Automated pseudo-invariant field optimization method and system for on-orbit calibration of remote sensing satellites

Abstract

This invention discloses an automated pseudo-invariant field selection method and system for on-orbit calibration of remote sensing satellites. The method includes: online loading and acquisition of apparent reflectance images, precipitation, and aerosol optical thickness data via a cloud platform; forced screening of images simultaneously satisfying AOD ≤ 0.2 and daily precipitation = 0 to eliminate atmospheric and humidity pseudo-fluctuations, establishing a multi-temporal dataset; independent calculation of spatial uniformity and temporal stability using the coefficient of variation (CV), combined with Getis-Ord statistics for spatial autocorrelation clustering to identify low-value clusters with both spatiotemporal continuity, constructing a collaborative screening model; introducing a spatial aggregation optimization algorithm based on sliding windows and statistical consistency, whereby a threshold percentage of valid candidate pixels within the window is considered a qualified spectral control point; and traversing and outputting a data product containing metadata, spatial boundaries, and multi-temporal spectral features. This invention achieves automated aggregation from discrete pixels into regular, uniform, and standardized regions directly usable for on-orbit satellite calibration.

Description

Technical Field

[0001] This invention relates to the field of interdisciplinary technology of remote sensing information technology and satellite calibration technology, and in particular to an automated pseudo-invariant field optimization method and system for on-orbit calibration of remote sensing satellites. Background Technology

[0002] With the continuous improvement of the global Earth observation system, satellite remote sensing has entered a new stage of quantitative application. Radiometric calibration, as the cornerstone of quantitative remote sensing, directly determines the quality and application value of remote sensing data products. Pseudo-Invariant Calibration Sites (PIS), as a natural radiometric calibration benchmark, has become an internationally recognized important means of on-orbit radiometric calibration due to its long-term stable surface reflection characteristics.

[0003] However, with the continuous improvement of satellite payload performance and the increasingly stringent requirements for calibration accuracy, existing PICS selection techniques are no longer sufficient to meet the needs of modern remote sensing calibration. Current pseudo-invariant field selection techniques have inherent limitations, mainly manifested in the following problems:

[0004] 1. At the data preprocessing level, existing methods generally lack rigorous control over the quality of raw observation data. The apparent reflectance received by remote sensors is a comprehensive reflection of surface and atmospheric characteristics. Existing technologies often directly use reflectance data that has not been filtered for atmospheric quality for subsequent analysis. Especially in areas with significant aerosol influences, high aerosol optical thickness leads to enhanced atmospheric scattering and absorption, severely distorting the true reflectance characteristics of the surface. Similarly, changes in surface humidity caused by precipitation events alter surface reflectance characteristics, resulting in short-term fluctuations in reflectance, reducing the stability of time series, and thus affecting the long-term usability of the site. These key atmospheric and meteorological parameters are not used as mandatory preconditions for filtering in existing technologies, leading to systematic errors introduced into the screening process from the outset.

[0005] 2. In terms of constructing evaluation systems, existing technologies exhibit significant limitations. Most studies use a single indicator or a simple, parallel combination of indicators for screening, lacking correlation and weight constraints among the indicators, and failing to form a comprehensive evaluation mechanism with multi-dimensional collaboration. For example, some methods focus only on the stability of time series while neglecting spatial uniformity; some methods, while examining spatiotemporal characteristics simultaneously, fail to establish an effective collaborative screening model. This one-sided evaluation system makes it difficult to ensure that the selected targets simultaneously possess ideal spatiotemporal stability, resulting in uncertainty in the screening results during practical applications.

[0006] 3. In terms of engineering applications, existing technologies have serious shortcomings. Practical satellite calibration tasks require large, continuous, and uniform areas with clearly defined boundaries. However, current methods typically output discrete sets of pixels that meet specific statistical conditions. These pixels often exhibit fragmented spatial distribution, lacking continuity and regularity. Furthermore, existing methods largely rely on manual experience for area selection, lacking automated engineering implementation paths, making it difficult to meet the needs of large-scale, operational satellite calibration tasks.

[0007] With the continuous launch and in-orbit operation of new-generation high-resolution Earth observation satellites, the ability to acquire and apply remote sensing data is constantly improving, placing higher demands on the accuracy and efficiency of radiometric calibration. Existing pseudo-invariant field selection techniques are no longer adequate to meet this development trend. Currently, there is a lack of a complete technical system encompassing data quality control, multi-dimensional collaborative evaluation, and engineering output. This not only affects the calibration accuracy and quantification capabilities of individual satellites but also impacts the radiometric consistency and cross-platform comparability of multi-satellite collaborative observation data. It is a key technical bottleneck restricting the development of long-term observation sequence analysis and global Earth observation systems towards higher precision and stronger collaboration. Summary of the Invention

[0008] In view of the aforementioned existing problems, the present invention is proposed.

[0009] Therefore, the purpose of this invention is to propose an automated pseudo-invariant field selection method and system for on-orbit calibration of remote sensing satellites. This method develops an automated, high-precision, and engineering-practical pseudo-invariant field selection method. It constructs a pre-data purification mechanism based on atmospheric and meteorological parameters, establishes a collaborative screening model integrating spatial uniformity and temporal stability, and employs a sliding window geometric optimization algorithm to aggregate discrete pixels into regular, uniform, and directly usable standardized regions for on-orbit satellite calibration. Ultimately, this forms a complete technology chain from multi-source data input to engineering output, meeting the urgent needs of modern remote sensing for large-scale, high-precision, and automated radiometric calibration.

