith angle of the satellite. Indicates relative azimuth. Indicates wavelength; , , All represent kernel coefficients; Represents an isotropic scattering nucleus; Represents the volume scattering nucleus; This represents the geometric optics scattering nucleus.

[0058] The specific form of each scattering nucleus is as follows:

[0059]

[0060]

[0061]

[0062] in, This represents an intermediate representation.

[0063] The definition of the parameters involved in the scattering kernel is as follows:

[0064]

[0065]

[0066]

[0067]

[0068]

[0069]

[0070] in, Indicates the area where the observation direction overlaps with the illumination direction; , , All of these represent the corresponding corrected parameters; , Both represent intermediate representations; This represents the distance from the center of the ellipsoid to the Earth's surface. Indicates the horizontal radius of the ellipsoid; This represents the vertical radius of the ellipsoid.

[0071] To clarify, in this embodiment, based on the RTLSR kernel-driven semi-empirical BRDF model, it is assumed that the vegetation region is an ellipsoid. and These are dimensionless quantities used to describe the relative height and shape of vegetation, respectively. The relative height refers to the extent of vegetation extension in the vertical direction, while the shape reflects the geometry of the ellipsoid and its spatial distribution, in order to improve the model's ability to capture and reproduce the reflective properties of vegetation at different observation angles.

[0072] As an optional implementation, in the MODIS BRDF product inversion algorithm of this embodiment, given... , .

[0073] Next, introduce To simplify the concept of surface reflectance, in reality, and The changes are minimal within a month. Substituting this into the calculation formula of the semi-empirical BRDF model further simplifies the process, resulting in the following calculation formula:

[0074]

[0075]

[0076] The corresponding calculation formula obtained by combining the results is as follows:

[0077]

[0078] Further derivation yields the corresponding calculation formula:

[0079]

[0080] Similarly, based on this calculation formula, the relevant parameter representations for MISR and MODIS angles are obtained, and the corresponding calculation formulas are:

[0081]

[0082] Further derivation yields the corresponding calculation formula:

[0083]

[0084] in, , These represent the BRDF at MODIS and MISR angles, respectively; This represents the surface reflectance of MODIS under the MISR observation geometry. This represents the MODIS surface reflectance.

[0085] Understandably, due to the different spectral response ranges and sensitivities of MISR and MODIS sensors, the reflectance characteristics of the same ground feature differ on different sensors. Therefore, in order to eliminate the differences in the spectral response functions of MISR and MODIS sensors, a spectral conversion model needs to be constructed to convert the MODIS surface reflectance into the surface reflectance of the corresponding MISR angle. That is, after obtaining the MODIS surface reflectance at different MISR angles, it is converted into the MISR surface reflectance to invert and obtain MISRAOD, and to estimate the surface reflectance of MISR at each angle. The spectral conversion model is a model that uses spectral conversion technology to convert a signal in one spectral range into a signal in another spectral range to ensure the consistency and comparability of the data.

[0086] Preferably, in this embodiment, using the standard spectral database of ENVI software employed by Chen et al., spectral data of 28 typical characteristics were selected to calculate the surface reflectance of different blue light bands of MODIS and MISR. The corresponding calculation formula is as follows:

[0087]

[0088] in, Indicates surface reflectance; , These represent the upper and lower limits of the integration wavelength, respectively; Indicates wavelength as The band response value of the sensor, Indicates the first One band; Indicates the first Surface reflectance in each band; It represents the difference between the upper and lower limits of the wavelength.

[0089] By comparing the reflectance of different ground features in the MISR and MODIS blue bands, the spectral conversion model is obtained by fitting the above calculation formula. The corresponding calculation formula is as follows:

[0090]

[0091] Based on the spectral conversion model, the surface reflectance of the obtained MODIS at the MISR angle is converted into the surface reflectance required for the inversion of MISRAOD, that is, the surface reflectance of MISR at each angle is obtained.

