A remote sensing satellite radiometric calibration method and system

Abstract

This application provides a method and system for radiometric calibration of remote sensing satellites, belonging to the field of remote sensing measurement and on-orbit satellite calibration technology. The method includes: selecting effective calibration points based on the footprint range of the observation platform on the ground; extracting the observed radiance at all effective calibration point locations on the platform; constructing atmospheric bottom reflectance under the observation conditions of the platform, combined with corresponding atmospheric parameters and geometric conditions; performing surface correction on the surface reflectance of the platform and the satellite under both the platform observation geometry and the satellite transit observation geometry; calculating the angle-corrected surface reflectance of the platform; performing positive radiometric reconstruction to obtain the satellite entrance pupil radiance; and calculating the calibration coefficient by combining the average DN value of all effective calibration points in the satellite image. This method can achieve accurate radiometric calibration of remote sensing satellites under conditions of asynchronous observation between the platform and the satellite platform.

Description

Technical Field

[0001] This application relates to the fields of remote sensing measurement and satellite on-orbit calibration technology, and in particular to a method and system for radiometric calibration of remote sensing satellites. Background Technology

[0002] A near-space platform is a flight-borne vehicle that operates in near-space (typically between 20 km and 100 km above sea level), capable of carrying scientific instruments, sensors, or communication equipment and performing specific tasks. In radiometric calibration missions, near-space platforms, equipped with high-precision radiometers, conduct observations in the stratosphere, providing satellites with reference data closer to the upper atmosphere than ground-based observations. Typical near-space platforms include high-altitude scientific balloons and stratospheric airships.

[0003] Existing radiometric calibration methods for satellite payloads involving near-space platforms typically compare the radiometric measurement (TOA) equipment onboard the near-space platform with the satellite payload's observations to retrieve or update the satellite payload's radiometric calibration coefficients. For example, the research scheme in the existing literature: "Radiometric Calibration Experiment and Preliminary Results of Near-Space Altitude Satellite Optical Payloads. Wang Ning et al. Journal of Remote Sensing. 2023." (refer to...) Figure 1 The overall process of radiometric calibration involving the airborne platform includes the following steps: acquiring high-altitude spherical on-Earth radiance observation data, performing radiometric correction and geometric footprint calculation, performing time and spatial matching based on satellite observation geometric information, then performing atmospheric compensation and channel response convolution in sequence to obtain simulated channel apparent radiance, and finally comparing the calibration with satellite observation data to obtain the calibration result.

[0004] However, the above method requires synchronizing the observation conditions of the airborne platform and the satellite platform to the same conditions through time matching and spatial matching. In engineering practice, this has the following shortcomings, which limit the final calibration accuracy and make the error easily amplified in complex scenarios: (1) The BRDF error caused by the inconsistency of observation geometry is difficult to suppress.

[0005] The observation time, observation azimuth and observation zenith angle of the airborne platform and the satellite payload are difficult to be strictly consistent. Especially when the ground surface has significant anisotropic reflection characteristics or obvious topographic relief, if there is no geometric equivalent transformation or BRDF normalization processing, a systematic deviation will occur between the platform observation and the satellite observation, which will then be transmitted as calibration coefficient deviation.

[0006] (2) The mismatch between the surface non-uniformity and the footprint scale leads to the amplification of error.

[0007] Surface inhomogeneity not only directly affects radiation uniformity but also further amplifies uncertainties such as footprint positioning errors and angular differences. Existing methods often use fixed-size windows (e.g., fixed pixel blocks) to calculate uniformity indices. However, when this scale is inconsistent with the actual coverage of the instantaneous field of view (footprint) of the airborne platform, the uniformity evaluation cannot truly reflect the consistency of the effective observed footprints, leading to the inclusion of non-uniform surface samples and causing error amplification.

[0008] (3) The treatment of time differences and atmospheric path differences is inconsistent.

[0009] There are generally inconsistencies in observation time and observation altitude between airborne platforms and satellite platforms. If only empirical corrections are used or different processing methods are applied to different regions, it is easy to cause inconsistencies in the processing of time difference and atmospheric difference terms under different scenarios. It is difficult to form a unified calibration result that can be experimentally verified and reproduced, which ultimately affects the calibration accuracy and stability.

[0010] Therefore, there is a need to provide an improved technical solution that addresses the shortcomings of the existing technology. Summary of the Invention

[0011] The purpose of this application is to provide a method and system for radiometric calibration of remote sensing satellites. The core of this method is a closed-loop process of "footprint-scale uniformity screening - geometric equivalent transformation - forward reconstruction from BOA to TOA". This method enables the reconstruction of on-sensor radiometric values ​​that can be compared and regressed under the conditions of satellite transit geometry and atmospheric conditions. This results in stable and reliable radiometric calibration coefficients, enabling radiometric calibration of remote sensing satellites under asynchronous observation conditions between the airborne platform and the satellite platform, and ensuring the stability and accuracy of the calibration.