[0010] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0011] In a first aspect, the present invention provides an automated pseudo-invariant field selection method for on-orbit calibration of remote sensing satellites, comprising the following steps:

[0012] S1. Based on the cloud processing platform for satellite imagery and Earth observation data, online loading and acquisition of remote sensing images of apparent reflectance of the top of the atmosphere, ERA5-Land precipitation data and aerosol optical thickness data are performed for batch data processing.

[0013] S2. For each remote sensing image, a judgment is made, and remote sensing images that simultaneously meet the requirements of aerosol optical thickness (AOD) ≤ 0.2 and daily precipitation = 0 are forcibly selected in order to eliminate pseudo fluctuations caused by atmospheric scattering and changes in surface humidity, and to establish a multi-temporal apparent reflectance dataset.

[0014] S3. Based on the multi-temporal apparent reflectance dataset, the coefficient of variation (CV) is used to independently calculate spatial homogeneity and temporal stability. The Getis-Ord statistical method is combined to perform spatial autocorrelation clustering analysis on the spatiotemporal CV to identify low-value clusters that simultaneously possess spatiotemporal continuity, thereby obtaining candidate regions for pseudo-invariant fields and constructing a collaborative screening model that integrates spatial homogeneity and temporal stability.

[0015] S4. A spatial aggregation optimization algorithm based on sliding window and statistical consistency is introduced into the collaborative screening model. Low-value clustering areas are used as input priors. A sliding window is used to systematically scan the candidate areas. When the proportion of valid candidate pixels belonging to low-value clustering areas in the window reaches the preset consistency threshold, the entire window is identified as a qualified spectral control point, and a set of spectral control points that conforms to satellite calibration specifications is generated.

[0016] S5. Traverse the spectral control point set, filter to obtain the pseudo-invariant field vector boundary, calculate the statistical characteristics of surface reflectance for each band by year, obtain multi-temporal stable reflectance characteristics, and output three types of data products including field selection metadata files, spatial boundary files, and multi-temporal spectral feature datasets to characterize the spatial properties and spectral stability characteristics of the pseudo-invariant field.

[0017] Furthermore, step S1 includes:

[0018] S11. Based on the set research scope and time period, load Landsat 8 TOA imagery, extract multispectral bands and quality bands, and remove low-quality images with cloud cover exceeding the threshold based on image metadata cloud cover index; at the same time, match ERA5-Land daily total precipitation and MODIS MCD19A2 AOD aerosol optical thickness data.

[0019] S12. Analyze the quality bands of Landsat 8 TOA imagery, identify clouds and shadows through bitwise operations, perform cloud masking, and generate clear-sky effective images; calculate daily cumulative precipitation based on ERA5-Land reanalysis data, and calculate the daily average aerosol optical thickness of the target area based on MODISMCD19A2 AOD data to form daily series of meteorological and atmospheric parameters.

[0020] S13. Project the cloud-masked apparent reflectance data, daily precipitation sequence, and daily AOD data into the EPSG:4326 coordinate system and output a spatiotemporal data cube containing apparent reflectance, precipitation, and AOD.

[0021] Furthermore, step S3 includes:

[0022] S31. The coefficient of variation (CV) is used to measure the spatiotemporal stability of the site. The spatial coefficient of variation is determined based on the ratio of the standard deviation to the average value of the reflectance of each pixel in the calculation window, and the temporal coefficient of variation is determined based on the ratio of the standard deviation to the average value of the average reflectance of the time series pixels in the calculation window.

[0023] S32. Calculate the spatial variation coefficient of each image in the effective dataset, generate a spatial image, and perform time series pixel spatial superposition mean processing on all spatial images of the multi-band to generate an average spatial variation coefficient image.

[0024] S33. Calculate the average value of all images in the effective dataset, calculate the time variation coefficient based on the average value, generate a time image, and perform mean processing on all time images of the multi-band dataset to obtain the average time variation coefficient image.

[0025] S34. Using the Getis-Ord statistical method, cluster analysis was performed on the mean spatial coefficient of variation image and the mean temporal coefficient of variation image to calculate the Gi* index; clusters with a confidence level of 99% and Z score < -2 and small P value were selected as the evaluation results of spatiotemporal consistency.

[0026] S35. Based on the calculated statistical results, select a spatial variation coefficient of less than 5% and a temporal variation coefficient of less than 10% as thresholds. Combine the spatiotemporal consistency evaluation results to obtain candidate regions for pseudo-invariant fields.

[0027] Furthermore, step S4 includes:

[0028] S41. Set a square with a side length of 1 kilometer as the basic unit of spatial aggregation. Use a fixed step size of 30 meters to drive the window to slide systematically within the study area to achieve a complete scan and spatial coverage of the entire pseudo-invariant field candidate region.

[0029] S42. For each window unit, count the number of valid pixels belonging to the pseudo-invariant field candidate region, and calculate the proportion of valid pixels in the total number of pixels in the window. Use the proportion to quantify the degree of spatial consistency within the window unit.

[0030] S43. Set the spatial consistency threshold to 95%. When the proportion of effective pixels in a window unit reaches or exceeds 95%, the window unit is determined to meet the spatial consistency requirements and is recognized as a qualified spectral control point. When the proportion is less than 95%, the window unit is removed entirely.

[0031] S44. Form a set of regions with clear geographical boundaries, fixed size and highly consistent internal optical properties by all spectral control points.

[0032] Secondly, this invention provides an automated pseudo-invariant field optimization system for on-orbit calibration of remote sensing satellites, comprising:

[0033] Data acquisition and preprocessing module: Used to load and acquire remote sensing images of apparent reflectance of the top of the atmosphere, ERA5-Land precipitation data and aerosol optical thickness data online based on the cloud processing platform of satellite imagery and earth observation data, and perform batch data processing;

[0034] Pre-processing purification and screening module: used to judge each remote sensing image and forcibly screen the remote sensing images that simultaneously meet the requirements of aerosol optical thickness (AOD) ≤ 0.2 and daily precipitation = 0, so as to eliminate pseudo fluctuations caused by atmospheric scattering and changes in surface humidity and establish a multi-temporal apparent reflectance dataset.