[0092] Further, step S2 includes:

[0093] Step S21: Obtain the apparent reflectance data and AERONET AOD data of the MISR sensor, perform atmospheric correction based on the 6S radiative transfer model, back-calculate the atmospheric corrected reflectance data of MISR, and compare to obtain the estimation error of the surface reflectance of MISR at various angles.

[0094] Step S22: Similarly, perform atmospheric correction on the MODIS data to obtain the corresponding MODIS atmospheric corrected reflectance;

[0095] It is explained that the 6S radiative transfer model is mainly used to simulate the propagation process of solar radiation in the Earth's atmosphere. Based on physical principles, it can accurately calculate the attenuation and scattering effects experienced by sunlight as it passes through the atmosphere.

[0096] Specifically, based on apparent reflectance data acquired by the MISR sensor and AERONET AOD data (AERONET AOD data, officially released by NASA and provided by the AERONET (Aerosol Robotic Network) global aerosol monitoring network), atmospheric correction was performed using the 6S radiative transfer model to eliminate the influence of the atmosphere on surface reflectance measurements. The atmospheric-corrected reflectance data was then calculated and compared with the estimated MISR surface reflectance at various angles. A significant deviation was found between the estimated data and the atmospheric-corrected reflectance data. Therefore, MODIS data was introduced for atmospheric correction to obtain the corresponding MODIS atmospheric-corrected reflectance. Specifically, the MODIS MCD19A2 AOD data product was used to atmospherically correct the MODIS MOD02 data. MCD19A2 refers to the remote sensing data product provided by NASA, primarily containing AOD data acquired by MODIS sensors on Terra and Aqua satellites. This atmospheric-corrected reflectance was then substituted into the surface reflectance conversion formula between MODIS and MISR to obtain the new MISR surface reflectance based on MODIS atmospheric correction.

[0097] Step S23: Compare and analyze the MODIS atmospheric corrected reflectance with the estimated MISR surface reflectance at various angles, and use regression analysis to construct a monthly dynamic error correction model for surface reflectance.

[0098] Specifically, the estimation error generated by the estimated surface reflectance of the MISR at various angles and the new MISR surface reflectance is regressed and fitted monthly, that is, the monthly dynamic error correction model of surface reflectance is constructed by regressing and fitting the old and new surface reflectance.

[0099] Step S24: Optimize the estimated surface reflectance of MISR at various angles using the monthly dynamic error correction model for surface reflectance; that is, reduce the estimation error of the original surface reflectance and improve the estimation accuracy of surface reflectance.

[0100] As explained, in step S3, the estimated surface reflectance of the MISR at each angle is corrected based on the monthly dynamic error correction model for surface reflectance. The corrected surface reflectance of the MISR at each angle is denoted as... .

[0101] As an optional implementation, the 6S radiative transfer model used in step S3 has been improved. Specifically, in the quantitative calculation of spectral absorption and scattering, an advanced approximation algorithm in the field is introduced. Combined with the successive scattering algorithm, a more accurate and comprehensive atmospheric radiative transfer simulation framework is constructed. By considering the multiple scattering effect, the description accuracy of the radiative transfer process under complex atmospheric conditions is effectively improved, and the fit between the input parameters and the actual atmospheric state is significantly enhanced.

[0102] Further, in step S3, a lookup table is constructed to determine the AOD inversion results for each angle, including:

[0103] Step S31: Simulate the atmospheric radiative transfer process using the 6S radiative transfer model, set the geometric parameters, atmospheric model parameters, aerosol model parameters, sensor spectral characteristic parameters, and surface reflectivity parameters, run the 6S radiative transfer model, and construct lookup tables under different atmospheric conditions.

