Cross-calibration method of low-light channel using moonlight illumination in complete earth shadow area
By cross-calibrating with a stable target illuminated by moonlight in the complete Earth shadow area, combined with the lunar phase model and multi-day observation data, the problem of low accuracy of on-orbit radiation calibration of low-light-level remote sensors was solved, and high-precision low-light-level channel radiation calibration was achieved.
Patent Information
- Application Number
- CN202411891745.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-20
- Publication Date
- 2025-09-16
- Estimated Expiration
- 2044-12-20
AI Technical Summary
The existing on-orbit radiation calibration method for satellite low-light level remote sensors has the problems of large interference from solar stray light, low calibration accuracy and high cost. The existing cross-calibration method has limited frequency and large error.
A stable target in a completely shadowed area illuminated by moonlight is used as the calibration site. A high-precision low-light-level remote sensor is used for cross-calibration. Combining the lunar phase model with multi-day observation data, the cross-calibration coefficient is obtained through spectral and angular correction and linear fitting, thus achieving the radiation calibration of the low-light-level channel.
It effectively reduces the influence of solar stray light, significantly improves the on-orbit radiation calibration accuracy of low-light-level remote sensors, expands the dynamic range of radiance, and improves the applicability and robustness of the calibration method.
Smart Images
Figure CN119828172B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of satellite remote sensing technology and radiation calibration, and in particular to a satellite payload micro-light channel cross-calibration method based on full earth shadow area moonlight illumination observation data. Background Art
[0002] Low-light-level remote sensing can enhance the information collection capabilities of optical remote sensors in low-light conditions such as at night and at dusk. It has applications in monitoring human activity, tracking urbanization and socioeconomic trends, assessing environmental pollution, and analyzing light pollution. Accurate radiometric calibration is crucial for the quantitative application of low-light-level remote sensor Earth observation data. After launch, changes in a satellite's characteristics and the surrounding environment can cause sensor aging and decreased sensitivity, severely impacting its radiometric performance. Therefore, regular on-orbit radiometric calibration of low-light-level remote sensors is essential.
[0003] Currently, the main operational calibration methods for spaceborne low-light-level imagers include: transfer calibration based on pre-launch calibration test data, which introduces significant uncertainty; post-launch field calibration using stable targets in ice, snow, or deserts, but these methods require precise calculations of the radiation transfer process and are subject to model error limitations; and calibration methods based on ground-based active light sources can avoid the reliance on models for stable targets, but are costly and have limited calibration frequency. Cross-calibration is a newer method that offers high calibration frequency, low cost, and can effectively reduce simulation errors, but research on this topic is limited. Summary of the Invention
[0004] The purpose of the present invention is to provide a low-light channel cross-calibration method based on full moonlight illumination in the Earth's shadow area. By using a stable target fully illuminated by moonlight in the Earth's shadow area as a calibration site, the problem of satellite low-light channels being interfered with by solar stray light is solved through cross-calibration with a high-precision low-light remote sensor, thereby improving the on-orbit radiation calibration accuracy of the low-light remote sensor.
[0005] To achieve the above objectives, the present invention proposes a low-light channel cross-calibration method using full moonlight illumination in the shadow area, comprising the following steps:
[0006] S1, selecting stable target observation data of a satellite low-light detection instrument to be calibrated and a reference instrument illuminated by moonlight in a completely shadowed area, wherein the satellite low-light detection instrument to be calibrated and the reference instrument are imaging radiometers;
[0007] S2, removing low-quality data from the observation data of the stable target completely illuminated by moonlight in the shadow area of the satellite to be calibrated and the reference instrument;
[0008] S3, selecting, from the observation data after removing low-quality data, matching data pairs between the to-be-calibrated satellite low-light instrument and the reference instrument under a cross-time matching condition that allows observations under multiple lunar phase conditions;
[0009] S4, calculating lunar radiation based on a lunar phase model, performing spectral and angular corrections on the data of the reference instrument in the matching data pair, and calculating a reference radiance of the instrument to be calibrated for a completely stable target in the shadow area;
[0010] S5, fitting the observation data of the satellite low-light level instrument to be calibrated in the complete earth shadow area under multiple lunar phase conditions in the matching data pair and the calculated reference radiance to obtain a cross-calibration coefficient, thereby completing the radiometric calibration of the satellite low-light level instrument.