[0035] Spatiotemporal collaborative screening module: Based on a multi-temporal apparent reflectance dataset, it independently calculates spatial homogeneity and temporal stability using the coefficient of variation (CV), and combines the Getis-Ord statistical method to perform spatial autocorrelation clustering analysis on the spatiotemporal CV to identify low-value clusters that simultaneously possess spatiotemporal continuity, thereby obtaining pseudo-invariant field candidate regions and constructing a collaborative screening model.

[0036] Geometric morphology optimization module: This module introduces a spatial aggregation optimization algorithm based on sliding window and statistical consistency into the collaborative screening model. It uses low-value clustering areas as input priors and uses a sliding window to systematically scan candidate regions. When the proportion of valid candidate pixels belonging to low-value clustering areas within the window reaches a preset consistency threshold, the entire window is identified as a qualified spectral control point and a set of spectral control points is generated.

[0037] The results output and data processing module is used to traverse the spectral control point set, filter out pseudo-invariant field vector boundaries, calculate the statistical characteristics of surface reflectance for each band by year, obtain multi-temporal stable reflectance characteristics, and output three types of data products, including field selection metadata files, spatial boundary files, and multi-temporal spectral feature datasets.

[0038] Furthermore, the data acquisition and preprocessing module includes:

[0039] Data acquisition unit: used to load Landsat 8 TOA imagery to extract multispectral and quality bands, remove high cloud cover images, and match ERA5-Land daily total precipitation and MODIS MCD19A2 AOD data;

[0040] Data preprocessing unit: used to analyze quality bands for cloud masking, calculate daily cumulative precipitation and daily average aerosol optical thickness, and form daily series of meteorological and atmospheric parameters;

[0041] Multi-source data construction unit: used to project apparent reflectance data, precipitation and AOD data into the EPSG:4326 coordinate system and output a spatiotemporal data cube.

[0042] Furthermore, the spatiotemporal collaborative filtering module includes:

[0043] Coefficient of variation (CV) calculation unit: used to measure the spatiotemporal stability of a site using the coefficient of variation (CV), and to calculate the spatial coefficient of variation and the temporal coefficient of variation respectively;

[0044] Spatiotemporal mean image generation unit: used to generate spatial and temporal coefficient of variation images with multi-band averages;

[0045] Getis-Ord statistical unit: Used to perform cluster analysis on spatiotemporal coefficient of variation images using the Getis-Ord statistical method, calculate the Gi* index, and select clusters with a confidence level of 99% and Z < -2 and small P-value as spatiotemporal consistency evaluation results;

[0046] Candidate region delineation unit: used to select pseudo-invariant field candidate regions with a spatial variation coefficient of less than 5% and a temporal variation coefficient of less than 10% as thresholds, combined with the spatiotemporal consistency evaluation results.

[0047] Furthermore, the geometry optimization module includes:

[0048] Sliding window traversal unit: used to set a 1-kilometer square basic unit, and uses a 30-meter step size to drive the window for a scan without omissions;

[0049] Consistency statistics unit: used to count the number and proportion of valid candidate pixels in each window unit, and quantify the degree of spatial consistency;

[0050] Control point screening unit: used to set a 95% consistency threshold, identify windows that reach the threshold as qualified spectral control points, and remove windows that are below the threshold;

[0051] Region set generation unit: used to form a region set with all spectral control points having clear geographical boundaries, fixed size and highly consistent internal optical properties.

[0052] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0053] Fourthly, the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method described above.

[0054] The beneficial effects of this invention are as follows: The automated pseudo-invariant field optimization method and system for on-orbit calibration of remote sensing satellites provided by this invention constructs a pre-data purification mechanism based on atmospheric and meteorological parameters, establishes a collaborative screening model integrating spatial uniformity and temporal stability, and employs a sliding window geometric optimization algorithm to aggregate discrete pixels into regular, uniform, and standardized regions that can be directly used for on-orbit satellite calibration. This ultimately forms a complete technology chain from multi-source data input to engineering output, achieving high automation, high precision, and engineering practicality, thus meeting the urgent needs of modern remote sensing for large-scale, high-precision, and automated radiometric calibration. Attached Figure Description

[0055] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. 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.

[0056] Figure 1 This is an overall flowchart of the automated pseudo-invariant field optimization method for on-orbit calibration of remote sensing satellites in this embodiment of the invention;

[0057] Figure 2 This is a spatial distribution map of the average spatial CV of Landsat-8 OLI data (2020-2025) according to an embodiment of the present invention;

[0058] Figure 3 This is a spatial distribution map of the average time CV of Landsat-8 OLI data (2020-2025) according to an embodiment of the present invention;

[0059] Figure 4 This is a spatial distribution map of Getis statistics based on the average spatial CV in an embodiment of the present invention;

[0060] Figure 5 This is a spatial distribution diagram of Getis statistics based on average time CV in an embodiment of the present invention;

[0061] Figure 6 This is a schematic diagram of the stable field screening results for a certain city according to an embodiment of the present invention;

[0062] Figure 7 This is a bar chart of the four-band spatial variation coefficient statistics of the stable field screening results in an embodiment of the present invention;

[0063] Figure 8 This is a bar chart of the statistical values ​​of the four-band time variation coefficients of the stable field screening results in an embodiment of the present invention;

[0064] Figure 9 This is a flowchart of the automated pseudo-invariant field selection method for on-orbit calibration of remote sensing satellites according to the present invention;

[0065] Figure 10 This is a schematic diagram of the configuration of a computer device according to an embodiment of the present invention. Detailed Implementation

[0066] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0067] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0068] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0069] Reference Figures 1-10 This is one embodiment of the present invention, which provides an automated pseudo-invariant field selection method for on-orbit calibration of remote sensing satellites, comprising the following steps:

[0070] S1. Based on the cloud processing platform for satellite imagery and Earth observation data (Google Earth Engine, GEE platform), online loading and acquisition of atmospheric top apparent reflectance (Landsat 8 TOA) remote sensing images, ERA5-Land precipitation data, and aerosol optical thickness (MODIS MCD19A2 AOD) data are performed for batch data processing.