[0104] Specifically, without considering gas absorption—that is, under ideal conditions where there is no interference from absorption by atmospheric gas components—the relationship between the apparent reflectance received by the top-level atmospheric remote sensing sensor and the corrected surface reflectance at various angles of the MISR is determined to understand and analyze the spatial distribution and variation patterns of surface reflectance characteristics. The corresponding calculation formula is as follows:

[0105]

[0106] in, This represents the apparent reflectance received by the remote sensing sensor at the top of the atmosphere; This represents the corrected surface reflectance of the MISR at various angles; Indicates the zenith angle of the sun. Indicates the zenith angle of the satellite. Indicates relative azimuth. This represents path radiation caused by atmospheric molecular scattering; This represents path radiation caused by aerosol scattering; Indicates single-scatter albedo; Indicates AOD; The cosine of the satellite's zenith angle; The cosine of the solar zenith angle; This represents the total downward radiation when the Earth's surface is normalized to zero. This represents the total transmittance in the upward field of view direction of the satellite sensor; It represents the atmospheric hemispheric albedo.

[0107] It can be noted that the 6S radiative transfer model incorporates seven standard aerosol modes: aerosol-free, continental, marine, urban, user-defined, desert, and biomass combustion aerosols. The appropriate aerosol type is selected based on the specific scenario. In this embodiment, the analysis focuses on Beijing, which is typically generated by multiple sources such as dust storms, industrial emissions, and biomass combustion, resulting in complex particle size distribution and chemical composition. Therefore, the continental aerosol mode is suitable. Then, if the apparent reflectance collected by the top-level atmospheric remote sensing sensor differs from the apparent reflectance calculated using a subsequent lookup table... If they are equal, then the AOD is obtained through inversion, i.e. .

[0108] Specifically, in the actual inversion process, the 6S radiative transfer model is used to simulate the atmospheric radiative transfer process. By setting relevant parameters such as geometric parameters, atmospheric model, aerosol type, and AOD, the scattering and absorption process of solar radiation by the atmosphere is simulated. A lookup table of parameters such as atmospheric transmittance and path radiation under different atmospheric conditions is constructed. Here, the atmospheric model refers to the key parameters in the 6S radiative transfer model used to describe the physical characteristics of the atmosphere. In this embodiment, the model provides seven default atmospheric models: no gas absorption, tropical atmosphere, mid-latitude summer, mid-latitude winter, subarctic summer, subarctic winter, and US standard atmosphere (62 years). The model is selected according to the actual situation. For example, if studying Beijing, mid-latitude winter or summer is selected according to the time.

[0109] During the simulation, different parameter combinations were set in the 6S radiative transfer model to simulate the radiative transfer process, and the radiative values ​​such as the top atmospheric reflectivity corresponding to each parameter combination were obtained to construct a lookup table.

[0110] Step S32: When performing AOD inversion at each angle, match the corresponding parameters from the lookup table and use the interpolation algorithm to obtain the AOD inversion results at each angle.

[0111] It can be explained that when performing AOD inversion at various angles, the corresponding parameters are quickly obtained from a lookup table based on the actual observed geometric angles and atmospheric conditions. In practice, the preprocessed MI1B2T data is substituted as the measured radiation data of MISR at each observation angle, and matched with the radiation values ​​obtained from the lookup table to find the AOD corresponding to the closest radiation value. Then, the AOD of the corresponding parameters is calculated by linear interpolation, and the MISR AOD value is obtained through inversion. The MI1B2T data comes from NASA's official release and has undergone preprocessing operations such as data cleaning, format conversion, outlier removal, and necessary data standardization.

[0112] Furthermore, a quality control standard is preset, and outliers and large error data in the AOD inversion results are removed based on the quality control standard.

[0113] The explanation is that the AOD inversion results are subject to quality control based on quality control standards to remove outliers and large error data. Specifically, data points with abnormally high or low surface reflectance, as well as data points with significant differences from surrounding data, are investigated and removed to ensure the reliability of the inversion results.

[0114] Outliers may be caused by sensor errors, sudden changes in atmospheric conditions, or other external factors; large error data may be caused by special environmental conditions or other factors in a local area, which will have a negative impact on the AOD inversion results; quality control standards refer to the criteria and thresholds used to evaluate and screen data quality during the processing, in order to ensure the accuracy and reliability of the AOD inversion results.