[0012] To achieve the above objectives, this application provides the following technical solution: This application provides a method for radiometric calibration of remote sensing satellites, including: Based on the footprints on the ground observed by the airborne platform, effective calibration points were selected. The observed radiance at all valid calibration points on the airborne platform is extracted. Under the observation conditions of the airborne platform, and combined with the corresponding atmospheric parameters and geometric conditions, the surface reflectance is obtained by using the radiometric inversion method. ; Using the BRDF parameters provided by satellite products, and based on the BRDF model, the surface reflectance under the observation geometry of the near-air platform was simulated. Surface reflectance under satellite overpass observation geometry and calculate and The ratio of to is used as the surface BRDF correction factor. ; Correction factor for surface BRDF Surface reflectivity of the air-to-ground platform Multiply to obtain the surface reflectance after angle correction of the air-to-ground platform. ; Regarding the surface reflectance Perform a forward radiation simulation to obtain the satellite's entrance pupil radiation. The calibration coefficient is calculated by combining the average DN value of all the effective calibration points in the satellite imagery.

[0013] Preferably, effective calibration points are selected based on the footprint range on the ground observed by the aerial platform, including: Based on the position, attitude, and sensor field of view parameters of the aerial platform, the range of the footprints on the ground observed by the aerial platform in this study is calculated. Construct a dynamic region of interest (ROI) on the satellite imagery corresponding to the footprint area; Calculate the uniformity index within the dynamic ROI. When the uniformity index meets the preset threshold condition, the dynamic ROI is determined to be a valid calibration point.

[0014] Preferably, constructing a dynamic region of interest (ROI) corresponding to the footprint range on the satellite imagery includes: Based on the flight altitude H and sensor field of view θ of the airborne platform at each moment, calculate the field of view diameter R of the sensor projected on the ground on the airborne platform; Based on the field of view diameter R, the side length L of the inscribed square of the footprint is calculated and converted with the pixel resolution of the satellite image to obtain the size of the extraction window; The region of pixels covered by the extraction window on the satellite image is taken as the dynamic region of interest (ROI).

[0015] Preferably, the surface reflectance is obtained by inverting using a radiometric inversion method. The specific formula is as follows: Calculate using the following formula : , In the formula, This represents the observed radiance of the airborne platform after spectral normalization. , , This is the atmospheric correction factor.

[0016] This embodiment provides a remote sensing satellite radiometric calibration system, which performs the steps of the method provided in any of the above embodiments, including: The effective calibration point screening unit is configured to screen out effective calibration points based on the footprint range on the ground observed by the airborne platform during this observation. The downlink inversion unit is configured to extract the observed radiance of all valid calibration points of the airborne platform and use the radiometric inversion method to obtain the surface reflectance. ; The surface correction factor calculation unit is configured to use BRDF parameters provided by satellite products and, based on the BRDF model, simulate surface reflectance under the observation geometry of the airborne platform. Surface reflectance under satellite overpass observation geometry and calculate and The ratio of to is used as the surface BRDF correction factor. ; The observation geometric correction unit is configured to adjust the surface BRDF correction factor. Surface reflectivity of the air-to-ground platform Multiply to obtain the surface reflectance after angle correction of the air-to-ground platform. ; Forward simulation unit, configured to simulate the surface reflectance Perform a forward radiation simulation to obtain the satellite's entrance pupil radiation. The calibration coefficient is calculated by combining the average DN value of all the effective calibration points in the satellite imagery.

[0017] This application also provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the remote sensing satellite radiometric calibration method provided in any of the above embodiments.

[0018] This application also provides a computer-readable storage medium storing a computer program / instructions thereon, which, when executed by a processor, implements the steps of the remote sensing satellite radiometric calibration method provided in any of the above embodiments.

[0019] This application also provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the remote sensing satellite radiometric calibration method provided in any of the above embodiments.

[0020] Beneficial effects: This application addresses the specific constraint of "non-synchronous ambient platform calibration" (i.e., the ambient platform and the satellite are not observed synchronously), and proposes and implements a closed-loop combination: platform radiance - platform time BOA (downlink correction) - satellite geometry BOA - satellite time TOA (uplink forward modeling), thereby transforming the incomparable radiance problem into a transferable common reference quantity, and ultimately achieving high-precision cross-calibration under asynchronous conditions. Attached Figure Description

[0021] Figure 1This is a flowchart illustrating the existing satellite payload radiometric calibration method involving an airborne platform.

[0022] Figure 2 A flowchart of the remote sensing satellite radiometric calibration method provided in this application.

[0023] Figure 3 This is a schematic diagram of dynamic ROI extraction, where (a) is the flight altitude of the airborne platform at different times, and (b) is a schematic diagram of the radiometer field of view (FOV) and the inscribed square.

[0024] Figure 4 This is a schematic diagram showing the distribution of effective calibration points in satellite imagery. Detailed Implementation

[0025] The embodiments of this application will now be described with reference to the accompanying drawings.

[0026] This embodiment provides a method for radiometric calibration of remote sensing satellites, including steps 1 to 5: Step 1: Based on the footprint range on the ground observed by the airborne platform, select effective calibration points.

[0027] The airborne platform is equipped with a hyperspectral radiometer (sensor) to acquire hyperspectral radiance data of the test area and simultaneously record the platform's latitude, longitude, altitude, attitude, and observation time. Data acquisition and preprocessing steps are used to obtain radiation measurement data from the airborne platform and the corresponding platform positioning / attitude information; acquire satellite payload transit observation data and its corresponding solar / observation geometry parameters; simultaneously acquire atmospheric and auxiliary data (such as reanalysis data, satellite atmospheric products, etc.) during the satellite transit period; and complete necessary preprocessing of the radiation data.