[0071] S11. Based on the set research scope and time period, load Landsat 8 TOA imagery, extract four bands (blue, green, red, and near-infrared bands) from B2 to B5, and extract the quality band (QA_PIXEL). Low-quality images with cloud cover exceeding the threshold (65%) are removed based on the image metadata cloud cover index (CLOUD_COVER). Simultaneously, match ERA5-Land daily total precipitation (total_precipitation_sum) and MODIS MCD19A2 AOD aerosol optical thickness (Optical_Depth_055) data for subsequent meteorological and atmospheric condition screening.

[0072] S12. Analyze the quality band (QA_PIXEL) of Landsat 8 TOA imagery, identify clouds and shadows through bitwise operations, perform cloud masking on the corresponding pixels, and generate clear-sky effective images; calculate the daily cumulative precipitation at each pixel location based on ERA5-Land reanalysis data; simultaneously, calculate the daily average aerosol optical thickness of the target area using spatial statistical methods based on MODIS MCD19A2 AOD data, forming a daily series of meteorological and atmospheric parameters covering the study period;

[0073] S13. The Landsat apparent reflectance data, daily precipitation series and daily AOD data after cloud masking are uniformly projected to the EPSG:4326 coordinate system, and the output includes a spatiotemporal data cube containing apparent reflectance, precipitation and AOD, providing a complete and reliable data foundation for subsequent quality control and screening of pseudo-invariant fields.

[0074] S2. For each remote sensing image, a judgment is made, and remote sensing images that simultaneously meet the requirements of aerosol optical thickness (AOD) ≤ 0.2 and daily precipitation = 0 are forcibly selected to eliminate pseudo fluctuations caused by atmospheric scattering and changes in surface humidity, and to establish a multi-temporal apparent reflectance dataset.

[0075] S3. Based on the multi-temporal apparent reflectance dataset, the coefficient of variation (CV) is used to independently calculate spatial homogeneity and temporal stability. The Getis-Ord statistical method is combined to perform spatial autocorrelation clustering analysis on the spatiotemporal CV to identify low-value clusters that simultaneously possess spatiotemporal continuity, thereby obtaining candidate regions for pseudo-invariant fields and constructing a collaborative screening model that integrates spatial homogeneity and temporal stability.

[0076] S31. The coefficient of variation (CV) is used to measure the spatiotemporal stability of the site. The formula for calculating the coefficient of variation (CV) is:

[0077] (1)

[0078] When equation (1) is used to evaluate spatial homogeneity, To calculate the standard deviation of the reflectance of each pixel in the window, It is the average value of all pixels participating in the calculation within the calculation window; the spatial variation coefficient is denoted as... When equation (1) is used to evaluate the stability of a time series, To calculate the standard deviation of the average reflectance of pixels in the time series within the calculation window, It is the average value of the time series pixels in the calculation window, and the coefficient of variation over time is denoted as . The value is directly proportional to the degree of change in reflectance of each pixel in the calculation window, that is... The larger the value, the worse the spatiotemporal stability;

[0079] S32. Calculate the spatial variation coefficient of each image in the Landsat8 valid dataset of the study area, and generate the spatial variation coefficient. Images, representing all spatial dimensions of the B2, B3, B4, and B5 bands (blue, green, red, and near-infrared bands). The images are subjected to time-series pixel spatial overlay mean processing to generate an average spatial variation coefficient image covering the entire study area;

[0080] S33. Calculate the average value of all images in the Landsat 8 valid dataset for the study area, and calculate the coefficient of temporal variation based on the average value. Images, for all times in the B2, B3, B4, and B5 bands (blue, green, red, and near-infrared bands). The images were mean-processed to obtain the average time variation coefficient images of the four bands B2-B5;

[0081] S34, Using Getis-Ord Statistical methods are used to perform cluster analysis on the mean spatial coefficient of variation image and the mean temporal coefficient of variation image to calculate... The exponent is calculated as follows:

[0082] (2)

[0083] In equation (2), S is the global mean, S is the global standard deviation, i represents the current calculated cell, the total number of cells is n, i=1,2,…n; j represents the other cells in the neighborhood of the current calculated center cell i. It is the spatial weight between the current computation center pixel and its neighboring elements (at a distance of ≤1.5 pixels from the computation center element), within the neighborhood. =1, otherwise 0;

[0084] The statistical results dataset contains three attribute fields: Z-score, P-value, and confidence interval; clusters with a confidence level of 99% and Z < -2 and small P-values ​​are selected as the evaluation results of spatial consistency.

[0085] S35, Regarding the calculated results Statistical results, selection Less than 5%, Less than 10% was used as the threshold for evaluating the spatiotemporal stability of the study area; clusters with a confidence level of 99%, Z < -2 and small P value were selected as the evaluation results of spatiotemporal consistency, and candidate regions for pseudo-invariant fields were obtained.