[0115] As explained, in step S4, the multi-angle imager (MISR) can conduct simultaneous observations of the ground from nine angles. If each pixel on the ground is regarded as a "station", then the connection between the pixel "station" and the sensor camera has similar characteristics to the rays in ionospheric tomography. Based on this, the position information of the sensor camera and each surface pixel is recorded, and combined with the value corresponding to each ray, a remote sensing aerosol tomography dataset is constructed.

[0116] Specifically, based on the AOD inversion results obtained in step S3, geometric parameters such as latitude and longitude and DEM data at each pixel are collected. The preprocessed MI1B2T data mentioned above contains latitude and longitude data for each pixel on the ground. The DEM data includes three resolutions: 1km, 500m, and 250m. The WGS84 (World Geodetic System 1984) ellipsoidal projection is used uniformly. In this embodiment, 1km resolution data is used. After projection conversion, the DEM values ​​of the corresponding pixels are extracted based on the latitude and longitude in the MI1B2T data. Then, the location information of the corresponding pixel station is determined according to the latitude and longitude of the pixel and the DEM data. The specific positioning of the ray path is realized through geometric parameters.

[0117] Preferably, based on the 6S atmospheric radiative transfer theory, a geometric mapping relationship between the tilted column AOD and the vertical column AOD is established through the trigonometric function relationship of the satellite zenith angle, generating a multi-view tilted column AOD dataset to obtain the aerosol parameter values ​​corresponding to each "ray"; then, the aerosol parameter values ​​of each pixel are obtained, and the aerosol tomography dataset is integrated to establish an aerosol tomography dataset.

[0118] Further, step S5 includes:

[0119] Step S51: Define the area to be chromatography as the chromatography region, divide the chromatography region into multiple voxel grids, and discretize them into several layers.

[0120] Specifically, the area to be analyzed for the task is determined based on the actual situation, i.e., the chromatography region is defined. For example, if it is necessary to determine the aerosol concentration of a certain city or a certain district within a city, then the city and that district within the city are the chromatography region. Then, the area to be inverted, i.e., the chromatography region, is divided into multiple voxel grids according to latitude, longitude, and altitude, denoted as […]. Next, based on the prior law of exponential decrease of aerosol concentration in the vertical direction and the horizontal resolution of the MISR sensor, the tomographic region was divided into a grid of 1.1 km × 1.1 km in the horizontal direction and discretized in the vertical direction according to the elevation [0 km, 0.5 km, 1 km, 1.5 km, 2 km, 3 km, 5 km, 15 km], resulting in a total of 7 layers.

[0121] Step S52: Establish an aerosol chromatography model and use the multiplicative algebraic reconstruction method to iteratively solve the aerosol chromatography model to obtain the aerosol concentration distribution in each voxel block.

[0122] Specifically, the aerosol concentration in each grid cell, i.e., each voxel block grid, is defined as an unknown quantity, denoted as . The aerosol concentration data for each voxel block is calculated separately, and the corresponding calculation formula is as follows:

[0123]

[0124] in, This indicates the solution obtained for the first... aerosol column concentration per pixel; Indicates the total number of voxel blocks; This represents the total number of pixels across all angles. Indicates the first The aerosol column of one pixel in the first Intercept within an individual pixel; Indicates the first Unknown quantities in the individual pixel grid; This includes the corresponding series expansion error and measurement noise.

[0125] To further simplify the calculation formula, it is converted into matrix form:

[0126]

[0127] in, This represents the column vector consisting of the aerosol column concentration values ​​at each pixel under each observation angle; The matrix representing the intercepts of the aerosol concentration column within the corresponding voxel; This represents the column vector representing the aerosol concentration of each voxel block; This represents a column vector consisting of discrete error and observation noise.

[0128] Understandably, constructing an aerosol tomography model essentially involves solving for the intercept of each ray within a voxel block. When the ray trajectory passes through a discrete grid cell, i.e. a voxel block, the equations of the ray's straight line and the corresponding elevation, longitude, and latitude surfaces can be solved based on spatial geometric relationships within the WGS84 (World Geodetic System 1984) geodetic coordinate system framework. This allows us to solve for the three-dimensional coordinates of the intersection points of the ray with each voxel block and obtain the intercept of each ray within the discretized spatial grid.