[0028] This observation refers to the entire process of Earth observation and data acquisition completed by the airborne platform. The footprint range on the ground is the ground area corresponding to the field of view of the hyperspectral radiometer carried on the airborne platform during this observation. This range is a continuous geographic spatial area and represents the surface area represented by the data from this observation.

[0029] An effective calibration point is a ground target point located within the ground footprint range of this observation and meeting a series of preset quality and applicability conditions. Its effectiveness depends on conditions such as spectral characteristics (e.g., high reflectivity, flat spectrum, good isotropy, etc.) and spatial uniformity (uniform surface material, small reflectivity variation within the footprint range).

[0030] In this embodiment, the selection of effective calibration points is based on the primary condition that the ground footprints observed are within the range of this observation, combined with other validity conditions, to select the set of points most suitable for this calibration.

[0031] The selection of valid calibration points includes the following sub-steps: Based on the position, attitude, and sensor field-of-view parameters of the aerial platform, calculate the footprint range of the aerial platform on the ground during this observation; construct a dynamic region of interest (ROI) corresponding to the footprint range on the satellite image; calculate the uniformity index within the dynamic ROI, and determine the dynamic ROI as a valid calibration point when the uniformity index meets a preset threshold condition. These sub-steps are explained in detail below.

[0032] I. Footprint mapping and dynamic ROI construction.

[0033] The position, attitude, and sensor field of view parameters of the airborne platform are obtained synchronously during the platform's flight. These parameters are used as input parameters, and the mapping relationship from the pixels of the sensors on the platform to the ground points is calculated using collinearity equations or geometric projection models. This allows the calculation of the footprint range of the airborne platform on the ground during this observation.

[0034] Specifically, the real-time flight altitude data recorded by the airborne platform during the observation process is first obtained. Then, combined with the field-of-view parameters of the radiometer, the size of the ground projection field of view corresponding to the radiometer at different times is calculated according to the geometric relationship. The ground projection field of view at all times is combined to obtain the footprint range of this observation.

[0035] Simultaneously, image data of the surface targets in the test area passed by the satellite to be calibrated were acquired, and areas with no clouds, no snow, and uniform surface were selected as calibration areas.

[0036] Considering that different bands of satellite sensors typically have different spatial resolutions, to ensure a consistent spatial correspondence among the bands, this embodiment selects the band with the higher spatial resolution from the satellite sensor as the reference resolution, and uses bilinear interpolation to resample the bands with lower resolution, unifying them to the reference resolution. Bilinear interpolation can better maintain the radiation continuity within a uniform area of ​​the Earth's surface, while not changing the average reflectance characteristics of the region, thereby avoiding systematic radiation biases introduced by resampling and providing a consistent data foundation for subsequent cross-band joint processing.

[0037] 2. Based on the calculated range of ground footprints, define dynamic ROIs on satellite imagery.

[0038] Since the platform's flight altitude ranges from approximately 19km to 32km, its ground projection field of view diameter varies significantly. Therefore, it is necessary to dynamically adjust the corresponding satellite image extraction area. This involves constructing a dynamic Region of Interest (ROI) on the satellite image corresponding to the footprint range. "Dynamic" means that the ROI changes with the platform's real-time flight altitude H and the sensor's field of view θ. The specific dynamic construction method is as follows: Based on the platform's flight altitude H and sensor's field of view θ at various times, calculate the field of view diameter R of the sensor's ground projection on the platform; based on the field of view diameter R, calculate the side length L of the inscribed square of the footprint, and convert it to the pixel resolution of the satellite image to obtain the size of the extraction window; the area of ​​pixels covered by the extraction window on the satellite image is taken as the dynamic ROI.

[0039] Specifically, after unifying the spatial resolution of each band of the satellite sensor to the reference resolution, the radiometer ground projection field of view diameter R is calculated based on the real-time flight altitude H of the airborne platform at each moment, and the size of the satellite image extraction window is dynamically determined by using a circular footprint inscribed in a square.

[0040] The following reference Figure 3 Please provide an explanation. Figure 3 In the diagram, (a) represents the flight altitude of the airborne platform at different times, and (b) is a schematic diagram of the field of view (FOV) of the radiometer on the airborne platform and the inscribed square. The footprint projected onto the ground by the radiometer's field of view on the airborne platform is considered as a circle. The diameter R of this circle is calculated, and then the side length L of the inscribed square is calculated. The relevant expressions are as follows: (1) In the formula, R is the field of view diameter of the sensor projected onto the ground on the airborne platform, H is the flight altitude of the airborne platform, θ is the field of view angle of the sensor on the airborne platform, and L is the side length of the inscribed square of the footprint.

[0041] Subsequently, the side length L of the inscribed square is converted to the pixel resolution of the satellite image to obtain the size of the extraction window (i.e., the number of rows and columns); and based on this, the pixel region corresponding to the radiometer observation footprint is extracted from the satellite image, that is, the region composed of pixels covered by the extraction window on the satellite image is taken as the dynamic region of interest (ROI).

