Cloud detection method suitable for FY-3 E star low-light imager

Abstract

The invention discloses a cloud detection method suitable for a FY-3 E star low-light imager. The method comprises the following steps: obtaining observation data of the low-light imager in a target observation area; constructing a cloud detection criterion; according to the cloud detection criterion, a same lighting condition reflectance comparison test and a flare area reflectance comparison test are designed, and whether each pixel in a target observation area is a low-cloud pixel or not is judged one by one based on obtained observation data of a low-light imager in the target observation area. Therefore, the low clouds in the morning and evening period can be effectively identified, and the cloud detection precision is improved.

Description

Technical Field

[0001] The present invention relates to a cloud detection method applicable to the low-light imager of Fengyun-3E satellite, belonging to the technical field of remote sensing satellite cloud detection data. Background Art

[0002] The detection instruments carried by meteorological satellites can measure the electromagnetic radiation emitted by the earth and the atmosphere. Under clear sky conditions, the radiance measured by meteorological satellites is a function of atmospheric temperature, humidity, and absorption gas profiles, as well as the earth's surface temperature and emissivity. Under cloudy conditions (especially under precipitation conditions), the radiation emitted by the atmosphere below the cloud layer will be absorbed by the cloud layer and cannot reach the satellite. At this time, for the inversion processes of atmospheric and surface variables (such as total precipitable water, temperature and humidity profiles, land or ocean surface temperature, snow / ice cover) and cloud properties (such as cloud top phase, height, particle size, etc.), it is necessary to first distinguish clear sky observations from cloudy observations. On the other hand, existing radiative transfer models can simulate the radiance under clear sky conditions relatively accurately, but there is still a large uncertainty in simulating the radiance under cloudy conditions. At present, the cloud area radiative data assimilation technology in numerical weather prediction is not yet mature, and in practice, satellite observations contaminated by clouds are excluded first. In summary, cloud detection is required for both the inversion of geophysical parameters and the assimilation of satellite data.

[0003] A few years ago, our team developed a cloud detection technology based on the optical imager of geostationary meteorological satellites. Zhuge&Zou (2016) and Li et al. (2020) constructed criteria based on the differences in cloud sensitivity and spatio-temporal characteristics of different channels and established a pure infrared cloud detection algorithm. When dealing with cloud detection at dawn and dusk, it is mainly based on the assumption that the infrared spectrum of clouds changes little within 10 - 15 minutes, but this assumption is not very applicable to fast-moving cloud clusters. Zhuge et al. (2017) established a fast cloud detection algorithm that only includes visible light channels. Although its operation efficiency is very high, it is not applicable to night or dawn and dusk periods. Geostationary meteorological satellites are fixed at a certain position above the equator and remain unchanged relative to the earth's position. The effective observation area is limited to within about 60° around the sub-satellite point and cannot achieve global observation.

[0004] In the process of global weather and environmental monitoring, polar-orbiting meteorological satellites have played an important role. These polar-orbiting meteorological satellites include Fengyun-1 (the first generation) and Fengyun-3 (the second generation) of China, the NOAA series of the United States, and the METOP series of Europe. At present, the overpass times of polar-orbiting meteorological satellites in Europe and the United States are mainly around 1:30 local time (afternoon satellite) or 9:30 (morning satellite), lacking observations of the atmospheric and cloud states during the dawn and dusk phases.

[0005] On July 5, 2021, China launched the fifth satellite of the second-generation polar-orbiting meteorological satellites, namely Fengyun-3E. Fengyun-3E is the world's first civil meteorological satellite in the dawn-dusk orbit. Its unique orbital position, advanced detection technology, and rich observation data enhance the performance of the global meteorological forecasting system in multiple aspects, especially in improving the prediction ability and refinement of extreme weather events. The medium-resolution spectral imager on Fengyun-3E, the low-light type (or directly called "low-light imager"), has 1 low-light channel and 6 infrared channels. How to make good use of the low-light imager and strengthen the monitoring of clouds during the dawn-dusk period is worthy of research.