[0011] In some optional implementations, the solar zenith angle of the observation data selected in step S1 is greater than a first specific value.
[0012] In some optional implementations, the observation data selected in step S1 further satisfies:
[0013] the lunar zenith angle is less than a second specified value; and / or
[0014] The moon phase angle is within a specific range.
[0015] In some optional implementations, step S2 includes:
[0016] Removing non-uniform pixels such as clouds, auroras, and lightning from the observation data of the satellite low-light instrument to be calibrated with target uniformity; and / or
[0017] The data with a certain number of columns of pixels with low spatial resolution on both sides of the scanning edge are removed from the observation data of the satellite low-light instrument to be calibrated.
[0018] In some optional implementations, a threshold is set by the standard deviation and coefficient of variation of an N×N window to select stable pixels that meet the preset threshold conditions:
[0019]
[0020] Where N is a positive integer, σ and L ave They represent the standard deviation and mean of the radiance of the window pixels respectively, and T is the preset threshold; pixels that do not meet the threshold conditions are removed.
[0021] In some optional implementations, step S4 includes:
[0022] Calculate the lunar spectral irradiance at different times and lunar phases based on the lunar irradiance model;
[0023] performing spectral correction on the observation data of the reference instrument in the matching data pair according to the spectral response functions of the to-be-calibrated satellite low-light instrument and the reference instrument;
[0024] The observation data of the reference instrument in the matching data pair is angle-corrected by using the lunar zenith angle at the observation time of the satellite twilight instrument to be calibrated, the lunar zenith angle at the observation time of the reference instrument, and the sun-moon-earth distance correction factor to obtain the reference radiance of the twilight instrument to be calibrated.
[0025] In some optional implementations, step S5 includes:
[0026] For the multi-month matching data pairs of the satellite low-light instrument to be calibrated and the reference instrument in the screened stable target area, the reference radiance values within a certain pixel area in each month are averaged. Through the linear regression analysis method, the averaged reference radiance of the low-light instrument to be calibrated and the actual observation data DN value are linearly fitted to obtain the radiation calibration coefficient.
[0027] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:
[0028] (1) By selecting a stable target in the complete earth shadow area without stray interference as the calibration site, the influence of solar stray light on the calibration accuracy is effectively reduced.
[0029] (2) Using the low-light channel of a high-precision reference instrument as a reference benchmark, the cross-calibration method is used to significantly improve the on-orbit radiation calibration accuracy of the low-light channel of the instrument to be calibrated.
[0030] (3) Combining multi-day observation data with the lunar irradiance model, the multi-lunar phase reference radiance of the instrument to be calibrated for a stable target in the complete Earth shadow area is obtained, which expands the dynamic range of the radiance of the low-light channel and improves the applicability and robustness of the calibration method. BRIEF DESCRIPTION OF THE DRAWINGS
[0031] Other features, objects and advantages of the present application will become more apparent upon reading the detailed description of non-limiting embodiments with reference to the following drawings:
[0032] Figure 1 The figure is a flow chart of an embodiment of a low-light channel cross-calibration technology using full moonlight illumination in the shadow area.
[0033] Figure 2 A schematic diagram of the satellite imaged earth shadow area provided in an embodiment of the present invention.
[0034] Figure 3 This is an image of low-light-level instrument observation data.
[0035] Figure 4Cross-calibration results for low-light-level instruments. DETAILED DESCRIPTION
[0036] In order to enable those skilled in the art to better understand the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.
[0037] my country has launched the Fengyun-3E (FY-3E), the world's first civilian meteorological satellite in a sun-synchronous dawn-dusk orbit. Its onboard low-light-level medium-resolution spectral imager (MERSI-LL) can acquire low-light images of Earth scenes in the visible spectrum during dawn, dusk, and nighttime, enabling visible light imaging and quantitative remote sensing applications across a wide dynamic range near the terminator. Due to its unique orbital characteristics and the high sensitivity of its low-light-level detector, the low-light-level channel's observation data on the nightside of the terminator are contaminated by stray light, affecting precise cross-calibration.
[0038] To enable on-orbit calibration of low-light-level detectors despite stray light contamination, this paper proposes a cross-calibration method based on moonlight illumination in the full Earth shadow. By utilizing the stable observation area in the full Earth shadow under polar night conditions and combining this cross-calibration technique with other high-precision low-light-level remote sensors, the technique effectively avoids interference from solar stray light and improves the calibration accuracy of the satellite's low-light channel.