[0042] As the flight altitude H of the airborne platform changes at different times, the values ​​of R, L, and the size of the extraction window also change synchronously. This enables dynamic matching and extraction of footprints observed under different altitude conditions, ensuring that the extracted pixel area falls completely within the effective observation range of the radiometer and avoiding deviations caused by mixed pixels at the edge of the field of view. In other words, the dynamic ROI determines the window size by incorporating a square within the circular area of ​​the footprint, and can adaptively select the pixel window (e.g., 3×3 or 4×4) as the platform altitude changes, ensuring that the ROI scale is consistent with the instantaneous footprint coverage.

[0043] III. Footprint Scale Uniformity Test and Sample Screening. Calculate the uniformity index of radiance / reflectance (such as the coefficient of variation, CV) within the dynamic ROI. When CV meets a preset threshold condition (e.g., CV is less than or equal to a certain threshold), the observed sample is determined to be a valid sample and is called the initial calibration point.

[0044] Due to the complex surface environment, to ensure calibration accuracy, the extracted dynamic regions of interest (ROIs) were used as initial calibration points, and rigorous quality screening was performed to remove invalid points affected by cloud cover or surface unevenness. The screening process was as follows: 1) Cloud and snow pixel removal: Using standard cloud mask products of the satellite to be calibrated or by using the band threshold method, identify and remove sample points (calibration points) containing clouds, cloud shadows or snow cover within the observation field of view (i.e., within the aforementioned ROI) to ensure that the radiation signal mainly comes from surface reflection.

[0045] 2) Surface Uniformity Verification (Surface Uniformity Control): To reduce the impact of geometric registration errors, the selected calibration points must have highly uniform surface uniformity. Therefore, a uniformity index is calculated within the dynamic ROI. When the uniformity index meets a preset threshold condition, the dynamic ROI is determined to be a valid calibration point. In this embodiment, the coefficient of variation is used as the uniformity index. The coefficient of variation of satellite image counts within the extraction window (i.e., within the ROI range) is calculated using the following formula: (2) In the formula, Represents the coefficient of variation. To extract the standard deviation of the pixel values ​​within the window, To extract the mean value of the pixels within the window.

[0046] For example, the threshold condition can be set as follows: That is, only when the coefficient of variation within the ROI is less than or equal to the threshold. Only when the observation point is considered a valid experimental area, that is, the ROI is considered a valid calibration point.

[0047] The changes in the number of calibration points for each screening step in the above screening process are shown in Table 1. Table 1 is as follows: Table 1. Changes in the number of calibration points

[0048] After quality screening, 45 valid calibration points were retained, referred to as the final calibration points, and their distribution is as follows: Figure 4 As shown.

[0049] Step 2: Extract the observed radiance at all valid calibration points on the airborne platform. Under the observation conditions of the airborne platform, and in conjunction with the corresponding atmospheric parameters and geometric conditions, use the radiometric inversion method to obtain the surface reflectance. .

[0050] This step aims to construct the BOA reflectance: under the observation conditions of the airborne platform, combined with the corresponding atmospheric parameters and geometric conditions, the airborne platform observations are processed downlink to obtain the reflectance / radiation characterization of the bottom surface (BOA), so that subsequent geometric equivalence and forward reconstruction are based on a unified surface radiation state.

[0051] Since the radiation measurements obtained from the airborne platform still include the influence of the downlink atmospheric path and platform observation conditions, they need to be converted into surface radiation representations with clear physical meaning. Therefore, the purpose of this step is to transfer the airborne platform radiation benchmark to the 45 retained calibration points.

[0052] Specifically, based on the observation geometry and atmospheric state parameters corresponding to the observation time of the airborne platform, the radiative transfer relationship under the observation conditions is constructed, and the surface reflectance is obtained from the observed radiance of the platform using the radiative inversion method.

[0053] Among them, the surface reflectance is obtained by inverting using the radiation inversion method. The specific calculation formula is as follows: (3) In the formula, This represents the radiance observed at the airborne platform after spectral normalization. , , This is the atmospheric correction factor.

[0054] Different optical remote sensing sensors differ in band settings, center wavelength, bandwidth, and spectral response functions, making it impossible to directly compare spectral signals acquired by different sensors even when observing the same surface target. To enable quantitative applications of multi-sensor data, it is necessary to unify the spectra of all sensors to the same sensor standard through interpolation and resampling. Therefore, this application uses radiance observed from a near-airborne platform after spectral normalization processing. To calculate surface reflectance, The acquisition steps are as follows: First, collect the wavelength and radiance data of the hyperspectral radiometer, the spectral response function (SRF) data of the corresponding satellite band, and the wavelength range of the band. Then, interpolate both to the same wavelength to achieve wavelength alignment. Next, use numerical integration to calculate the integral of the product of SRF and radiance within the band, and the integral of SRF itself. Finally, divide the two integral results to obtain the radiance observed by the hyperspectral radiometer after spectral normalization. ,Should Equivalent radiance that can be directly used for satellite radiometric calibration and authenticity verification.

[0055] Atmospheric correction factor , , The acquisition steps are as follows: collect synchronous atmospheric data and geometric parameters from the airborne platform, and use the 6S radiative transfer model to calculate the radiative transfer at each observation point to obtain... , , These three atmospheric correction parameters.