[0006] As is well known, the solar radiation process during the dawn-dusk stage is relatively complex. At this time, due to the existence of the curvature of the Earth's atmosphere, the assumption of a plane-parallel atmosphere for radiation transfer is no longer satisfied, and the processes of refraction, scattering, and absorption need to be re-described. The cloud detection criteria based on the visible light threshold (applicable to daytime) and the criteria based on the infrared brightness temperature difference (only applicable when there is no solar radiation influence at night) are no longer applicable to the detection of low clouds. To strengthen the utilization rate of the global observation data during the dawn-dusk period of Fengyun-3E and improve the inversion accuracy of geophysical parameters and the assimilation accuracy of satellite data based on various instruments of Fengyun-3E, it is necessary to establish a reliable and feasible new cloud detection scheme applicable to the low-light imager of Fengyun-3E. Summary of the Invention

[0007] In view of the deficiencies of the prior art, the present invention proposes a cloud detection method applicable to the low-light imager of Fengyun-3E, aiming to improve the cloud detection accuracy of the low-light imager of Fengyun-3E. For this purpose, the present invention designs a reflectivity contrast test under the same illumination conditions (SI_RCT) and a reflectivity contrast test in the flare area (SG_RCT). Compared with the prior art, the present invention provides algorithm support for the subsequent inversion of geophysical parameters and the assimilation of satellite data based on various instruments of Fengyun-3E, can effectively identify low clouds during the dawn-dusk period, improves the cloud detection accuracy, and ultimately improves the level of weather monitoring and forecasting.

[0008] To achieve the above technical objectives, the present invention will adopt the following technical solutions:

[0009] A cloud detection method applicable to the low-light imager of Fengyun-3E, comprising:

[0010] Obtaining the observation data of the low-light imager in the target observation area;

[0011] Constructing a cloud detection criterion, and based on the observation data of the low-light imager obtained in the target observation area, judging one by one whether each pixel in the target observation area is a low-cloud pixel; when any pixel in the target observation area satisfies:

[0012] Or

[0013] When it is, it indicates that the corresponding pixel is a low cloud pixel;

[0014] In the formula: M ch1 represents the observed reflectance value, which is obtained by observing through the low-light channel of the low-light imager;

[0015] represents all the observed reflectance values with the same solar zenith angle θ sz and excluding the influence of solar flares and the area where M ch6 > 250K;

[0016] M ch6 represents the observed brightness temperature in the infrared window region, which is obtained by observing through the infrared window region channel of the low-light imager;

[0017] n represents the total number of all the observed reflectance values with the same solar zenith angle θ sz and excluding the influence of solar flares and the area where M ch6 > 250K;

[0018] P90(·) and P10(·) respectively represent the 90th and 10th percentiles;

[0019] represents all the reflectance values with the same solar flare angle θ sg and excluding land and the area where M ch6 > 250K; m represents the total number of all the observed reflectance values with the same solar flare angle θ sg and excluding land and the area where M ch6 > 250K.

[0020] Preferably, it further includes obtaining the output result of the radiative transfer model and interpolating the obtained output result of the radiative transfer model onto each pixel of the target observation area; the output result of the radiative transfer model is the output obtained by simulating the observation of the low-light imager in the target observation area by the radiative transfer model;

[0021] The cloud detection criterion determines whether each pixel in the target observation area is a low cloud pixel according to the obtained output result of the radiative transfer model and the observation data of the low-light imager in the target observation area; when any pixel in the target observation area satisfies:

[0022] or M ch2 -S ch2 < ε SG_DLS When it is, it indicates that the corresponding pixel is a low cloud pixel;

[0023] In the formula: R 3.8μm(T) represents the emissivity at a wavelength of 3.8 μm calculated from the temperature T according to the Planck function;

[0024] ε ULST is the preset ULST test threshold; M ch2 represents the bright temperature of short-wave infrared observation, obtained from the observation of the short-wave infrared channel of the low-light imager in the target observation area; M ch6 represents the bright temperature of infrared window area observation, obtained from the observation of the infrared window area channel of the low-light imager in the target observation area; S ch2 represents the simulated bright temperature of short-wave infrared clear sky, obtained by simulating the observation of the short-wave infrared channel of the low-light imager in the target observation area through the radiative transfer model; S ch6 represents the simulated bright temperature of infrared window area clear sky, obtained by simulating the observation of the infrared window area channel of the low-light imager in the target observation area through the radiative transfer model; ε SG_DLS represents the preset SG_DLS test threshold.