[0086] S4. A spatial aggregation optimization algorithm based on sliding window and statistical consistency is introduced into the collaborative screening model. Low-value clustering areas are used as input priors. A sliding window is used to systematically scan the candidate areas. When the proportion of valid candidate pixels belonging to low-value clustering areas in the window reaches the preset consistency threshold, the entire window is identified as a qualified spectral control point, and a set of spectral control points that conforms to satellite calibration specifications is generated.

[0087] S41. A sliding window geometric optimization algorithm is adopted, and a square with a side length of 1 kilometer is set as the basic unit for spatial aggregation. A fixed step size of 30 meters is used to drive the window to slide systematically within the study area, so as to achieve a complete scan and spatial coverage of the entire pseudo-invariant field candidate region.

[0088] S42. For each window unit, count the number of valid pixels belonging to the pseudo-invariant field candidate region, and calculate the proportion of valid pixels in the total number of pixels in the window, and use the proportion to quantify the degree of spatial consistency within the window unit.

[0089] S43. Set the spatial consistency threshold to 95%. When the percentage of effective pixels in a window unit reaches or exceeds this threshold, the window unit is determined to meet the spatial consistency requirements and is identified as a qualified spectral control point. When the percentage of effective pixels in a window unit is lower than this threshold, the window unit is removed from the candidate list.

[0090] S44. Form a set of regions with clear geographical boundaries, fixed size (1km×1km) and highly consistent internal optical properties by forming all spectral control points.

[0091] S5. Traverse the spectral control point set, filter to obtain the pseudo-invariant field vector boundary, calculate the statistical characteristics of surface reflectance for each band by year, obtain multi-temporal stable reflectance characteristics, and output three types of data products including field selection metadata files, spatial boundary files, and multi-temporal spectral feature datasets to characterize the spatial properties and spectral stability characteristics of the pseudo-invariant field.

[0092] Secondly, this invention provides an automated pseudo-invariant field optimization system for on-orbit calibration of remote sensing satellites, comprising:

[0093] Data Acquisition and Preprocessing Module: This module is used to load and acquire remote sensing images of apparent reflectance at the top of the atmosphere, ERA5-Land precipitation data, and aerosol optical thickness data online based on a cloud-based satellite imagery and Earth observation data platform, and to perform batch data processing.

[0094] The data acquisition and preprocessing module includes:

[0095] Data acquisition unit: Based on the set research scope and time period, it loads Landsat 8 TOA imagery, extracts four bands (B2–B5) and quality bands, and removes low-quality images with cloud cover exceeding the threshold based on image metadata cloud cover index; it also matches ERA5-Land daily total precipitation and MODIS MCD19A2 AOD aerosol optical thickness data for subsequent meteorological and atmospheric condition screening.

[0096] The data preprocessing unit is used to analyze the quality bands of Landsat 8 TOA imagery, identify clouds and shadows through bitwise operations, perform cloud masking on the corresponding pixels, and generate clear-sky effective images. Based on ERA5-Land reanalysis data, it calculates the daily cumulative precipitation at each pixel location. Simultaneously, based on MODIS MCD19A2 AOD data, it uses spatial statistical methods to calculate the daily average aerosol optical thickness of the target area, forming a daily series of meteorological and atmospheric parameters covering the study period.

[0097] Multi-source data construction unit: This unit projects Landsat apparent reflectance data, daily precipitation sequences, and daily AOD data through cloud masking onto the EPSG:4326 coordinate system, outputting a spatiotemporal data cube containing apparent reflectance, precipitation, and AOD data. This provides a complete and reliable data foundation for subsequent quality control and screening of pseudo-invariant fields.

[0098] Pre-processing purification and screening module: used to judge each remote sensing image and forcibly screen the remote sensing images that simultaneously meet the requirements of aerosol optical thickness (AOD) ≤ 0.2 and daily precipitation = 0, so as to eliminate pseudo fluctuations caused by atmospheric scattering and changes in surface humidity and establish a multi-temporal apparent reflectance dataset.

[0099] Spatiotemporal collaborative screening module: Based on a multi-temporal apparent reflectance dataset, it independently calculates spatial homogeneity and temporal stability using the coefficient of variation (CV), and combines the Getis-Ord statistical method to perform spatial autocorrelation clustering analysis on the spatiotemporal CV to identify low-value clusters that simultaneously possess spatiotemporal continuity, thereby obtaining pseudo-invariant field candidate regions and constructing a collaborative screening model.

[0100] The spatiotemporal collaborative screening module includes: Coefficient of variation unit: used to measure the spatiotemporal stability of the site using the coefficient of variation (CV);

[0101] Mean Spatial Coefficient of Variation Image Unit: Used to calculate the spatial coefficient of variation of each image in the valid Landsat8 dataset of the study area, generating spatial... Images, for all spatial bands B2, B3, B4, and B5 respectively. The images are subjected to time-series pixel spatial overlay mean processing to generate an average spatial variation coefficient image covering the entire study area;

[0102] Mean Time Coefficient of Variation Image Unit: Used to calculate the mean of all images in the valid Landsat 8 dataset for the study area, and then calculate the time coefficient of variation based on the mean to generate the time. Images, for all times in bands B2, B3, B4, and B5 respectively. The images were mean-processed to obtain the average time variation coefficient images of the four bands B2-B5;

[0103] Getis-Ord Statistical unit: used when using Getis-Ord Statistical methods are used to perform cluster analysis on the mean spatial coefficient of variation image and the mean temporal coefficient of variation image to calculate... index;

[0104] Pseudo-invariant field candidate region cells: used for the calculated Statistical results, selection Less than 5%, Less than 10% was used as the threshold for evaluating the spatiotemporal stability of the study area; clusters with a confidence level of 99%, Z < -2 and small P value were selected as the evaluation results of spatiotemporal consistency, and candidate regions for pseudo-invariant fields were obtained.