[0056] Through the inversion process, the radiation observation data of the airborne platform are uniformly converted into intermediate physical quantities that characterize the surface radiation properties.

[0057] Step 3: Using the BRDF parameters provided by the satellite products, and based on the BRDF model, simulate the surface reflectance under the observation geometry of the airborne platform. Surface reflectance under satellite overpass observation geometry and calculate and The ratio of to is used as the surface BRDF correction factor. .

[0058] Step 4, adjust the surface BRDF correction factor. Surface reflectivity of the air-to-ground platform Multiply to obtain the surface reflectance after angle correction of the air-to-ground platform. .

[0059] Steps 3-4 aim to perform surface reflectance conversion based on observational geometry equivalence. This involves introducing a BRDF model or kernel-driven parameters to calculate surface directional reflectance characteristics under both near-air platform observation geometry and satellite overflight observation geometry. Geometric normalization coefficients are then constructed to convert the BOA reflectance under near-air platform conditions. Equivalent BOA reflectance converted to satellite overpass geometry .

[0060] Because airborne platforms typically observe the Earth's surface near vertically, while satellite imaging has specific zenith and azimuth angles, this difference in observation geometry can lead to directional differences in the reflectivity of the same surface feature for non-Lambertian surfaces. Therefore, surface BRDF correction is necessary.

[0061] In this embodiment, under asynchronous observation conditions, the transit times of the two platforms, the near-station platform and the satellite, cannot be unified, resulting in significant differences in both solar geometry and observation geometry. Therefore, the BRDF (Radiative Radiation Flow Function) of the Earth's surface must be corrected to reduce the impact of directional radiation from ground features. Specifically, a linear kernel-driven BRDF model can be used, expressed by the following formula: (4) In the formula, To observe the zenith angle, The zenith angle of the sun. The relative azimuth angle. For surface reflectance, , , These are the kernel coefficients obtained through fitting. It is a geometric optics kernel used to describe the geometric effects of terrain and shadow. It serves as the volume scattering kernel, used to describe the volume scattering characteristics of vegetation.

[0062] Using the BRDF parameters provided by the satellite products, the observation geometry of the satellite payload and the radiometer is simulated using the above formula (4) under the satellite payload geometry conditions. and the geometric conditions of the radiance meter The surface reflectance is then calculated. and The ratio of to is used as the surface BRDF correction factor. ;Right now: (5) Using surface BRDF correction factor and the surface reflectivity of the air-to-ground platform The surface reflectance observed vertically by the radiance meter is converted into a satellite payload. The formula for calculating surface reflectance under geometric conditions is as follows: (6) Step 5, assess surface reflectance Perform a forward radiation simulation to obtain the satellite's entrance pupil radiation. The radiometric calibration coefficient of the satellite payload is calculated by combining the average DN value of all valid calibration points in the satellite imagery.

[0063] Step 5 aims to solve for satellite sensor calibration parameters based on entrance pupil radiance, addressing the inconsistency in handling altitude / time differences by implementing BOA→TOA forward reconstruction and residual atmospheric path difference compensation. Specifically, the equivalent BOA reflectance under the satellite transit geometry obtained in the preceding steps is... As input, a forward radiative transfer model is used to calculate the satellite's TOA radiance at the sensor under atmospheric transit conditions. Then, the satellite's entrance pupil radiance (i.e., the radiance obtained after calibration and simulation calculation) is used. Using the simulated true value as the radiation reference, and combining it with the DN value extracted from the satellite sensor under strict spatiotemporal matching conditions, a linear response model between the sensor's radiation input and quantized output is constructed. By calculating the ratio of the simulated true value to the satellite's observed count value, the absolute radiation calibration coefficient of the satellite is calculated, thus completing the transfer of radiation values ​​from a high-precision near-space reference to the satellite sensor, achieving high-precision calibration of the satellite sensor. Detailed steps are as follows: First, upward atmospheric correction and calibration coefficient derivation are performed. This step aims to derive the angle-corrected surface reflectance obtained in the previous step. The input background atmospheric values ​​are switched to the satellite's synchronous atmospheric products to obtain the TOA radiance of the satellite payload observation geometry. Then, the average DN value within the central area of ​​all calibration points is obtained from the satellite imagery of that day, and recorded as... Finally, the radiometric calibration coefficients of the satellite payload are determined according to the following formula. : (7) This completes the radiometric calibration of the satellite payload. As an example, the method provided in this application can be followed... Figure 2 The process shown includes the following steps: S1: Data Acquisition and Preprocessing: Acquire radiation measurement data of the airborne platform and the platform's positioning / attitude information at the corresponding time; acquire satellite payload transit observation data and its corresponding solar / observation geometric parameters; simultaneously acquire atmospheric and auxiliary data (such as reanalysis data, satellite atmospheric products, etc.) during the satellite transit period, and complete the necessary preprocessing of the radiation data.

[0064] S2: Footprint Mapping and Dynamic ROI Construction: Based on the location, attitude, and sensor field-of-view parameters of the aerial platform, the footprint range on the ground during a single observation by the aerial platform is calculated; a dynamic ROI corresponding to the footprint is constructed on the satellite imagery. Preferably, the dynamic ROI uses a square within the circular area of ​​the footprint to determine the window size, and the pixel window can be adaptively selected according to the platform height (e.g., 3×3 or 4×4, etc.) to ensure that the ROI scale is consistent with the instantaneous footprint coverage.