[0025] Preferably, the input of the radiative transfer model is the surface infrared emissivity, atmospheric temperature and humidity, and surface elements within the observation range extracted from the forecast field data.

[0026] Preferably, when the cloud detection criterion judges each pixel in the target observation area, when any pixel in the target observation area simultaneously satisfies:

[0027] M ch7 -M ch6 >0 and (M ch6 -M ch2 +5) / 10 - (M ch7 -M ch6 +4) / 6 < 0.16, it indicates that the corresponding pixel is a low cloud pixel;

[0028] In the formula: M ch7 represents the bright temperature of infrared dirty window observation, obtained from the observation of the infrared dirty window channel of the low-light imager in the target observation area.

[0029] Preferably, when the cloud detection criterion judges each pixel in the target observation area, when any pixel in the target observation area satisfies:

[0030] or

[0031] ρ(M ch6 , M ch4 ) > 0.7 or M ch2 -M ch7 > ε N-OTC when, it indicates that the corresponding pixel is a cirrus cloud pixel;

[0032] In the formula: is the preset first PFMFT test threshold; S ch7 represents the infrared dirty window clear sky simulated brightness temperature, obtained by observing the infrared dirty window channel of the low-light imager in the target observation area through a radiative transfer model; ρ(M ch6 ,M ch4 ) represents the observed brightness temperature M of the infrared window area channel of any pixel ch6 and the observed brightness temperature M of the water vapor infrared channel ch4 correlation coefficient; M ch4 represents the water vapor infrared observed brightness temperature, obtained by observing the water vapor infrared channel of the low-light imager in the target observation area; ε N-OTC is the preset N-OTC test threshold.

[0033] Preferably, when the cloud detection criterion judges each pixel in the target observation area, if any pixel in the target observation area satisfies:

[0034] or it indicates that the corresponding pixel is a cloudy pixel;

[0035] In the formula: is the maximum observed brightness temperature observed through the infrared window area channel by any pixel in the preset neighborhood window B in the observation area; σ z is the terrain height standard deviation within the neighborhood window B, calculated based on the geographic information data read by the low-light imager; γ represents the temperature lapse rate; ε RTCT is the preset RTCT test threshold;

[0036] R2(T) represents the infrared window area channel emissivity calculated from the temperature T according to Planck's law; is the radiation emitted by the atmosphere above the tropopause and above, calculated based on the atmospheric temperature and humidity profile of the forecast field and the atmospheric layer optical thickness; ε ETROP is the preset ETROP test threshold.

[0037] Preferably, when the cloud detection criterion judges each pixel in the target observation area, if any pixel in the target observation area satisfies:

[0038] or it indicates that the corresponding pixel is a cloudy pixel;

[0039] In the formula: respectively represent the brightness temperatures observed through the infrared window area channel and the infrared dirty window channel by the pixel at the NWC position; the NWC position represents the adjacent warm center of any pixel in the observation area, and represents the warmest pixel of any pixel in its preset neighborhood window A; is the preset second RFMFT test threshold;

[0040] Indicates the correction value of the low-light channel observation, obtained by correcting the solar altitude angle; Indicates the minimum value of the correction value of the low-light channel observation of any pixel in the preset neighborhood window D.

[0041] Another technical object of the present invention is to provide an electronic device, including a memory, a processor, and a computer program stored on the memory and running on the processor, and the computer program runs to execute the above-mentioned cloud detection method applicable to the low-light imager of Fengyun-3 E satellite.

[0042] Based on the above technical objects, compared with the prior art, the present invention has the following advantages:

[0043] 1. In the cloud detection criterion of the cloud detection method of the present invention, a reflectivity contrast test under the same illumination condition (SI_RCT) and a reflectivity contrast test in the flare area (SG_RCT) are specifically designed. Compared with the prior art, the new scheme can effectively identify low clouds during the dawn and dusk periods and improve the cloud detection accuracy.

[0044] 2. The cloud detection result obtained by using the cloud detection method of the present invention is closer to the true value. Based on the comparison chart of the observation and cloud detection result of the low-light imager of Fengyun-3 E satellite at 07:25 UTC on June 11, 2023 ( Figure 2 ), it can be found that there is a large difference between the official cloud detection result of the National Satellite Meteorological Center (prior art) and the cloud distribution map of the NJIAS HCFD dataset (representing the true value result); the cloud detection result of the new scheme is in good agreement with the cloud distribution map of the NJIAS HCFD dataset.