[0105] Geometric morphology optimization module: This module introduces a spatial aggregation optimization algorithm based on sliding window and statistical consistency into the collaborative screening model. It uses low-value clustering areas as input priors and uses a sliding window to systematically scan candidate regions. When the proportion of valid candidate pixels belonging to low-value clustering areas within the window reaches a preset consistency threshold, the entire window is identified as a qualified spectral control point, and a set of spectral control points is generated.

[0106] The geometry optimization module includes: a sliding window configuration and spatial traversal unit: used to use the sliding window geometry optimization algorithm, setting a square with a side length of 1 kilometer as the basic unit for spatial aggregation, using a fixed step size of 30 meters, driving the window to slide systematically within the study area, so as to achieve a complete scan and spatial coverage of the entire pseudo-invariant field candidate region;

[0107] Intra-window spatial consistency statistics unit: For each window unit, it is used to count the number of valid pixels belonging to the pseudo-invariant field candidate region and calculate the proportion of valid pixels in the total number of pixels in the window, and use the proportion to quantify the degree of spatial consistency within the window unit.

[0108] Unit filtering based on consistency threshold: The threshold for spatial consistency is set to 95%. When the proportion of effective pixels in a window unit reaches or exceeds this threshold, the window unit is determined to meet the spatial consistency requirements and is identified as a qualified spectral control point. When the proportion of effective pixels in a window unit is lower than this threshold, the window unit is removed from the candidate list.

[0109] Standardized control point generation unit: used to form a set of regions with clear geographical boundaries, fixed size and highly consistent internal optical properties from all spectral control points.

[0110] The results output and data processing module is used to traverse the spectral control point set, filter out the pseudo-invariant field vector boundary, calculate the statistical characteristics of surface reflectance for each band by year, obtain multi-temporal stable reflectance characteristics, and output three types of data products, including field selection metadata files, spatial boundary files, and multi-temporal spectral feature datasets. This module characterizes the spatial properties and spectral stability characteristics of the pseudo-invariant field, providing reliable prior knowledge and standardized input for satellite on-orbit radiometric calibration.

[0111] This invention utilizes an automated pseudo-invariant field selection method based on on-orbit calibration of remote sensing satellites to screen pseudo-invariant fields for a city (e.g., Figures 2-8 As shown in the image, Landsat 8 multi-temporal remote sensing datasets from 2020 to 2025 were selected for the city with cloud cover less than 65%. Based on this, images with daily precipitation of 0 and aerosol optical thickness less than 0.2 were further selected, resulting in 480 remote sensing images for subsequent analysis. Less than 5%, The selection criteria were: less than 10%, and Gi* results satisfying Z < −2 and P < 0.05. After geometric screening, 20 spectral control points were selected. The average temporal stability of the candidate fields was 6.09%, the average spatial stability was 2.55%, and the spectral changes in the region were not significant in the past ten years, remaining stable over a long period of time. The selected region can be used as a reference for stable field radiation in China.

[0112] This invention also provides a computer device. Figure 10 This is a schematic diagram of the structure of a computer device provided in an embodiment of the present invention; see the accompanying drawings. Figure 10As shown, the computer device includes: an input device 23, an output device 24, a memory 22, and a processor 21; the memory 22 is used to store one or more programs; when the one or more programs are executed by the one or more processors 21, the one or more processors 21 implement the automated pseudo-invariant field optimization method for on-orbit calibration of remote sensing satellites as provided in the above embodiments; wherein the input device 23, the output device 24, the memory 22, and the processor 21 can be connected via a bus or other means. Figure 10 Taking the example of a connection between China and Israel via a bus.

[0113] The memory 22, as a read / write storage medium for a computing device, can be used to store software programs and computer-executable programs, such as the program instructions corresponding to the automated pseudo-invariant field optimization method for on-orbit calibration of remote sensing satellites described in this embodiment of the invention. The memory 22 may primarily include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a function; the data storage area may store data created based on the use of the device. Furthermore, the memory 22 may include high-speed random access memory and non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some instances, the memory 22 may further include memory remotely located relative to the processor 21, and these remote memories can be connected to the device via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0114] Input device 23 can be used to receive input digital or character information, and generate key signal inputs related to user settings and function control of the device; output device 24 may include display devices such as a display screen.

[0115] The processor 21 executes various functional applications and data processing of the device by running software programs, instructions and modules stored in the memory 22, thereby realizing the above-mentioned automated pseudo-invariant field optimization method for on-orbit calibration of remote sensing satellites.

[0116] The computer equipment provided above can be used to execute the automated pseudo-invariant field optimization method for on-orbit calibration of remote sensing satellites provided in the above embodiments, and has corresponding functions and beneficial effects.

[0117] This invention also provides a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform an automated pseudo-invariant field optimization method for on-orbit calibration of remote sensing satellites as provided in the above embodiments. The storage medium can be any type of memory device or storage device, including: mounting media such as CD-ROM, floppy disk, or magnetic tape; computer system memory or random access memory such as DRAM, DDR RAM, SRAM, EDO RAM, Rambus RAM, etc.; non-volatile memory such as flash memory, magnetic media (e.g., hard disk or optical storage); registers or other similar types of memory elements; the storage medium may also include other types of memory or combinations thereof; furthermore, the storage medium may reside in a first computer system in which the program is executed, or it may reside in a different second computer system connected to the first computer system via a network (such as the Internet); the second computer system can provide program instructions to the first computer for execution. The storage medium includes two or more storage media that may reside in different locations (e.g., in different computer systems connected via a network). The storage medium may store program instructions (e.g., specifically implemented as a computer program) executable by one or more processors.