[0129] Next, the constructed aerosol chromatography model, i.e., the chromatography equation, is solved iteratively. Specifically, the multiplicative algebraic reconstruction method is used to iteratively solve the established chromatography equation. First, initial values ​​are assigned to each voxel block in the chromatography region. The initial values ​​are assigned different values ​​according to the prior law of exponential decrease of aerosol concentration in the vertical direction. The initial estimate of the voxel blocks in the region to be reconstructed, i.e. the chromatography region, is improved step by step through iteration.

[0130] Furthermore, in step S52, during the iterative solution process, the aerosol concentration distribution is corrected in each iteration, wherein the ratio of the aerosol concentration obtained through iteration to the aerosol concentration in the aerosol chromatography dataset is corrected.

[0131] Specifically, in actual operation, each iteration corresponds to one slant distance aerosol inclined column measurement, that is, it is performed for one equation, and the correction basis for each round of iteration is the first... The ratio of the slant aerosol column concentration calculated by the aerosol density in the next iteration to the aerosol column concentration obtained from actual observation, where the aerosol column concentration obtained from actual observation refers to the data that can be directly obtained from the aerosol chromatography dataset in step S4. The aerosol density distribution image is corrected accordingly so that the iteration result converges to the solution of the aerosol concentration data in matrix form.

[0132] In the In this iteration, the correction formula is:

[0133]

[0134] in, Indicates the first Individual block number Step iteration value; Represents the first in matrix A Row vectors; Represents the relaxation factor for step-by-step iteration. .

[0135] It can be noted that MART converges quickly and all its solutions are positive, which meets the physical constraint that the aerosol optical thickness value in aerosol chromatography reconstruction must be positive; thus, the aerosol concentration distribution of each voxel block can be calculated.

[0136] Step S53: Extract the three-dimensional distribution information of aerosols based on the aerosol concentration distribution.

[0137] It is explained that the aerosol concentration distribution of each voxel block can be used to obtain the three-dimensional distribution information of aerosols in the corresponding chromatography region, and then integrated to obtain the long-term three-dimensional distribution information of aerosols in the entire region, thus forming an intuitive multi-temporal three-dimensional image that shows the flow trajectory and focusing area of ​​aerosols in the air.

[0138] Understandably, this invention estimates the surface reflectance of MISR at various angles using spatiotemporal filtering and spectral conversion techniques, obtains the estimation error, and constructs a monthly dynamic error correction model for surface reflectance by analyzing the estimation error to optimize the estimated surface reflectance of MISR at various angles, thereby improving the accuracy of surface reflectance data. Then, it constructs a lookup table in conjunction with the 6S radiative transfer model to determine the AOD inversion results at each angle, thus achieving AOD inversion at each angle and ensuring the accuracy of AOD inversion. Finally, it constructs an aerosol tomography dataset for subsequent tomography and discretization analysis to extract aerosol three-dimensional distribution information. The entire invention couples multi-angle AOD inversion algorithms with tomography techniques, effectively utilizing existing passive satellite sensor observation data without relying on large amounts of active remote sensing data or ground observation data, reducing data acquisition costs. This has significant theoretical value and practical guiding significance for deepening the study of the vertical distribution structure of aerosol optical properties and revealing the formation mechanism of pollution events.

[0139] Therefore, effectively utilizing existing passive satellite sensor observation data to extract aerosol vertical observation information and deepen the research on the vertical distribution structure of aerosol optical properties is of great scientific significance for a more comprehensive study of the spatiotemporal variations of aerosols and the formation mechanism of pollution events, and for better revealing the climate and environmental effects of aerosols.