[0045] 3. The cloud detection method of the present invention significantly improves the hit rate of detecting broken clouds and water clouds. Whether on the ocean surface or on land, an existing cloud detection scheme and the cloud detection method of the present invention can effectively identify ice clouds and convective clouds, and the hit rate is close to 100%. However, the existing cloud detection scheme has poor effects in detecting broken clouds and water clouds, and the present invention significantly improves the hit rate of detecting water clouds. Especially on land, the present invention increases the hit rate of identifying water clouds from less than about 20% in the prior art to more than 60%, as shown in Figure 3 .

[0046] 4. Evaluating the cloud detection result with the NJIAS HCFD dataset as the true value, the present invention obtains a higher technical score. Whether it is during the day (including dawn and dusk) or at night, the technical score of the cloud detection result of the present invention is significantly higher than that of the existing cloud detection scheme, as shown in Figure 4 . On the ocean surface, the improvement range of the cloud detection technical score is about 25%; on land, the improvement range of the cloud detection technical score exceeds 50%.

[0047] 5. The improvement of the cloud detection accuracy of the FY-3E satellite's low-light imager provides algorithmic support for subsequent inversion of geophysical parameters and satellite data assimilation based on various instruments of the FY-3E satellite, and improves the level of weather monitoring and forecasting. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 is a schematic flow chart of the cloud detection method applicable to the low-light imager of the FY-3E satellite according to the present invention;

[0049] Figure 2 is an effect diagram of cloud area and clear sky division for a single observation of the low-light imager of the FY-3E satellite based on the cloud detection method of the present invention;

[0050] Figure 3 is a comparison chart of hit rates obtained by the cloud detection method of the present invention for different cloud types compared with the prior art.

[0051] Figure 4 is an objective evaluation result chart of the cloud detection method of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0052] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. The description of at least one exemplary embodiment is actually only illustrative and in no way limits the present invention and its application or use. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention. Unless otherwise specifically stated, the relative arrangements, expressions, and numerical values of the components and steps described in these embodiments do not limit the scope of the present invention. Technologies, methods, and devices known to those of ordinary skill in the relevant art may not be discussed in detail, but should be regarded as part of the specification when appropriate. In all the examples shown and discussed here, any specific value should be construed as merely exemplary and not as a limitation. Therefore, other examples of the exemplary embodiments may have different values.

[0053] As Figure 1 shown, the cloud detection method applicable to the low-light imager of the FY-3E satellite according to the present invention includes:

[0054] The first step is to decode the observation data of the low-light imager.

[0055] The observation of the low-light imager every 5 minutes is defined as one scene. The geographical information (longitude, latitude, altitude, surface type) and observation geometric information (observation zenith angle, observation azimuth angle, solar zenith angle, solar azimuth angle, solar flare angle) of each scene are read, as well as the measurement values of 7 channels, including 1 low-light channel and 6 infrared channels. The microscopic channel can observe the reflectance of the low-light channel, and the 6 infrared channels can observe the brightness temperature measurement values. Among them, the 7 channels are sequentially denoted as channel 1 to channel 7. Among them, channel 1 is denoted as the low-light channel, and the center wavelength can be selected as 0.7μm, and its observation is denoted as the reflectance observation value M ch1 ; Channel 2 is denoted as the short-wave infrared channel, and the center wavelength can be selected as 3.8μm, and its observation is denoted as the short-wave infrared observation brightness temperature M ch2 ; Channel 3 is denoted as the mid-infrared channel, and the center wavelength can be selected as 4.05μm; Channel 4 is denoted as the water vapor infrared channel, and the center wavelength can be selected as 7.2μm, and its observation is denoted as the water vapor infrared observation brightness temperature M ch4 ; Channel 5 is denoted as the infrared phase channel, and the center wavelength can be selected as 8.55μm; Channel 6 is denoted as the infrared window channel, and the center wavelength can be selected as 10.8μm, and its observation is denoted as the infrared window observation brightness temperature M ch6 ; Channel 7 is denoted as the infrared dirty window channel, and the center wavelength can be selected as 12.0μm, and its observation is denoted as the infrared dirty window observation brightness temperature M ch7 .