[0065] S3: Footprint scale uniformity test and sample screening: Calculate the uniformity index of radiation / reflectivity (such as coefficient of variation CV) within the dynamic ROI. When CV meets the preset threshold condition (e.g., CV is less than or equal to a certain threshold), the observed sample is determined to be a valid sample.

[0066] S4: BOA reflectivity construction: Under the observation conditions of the airborne platform, combined with the corresponding atmospheric parameters and geometric conditions, the airborne platform observations are processed by downlink to obtain the reflectivity / radiative characterization of the bottom surface (BOA), so that subsequent geometric equivalence and forward reconstruction are based on a unified surface radiation state.

[0067] S5: BRDF Correction: Introduce the BRDF model or kernel driving parameters to calculate the surface directional reflection characteristics under the observation geometry of the airborne platform and the satellite transit observation geometry, respectively, construct the geometric normalization coefficient, and convert the BOA reflectance under the airborne platform condition into the equivalent BOA reflectance under the satellite transit geometry condition.

[0068] S6: BOA→TOA forward reconstruction and residual atmospheric path difference compensation (solving the problem of inconsistent handling of altitude difference / time difference): Using the equivalent BOA reflectivity obtained from S5 under the satellite transit geometry as input, the radiative transfer forward model is used to calculate the satellite's TOA radiance at the sensor under the satellite transit atmospheric conditions.

[0069] S7: Solving and outputting calibration coefficients: Solving the TOA radiation at the sensor obtained from the reconstruction of S6 with the actual observed DN value of the satellite to obtain the radiation calibration coefficients of the satellite payload.

[0070] In summary, the method provided in this embodiment, under the engineering constraint of inconsistency between observations from the near-field platform and satellite, proposes an equivalent conversion mechanism for radiometric values ​​centered on "downlink inversion—geometric equivalence—uplink forward simulation." This method first converts radiance into BOA reflectance, which is closer to the inherent properties of the Earth's surface and can be aligned across time and angles, using BOA as a common reference quantity. Then, under the atmospheric and observational geometry conditions at the time of satellite transit, forward simulation is performed to obtain the TOA radiance that the satellite should observe. This is ultimately used to solve for calibration coefficients. A dynamic field-of-view matching and quality control process based on platform altitude changes is also constructed to reliably convert near-field platform observations into entrance pupil radiance consistent with satellite imaging conditions, thereby completing the on-orbit radiometric calibration of optical remote sensing satellite sensors.

[0071] Traditional methods typically assume that the observation conditions—including observation time, azimuth, and zenith angle—are consistent between the airborne platform and the satellite payload. This consistency is a fundamental prerequisite for effective calibration comparison between the airborne platform's radiance observation data and the satellite's observation data. Existing literature, as described in the background section, addresses the inconsistency in these observation conditions through temporal and spatial matching. Only observation data points that simultaneously meet both high temporal synchronization and high spatial overlap can be considered valid calibration points. This data selection method results in very scarce usable observation data, leading to insufficient spatiotemporal representativeness of the calibration results. The calculation of calibration coefficients is essentially based on statistical analysis of a set of data pairs; insufficient observation points may lead to decreased statistical significance and increased uncertainty. Furthermore, the planning and implementation of airborne platform flight experiments are usually very costly. The dual-matching mechanism of temporal and spatial matching used in traditional methods imposes strict requirements on weather conditions (cloud-free), path control, and the spatiotemporal rendezvous between the airborne platform and the satellite. Problems in any of these aspects can lead to mission failure, increasing the difficulty and risk of the experiment.

[0072] The method provided in this application decouples the synchronization between the observation time of the airborne platform and the satellite's transit time. This method does not require the airborne platform and satellite to observe strictly simultaneously and from the same location. All observation points within the effective footprint range of the airborne platform on a single flight trajectory can be used as effective calibration points in subsequent calculations, expanding the calibration data source for effective calibration points. By introducing BRDF correction and using geometric equivalence and BRDF normalization, the method effectively separates and compensates for changes in surface reflectivity caused by differences in observation geometry. This allows the airborne platform to conduct independent observations within more flexible time windows and with more relaxed flight paths, as long as its observation area (footprint range) is covered by subsequent satellites. This reduces the requirements for flight control, dependence on perfect weather, and the complexity of mission planning, thereby reducing the risk of experimental failure, technical difficulty, and overall implementation cost. This makes the airborne platform radiometric calibration technology more engineering-practical and scalable.

[0073] The following example, using the calibration of sensors on a MODIS satellite, details the calibration accuracy calculation process of the method provided in this embodiment, and compares it with existing literature (i.e., the literature described in the background art) to illustrate the improvement in calibration accuracy of the method provided in this application.