[0140] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0141] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A three-dimensional aerosol chromatography method based on a multi-angle imager, characterized in that, The method includes: Geometric parameters and kernel-driven model coefficients were obtained. Based on spatiotemporal filtering and spectral conversion techniques, a prior dataset of surface BRDF parameters was constructed. The surface reflectance of MISR at various angles was estimated by combining a semi-empirical kernel-driven model. The estimation errors of surface reflectance at various angles of the MISR were obtained. These errors were analyzed to construct a monthly dynamic error correction model for surface reflectance, optimizing the estimated surface reflectance of the MISR at various angles, including: Acquire apparent reflectance data and AERONET AOD data from the MISR sensor, perform atmospheric correction based on the 6S radiative transfer model, back-calculate the atmospheric-corrected reflectance data of MISR, and compare the estimation errors of surface reflectance of MISR at various angles. Based on the method of obtaining MISR atmospheric corrected reflectance data, atmospheric correction is similarly performed on MODIS data to obtain the corresponding MODIS atmospheric corrected reflectance. The MODIS atmospheric corrected reflectance was compared and analyzed with the estimated MISR surface reflectance at various angles. A monthly dynamic error correction model for surface reflectance was constructed using regression analysis. Surface reflectance of MISR at various angles was optimized and estimated using a monthly dynamic error correction model for surface reflectance. Based on the optimized MISR surface reflectance at each angle, a lookup table was constructed in conjunction with the 6S radiative transfer model to determine the AOD inversion results at each angle. An aerosol tomography dataset was constructed by combining geometric parameters and AOD inversion results from various angles; The aerosol chromatography region is planned and discretized, an aerosol chromatography model is established, and the three-dimensional distribution information of aerosols is extracted by iterative solution.

2. The three-dimensional aerosol chromatography method based on a multi-angle imager according to claim 1, characterized in that, The geometric parameters include solar zenith angle, satellite zenith angle, solar azimuth angle, and satellite azimuth angle.

3. The three-dimensional aerosol chromatography method based on a multi-angle imager according to claim 2, characterized in that, Constructing a prior dataset of BRDF parameters for the land surface, specifically as follows: MCD43A1 data was acquired and processed using a penalized least squares estimation spatiotemporal filtering algorithm based on discrete cosine transform to obtain a prior dataset of surface BRDF parameters.

4. The three-dimensional aerosol chromatography method based on a multi-angle imager according to claim 1, characterized in that, Construct a lookup table to determine the AOD inversion results from each angle, including: The atmospheric radiative transfer process was simulated using the 6S radiative transfer model. Geometric parameters, atmospheric model parameters, aerosol model parameters, sensor spectral characteristic parameters, and surface reflectivity parameters were set. The 6S radiative transfer model was run, and lookup tables were constructed under different atmospheric conditions. When performing AOD inversion at various angles, the corresponding parameters are matched from the lookup table, and the interpolation algorithm is used to obtain the AOD inversion results at each angle.

5. The three-dimensional aerosol chromatography method based on a multi-angle imager according to claim 4, characterized in that, Preset quality control standards, and remove outliers and large error data from the AOD inversion results based on these standards.

6. The three-dimensional aerosol chromatography method based on a multi-angle imager according to claim 1, characterized in that, The tomographic region is planned and discretized, an aerosol tomographic model is established, and the three-dimensional distribution information of aerosols is extracted through iterative solution, including: The area to be chromatography is defined as the chromatography region, which is then divided into multiple voxel grids and discretized into several layers. An aerosol chromatography model was established, and the multiplicative algebraic reconstruction method was used to iteratively solve the aerosol chromatography model to obtain the aerosol concentration distribution in each voxel block. The three-dimensional distribution information of aerosols in the chromatography region is reconstructed based on the aerosol concentration distribution of each grid.

7. The three-dimensional aerosol chromatography method based on a multi-angle imager according to claim 6, characterized in that, During the iterative solution process, the aerosol concentration distribution is corrected in each iteration, specifically by correcting the ratio of the aerosol concentration obtained through iteration to the aerosol concentration in the aerosol chromatography dataset.

Images