Optical scanning radiometer in-orbit dark background long time series data extraction method and system

Abstract

The invention discloses an optical scanning radiometer in-orbit dark background long time series data extraction method and system, and the method comprises the steps: obtaining cold air observation data and solar zenith angle data from an optical scanning radiometer, and generating a channel-level cold air observation data set and a channel-level solar zenith angle data set. According to the channel-level cold air observation data set and the channel-level solar zenith angle data set, various screening methods are adopted to generate to-be-selected schemes and screening data sets of the to-be-selected schemes. For each to-be-selected scheme, calculating a statistical mean value and data volatility based on a screening data set of the to-be-selected scheme, calculating a dark background representative value of the to-be-selected scheme, and determining a mean value sorting sequence number and a volatility sorting sequence number; and determining an optimal dark background value scheme by adopting a weighted assignment method according to the mean sorting sequence number and the volatility sorting sequence number. Therefore, according to the in-orbit dark background long-time-sequence data extraction method of the optical scanning radiometer, long-time-sequence stable extraction is achieved, and the remote sensing data radiometric calibration precision is remarkably improved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of satellite optical remote sensing technology, and in particular to a method and system for extracting long-term time-series data with dark background from an optical scanning radiometer in orbit. Background Technology

[0002] In the field of satellite optical remote sensing, the photoelectric converted digital values ​​recorded by satellite instruments need to be processed through radiometric calibration to obtain the actual radiometric readings observed by the remote sensor. For instruments with linear responses, the following formula is generally used for processing:

[0003] L = k(DN - DC);

[0004] Where L is reflectivity or radiance, DN is the observed count (Digital Number, DN), k is the radiometric calibration gain, and DC is the dark background value, commonly referred to as the dark current count (DDC). During calibration, k and DC are two key parameters. k is related to the instrument's radiometric response characteristics; DC is the instrument's (mainly the detector's) background output when there is no incident energy, which needs to be subtracted during data processing. Since the satellite instrument's operating environment and status change in orbit, these two parameters also need to be updated in orbit.

[0005] Because instrument radiation response decays in orbit, numerous methods exist for estimating radiation calibration gain in orbit and compensating for changes in the instrument's radiation response through calibration. However, instruments generally do not have the capability to measure background output in orbit. Therefore, the instrument's dark background value is typically determined before launch or during calibration by data fitting, along with the calibration gain. This value usually remains constant in orbit or is updated along with the in-orbit calibration gain. A fixed dark background value cannot reflect the actual in-orbit situation, and updating it along with the in-orbit calibration gain is subject to random fluctuations due to data noise during data fitting, resulting in low reliability and failing to reflect the actual in-orbit situation. Furthermore, the frequency of calibration updates limits the timely reflection of changes in instrument status. Therefore, a technical method is needed to extract long-term in-orbit dark background data from instruments. Summary of the Invention

[0006] This invention provides a method and system for extracting long-term time-series data from an optical scanning radiometer in orbit against a dark background. It achieves stable long-term extraction and significantly improves the radiometric calibration accuracy of remote sensing data.

[0007] To achieve the above objectives, in a first aspect, the present invention provides a method for extracting long-term time-series dark background data from an optical scanning radiometer in orbit, comprising: acquiring cold-space observation data and solar zenith angle data from the optical scanning radiometer, generating a channel-level cold-space observation dataset and a channel-level solar zenith angle dataset; generating candidate schemes and their selected datasets based on the channel-level cold-space observation dataset and the channel-level solar zenith angle dataset using multiple screening methods; for each candidate scheme, calculating the statistical mean and data volatility based on its selected dataset, and calculating the representative value of the dark background for that scheme, determining the mean ranking number and volatility ranking number; determining the preferred dark background value scheme using a weighted assignment method based on the mean ranking number and volatility ranking number; and generating a long-term output dataset containing daily dark background values, observation dates, days from launch date, a first reference value, and a second reference value based on the preferred dark background value scheme.

[0008] Secondly, this invention provides a long-term data extraction system for on-orbit dark background data from an optical scanning radiometer, comprising: a first generation module, a second generation module, a first determination module, a second determination module, and a third generation module. The first generation module is used to acquire cold-space observation data and solar zenith angle data from the optical scanning radiometer, generating a channel-level cold-space observation dataset and a channel-level solar zenith angle dataset. The second generation module is used to generate candidate schemes and their selected datasets using multiple filtering methods based on the channel-level cold-space observation dataset and the channel-level solar zenith angle dataset. The first determination module is used to calculate the statistical mean and data volatility for each candidate scheme based on its selected dataset, and to calculate the representative dark background value of the scheme, determining the mean ranking number and volatility ranking number. The second determination module is used to determine the preferred dark background value scheme using a weighted assignment method based on the mean ranking number and volatility ranking number. The third generation module is used to generate a long-term output dataset containing daily dark background values, observation dates, days from launch, a first reference value, and a second reference value, based on the preferred dark background value scheme.

[0009] Thirdly, the present invention provides an electronic device, comprising: at least one processor; and

[0010] A memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the on-orbit dark background long-term data extraction method of the optical scanning radiometer as described above.

[0011] Fourthly, the present invention provides a computer-readable storage medium including a computer program and instructions, which, when the computer program or the instructions are executed on a computer, cause the computer to perform the on-orbit dark background long-term data extraction method of an optical scanning radiometer as described above.

[0012] As shown above, the method and system for extracting long-term dark background data from an optical scanning radiometer in orbit provided by this invention effectively selects the dark background value scheme with the best overall performance by combining the ranking results of two key indicators: mean magnitude and fluctuation degree. This approach ensures that the final selected dark background value remains at a low level while maintaining its stability over long-term operation, avoiding significant fluctuations. This dual consideration overcomes the limitations of previous single methods (such as only considering the minimum value or only data from a specific angle), which are either easily affected by outliers leading to unstable results or ignore the magnitude of the values ​​themselves. Ultimately, the output long-term dataset (including daily recommended values ​​and reference upper and lower limits) provides a more reliable and consistent dark current subtraction benchmark for radiometric calibration. Attached Figure Description

[0013] Figure 1 This is a flowchart illustrating a method for extracting long-term time-series data from an optical scanning radiometer in orbit against a dark background, according to Embodiment 1 of the present invention.