[0118] Of course, the computer-executable instructions provided in the embodiments of the present invention are not limited to the automated pseudo-invariant field selection method for on-orbit calibration of remote sensing satellites as described in the above embodiments, but can also perform related operations in the automated pseudo-invariant field selection method for on-orbit calibration of remote sensing satellites provided in any embodiment of the present invention.

[0119] In summary, the present invention has the following significant beneficial effects:

[0120] 1. Source data purification to eliminate systematic errors and improve benchmark reliability.

[0121] This invention overcomes the limitations of traditional methods that directly use data without atmospheric quality screening. It creatively uses aerosol optical thickness (AOD≤0.2) and daily precipitation (=0) as mandatory pre-constraints. This completely eliminates signal distortion caused by atmospheric aerosol scattering and absorption, as well as short-term pseudo-fluctuations in reflectivity due to changes in surface precipitation humidity, from a physical mechanism perspective. This provides pure, true radiation input for subsequent spatiotemporal stability calculations, fundamentally ensuring the absolute reliability of the pseudo-invariant field as a calibration benchmark.

[0122] 2. Spatiotemporal depth collaboration accurately eliminates pseudo-stable regions, improving screening precision.

[0123] This invention abandons the single-indicator or simple parallel evaluation methods of existing technologies, and constructs a spatiotemporal collaborative screening model of "coefficient of variation (CV) + Getis-Ord spatial autocorrelation". This model not only quantifies the dispersion of pixels themselves, but also identifies low-value continuous clusters that show statistical significance in both spatial and temporal dimensions through Getis-Ord clustering. This deeply coupled collaborative mechanism effectively excludes isolated or "pseudo-invariant" pixels that only possess single-dimensional stability, ensuring that the selected region has true spatiotemporal homogeneity.

[0124] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An automated pseudo-invariant field selection method for on-orbit calibration of remote sensing satellites, characterized in that, Includes the following steps: S1. Based on the cloud processing platform for satellite imagery and Earth observation data, online loading and acquisition of remote sensing images of apparent reflectance of the top of the atmosphere, ERA5-Land precipitation data and aerosol optical thickness data are performed for batch data processing. S2. For each remote sensing image, a judgment is made, and remote sensing images that simultaneously meet the requirements of aerosol optical thickness (AOD) ≤ 0.2 and daily precipitation = 0 are forcibly selected in order to eliminate pseudo fluctuations caused by atmospheric scattering and changes in surface humidity, and to establish a multi-temporal apparent reflectance dataset. S3. Based on the multi-temporal apparent reflectance dataset, the coefficient of variation (CV) is used to independently calculate spatial homogeneity and temporal stability. The Getis-Ord statistical method is combined to perform spatial autocorrelation clustering analysis on the spatiotemporal CV to identify low-value clusters that simultaneously possess spatiotemporal continuity, thereby obtaining candidate regions for pseudo-invariant fields and constructing a collaborative screening model that integrates spatial homogeneity and temporal stability. S4. Introduce a spatial aggregation optimization algorithm based on sliding window and statistical consistency to the collaborative screening model. Use the low-value cluster area as the input prior and use a sliding window to systematically scan the candidate area. When the proportion of valid candidate pixels belonging to the low-value cluster area in the window reaches the preset consistency threshold, the window as a whole is identified as a qualified spectral control point, and a set of spectral control points that conforms to the satellite calibration specifications is generated. S5. Traverse the set of spectral control points, filter to obtain the pseudo-invariant field vector boundary, calculate the statistical characteristics of surface reflectance for each band by year, obtain multi-temporal stable reflectance characteristics, and output three types of data products including field selection metadata files, spatial boundary files, and multi-temporal spectral feature datasets to characterize the spatial properties and spectral stability characteristics of the pseudo-invariant field.

2. The automated pseudo-invariant field optimization method for on-orbit calibration of remote sensing satellites according to claim 1, characterized in that, Step S1 includes: S11. Based on the set research scope and time period, load Landsat 8 TOA imagery, extract multispectral bands and quality bands, and remove low-quality images with cloud cover exceeding the threshold based on image metadata cloud cover index; at the same time, match ERA5-Land daily total precipitation and MODIS MCD19A2 AOD aerosol optical thickness data. S12. Analyze the quality bands of Landsat 8 TOA imagery, identify clouds and shadows through bitwise operations, perform cloud masking, and generate clear-sky effective images; calculate daily cumulative precipitation based on ERA5-Land reanalysis data, and calculate the daily average aerosol optical thickness of the target area based on MODIS MCD19A2AOD data to form daily series of meteorological and atmospheric parameters. S13. Project the cloud-masked apparent reflectance data, daily precipitation sequence, and daily AOD data into the EPSG:4326 coordinate system and output a spatiotemporal data cube containing apparent reflectance, precipitation, and AOD.

3. The automated pseudo-invariant field optimization method for on-orbit calibration of remote sensing satellites according to claim 1, characterized in that, Step S3 includes: S31. The coefficient of variation (CV) is used to measure the spatiotemporal stability of the site. The spatial coefficient of variation is determined based on the ratio of the standard deviation to the average value of the reflectance of each pixel in the calculation window, and the temporal coefficient of variation is determined based on the ratio of the standard deviation to the average value of the average reflectance of the time series pixels in the calculation window. S32. Calculate the spatial variation coefficient of each image in the effective dataset, generate a spatial image, and perform time series pixel spatial superposition mean processing on all spatial images of the multi-band to generate an average spatial variation coefficient image. S33. Calculate the average value of all images in the effective dataset, calculate the time variation coefficient based on the average value, generate a time image, and perform mean processing on all time images of the multi-band dataset to obtain the average time variation coefficient image. S34. Using the Getis-Ord statistical method, cluster analysis is performed on the mean spatial coefficient of variation image and the mean temporal coefficient of variation image to calculate the Gi* index; clusters with a confidence level of 99% and Z score < -2 and small P value are selected as the evaluation results of spatiotemporal consistency. S35. Based on the calculated statistical results, select a spatial variation coefficient of less than 5% and a temporal variation coefficient of less than 10% as thresholds. Combine the spatiotemporal consistency evaluation results to obtain candidate regions for pseudo-invariant fields.