[0056] The radiation indexes of each channel of the low-light imager on Fengyun-3 E satellite (abbreviation: MERSI-LL) refer to Table 1 below:

[0057] Table 1

[0058]

[0059] Second step, brightness temperature observation and auxiliary information data processing.

[0060] According to the measurement values of the 7 channels read, calculate the local correlation coefficient of the brightness temperature of the low-light channel (channel 4) and the infrared window channel (channel 6) at each pixel, and find the position of the neighboring warm center (NWC). The neighboring warm center NWC is defined as the warmest pixel of the target pixel in the preset neighborhood window A (that is, the 21×21 pixel box around the target pixel), which represents the position of the minimum optical thickness in the local area (clear sky or thin cloud). According to the geographical information data read, calculate the local standard deviation of the altitude at each pixel.

[0061] Third step, GFS forecast field data preprocessing.

[0062] Extract parameters such as atmospheric temperature and humidity and surface elements from the global GFS forecast field data with a resolution of 0.25°. Obtain the northernmost latitude, southernmost latitude, westernmost longitude, and easternmost longitude according to geographical information, and extract the surface infrared emissivity, atmospheric temperature and humidity, and surface elements within the observation range.

[0063] Step 4, simulation of clear-sky brightness temperature of the low-light imager.

[0064] Based on the surface infrared emissivity, atmospheric temperature and humidity, and surface elements, the FY-RTM radiative transfer model is used to simulate the brightness temperature of 6 infrared channels and the atmospheric layer optical thickness.

[0065] Step 5, secondary processing of the brightness temperature simulation results.

[0066] Interpolate the simulated brightness temperature and the total atmospheric optical thickness to each pixel of the low-light imager.

[0067] Step 6, cloud detection:

[0068] To achieve cloud detection, the present invention constructs a total of 12 sets of criteria and divides them into three categories, which respectively belong to the cirrus cloud recognition algorithm, the low cloud recognition algorithm, and the ordinary cloud recognition algorithm. Among them:

[0069] The cirrus cloud recognition algorithm includes 3 sets of criteria, corresponding to: Split window test 1 (the first PFMFT test) at 11 and 12 μm, Cirrus cloud water vapor test (CIRH2O test), New optically thin cloud test (N-OTC test);

[0070] The low cloud recognition algorithm includes 5 sets of criteria, corresponding to: Uniform layer cloud test (ULST test), Desert night low cloud test (DZT_NLS test), Reflectivity contrast test under the same illumination condition (SI_RCT test), Low cloud test in the flare area (SG_DLS test), Reflectivity contrast test in the flare area (SG_RCT test);

[0071] The ordinary cloud recognition algorithm includes 4 sets of criteria, corresponding to: Relative thermal contrast test (RTCT test), Tropopause emissivity test (ETROP test), Relative visible light contrast test (RVCT test), Split window test 2 (the second RFMFT test) at 11 and 12 μm.

[0072] For each pixel, perform 12 sets of tests respectively. If any one of the 12 criteria results in a cloudy condition, the final determination result of this pixel is cloudy; otherwise, it is clear sky.

[0073] The explanations of the 12 tests are as follows:

[0074] (1) Relative thermal contrast test (RTCT):

[0075] Analyze whether the corresponding target pixel is in a cloud area by analyzing the local difference in the brightness temperature of the infrared window channel (channel 6) of the target pixel. If the local difference in the brightness temperature of the infrared window channel of the target pixel satisfies it indicates that the target pixel is affected by clouds. Among them, M ch6 is the observed brightness temperature of the infrared window channel, is the maximum observed brightness temperature of the infrared window channel within the preset neighborhood window B (i.e., the 3×3 pixel frame around the target pixel), σ z is the standard deviation of the terrain height within the neighborhood window B, γ represents the temperature lapse rate, which is a constant of 7 K / km -1 , ε RTCT is the threshold of the RTCT test, with a value of 3.2 K at sea and 4.1 K on land.