[0014] Figure 2 This is a schematic diagram of the structure of an on-orbit dark background long-term data extraction system for an optical scanning radiometer according to Embodiment 2 of the present invention;

[0015] Figure 3 This is a schematic diagram of the structure of an electronic device according to Embodiment 3 of the present invention;

[0016] Figure 4 This is a flowchart illustrating a method for extracting long-term time-series data from an optical scanning radiometer in orbit against a dark background, according to a specific embodiment of the present invention.

[0017] Figure 5 This is a schematic diagram illustrating the short-term fluctuations of the channel 1SV value in the existing scheme A.

[0018] Figure 6 This is a schematic diagram illustrating the short-term fluctuations of the channel 1EV value in the existing scheme B.

[0019] Figure 7 A schematic diagram illustrating the change in the lifetime of channel 1EV values ​​after threshold filtering in the existing scheme B;

[0020] Figure 8 Histogram of the lifetime distribution of channel 1EV values ​​after threshold filtering in existing scheme B;

[0021] Figure 9A comparative diagram of different SV screening schemes throughout their lifecycle;

[0022] Figure 10 This is a schematic diagram showing that the optimal solution for channel 2 is the filtering solution numbered 5.

[0023] Figure 11 This is a schematic diagram showing that the optimal solution for channel 3 is the selection scheme numbered 19.

[0024] Figure 12 This is a schematic diagram showing that the optimal solution for channel 1 is the selection scheme numbered 13. Detailed Implementation

[0025] The embodiments of the present invention will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely illustrative of the present invention and not intended to limit the scope of the invention. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the embodiments of the present invention, and not all structures.

[0026] To facilitate understanding, the main implementation concepts of the various embodiments of the present invention will be briefly described first.

[0027] This invention primarily targets the solar reflectance spectrum of optical scanning radiometers with cold-space observation capabilities. In the visible spectrum, silicon detectors are generally used, as they are widely considered to have stable background output values ​​with minimal variation. Two main methods exist for extracting on-orbit dark background data in the solar reflectance spectrum:

[0028] Existing scheme A: In a complete scan cycle, the optical scanning radiometer performs both Earth observations and deep cold space observations. For meteorological satellites, the time difference between the two is less than 0.1 seconds. Since the cold space field of view points towards deep space, the cold space observation count (SV) can be used as the dark current count under energy-free conditions, barring other optical stray light or contamination. The average of the cold space observation counts within a scan cycle can be used as the dark background value of the Earth observation data for that cycle.

[0029] Existing Solution B: When the optical scanning radiometer is operating on the far side of the Earth, the Earth observation values ​​(EV) are less affected by optical contamination and the values ​​are smaller. Therefore, if there are no other optical strays or contamination, the Earth observation counts can be used as the dark current counts under the instrument's energy-free incident state. The average of the Earth observation counts on the far side of the Earth within one scanning cycle can be used as the dark background value of the Earth observation data for that cycle.

[0030] When performing radiation calibration

[0031] Refi=100cos(θs)Slopei(EVi-SVi)*d2 ;

[0032] Refi is the apparent reflectance at the top of the atmosphere, θs is the solar zenith angle, d is the Earth-Sun distance, Slopei is the calibration coefficient for band i, EVi is the DN value of the observed target, and SVi is the DN value of the cold space observation. Cold space observation refers to the dim or no-light observation of the remote sensor with the cosmic blackbody as the background. At this time, the output signal of the instrument is the reference current value, and the corresponding reflectance should be 0. In the formula, the part of EVi - SVi represents subtracting the DN value of the cold space observation from the DN value of the observed target, so as to obtain the DN value after cold space background correction.

[0033] The above two existing schemes have unstable results and are affected by stray light, etc., and cannot represent the observation background. The dark background (dark current) is determined by the detector and the electronics system itself, and there are changes caused by aging over time, temperature changes, etc.

[0034] In the existing scheme A, the original cold space observation has the influence of stray light reflected from the Earth and other devices entering the cold space field of view, and the pollution influence caused by external targets such as the Moon entering the cold space field of view. Moreover, if the mean value of the cold space observation count values within a scanning period is used, there is a characteristic of periodic fluctuation. If it is selected as the dark background value of the Earth observation data within this period, a fluctuation error will be introduced. (As shown in Figure 5 for the fluctuation of the channel 1 SVi value within 2 days).

[0035] In the existing scheme B, the Earth observation value (EV) of the scanning radiometer is affected by the Earth target, external stellar targets, and scattered light, and the data shows periodic fluctuations. Its fluctuation has a certain relationship with the solar zenith angle and latitude. If an ideal EV is wanted as the dark background value, the EV value must be screened. Screen the EV when the optical scanning radiometer runs to the back of the Earth, but the extraction process requires multiple parameters and complex calculations. As shown in Figure 6 for the fluctuation of the channel 1 EV value within 2 days; Figure 7 is the time series distribution after screening using the solar zenith angle 130 degrees < SolarZ < 140 degrees as the threshold, Figure 8 is the corresponding frequency histogram. And using the new method, the processing process is relatively simple and better processing results can be obtained (as shown in Figure 9 ).

[0036] The present invention aims at the solar reflection spectral band of the optical scanning radiometer with cold space observation, solves the problem of stable extraction of the instrument's dark background information in orbit with long time series and high frequency, avoids the interference effects of the Earth and external stellar targets, etc., and reflects the time change of the instrument state, and better represents the instrument's background state.

[0037] Example 1 Figure 1 This is a flowchart illustrating a method for extracting long-term time-series data from an optical scanning radiometer in orbit against a dark background, as described in Embodiment 1 of the present invention. Figure 1 As shown, Embodiment 1 provides a method for extracting long-term time-series data with a dark background from an optical scanning radiometer in orbit, including:

[0038] Step S100: Acquire cold air observation data and solar zenith angle data from the optical scanning radiometer to generate channel-level cold air observation dataset and channel-level solar zenith angle dataset.

[0039] In one implementation, cold-sky observation data and solar zenith angle data are acquired from an optical scanning radiometer to generate a channel-level cold-sky observation dataset and a channel-level solar zenith angle dataset. The method step S100 provided by this invention includes:

[0040] Step S101: Extract the original cold space observation count values ​​and corresponding solar zenith angle values ​​one by one according to the probe position, and store them as the initial dataset.

[0041] Step S102: Based on the channel information, distinguish the original cold air observation count value and the solar zenith angle value in the initial dataset, and generate channel-level cold air observation data and channel-level solar zenith angle data.

[0042] Step S103: The channel-level cold air observation data are grouped and stored according to the channel identifier to form a channel-level cold air observation dataset.