[0039] like Figure 1 As shown, the embodiment of the present application provides a low-light channel cross-calibration method using full moonlight illumination in the shadow area, including the following steps:
[0040] S1, select the stable target observation data of the low-light detection instrument of the satellite to be calibrated and the reference instrument in the complete moonlight shadow area.
[0041] In an embodiment of the present invention, the satellite low-light instrument to be calibrated and the reference instrument are imaging radiometers. The satellite low-light instrument to be calibrated can be, for example, the MERSI-LL low-light channel onboard the FY-3E satellite. The reference instrument is also a satellite low-light detector with high-precision on-orbit radiometric calibration. For example, the Day / Night band DNB channel of the Visible Infrared Imaging Radiometer (VIIRS) onboard the National Oceanic and Atmospheric Administration (NOAA-20) satellite can serve as a calibration benchmark for the FY-3E low-light channel. The spectral range (500-900 nm) and spatial resolution of the VIIRS / DNB low-light channel are similar to those of the FY-3E MERSI-LL low-light channel, ensuring that the calibration results of the FY-3E MERSI-LL are not affected by differences in low-light channels. A fully shadow-stabilized target refers to a target whose shadow is stable during the nighttime period. It can be a regional target on the Earth's surface with a certain range. The low-light instrument has uniform and stable radiance within the range of the shadow-stabilized target during the nighttime period. For example, the Antarctic ice target Dome C (center coordinates 74.5°S, 123°E) can be selected as a target in a completely stable shadow area. The formation and imaging area of the shadow area in this area are shown in the figure below. Figure 2 shown. Figure 2 Middle: (a) Schematic diagram of the Earth's shadow area formed based on the positional relationship between the Sun, Earth, and satellites; (b) Schematic diagram of the Earth's shadow area within the imaging area of the low-light instrument.
[0042] The low-light channel detects weak visible light at night. In some embodiments, the solar zenith angle can be restricted to select purely nighttime data to ensure that only the moon provides incident radiation and there is no interference from stray light. Specifically, observation data with a solar zenith angle greater than a first specified value can be selected as stray-free observation data. Alternatively, the lunar zenith angle can be used to filter nighttime moonlight data. In some embodiments, observation data with a lunar zenith angle less than a second specified value can be selected.
[0043] At the same time, nighttime low-light observation data must meet certain lunar incident radiation requirements, with the incident radiation exceeding the detector's signal-to-noise ratio. Because lunar radiation varies periodically with the lunar phase angle, the lunar phase angle can be used as a constraint parameter. In some embodiments, data with lunar phase angles within a specific range can be selected.
[0044] According to the longitude and latitude of the stable target Dome C, the time when the two satellites pass through the Dome C site is obtained through orbit prediction. The L1 remote sensing data file of the corresponding time is selected, and the appropriate angle threshold is found through experiments.
[0045] For example, the screening conditions for the observation data of the twilight channel of the calibrated satellite twilight detection instrument and the reference instrument ("two instruments") in the complete shadow area of the ice and snow target are: (1) the solar zenith angle is greater than 118.4° to ensure that there is no solar radiation and stray light, and the observation data is in polar night conditions; (2) the lunar zenith angle is less than 70° to ensure that the lunar radiation can fully illuminate the surface target; (3) the lunar phase angle is between 2° and 65° to ensure that there is sufficient moonlight radiation at night and lunar radiation data under different lunar phases. Through screening, the complete shadow area observation data files of MERSILL and VIIRS / DNB passing through the Dome C site are obtained. The pixel closest to the longitude and latitude of Dome C in the file is taken as the center pixel, and the 20×20 observation pixels around the center pixel of the two remote sensors are extracted for subsequent processing.
[0046] S2, remove low-quality data from the observation data after screening the stable targets illuminated by moonlight in the complete shadow area of the two instruments.
[0047] In this embodiment, the quality of the observation data of the two instruments can be evaluated according to a pre-set quality evaluation rule, and low-quality data can be removed.
[0048] In some embodiments, the observation data of the satellite low-light instrument to be calibrated can be used to remove non-uniform pixels such as clouds, auroras, and lightning with target uniformity.