4. The automated pseudo-invariant field optimization method for on-orbit calibration of remote sensing satellites according to claim 1, characterized in that, The S4 step includes: S41. Set a square with a side length of 1 kilometer as the basic unit of spatial aggregation. Use a fixed step size of 30 meters to drive the window to slide systematically within the study area to achieve a complete scan and spatial coverage of the entire pseudo-invariant field candidate region. S42. For each window unit, count the number of valid pixels belonging to the pseudo-invariant field candidate region, and calculate the proportion of valid pixels in the total number of pixels in the window, and use the proportion to quantify the degree of spatial consistency within the window unit. S43. Set the spatial consistency threshold to 95%. When the proportion of effective pixels in a window unit reaches or exceeds 95%, the window unit is determined to meet the spatial consistency requirements and is recognized as a qualified spectral control point. When the proportion is less than 95%, the window unit is removed entirely. S44. Form a set of regions with clear geographical boundaries, fixed size and highly consistent internal optical properties by all spectral control points.

5. An automated pseudo-invariant field optimization system for on-orbit calibration of remote sensing satellites, characterized in that, include: Data acquisition and preprocessing module: Used to load and acquire remote sensing images of apparent reflectance of the top of the atmosphere, ERA5-Land precipitation data and aerosol optical thickness data online based on the cloud processing platform of satellite imagery and earth observation data, and perform batch data processing; Pre-processing purification and screening module: used to judge each remote sensing image and forcibly screen the remote sensing images that simultaneously meet the requirements of aerosol optical thickness (AOD) ≤ 0.2 and daily precipitation = 0, so as to eliminate pseudo fluctuations caused by atmospheric scattering and changes in surface humidity and establish a multi-temporal apparent reflectance dataset. Spatiotemporal collaborative screening module: Based on the multi-temporal apparent reflectance dataset, it independently calculates spatial homogeneity and temporal stability using the coefficient of variation (CV), and combines the Getis-Ord statistical method to perform spatial autocorrelation clustering analysis on the spatiotemporal CV to identify low-value clustering regions that simultaneously possess spatiotemporal continuity, thereby obtaining pseudo-invariant field candidate regions and constructing a collaborative screening model. Geometric morphology optimization module: This module is used to introduce a spatial aggregation optimization algorithm based on sliding window and statistical consistency into the collaborative screening model. It uses the low-value cluster area as the input prior and uses a sliding window to systematically scan the candidate area. When the proportion of valid candidate pixels belonging to the low-value cluster area in the window reaches a preset consistency threshold, the entire window is identified as a qualified spectral control point and a spectral control point set is generated. The results output and data processing module is used to traverse the spectral control point set, filter out pseudo-invariant field vector boundaries, calculate the statistical characteristics of surface reflectance for each band by year, obtain multi-temporal stable reflectance characteristics, and output three types of data products, including field selection metadata files, spatial boundary files, and multi-temporal spectral feature datasets.

6. The automated pseudo-invariant field optimization system for on-orbit calibration of remote sensing satellites according to claim 5, characterized in that, The data acquisition and preprocessing module includes: Data acquisition unit: used to load Landsat 8 TOA imagery to extract multispectral and quality bands, remove high cloud cover images, and match ERA5-Land daily total precipitation and MODIS MCD19A2 AOD data; Data preprocessing unit: used to analyze quality bands for cloud masking, calculate daily cumulative precipitation and daily average aerosol optical thickness, and form daily series of meteorological and atmospheric parameters; Multi-source data construction unit: used to project apparent reflectance data, precipitation and AOD data into the EPSG:4326 coordinate system and output a spatiotemporal data cube.

7. The automated pseudo-invariant field optimization system for on-orbit calibration of remote sensing satellites according to claim 5, characterized in that, The spatiotemporal collaborative filtering module includes: Coefficient of variation (CV) calculation unit: used to measure the spatiotemporal stability of a site using the coefficient of variation (CV), and to calculate the spatial coefficient of variation and the temporal coefficient of variation respectively; Spatiotemporal mean image generation unit: used to generate spatial and temporal coefficient of variation images with multi-band averages; Getis-Ord statistical unit: Used to perform cluster analysis on spatiotemporal coefficient of variation images using the Getis-Ord statistical method, calculate the Gi* index, and select clusters with a confidence level of 99% and Z < -2 and small P-value as spatiotemporal consistency evaluation results; Candidate region delineation unit: used to select pseudo-invariant field candidate regions with a spatial variation coefficient of less than 5% and a temporal variation coefficient of less than 10% as thresholds, combined with the spatiotemporal consistency evaluation results.

8. The automated pseudo-invariant field optimization system for on-orbit calibration of remote sensing satellites according to claim 5, characterized in that, The geometry optimization module includes: Sliding window traversal unit: used to set a 1-kilometer square basic unit, and uses a 30-meter step size to drive the window for a scan without omissions; Consistency statistics unit: used to count the number and proportion of valid candidate pixels in each window unit, and quantify the degree of spatial consistency; Control point screening unit: used to set a 95% consistency threshold, identify windows that reach the threshold as qualified spectral control points, and remove windows that are below the threshold; Region set generation unit: used to form a region set with all spectral control points having clear geographical boundaries, fixed size and highly consistent internal optical properties.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the method described in any one of claims 1-4.

10. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method described in any one of claims 1-4.

Images