[0076] (2) Tropopause Emissivity Test (ETROP):

[0077] Assume that under clear-sky conditions with clouds, the observed brightness temperature of the infrared window channel is much lower than the simulated clear-sky temperature. If a pixel satisfies it will be determined to have clouds. Among them, R2(T) represents the emissivity of the infrared window channel calculated from the temperature T according to Planck's law, S ch6 is the simulated clear-sky brightness temperature of the infrared window channel, is the radiation emitted by the clouds located at the tropopause height and the atmosphere above it. ε ETROP is the threshold of the ETROP test, with a value of 0.1 at sea and 0.3 on land.

[0078] (3) 11 and 12 μm Split Window Test One (PFMFT):

[0079] Judge whether the pixel is a semi-transparent cloud according to whether the brightness temperature difference (M ch6 -M ch7 ) is relatively large. If a pixel satisfies it will be determined to have clouds. Among them, S ch6 and S ch7 are the simulated clear-sky brightness temperatures of channels 6 and 7 respectively. is the threshold of PFMFT, with a value of 0.8 at sea and 2.5 on land.

[0080] (4) 11 and 12 μm Split Window Test Two (RFMFT):

[0081] If there is a large M ch6 -M ch7 deviation between the target pixel and the pixel at the NWC position, it indicates the presence of clouds. If the target pixel satisfies it will be determined to have clouds. Among them, |·| represents the absolute value, and are the brightness temperature observation values of the infrared window channel and the infrared dirty window channel of the pixels at the NWC positions, respectively. is the second threshold of RFMFT, with a value of 0.7K over the sea and 1.0K over land.

[0082] (5) Cirrus water vapor test (CIRH2O):

[0083] High clouds are detected through the correlation between channel 6 and channel 4. If a pixel satisfies ρ(M ch6 , M ch4 ) > 0.7, it will be marked as cloudy. Among them, ρ(M ch6 , M ch4 ) represents the correlation coefficient of the brightness temperatures of channel 4 and channel 6 for each pixel within a 5×5 pixel box.

[0084] (6) Uniform low cloud test (ULST):

[0085] It is considered that the emissivity of low clouds is lower than that of the surface. Based on this consideration, if a pixel satisfies it will be marked as fog or low cloud. Among them, R1(T) represents the radiance of the shortwave infrared channel calculated from temperature T according to the Planck function. is the observed emissivity of the shortwave infrared channel, is the estimated value of the emissivity of the shortwave infrared channel under clear sky conditions, ε ULST is the threshold of ULST, with a value of 0.04 over the sea and 0.09 over land. Since the uncertainty of reflected solar radiation is greater than the emissivity difference between low clouds and the surface during the day, ULST is only used for cloud detection at night.

[0086] (7) Relative visible contrast test (RVCT):

[0087] The basic assumption is that in a small area, a pixel that is much brighter than the darkest pixel in the neighborhood may be a cloudy pixel. The RVCT criterion used by the low light imager is the observed reflectance of the low light channel after correction for the solar elevation angle minus the minimum value on the surrounding 3×3 pixel array RVCT can be used to identify small-scale clouds and cloud edges. However, misjudgment is likely to occur near the coastline or in the case of strong surface reflectance gradients. Therefore, this test is not applicable to ice and snow surfaces or coastal areas. The advantage of this test is that it does not require knowledge of the clear sky reflectance value. The condition for RVCT to determine whether a pixel is a cloudy pixel is

[0088] represents the corrected value of the observed low light channel after correction for the solar elevation angle and can be expressed as θsz denotes the solar zenith angle, denotes the minimum value of the corrected observation value of the low-light channel in the preset neighborhood window D (referring to the 3×3 pixel array around any pixel).

[0089] (8) Desert Night Low Cloud Test (DZT_NLS):

[0090] The condition for judging whether a pixel is a cloudy pixel is similar to ULST. Only two additional criteria (M ch7 -M ch6 > 0 and (M ch6 -M ch2 +5) / 10 - (M ch7 -M ch6 +4) / 6 < 0.16) are used so that clear-sky desert pixels will not be mislabeled as cloudy.

[0091] (9) New Optically Thin Cloud Test (N-OTC):

[0092] The condition for N-OTC to judge whether a pixel is a cloudy pixel is M ch2 -M ch7 > ε N-OTC , ε N-OTC is the threshold of N_OTC.