[0043] Step S104: Group and store the channel-level solar zenith angle data according to the channel identifier to form a channel-level solar zenith angle dataset, and establish the channel-level correspondence between the channel-level cold air observation data and the channel-level solar zenith angle dataset.

[0044] Specifically, the process begins with step S101, which involves reading and matching the original cold space observation count (SV value) and its corresponding solar zenith angle value one by one according to the position of each independent detection unit (detector) of the instrument, and temporarily storing these original data pairs as an initial, undifferentiated set.

[0045] Subsequently, in step S102, the system analyzes and distinguishes the initial dataset based on the channel information (usually corresponding to different spectral bands) to which each probe belongs, separating the mixed data according to channel affiliation, thereby initially generating cold space observation data sequences and solar zenith angle data sequences divided by channel.

[0046] To further optimize the data structure and facilitate independent processing by channel, steps S103 and S104 group and store the differentiated channel-level data. Step S103 categorizes the cold air observation data of each channel using its channel identifier as the keyword, forming a clearly structured channel-level cold air observation dataset. Similarly, step S104 performs the same grouping operation on the solar zenith angle data, forming a channel-level solar zenith angle dataset. Crucially, during this process, the system also establishes and maintains a strict channel-level correspondence between the two datasets, ensuring that cold air observation values ​​and solar zenith angle values ​​from the same channel and the same observation time can be accurately associated and retrieved, providing precise data support for subsequent applications based on various filtering methods such as solar angle.

[0047] Step S200: Based on the channel-level cold air observation dataset and the channel-level solar zenith angle dataset, generate candidate schemes and their selection datasets using multiple screening methods.

[0048] In one implementation, based on the channel-level cold air observation dataset and the channel-level solar zenith angle dataset, multiple screening methods are used to generate candidate schemes and their screening datasets. Step S200 of the method provided by this invention includes:

[0049] Step S201: Obtain the observation count value and the corresponding solar zenith angle value from the channel-level cold air observation dataset and the channel-level solar zenith angle dataset.

[0050] Step S202: For the daily data, the following four independent filtering methods are used in parallel to generate candidate schemes and their daily filtering datasets:

[0051] Angle filtering method: Use a preset solar zenith angle threshold to filter data, take the median of the data that meets the conditions, and generate the first candidate scheme and its filtering dataset;

[0052] Maximum and minimum value selection method: After filtering outliers, extract the minimum and second smallest observations to generate a second candidate scheme and its selection dataset;

[0053] Frequency screening method: Calculate the frequency distribution of observations and extract the maximum and second largest frequency values ​​to generate a third candidate scheme and its screening dataset;

[0054] Proportional screening method: Select data in segments according to the 1%-Pmax proportional threshold and take the median to generate multiple fourth candidate schemes and their screening datasets;

[0055] Among them, Pmax is determined based on the cumulative distribution of channel-level cold air observation data, and can be taken as the maximum value of the cumulative probability that is not greater than the mean of cold air observation.

[0056] Specifically, step S200 is the core data processing step in the method described in this invention. Its purpose is to generate a series of candidate dark background value calculation schemes and their corresponding filtering datasets from the original channel-level cold space observation data (SV) and channel-level solar zenith angle (SolarZ) data through multiple parallel filtering strategies. This step begins with S201, which extracts the count value of each observation point and its corresponding solar zenith angle value from the constructed dataset, preparing the basic data for subsequent filtering. Following this, step S202 processes these data using four independent and parallel filtering methods.

[0057] The first method is the angle-based selection method, which filters data based on a preset solar zenith angle threshold (e.g., 130° to 140°). Its core assumption is that when the sun is within this specific angle range, stray light interference from the sun and earth is minimized, and the cold-space observations obtained at this time best represent the true instrument background. The median of the selected data is then used to generate the first candidate scheme.

[0058] The second method is the minimum value screening method. This method first uses statistical methods (such as the Rhineda criterion, i.e. ±3σ) to filter out obvious outliers, and then extracts the minimum and second minimum values ​​from the remaining "clean" data. The logic is that the dark background signal should be the lowest intrinsic background of the instrument output, thus generating a second candidate scheme.

[0059] The third method is the frequency selection method, which calculates a frequency distribution histogram of all observations and extracts the most frequent (maximum frequency) and second most frequent (second largest frequency) counts. This method is based on the assumption that, under ideal, uncontaminated conditions, the purest dark background value should be the most frequently occurring value, and thus generates a third candidate.

[0060] The fourth method is the proportion selection method, which is a more systematic traversal strategy. It sorts all observed data in ascending order of value, and then starts from 1%, gradually increasing the proportion threshold in fixed steps (e.g., 1%) until a maximum proportion Pmax (e.g., P(SV mean)) is determined based on the cumulative distribution of the data. At each proportion threshold, all data below that threshold are selected and their median is taken, thus generating a series (or more) of fourth candidate schemes.

[0061] By using these four methods with different focuses to perform parallel computation, step S200 finally outputs a set containing multiple candidate schemes and their corresponding datasets, laying a solid foundation for subsequent steps to select the dark background value with the best overall performance (i.e., the value is both low and stable).

[0062] Step S300: For each candidate scheme, based on its selected dataset, calculate the statistical mean and data volatility, and calculate the dark background representative value of the scheme to determine the mean ranking number and volatility ranking number.

[0063] In one implementation, for each candidate scheme, based on its selected dataset, a statistical mean and data volatility are calculated, and a dark background representative value for the scheme is calculated to determine the mean ranking number and volatility ranking number. Step S300 of the method provided by this invention includes:

[0064] Step S301: For each candidate scheme, calculate the statistical mean of the observation counts for that scheme based on its selected dataset.

[0065] Step S302: For each of the candidate schemes, based on its selected dataset, calculate the variance between the observed count value of the scheme and the statistical mean, and generate the data volatility of the scheme.

[0066] Step S303: Sort the statistical means of all candidate schemes in ascending order and determine the sorting number of the mean of each candidate scheme.

[0067] Step S304: Sort the data volatility of all candidate schemes in ascending order and determine the volatility ranking number of each candidate scheme.

[0068] Specifically, for each candidate scheme and its corresponding screening dataset, two basic calculations are first performed (steps S301 and S302): First, the statistical mean of all observation counts in the dataset is calculated to represent the average level of the dark background value suggested by the scheme; second, the variance between these observations and the above statistical mean is calculated to characterize the data volatility of the output results of the scheme. The smaller the variance, the more stable the dark background value extracted by the scheme changes with the time series, and the higher the reliability.