[0074] In calibrating MODIS sensors using a balloon-borne radiometer, several independent factors contribute to the overall uncertainty. This application considers seven main sources (i.e., seven influencing factors): radiometer observations, footprint positioning uncertainty, BRDF correction, surface heterogeneity, down-atmosphere influence, up-atmosphere influence, and radiative transfer model. For each influencing factor, sensitivity experiments or product validations were conducted to derive the standard uncertainty of its relative radiometric difference. The total uncertainty was calculated using the sum of the square roots of the contributions of all seven influencing factors, as shown in Equation (8), assuming that the terms are independent of each other. This method follows standard uncertainty propagation practices and provides a conservative but realistic estimate of end-to-end calibration accuracy.

[0075] (8) In the formula, For total uncertainty, Indicates radiometer observations, Indicates the location of the footprint. Indicates BRDF correction. Indicates the impact of downward atmospheric conditions. Indicates the influence of rising atmosphere. For the radiative transfer model, This indicates surface heterogeneity.

[0076] 1) Radiometer observation ( The radiation calibration uncertainty of the spherical radiation reference radiometer does not exceed 1.42%. Given that the radiometer operated stably during the flight experiment, the laboratory calibration results were directly adopted, and the relevant uncertainties were included in the total uncertainty budget.

[0077] 2) Footprint positioning ( In this flight experiment, the geolocation accuracy of the radiometer footprint was approximately 90 meters. Such positioning errors can lead to spatial misalignment between the radiometer footprint and the corresponding MODIS pixels, introducing position matching uncertainty into radiometric calculations and subsequent calibration results. To quantify this effect, a sliding window method was employed. Specifically, the footprint matching window was perturbed by a one-pixel offset, consistent with the magnitude of the MODIS orbital positioning error. For each sliding window, the average MODIS DN value was extracted and used to recalculate the inter-band validation results. Subsequently, the maximum relative error of all moving scenarios relative to the baseline was adopted to estimate the position matching uncertainty.

[0078] 3) BRDF parameters ( When correcting the BOA reflectance using the BRDF correction coefficients provided by MCD43A1 (MODIS satellite radiometer data product), the inherent uncertainties of the product itself need to be considered. Therefore, based on the relative error between the MOD09GA reflectance and the simulated reflectance at all calibration points, this error is introduced into the corrected BOA reflectance. The maximum relative error between the TOA radiance and the MODIS sensor radiance is thus the BRDF uncertainty.

[0079] 4) Surface heterogeneity ( To quantify the uncertainty introduced by surface heterogeneity, we analyze spatial variability at the observation scale. Specifically, we extract all MODIS pixels covered by a single radiometer observation footprint and calculate the CV of their radiometric values. The maximum CV between calibration points is adopted as the observation uncertainty related to surface heterogeneity.

[0080] 5) Downward atmospheric influence ( Due to the inherent uncertainties of various atmospheric parameters in the ERA5 reanalysis dataset, these uncertainties are introduced into all atmospheric parameters. The BOA reflectance is simulated using a radiative transfer model, and the maximum relative error compared to the baseline BOA reflectance represents the downlink atmospheric uncertainty, as shown in Table 2.

[0081] 6) Influence of the updraft atmosphere ( The AOT@550 nm calibration point was derived from Sunflower 8 aerosol products, and other atmospheric parameters were taken from MOD07_L2 products. Considering the uncertainties associated with these products, the uplink atmospheric uncertainty was quantified as the maximum relative deviation between the disturbed calibration coefficient and the baseline calibration coefficient relative to the calculated baseline calibration coefficient, as shown in Table 2. Table 2 is as follows: Table 2. Uncertainty of Atmospheric Parameters

[0082] 7) Radiative transfer model ( The 6S code relies on an approximation of the radiative transfer equation. Based on previous research, the inherent accuracy of the vector 6S model is typically between 0.5% and 1.5% relative to high-precision Monte Carlo simulations. To ensure a conservative estimate and account for potential residual errors during modeling, we adopted a standard uncertainty of 1.5% in the final error budget.

[0083] Finally, the total uncertainty is shown in Table 3, which is as follows: Table 3 Total Uncertainty

[0084] As can be seen from Table 3, the total uncertainty of the method provided in this application on a uniform surface is less than 3.52%.

[0085] Furthermore, the calibration deviations for different spectral bands were analyzed, and the results are shown in Table 4. Table 4 is as follows: Table 4. Calibration accuracy of MODIS satellites by the airborne platform.

[0086] The results of the calibration accuracy analysis of MODIS satellites by the airborne platform in the existing literature described in the background art are shown in Table 5. Table 5 is as follows: Table 5. Calibration accuracy of MODIS satellites by the airborne platform in existing literature.

[0087] Comparing Tables 4 and 5, it can be seen that the calibration bias of the method provided in this application is significantly better than that of existing literature in most bands, and is closer to the range of 0. In particular, for the B1 band: existing literature shows -7.18%, while this application shows 1.69%; for the B4 band: existing literature shows -6.72%, while this application shows 1.97%. These two bands originally exhibited a large negative bias (close to -7%) when using the research methods in existing literature, but under the method provided in this application, it has been reduced to about 2%, indicating that the method provided in this application has stronger suppression of systematic bias.