[0049] Specifically, because the difference between surface temperature and cloud temperature in polar regions is small, traditional infrared channel thresholding methods are ineffective in identifying non-uniform pixels in polar regions. Therefore, uniformity detection methods are used to remove non-uniform pixels such as clouds, auroras, and lightning. Uniformity detection specifically refers to the uniformity of pixel radiance, which can be measured, for example, by measuring the deviation of radiance from the mean.
[0050] In some optional implementations, the standard deviation and coefficient of variation of the N×N window can be used to select stable pixels that meet the threshold condition of the following formula (1):
[0051]
[0052] Where N is a positive integer, σ and L ave They represent the standard deviation and mean of the radiance of the window pixels respectively, and T is a preset threshold, for example, set to 0.3; the pixels that do not meet the threshold condition of formula (1) are removed.
[0053] For example, within the range of 20×20 pixels, the radiance standard deviation (σ) and mean value (L ave ), set the threshold Select stable pixels with high uniformity that can be retained. Remove observation data that may be contaminated by stray light to ensure data quality.
[0054] In addition, the spatial resolution of the pixels at the scanning edge is reduced, and their quality is not as good as the pixels at the subsatellite point. To ensure the calibration accuracy, a certain number of columns (for example, 200 columns) of pixels on both sides of the scanning edge of the observation data of the satellite low-light instrument to be calibrated are removed.
[0055] The image of the MERSI-LL observation data near the Dome-C area after removing stray light, clouds, edges and selected areas is shown below. Figure 3 As shown in (a)-(d). Figure 3 Middle: (a), (b), and (c) are low-quality observation data after removing stray light pollution, cloud pollution, and scanning edge pixels. (d) is the remaining data used for calibration. The blue box selects the stable target area completely within the shadow area.
[0056] S3, select matching data pairs under the cross-time matching condition that the two instruments allow observations under multiple lunar phase conditions from the observation data after removing the low-quality data.
[0057] The crossover time matching condition defines the permissible time difference between two satellite instruments observing the same stable target. Due to the difference in Earth observation times between the FY-3E and NOAA-20 instruments, the stable target exhibits pseudo-invariance in surface and atmospheric characteristics. Within a certain time difference threshold, atmospheric conditions at Dome C can be considered invariant. Therefore, the time matching constraint for the crossover between the two satellites can be appropriately relaxed. Taking into account the characteristics of the satellite orbits, a time matching threshold of 30 minutes is used to select crossover observations of the Dome C region from the two instruments' respective satellites to obtain matching data pairs. A matching data pair consists of observations from both instruments within a certain time difference range for the stable target, after low-quality data has been removed. Variations in lunar incident radiation caused by this time difference will be corrected using a lunar radiation model in subsequent steps.
[0058] S4, calculate the lunar radiation based on the lunar phase model, perform spectral and angular corrections on the matching data of the reference instrument, and obtain the reference radiance of the instrument to be calibrated for the stable target in the complete shadow area.
[0059] The lunar spectral irradiance at multiple different times and lunar phases can be calculated based on the lunar irradiance model. Then, the observation data of the reference instrument in the matching data pair can be spectrally corrected according to the spectral response functions of the two instruments. The observation data of the reference instrument in the matching data pair can be angle-corrected using the lunar zenith angle at the observation time of the to-be-calibrated satellite twilight instrument, the lunar zenith angle at the observation time of the reference instrument, and the Sun-Moon-Earth distance correction factor to obtain the reference radiance of the to-be-calibrated twilight instrument.
[0060] Specifically, due to the different observed radiance units of MERSI-LL and VIIRS / DNB (hereinafter referred to as VIIRS), it is necessary to convert the units to nw / (cm 2 ·sr). Radiance unit conversion formula:
[0061] For MERSI-LL: L' MERSI-LL =L MERSI-LL ×10 5
[0062] For VIIRS: L' VIIRS =L VIIRS ×10 9
[0063] In an embodiment of the present invention, the ROLO lunar radiation model is selected to calculate the lunar radiation intensity under different time and lunar phase conditions.