[0069] After completing the basic calculations for all schemes, steps S303 and S304 perform two independent sorting operations: the statistical means of all candidate schemes are sorted in ascending order, and each scheme is assigned a "mean sorting number". The smaller the number, the lower the average level of the dark background value output by the scheme, which is in line with the physical expectation that the dark background value should be as small as possible; at the same time, the data volatility (variance) of all schemes is also sorted in ascending order, and each scheme is assigned a "volatility sorting number". The smaller the number, the more stable the result of the scheme and the less affected it is by random noise and abnormal interference.

[0070] Through this step, each candidate scheme obtains two key ranking numbers. This lays a solid foundation for the subsequent step (S400) to use a weighted algorithm for comprehensive optimization. It abandons the limitations of traditional single indicators (such as only pursuing the minimum value or specific condition value), and aims to screen out high-quality dark background value schemes that can accurately reflect the instrument background and remain robust over long time series by taking into account both "small numerical value" and "low fluctuation".

[0071] Step S400: Based on the mean sorting number and the volatility sorting number, a weighted assignment method is used to determine the preferred dark background value scheme.

[0072] In one implementation, a preferred dark background value scheme is determined using a weighted assignment method based on the mean sorting number and the volatility sorting number. Step S400 of the method provided by this invention includes:

[0073] Step S401: For each candidate scheme, obtain its mean sorting number and volatility sorting number.

[0074] Step S402: Calculate the comprehensive weight value of each candidate scheme using a weighted assignment method, where the comprehensive weight value is expressed by the formula W = a·R. m +b·R v Calculate, R m R represents the sorting index of the mean of the scheme. v This indicates the volatility ranking number of the scheme, and a and b represent the preset weight coefficients.

[0075] Step S403: Sort the comprehensive weight values ​​of all candidate schemes in ascending order.

[0076] Step S404: Select the candidate scheme with the smallest comprehensive weight value as the preferred dark background value scheme.

[0077] Specifically, the system first gathers two key sorting numbers for all candidate solutions: the mean sorting number (Rmean). m ) and volatility sorting number (R v These serial numbers originate from step S300, where the statistical mean of all schemes (representing the magnitude of the dark background value) is sorted in ascending order, with the scheme having the smallest value ranked first (R). m =1), the second smallest ranked second (R) m =2), and so on; similarly, the data volatility (such as variance, representing the degree of data dispersion) of all schemes is also ranked in ascending order, with the most stable scheme ranking first (R²). v =1). That is to say, for an ideal solution, its R = 1. m and R v The serial numbers should be as small as possible.

[0078] Next, the system calculates a comprehensive weight value (W) for each scheme, using the formula W = a·R m +b·R v Here, 'a' and 'b' are pre-set weighting coefficients used to adjust the relative importance of the two objectives, "low numerical value" and "high stability," in the final decision. In a common implementation, if both are considered equally important, a = b = 1 can be set. That is, this formula transforms two ranking numbers with different dimensions and meanings into a single, comparable comprehensive score through weighted summation. The smaller the W value of a solution, the better its overall performance.

[0079] Then, the system sorts all the candidate solutions in ascending order according to the calculated comprehensive weight value W, generating a comprehensive ranking list from best to worst.

[0080] Finally, the candidate scheme ranked first in the list, that is, the one with the smallest comprehensive weight value W, is directly selected and determined as the final "preferred dark background value scheme".

[0081] For example, suppose three candidate options, A, B, and C, are generated for a certain channel. Their mean sorting index (R) is calculated. m The values ​​are 1 (minimum), 2, and 3 respectively; the volatility sorting index (R) v The weights are 2 (second most stable), 1 (most stable), and 3 respectively. If we take a = b = 1, then their combined weight values ​​W are: W A =1*1+1*2=3; W B =1*2+1*1=3; W C =1*3+1*3=6. Although scheme A has the smallest value and scheme B is the most stable, A and B have the same overall score (W=3). The final preferred scheme will be determined between A and B according to specific ranking rules (such as prioritizing the one with the smaller mean when they are tied), but scheme C is excluded because it has the largest W value.

[0082] Based on the above analysis, it can be seen that by introducing a weighted assignment model, the qualitative ranking of the two dimensions of mean and volatility is transformed into a quantitative comprehensive evaluation, thereby scientifically and objectively selecting the best scheme automatically. This ensures that the extracted dark background value sequence has both low value and stable characteristics, providing a reliable data foundation for subsequent radiometric calibration.

[0083] Step S500: Based on the preferred dark background value scheme, generate a long-term series output dataset containing daily dark background values, observation dates, days from launch date, first reference value, and second reference value.

[0084] In one implementation, according to the preferred dark background value scheme, a long-term series output dataset is generated, including daily dark background values, observation dates, days from the launch date, a first reference value, and a second reference value. The method step S500 provided by this invention includes:

[0085] Step S501: Obtain the daily dark background value from the preferred dark background value scheme, and match the corresponding observation date and the number of days since launch using the date association method;

[0086] Step S502: Obtain the screening result corresponding to the 1% ratio screening scheme as the first reference value;

[0087] Step S503: Obtain the filtering result corresponding to the angle filtering scheme as the second reference value;

[0088] Step S504: Combine the daily dark background value, observation date, number of days since launch, first reference value and second reference value to generate a long-term series output dataset.

[0089] Specifically, step S500, based on the aforementioned preferred dark background value scheme, generates a structured long-term time series output dataset. This dataset is organized by day and contains five core fields: daily dark background value, observation date, days until launch, first reference value, and second reference value. The implementation includes four steps: First, daily dark background values ​​are extracted from the determined preferred scheme, and the corresponding observation dates are matched through date association. The number of days until the satellite launch is calculated to establish a time reference. Second, the filtering result obtained using a 1% ratio filtering scheme is used as the first reference value for that day, representing the theoretical lower limit that can be obtained under extremely stringent filtering conditions. Simultaneously, the filtering result of the angle filtering scheme (usually based on the solar zenith angle threshold, such as 130-140 degrees) is used as the second reference value. Finally, the data from the above five dimensions are integrated daily to form a continuous and complete long-term time series dataset. This dataset not only provides daily recommended dark background values ​​but also characterizes a reasonable fluctuation range using a first and second reference value. This effectively supports accurate subtraction and reliability assessment of instrument background signals during radiometric calibration, improving the accuracy and consistency of quantitative processing of remote sensing data. It should be noted that the first and second reference values ​​are reference ranges provided based on different screening methods, used to assist users in evaluating the rationality of the preferred scheme. Angle screening, as a relatively stable method, typically represents an upper limit value with minimal interference, but the preferred scheme may vary depending on channel characteristics.