[0088] Comparing the overall deviation magnitude, the average deviation magnitude of the four bands in existing literature is approximately 4.58%, while that in this application is approximately 2.33%, representing an overall reduction of about 49%. Specifically, the deviation in band B2 also decreased from 3.76% to 2.52%, demonstrating a certain improvement. However, for band B3, existing literature shows a deviation of -0.64%, while this application shows 3.13%, indicating that this band is not superior to existing literature. Therefore, the method provided in this application achieves a significant improvement in accuracy in bands B1, B2, and B4, and significantly reduces the overall deviation magnitude.

[0089] In summary, this application provides a method for on-orbit radiometric calibration of optical remote sensing satellite sensors under conditions where the observation conditions of the airborne platform and the satellite are inconsistent. This method is applicable to different types of optical remote sensing satellite sensors and does not depend on specific data sources, atmospheric models, or radiative transfer models. It has the following advantages: (1) Footprint scale uniformity and uniformity screening: By using dynamic footprint ROI and CV threshold screening, the uniformity evaluation is upgraded from a fixed window to "dynamic footprint consistency" quality control, which significantly reduces the error amplification caused by the introduction of non-uniform surface samples and improves the stability and repeatability of calibration regression.

[0090] (2) BRDF correction: By using BRDF geometric normalization, the observations of the airborne platform are converted into equivalent observations under the geometric conditions of satellite transit, reducing the systematic bias caused by the difference in observation geometry. In particular, it can improve the consistency of calibration under conditions of significant directional reflection or complex surface.

[0091] (3) BOA→TOA forward reconstruction unified processing time / atmospheric difference: By realizing the reconstruction from BOA to TOA in the unified forward radiation transmission link, the atmospheric path difference caused by the altitude difference between the airborne platform and the satellite platform is uniformly compensated, and the atmospheric conditions of the satellite transit are used as the benchmark.

[0092] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for radiometric calibration of remote sensing satellites, characterized in that, include: Based on the footprints on the ground observed by the airborne platform, effective calibration points were selected. The observed radiance at all valid calibration points of the airfield platform is extracted. Under the observation conditions of the airfield platform, and combined with the corresponding atmospheric parameters and geometric conditions, the surface reflectance of the airfield platform is obtained by using the radiometric inversion method. ; Using the BRDF parameters provided by satellite products, and based on the BRDF model, the surface reflectance under the observation geometry of the airborne platform was simulated. Surface reflectance under satellite overpass observation geometry and calculate and The ratio of to is used as the surface BRDF correction factor. ; Correction factor for surface BRDF Surface reflectivity of the air-to-ground platform Multiply to obtain the surface reflectance after angle correction of the air-to-ground platform. ; Regarding the surface reflectance Perform a forward radiation simulation to obtain the satellite's entrance pupil radiation. The calibration coefficient is calculated by combining the average DN value of all the effective calibration points in the satellite imagery.

2. The method according to claim 1, characterized in that, Based on the footprint range observed from the airborne platform, valid calibration points were selected, including: Based on the position, attitude, and sensor field of view parameters of the aerial platform, the range of the footprints on the ground observed by the aerial platform in this study is calculated. Construct a dynamic region of interest (ROI) on the satellite imagery corresponding to the footprint area; Calculate the uniformity index within the dynamic ROI. When the uniformity index meets the preset threshold condition, the dynamic ROI is determined to be a valid calibration point.

3. The method according to claim 2, characterized in that, Constructing a dynamic region of interest (ROI) on the satellite imagery corresponding to the footprint area includes: Based on the flight altitude H and sensor field of view θ of the airborne platform at each moment, calculate the field of view diameter R of the sensor projected on the ground on the airborne platform; Based on the field of view diameter R, the side length L of the inscribed square of the footprint is calculated and converted with the pixel resolution of the satellite image to obtain the size of the extraction window; The region of pixels covered by the extraction window on the satellite image is taken as the dynamic region of interest (ROI).

4. The method according to claim 1, characterized in that, Earth's surface reflectance is obtained by inverting using the radiometric inversion method. The specific formula is as follows: Calculate using the following formula : ， In the formula, This represents the observed radiance of the airborne platform after spectral normalization. , , This is the atmospheric correction factor.

5. A remote sensing satellite radiometric calibration system, said system being used to perform the steps of the method according to any one of claims 1 to 4, comprising: The effective calibration point screening unit is configured to screen out effective calibration points based on the footprint range on the ground observed by the airborne platform during this observation. The downlink inversion unit is configured to extract the observed radiance of all valid calibration points of the airborne platform and use the radiometric inversion method to obtain the surface reflectance. ; The surface correction factor calculation unit is configured to use BRDF parameters provided by satellite products and, based on the BRDF model, simulate the surface reflectance under the observation geometry of the airborne platform. Surface reflectance under satellite overpass observation geometry and calculate and The ratio of to is used as the surface BRDF correction factor. ; The observation geometric correction unit is configured to adjust the surface BRDF correction factor. Surface reflectivity of the air-to-ground platform Multiply to obtain the surface reflectance after angle correction of the air-to-ground platform. ; Forward simulation unit, configured to simulate the surface reflectance Perform a forward radiation simulation to obtain the satellite's entrance pupil radiation. The calibration coefficient is calculated by combining the average DN value of all the effective calibration points in the satellite imagery.

6. A computer device comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program / instructions stored thereon, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method described in any one of claims 1 to 4.

8. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method described in any one of claims 1 to 4.

Images