[0064] The apparent radiance L of a stable Earth target observed by the low-light channel of a spaceborne imager with only lunar radiation incident l for:
[0065]
[0066] Among them, ρ e is the apparent reflectivity of the Earth observation target, θ m is the lunar zenith angle, d m is the Sun-Moon-Earth distance correction factor, E m is the TOA downlink lunar irradiance, which is calculated as:
[0067]
[0068] Where λ represents the wavelength, λ1 represents the starting wavelength of the spectral response, λ2 represents the ending wavelength of the spectral response, SRF(λ) represents the spectral response function of the load, and I m (λ) is the TOA downlink lunar spectral irradiance derived from the ROLO lunar model at different times. The spectral reflectance of the Dome C target area is relatively flat in the 500-900 nm band, with high reflectivity. Therefore, the variation of reflectivity with wavelength is not considered.
[0069] To fully correct for differences in the observation data between the reference instrument and the instrument being calibrated due to differences in sensor spectral response, atmospheric conditions, and observation geometry, corresponding corrections are required. This embodiment of the application performs cross-calibration based on a stable target, omitting atmospheric correction. This allows the observed atmospheric conditions to be assumed to be consistent and negligible at the time threshold, while also avoiding bias introduced by using inaccurate atmospheric parameters.
[0070] The apparent reflectivity of the Dome C stabilized target has the following relationship with the observation data of the satellite payload:
[0071]
[0072] Among them, ρ DomeC is the apparent reflectivity of the Dome C stabilized target, θ m is the lunar zenith angle, d m is the Sun-Moon-Earth distance correction factor, E m is the TOA downlink lunar irradiance, L l is the apparent radiance of the stable target.
[0073] The surface target reflectivities corresponding to the two satellite payloads VIIRS and MERSI-LL are:
[0074]
[0075] in, is the reflectivity of the surface target observed by VIIRS, θ m-V is the lunar zenith angle at the time of VIIRS observation, d m-V is the Sun-Moon-Earth distance correction factor for the VIIRS observation time, E m (λ V ,t V ) is based on the VIIRS observation time t V and the spectral response function for a specific wavelength λ V Calculated TOA downlink lunar irradiance, L VIIRS VIIRS observed radiance; is the reflectivity of the surface target observed by MERSI-LL, θ m-M is the lunar zenith angle at the time of MERSI-LL observation, d m-M is the Sun-Moon-Earth distance correction factor for the MERSI-LL observation time, E m (λ M ,t M ) is based on the MERSI-LL observation time t M and the spectral response function for a specific wavelength λ M Calculated TOA downlink lunar irradiance, L MERSI-LL is the MERSI-LL observation radiance.
[0076] The Dome C ice and snow target has a highly reflective, stable, and uniform reflectivity. It can be approximately considered a Lambertian reflector in the southern hemisphere from May to July and in the visible light detection band. Based on the fact that the reflectivity of the same stable target observed by VIIRS and MERSI-LL is the same, the radiation reference transfer formula between cross-calibrations can be obtained as follows:
[0077]
[0078] Among them, L' MERSI-LL is the MERSI-LL reference radiance derived from the VIIRS radiance.
[0079] S5, fit the observation data of the low-light-level instrument of the target satellite in the complete Earth shadow area under multiple lunar phase conditions in the matching data pair with the calculated reference radiance to obtain the cross-calibration coefficient and complete the radiometric calibration of the low-light-level instrument of the target satellite.
[0080] The MERSI-LL data were calibrated using the VIIRS radiance as a reference. For the multi-month matched data pairs of two instruments in the selected stable target area, the reference radiance values within the stable target range of a certain pixel area (e.g., a 20×20 pixel area) were averaged for each monthly matched data pair. Using a linear regression analysis method, a linear fit was performed between the averaged reference radiance of the MERSI-LL and the actual observed data DN value to obtain the radiation calibration coefficient. The linear fit can be achieved using the following regression equation (7):
[0081] L' MERSI-LL =a×DN MERSI-LL +b (7)
[0082] Among them, DN MERSI-LL represents the original MERSI-LL DN observations; a and b represent the slope and intercept of the linear fit, respectively.
[0083] Using the above method of the embodiment of the present application, the cross-matching data pairs of the two instruments MERSI-LL and VIIRS / DNB that meet all the constraints are obtained for a total of 6 lunar cycles. There is only one lunar cycle in 2021, the first year of launch, and 2023 and 2024 each year include 2 to 3 lunar cycles in the complete shadow of Dome C in Antarctica; the data of each cycle are distributed 4 to 5 days before and after the full moon on the 15th day of the lunar calendar; the specific ranges corresponding to these 6 lunar cycles are July 19 to 29, 2021, May 14 to 21, June 10 to 19, July 8 to 18, 2022, May 29 to June 9, and June 27 to July 7, 2023.