[0090] In a specific embodiment of the present invention, the method for extracting long-term time-series data from an optical scanning radiometer in orbit against a dark background involves SolarZ and SV data extraction, construction of candidate schemes, selection of candidate schemes, output of the optimal scheme, and comparison of the optimal schemes. Based on FY-3MERSI SV observation data as an example, this invention analyzes how to filter SV observation data to obtain SV values ​​with lower values ​​and smaller long-term series fluctuations, thereby optimizing the calibration effect. The SV data filtering process is as follows: Figure 4 As shown, the main steps of the present invention include:

[0091] Step 1: SolarZ and SV data extraction

[0092] SV raw observation data and SolarZ data were selected according to the position of each of the 25 channels of MERSI.

[0093] Step 2: Construct candidate solutions

[0094] Due to the differences in data, different data filtering methods are suitable for different data. There are four types of SV background data filtering methods: SolarZ angle value filtering, SV maximum value filtering, SV maximum occurrence frequency filtering, and SV sorting proportion filtering.

[0095] SolarZ Angle Value Screening: Assuming that the ideal SV background value should minimize the influence of solar stray light and slow ground reflection, a solar zenith angle threshold of 130 to 140 degrees is used as the screening condition. Data from one day is screened, and then the median of the selected data is taken as the representative SV background value.

[0096] Minimum value screening: Assuming that there are relatively few outliers in the data, the SV baseline value may be the minimum value in the data. After filtering outliers (using the Rhineda criterion to obtain data within three standard deviations (±3σ) of the SV mean), the minimum value in the data can be selected as the representative value of the SV baseline.

[0097] Maximum frequency selection: Assuming the data is relatively pure, the SV background value is very likely to be the value with the highest frequency. Select the value with the highest or second highest frequency in the data as the representative value of the SV background.

[0098] Data is filtered by percentage: Sort the data from smallest to largest, select all data less than a given percentage, with the data extraction percentage ranging from 1% to Pmax, and then take the median of the selected data as the SV background representative value.

[0099] Selection of the percentage threshold Pmax for SV background value data:

[0100] The SV data is grouped into k groups with a DN value interval of 1. The number of DN values ​​in each group is n1, n2, n3...nk.

[0101] Calculate the total number of SV data points SVnum, then the statistical probability of each data point is PP1 = n1 / SVnum, PP2 = n2 / SVnum, PP3 = n3 / SVnum…PPk = nk / SVnum.

[0102] If r represents a DN value in SV and P(r) represents the cumulative probability of a DN value less than or equal to r, then P(r) = PP1 + PP2 + PP3 + ... + PPr.

[0103] Assuming the baseline data of SV should be less than the mean of SV, SvMean, the maximum threshold for data filtering, Pmax, should satisfy Pmax <= P(SvMean). Based on the data from SV channel 1, Pmax is calculated to be 50%.

[0104] The following is a comparison of time series graphs extracted using the above methods, with the data granularity being a representative value calculated for each day:

[0105] P1 represents filtering data by 1% of the total data. P5, P10, P20, P30, and P40 are the same as P1. MAXprobability represents filtering by the highest frequency of occurrence. MAXprobability-1 represents filtering by the second highest frequency of occurrence. Min represents filtering by the lowest value. Min-1 represents filtering by the second lowest value. Angle represents filtering by angle.

[0106] Depend on Figure 9 It is evident that further evaluation of the results of different data screening schemes is needed to obtain the optimal data screening scheme.

[0107] Step 3, SV background screening scheme optimization:

[0108] Further optimization of the different screening schemes constructed in step 2.

[0109] Filter scheme number: Number each data filter scheme accordingly.

[0110] Numbering rules:

[0111] Percentage screening scheme:

[0112] The data extraction percentages range from 1% to Pmax, with 1% intervals, for a total of n (n = max) selection schemes. The selection scheme numbers are as follows:

[0113] Plan(1),Plan(2),…,Plan(max);

[0114] Maximum frequency filtering and second-highest frequency filtering schemes:

[0115] The filtering schemes for the highest and second highest frequency of occurrence are as follows:

[0116] Plan(n+1), Plan(n+2);

[0117] Minimum value filtering and second-smallest value filtering schemes:

[0118] Minimum value filtering and second-smallest value filtering schemes, with the following filtering numbers:

[0119] Plan(n+3), Plan(n+4);

[0120] Angle filtering, angle filtering scheme, filter number as follows:

[0121] Plan(n+5), a total of n+5 plans; taking channel 1 data as an example, the rules are as follows: Plan numbers 1-50 represent filtering according to data percentage (1%-50%). Plan numbers 51-52 represent filtering based on the highest and second highest frequency of occurrence. Plan numbers 53-54 represent filtering based on the lowest and second lowest values. Plan number 55 represents angle filtering.

[0122] Weighted ranking of different schemes: Assuming the SV background count should be sufficiently small, the results after data filtering should have a small statistical mean. Therefore, the evaluation criterion 1 for the filtering scheme is calculated as follows: In n+5 data filtering schemes, for the i-th data filtering scheme (assuming there are n... j There are data points, and each data point is x. i1 ,x i2 ,…,x ij mean μ i The calculation formula is: Assuming that the SV baseline count should be relatively stable, the results after data filtering should exhibit minimal data volatility. Therefore, the evaluation criterion 2 for the filtering scheme is calculated as follows:

[0123] In the n+5 data filtering schemes, the formula for calculating the i-th data filtering scheme is:

[0124] The results of evaluation criteria 1 and 2 are ranked. For different data screening schemes, the results are calculated according to evaluation criteria 1 and 2 respectively, and the calculated results are ranked to obtain the results of the i screening schemes after evaluation criterion 1: μ1, μ2, ..., μ i The results of evaluating the i selection schemes according to criterion 2: Sort the results of evaluation criterion 1 and evaluation criterion 2 separately in ascending order: Sort by evaluation criterion 1, and the resulting sequence numbers are... Sort according to evaluation criterion 2, and the sorted sequence number is as follows.

[0125] Weighted assignment of data filtering schemes: The ranking results of evaluation criteria 1 and evaluation criteria 2 are used as the weight values ​​of the data filtering schemes, and the weight value w is calculated. i Here, we consider that evaluation criteria 1 and evaluation criteria 2 are of equal importance, therefore the weights of screening scheme i are as follows:

[0126] To obtain the optimal data filtering scheme: assign weights w1, w2, ..., w to different schemes. i Sort the data in ascending order to obtain the sorted filter scheme numbers: The optimal selection scheme number is then x. Finally, the weight value w... i Sort the data to obtain the sorted results. Here, a weighted optimal solution is selected for each channel. The results for channel 1 are shown in Table 1:

[0127] Table 1

[0128]

[0129]

[0130]

[0131] The optimal selection schemes for the remaining 19 channels are numbered 13, 5, 19, 5, 8, 10, 7, 15, 3, 13, 9, 2, 4, 9, 5, 3, 21, 9, 3. Due to the different data distribution characteristics of each channel, the optimal scheme after selection is not the same.

[0132] The optimal solution for channel 2 is the filtering solution numbered 5, such as... Figure 10 As shown, the optimal solution for channel 3 is the selection scheme numbered 19, as follows: Figure 11 As shown.

[0133] Step 4, Optimal Scheme Output: The selected scheme at the top of the weighted sort is the optimal scheme. Simultaneously, the 1% ratio selection scheme and the angle selection scheme are chosen as reference values. The output includes the date of the background value, days since the satellite launch, the optimal scheme value, the 1% ratio selection scheme value, and the angle selection scheme value. The 1% ratio selection scheme serves as the first reference value (providing a lower limit), the angle selection scheme as the second reference value (providing an upper limit), and the optimal scheme value as the recommended value. These reference values ​​help understand the possible fluctuation range of the dark background value, but the optimal scheme is the primary basis.

[0134] Step 5, Optimal Solution Comparison: Finally, comparing the optimal screening solution with the angle screening solution, the maximum frequency screening solution, and the minimum value screening solution, it can be seen that the optimal screening solution not only has a lower background value but also has better stability. Comparing the optimal screening solution with the angle screening solution, taking channel 1 data as an example, the mean and variance of the SV optimal solution (97.65 and 6.74) are lower than those of the angle screening solution (97.83 and 6.75), exhibiting lower values ​​and less volatility.

[0135] Comparison of the optimal selection scheme and the maximum frequency selection scheme: Taking channel 1 data as an example, the mean and variance of the SV optimal scheme are (97.65 and 6.74) respectively, which are lower than those of the maximum frequency selection scheme (102.68 and 8.29) and have less volatility.

[0136] Comparison of the optimal screening scheme and the minimum screening scheme: Taking channel 1 data as an example, the mean and variance of the SV optimal scheme (97.65 and 6.74) are better than those of the minimum screening scheme (97.11 and 8.34).

[0137] In summary, the method for extracting long-term time-series data of an optical scanning radiometer in orbit against a dark background, as described in this invention, involves SolarZ and SV data extraction, construction of candidate schemes, selection of candidate schemes, output of the optimal scheme, and comparison of the optimal schemes. The method is simple and ultimately yields a more ideal SV value than before screening, with a lower value and a more stable time series.

[0138] Using FY-3MERSI SV observation data as an example, this paper analyzes how to filter SV observation data to obtain SV values ​​with lower values ​​and more stable long-term series, thereby optimizing the calibration effect. The results are detailed below:

[0139] 1. SV data filtering results:

[0140] Output a daily long-term series relationship graph with a 1% screening scheme as the first reference value, an angle screening scheme as the second reference value, and the optimal data screening scheme value as the preferred value, taking channel 1 as an example. Figure 12 As shown, P1 is the first reference value, angle is the second reference value, and P13 is the preferred value for channel 1.

[0141] 2. Ideal daily SV background data and upper and lower limit reference data for long-term series:

[0142] The SV optimal filtering output data is composed of the 1% screening result as the first reference value, the angle screening result as the second reference value, the optimal data screening result value, the date, and the count value of the days since the launch.

[0143] Table 2 below shows the output data for channel 1:

[0144] Table 2

[0145]

[0146] Example 2, Figure 2 This is a schematic diagram of the structure of an on-orbit dark background long-term data extraction system for an optical scanning radiometer according to Embodiment 2 of the present invention, as shown below. Figure 2 As shown in Embodiment 2, an on-orbit dark background long-term data extraction system for an optical scanning radiometer is provided, comprising: a first generation module, a second generation module, a first determination module, a second determination module, and a third generation module. The first generation module is used to acquire cold-space observation data and solar zenith angle data from the optical scanning radiometer, generating a channel-level cold-space observation dataset and a channel-level solar zenith angle dataset. The second generation module is used to generate candidate schemes and their selected datasets using multiple filtering methods based on the channel-level cold-space observation dataset and the channel-level solar zenith angle dataset. The first determination module is used to calculate the statistical mean and data volatility for each candidate scheme based on its selected dataset, and to calculate the representative dark background value of the scheme, determining the mean ranking number and volatility ranking number. The second determination module is used to determine the preferred dark background value scheme using a weighted assignment method based on the mean ranking number and volatility ranking number. The third generation module is used to generate a long-term output dataset containing daily dark background values, observation dates, days from launch date, a first reference value, and a second reference value, based on the preferred dark background value scheme.

[0147] In this embodiment, the first generation module includes a storage unit, a first generation unit, a forming unit, and an establishment unit. The storage unit is used to extract the original cold-sky observation counts and corresponding solar zenith angle values ​​one by one according to the probe position, and store them as an initial dataset. The first generation unit is used to distinguish the original cold-sky observation counts and solar zenith angle values ​​in the initial dataset according to channel information, and generate channel-level cold-sky observation data and channel-level solar zenith angle data. The forming unit is used to group and store the channel-level cold-sky observation data according to channel identifiers, forming a channel-level cold-sky observation dataset. The establishment unit is used to group and store the channel-level solar zenith angle data according to channel identifiers, forming a channel-level solar zenith angle dataset, and establish the channel-level correspondence between the channel-level cold-sky observation data and the channel-level solar zenith angle dataset.