[0084] The calibration is performed using multiple lunar phase data of each lunar cycle. The final calibration results for the six lunar cycles are as follows: Figure 4 shown. Figure 4 Middle: (a) to (f) represent different lunar cycles.
[0085] The Pearson correlation coefficient (R 2 ) are all above 0.98, Figure 4(a) Taking July 2021 as an example, the calibration formula is y = 0.1119x-5.264, and the calibration coefficients are 0.1119 and -5.264 respectively. Compared with the official business calibration results, the method of the embodiment of the present application achieved a calibration effect that is more consistent with VIIRS, with an average relative deviation of less than 10%, and has a high calibration accuracy. This method can be used for absolute calibration and calibration result verification of the FY3E low-light channel affected by stray light, and can track the stability of the subsequent long-term radiation response of the instrument based on a stable target.
Claims
1. A cross-calibration method for low-light channels using full moonlight illumination in the shadow area, characterized in that: The steps include: Step 1: Select stable target observation data of the satellite low-light instrument to be calibrated and the reference instrument under moonlight in a completely shadowed area, wherein the satellite low-light instrument to be calibrated and the reference instrument are imaging radiometers; Step 2: removing low-quality data from the observation data of the stable target in the complete moonlight shadow area of the calibrated satellite low-light instrument and the reference instrument; Step 3, selecting, from the observation data after removing low-quality data, matching data pairs between the to-be-calibrated satellite low-light instrument and the reference instrument under the cross-time matching constraint condition of observations allowed under multiple lunar phase conditions; Step 4: Calculate lunar radiation based on the lunar phase model, perform spectral and angular corrections on the data of the reference instrument in the matching data pair, and obtain the reference radiance of the instrument to be calibrated with the stable target in the complete shadow area; Step 5: Fit the observation data of the satellite low-light instrument to be calibrated in the complete earth shadow area under multiple lunar phase conditions in the matching data pair and the calculated reference radiance to obtain the cross-calibration coefficient and complete the radiation calibration of the satellite low-light instrument.
2. The method according to claim 1, characterized in that The solar zenith angle of the observation data selected in step 1 is greater than a first specific value.
3. The method according to claim 2, characterized in that The observation data selected in step 1 also meet the following requirements: the lunar zenith angle is less than a second specified value; and / or The moon phase angle is within a specific range.
4. The method according to claim 1, wherein The step 2 includes: Removing non-uniform pixels such as clouds, auroras, and lightning from the observation data of the satellite low-light instrument to be calibrated with target uniformity; and / or The data with a certain number of columns of pixels with low spatial resolution on both sides of the scanning edge are removed from the observation data of the satellite low-light instrument to be calibrated.
5. The method according to claim 4, characterized in that The non-uniform pixels are removed in the following manner: Through the standard deviation and coefficient of variation of the N×N size window, stable pixels that meet the preset threshold conditions are selected. Where N is a positive integer, σ and L ave They represent the standard deviation and mean value of the radiance of the window pixels respectively, and T is the preset threshold; Pixels that do not meet the threshold condition are removed.
6. The method according to claim 1, characterized in that The step 4 comprises: Calculate the lunar spectral irradiance at different times and lunar phases based on the lunar irradiance model; performing spectral correction on the observation data of the reference instrument in the matching data pair according to the spectral response functions of the to-be-calibrated satellite low-light instrument and the reference instrument; The observation data of the reference instrument in the matching data pair is angle-corrected by using the lunar zenith angle at the observation time of the satellite twilight instrument to be calibrated, the lunar zenith angle at the observation time of the reference instrument, and the sun-moon-earth distance correction factor to obtain the reference radiance of the twilight instrument to be calibrated.
7. The method according to claim 1, characterized in that The step 5 comprises: For the multi-month matching data pairs of the satellite low-light instrument to be calibrated and the reference instrument in the screened stable target area, the reference radiance values in a certain pixel area of each lunar phase are averaged. Through the linear regression analysis method, the averaged reference radiance of the low-light instrument to be calibrated and the actual observation data DN value are linearly fitted to obtain the radiation calibration coefficient.