[0148] In this embodiment, the second generation module includes a first acquisition unit and a second generation unit. The first acquisition unit is used to acquire observation counts and corresponding solar zenith angle values ​​from the channel-level cold air observation dataset and the channel-level solar zenith angle dataset. The second generation unit is used to generate candidate schemes and their daily selection datasets in parallel using the following four independent filtering methods for daily data: Angle filtering method: Data is filtered using a preset solar zenith angle threshold, and the median of the data meeting the conditions is taken to generate a first candidate scheme and its selection dataset. Maximum / minimum value filtering method: After filtering outliers, the minimum and second-smallest observation values ​​are extracted to generate a second candidate scheme and its selection dataset. Frequency filtering method: The frequency distribution of observation values ​​is calculated, and the maximum and second-largest frequency values ​​are extracted to generate a third candidate scheme and its selection dataset. Proportion filtering method: Data is segmented according to a 1%-Pmax proportion threshold, and the median is taken to generate multiple fourth candidate schemes and their selection datasets. Wherein, Pmax is determined based on the cumulative distribution of the channel-level cold air observation data.

[0149] In this embodiment, the first determining module includes: a first calculation unit, a second calculation unit, a first determining unit, and a second determining unit. The first calculation unit is used to calculate the statistical mean of the observed counts for each candidate scheme based on its selected dataset. The second calculation unit is used to calculate the variance between the observed counts and the statistical mean for each candidate scheme based on its selected dataset, generating the data volatility of the scheme. The first determining unit is used to sort the statistical means of all candidate schemes in ascending order and determine the mean ranking number for each candidate scheme. The second determining unit is used to sort the data volatility of all candidate schemes in ascending order and determine the volatility ranking number for each candidate scheme.

[0150] In this embodiment, the second determining module includes a second acquisition unit, a third calculation unit, a sorting unit, and a selection unit. The second acquisition unit is used to acquire the mean sorting number and volatility sorting number for each candidate scheme. The third calculation unit is used to calculate the comprehensive weight value of each candidate scheme using a weighted assignment method, wherein the comprehensive weight value is obtained using the formula W = a·R. m +b·R v Calculate, R m R represents the sorting index of the mean of the scheme. v This indicates the volatility ranking number of the scheme, where 'a' and 'b' represent preset weight coefficients. The ranking unit is used to sort all candidate schemes in ascending order of their comprehensive weight values. The selection unit is used to select the candidate scheme with the smallest comprehensive weight value as the preferred dark background value scheme.

[0151] In this embodiment, the third generation module includes a matching unit, a third acquisition unit, a fourth acquisition unit, and a third generation unit. The matching unit is used to acquire daily dark background values ​​from the preferred dark background value scheme and match the corresponding observation date and the number of days since launch using a date association method. The third acquisition unit is used to acquire the screening result corresponding to the 1% ratio screening scheme as a first reference value. The fourth acquisition unit is used to acquire the screening result corresponding to the angle screening scheme as a second reference value. The third generation unit is used to combine the daily dark background values, observation dates, the number of days since launch, the first reference value, and the second reference value to generate a long-term series output dataset.

[0152] The various variations and specific examples of the on-orbit dark background long-term data extraction method for optical scanning radiometers provided in Embodiment 1 are also applicable to the on-orbit dark background long-term data extraction system for optical scanning radiometers provided in this embodiment. Through the foregoing detailed description of an on-orbit dark background long-term data extraction method for optical scanning radiometers, those skilled in the art can clearly understand the implementation method of the on-orbit dark background long-term data extraction system for optical scanning radiometers in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here.

[0153] Example 3, Figure 3 This is a schematic diagram of the structure of an electronic device according to Embodiment 3 of the present invention, as shown below. Figure 3 As shown, Embodiment 3 also provides an electronic device 300, which may include a processor 301 and a memory 302.

[0154] Memory 302 is used to store programs. Memory 302 may include volatile memory, such as random-access memory (RAM), such as static random-access memory (SRAM), double data rate synchronous dynamic random-access memory (DDR SDRAM), etc.; memory may also include non-volatile memory, such as flash memory. Memory 302 is used to store computer programs (such as application programs, functional modules, etc. that implement the above methods), computer instructions, etc. The computer programs, computer instructions, etc., can be partitioned and stored in one or more memories 302. Furthermore, the computer programs, computer instructions, data, etc., can be accessed by processor 301.

[0155] The aforementioned computer programs and instructions can be stored in one or more partitions of memory 302. Furthermore, the aforementioned computer programs and instructions can be invoked by processor 301.

[0156] The processor 301 is configured to execute the computer program stored in the memory 302 to implement the various steps of the methods described in the above embodiments. For details, please refer to the relevant descriptions in the preceding method embodiments.

[0157] The processor 301 and the memory 302 can be independent structures or integrated structures. When the processor 301 and the memory 302 are independent structures, the memory 302 and the processor 301 can be coupled together via bus 303.

[0158] The electronic device in this embodiment can execute the technical solution in the above method. Its specific implementation process and technical principle are the same, and will not be repeated here.

[0159] Example 4 also provides a computer-readable storage medium including a computer program and instructions, which, when executed on a computer, cause the computer to perform the on-orbit dark background long-term data extraction method of any embodiment of the present invention for optical scanning radiometers.

[0160] Computer-readable storage media include various media that can store program code, such as USB flash drives, external hard drives, ROM, RAM, magnetic disks, or optical disks.

[0161] This embodiment also provides a computer program product, which includes: a computer program stored in a readable storage medium, at least one processor of an electronic device can read the computer program from the readable storage medium, and the at least one processor executes the computer program to cause the electronic device to perform the solution provided in any of the above embodiments.

[0162] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this invention disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this invention can be achieved, and this is not limited herein.

[0163] Note that the above description is merely a preferred embodiment of the present invention and the technical principles employed. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of the present invention, the scope of which is determined by the scope of the appended claims.

Claims

1. A method for extracting long-term time-series data from an optical scanning radiometer in orbit against a dark background, characterized in that, include: Step S100: Acquire cold air observation data and solar zenith angle data from the optical scanning radiometer to generate channel-level cold air observation dataset and channel-level solar zenith angle dataset; Step S200: Based on the channel-level cold air observation dataset and the channel-level solar zenith angle dataset, generate candidate schemes and their selection datasets using multiple screening methods; Step S300: For each candidate scheme, based on its selected dataset, calculate the statistical mean and data volatility, and calculate the dark background representative value of the scheme to determine the mean ranking number and volatility ranking number. Step S400: Based on the mean sorting number and the volatility sorting number, a weighted assignment method is used to determine the preferred dark background value scheme. Step S500: Based on the preferred dark background value scheme, generate a long-term series output dataset containing daily dark background values, observation dates, days from launch date, first reference value, and second reference value.

2. The method for extracting long-term time-series data from an optical scanning radiometer in orbit against a dark background as described in claim 1, characterized in that, The process of acquiring cold-sky observation data and solar zenith angle data from an optical scanning radiometer to generate channel-level cold-sky observation datasets and channel-level solar zenith angle datasets includes: The original cold space observation counts and corresponding solar zenith angle values ​​were extracted one by one according to the probe location and stored as the initial dataset. Based on the channel information, the original cold air observation counts and solar zenith angles in the initial dataset are distinguished to generate channel-level cold air observation data and channel-level solar zenith angle data. The channel-level cold air observation data are grouped and stored according to the channel identifier to form a channel-level cold air observation dataset. The channel-level solar zenith angle data are grouped and stored according to the channel identifier to form a channel-level solar zenith angle dataset, and a channel-level correspondence is established between the channel-level cold air observation data and the channel-level solar zenith angle dataset.

3. The method for extracting long-term time-series data from an optical scanning radiometer in orbit against a dark background as described in claim 1, characterized in that... The process of generating candidate schemes and their selection datasets using multiple screening methods based on the channel-level cold air observation dataset and the channel-level solar zenith angle dataset includes: Obtain the observation count values ​​and corresponding solar zenith angle values ​​from the channel-level cold air observation dataset and the channel-level solar zenith angle dataset; For daily data, the following four independent filtering methods are used in parallel to generate candidate solutions and their daily filtered datasets: Angle filtering method: Use a preset solar zenith angle threshold to filter data, take the median of the data that meets the conditions, and generate the first candidate scheme and its filtering dataset; Maximum and minimum value selection method: After filtering outliers, extract the minimum and second smallest observations to generate a second candidate scheme and its selection dataset; Frequency screening method: Calculate the frequency distribution of observations and extract the maximum and second largest frequency values ​​to generate a third candidate scheme and its screening dataset; Proportional screening method: Select data in segments according to the 1%-Pmax proportional threshold and take the median to generate multiple fourth candidate schemes and their screening datasets; Pmax is determined based on the cumulative distribution of channel-level cold air observation data.

4. The method for extracting long-term time-series data from an optical scanning radiometer in orbit against a dark background as described in claim 1, characterized in that, For each candidate scheme, based on its selected dataset, the statistical mean and data volatility are calculated, and the dark background representative value of the scheme is calculated. The determination of the mean ranking number and volatility ranking number includes: For each of the candidate schemes, the statistical mean of the observation counts for that scheme is calculated based on its selected dataset; For each of the candidate schemes, based on its selected dataset, the variance of the observed counts of that scheme to the statistical mean is calculated to generate the data volatility of that scheme; Sort the statistical means of all candidate options in ascending order and determine the sorting number of the mean for each candidate option. Sort the volatility of all candidate schemes in ascending order and determine the volatility ranking number of each candidate scheme.

5. The method for extracting long-term time-series data from an optical scanning radiometer in orbit against a dark background as described in claim 1, characterized in that, The step of determining the preferred dark background value scheme using a weighted assignment method based on the mean sorting number and the volatility sorting number includes: For each candidate solution, obtain its mean sorting number and volatility sorting number; The weighted average method is used to calculate the comprehensive weight value of each candidate option, where the comprehensive weight value is expressed by the formula W = a·R. m +b·R v Calculate, R m R represents the sorting index of the mean of the scheme. v This indicates the volatility ranking number of the scheme, and a and b represent the preset weight coefficients; Sort all candidate options in ascending order by their combined weight values; The candidate with the smallest overall weight value is selected as the preferred dark background value scheme.

6. The method for extracting long-term time-series data from an optical scanning radiometer in orbit against a dark background as described in claim 1, characterized in that, The step of generating a long-term series output dataset containing daily dark background values, observation dates, days from launch date, a first reference value, and a second reference value, based on a preferred dark background value scheme, includes: The daily dark background value is obtained from the preferred dark background value scheme, and the corresponding observation date and the number of days since the launch date are matched by the date association method. The screening results corresponding to the 1% screening scheme are used as the first reference value; The filtering results corresponding to the angle filtering scheme are used as the second reference value; The daily dark background value, observation date, number of days since launch, first reference value, and second reference value are combined to generate a long-term series output dataset.

7. A system for extracting long-term time-series data from an optical scanning radiometer in orbit against a dark background, characterized in that, include: The first generation module is used to acquire cold air observation data and solar zenith angle data from the optical scanning radiometer, and generate channel-level cold air observation dataset and channel-level solar zenith angle dataset. The second generation module is used to generate candidate schemes and their filtering datasets based on the channel-level cold air observation dataset and the channel-level solar zenith angle dataset using multiple filtering methods. The first determining module is used to, for each of the candidate schemes, calculate the statistical mean and data volatility based on its filtered dataset, calculate the dark background representative value of the scheme, and determine the mean sorting number and volatility sorting number. The second determining module is used to determine the preferred dark background value scheme by using a weighted assignment method based on the mean sorting number and the volatility sorting number. The third generation module is used to generate a long-term output dataset containing daily dark background values, observation dates, days from launch date, a first reference value, and a second reference value, based on the preferred dark background value scheme.

8. The on-orbit dark background long-term data extraction system for an optical scanning radiometer as described in claim 1, characterized in that, The first generation module includes: The storage unit is used to extract the original cold space observation count values ​​and the corresponding solar zenith angle values ​​one by one according to the probe position and store them as the initial dataset; The first generation unit is used to distinguish the original cold air observation count value and the solar zenith angle value in the initial dataset according to the channel information, and generate channel-level cold air observation data and channel-level solar zenith angle data. A forming unit is used to group and store the channel-level cold air observation data according to the channel identifier to form a channel-level cold air observation dataset. A unit is established to group and store the channel-level solar zenith angle data according to the channel identifier, forming a channel-level solar zenith angle dataset, and to establish the channel-level correspondence between the channel-level cold air observation data and the channel-level solar zenith angle dataset.

9. An electronic device, characterized in that, include: At least one processor; as well as A memory that is communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor, which are executed by the at least one processor to enable the at least one processor to perform the on-orbit dark background long-term data extraction method of the optical scanning radiometer according to any one of claims 1-6.

10. A computer-readable storage medium, characterized in that, It includes computer programs and instructions that, when the computer program or the instructions are run on a computer, cause the computer to perform the on-orbit dark background long-term data extraction method for optical scanning radiometers as described in any one of claims 1-6.

Images