[0093] (10) Reflectance Contrast Test under the Same Illumination Condition (SI_RCT):

[0094] SI_RCT is mainly for the twilight zone.

[0095] Considering the characteristics that the coverage area of each scene of the low-light imager is small and the low-light imager has no near-infrared channel, the reflectance value of the low-light channel (M ch1 ) is compared with the mean value of all reflectance values (denoted as sz ) with the same solar zenith angle θ ch6 (excluding the area affected by solar flares and M > 250K). Pixels that satisfy are judged as cloudy pixels. Among them, P90(·) and P10(·) represent the 90th and 10th percentiles respectively.

[0096] The area affected by solar flares refers to the flare area, which is defined as the ocean surface where the solar flare angle is less than 40°.

[0097] (11) Low Cloud Test in the Flare Area (SG_DLS):

[0098] Pixels in the flare area that satisfy M ch2 -S ch2 < ε SG_DLS are judged as cloudy pixels.

[0099] (12) Solar Flare Region Reflectivity Contrast Test (SG_RCT):

[0100] Compare the reflectivity value of the low-light channel (M ch1 ) with the mean value of all reflectivity values (denoted as sg ) having the same solar flare angle θ (excluding land and regions where M ch6 > 250K). Pixels that satisfy are determined as cloud pixels.

[0101] The present invention also provides a storage medium. When the program stored in the storage medium runs, it executes the above cloud detection method applicable to the low-light imager of Fengyun-3 E satellite.

[0102] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. The processor runs the computer program to execute the above cloud detection method applicable to the low-light imager of Fengyun-3 E satellite.

[0103] In the above embodiments of the present invention, the descriptions of the respective embodiments have their own emphases. For parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0104] In the several embodiments provided by the present application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are merely illustrative. For example, the division of the units can be a logical function division. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of the units or modules can be in an electrical or other form.

[0105] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0106] In addition, in each embodiment of the present invention, the functional units can be integrated in one processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units. ​

[0107] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes: various media such as USB flash drives, read-only memories (ROMs), random access memories (RAMs), mobile hard disks, magnetic disks, or optical discs that can store program codes.

[0108] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A cloud detection method applicable to the low-light imager of Fengyun-3E satellite, characterized in that: include: Obtain observation data of the low-light-level imager in the target observation area; Construct cloud detection criteria, and based on the observation data of the low-light imager in the target observation area, judge whether each pixel in the target observation area is a low cloud pixel one by one; when any pixel in the target observation area meets: or When , it indicates that the corresponding pixel is a low cloud pixel; Where: M ch1 It represents the reflectivity observation value, which is obtained by observing the low-light channel of the low-light imager; represents the same solar zenith angle θsz and excludes the influence of solar flares and M ch6 All reflectivity observations in the region >250K, i represents one of the elements; M ch6 It represents the observed brightness temperature in the infrared window area, which is obtained by observing the infrared window channel of the low-light imager; n represents the same solar zenith angle θsz and excludes the influence of solar flares and M ch6 The total number of all reflectivity observations in the region >250K; P90(·) and P10(·) indicate the 90th and 10th percentiles, respectively; represents the same solar flare angle θsg and excludes land and M ch6 All reflectivity values ​​for the area with a temperature greater than 250K; m represents the values ​​for the area with the same solar flare angle θsg and excluding the land and M ch6 The total number of all reflectivity observations in the region >250K, j represents one of the elements.

2. The cloud detection method applicable to the FY-3E low-light imager according to claim 1, characterized in that: It also includes obtaining the output result of the radiation transfer model and interpolating the obtained output result of the radiation transfer model to each pixel of the target observation area; The output result of the radiation transfer model is the output obtained by simulating the observation of the low-light-level imager in the target observation area by the radiation transfer model; The cloud detection criterion determines whether each pixel in the target observation area is a low cloud pixel based on the obtained radiation transfer model output results and the observation data of the low-light imager in the target observation area; when any pixel in the target observation area meets: or M ch2 -S ch2 <ε SG_DLS When , it indicates that the corresponding pixel is a low cloud pixel; Where: R1(T) represents the radiance at the center wavelength of the short-wave infrared channel calculated from the temperature T according to the Planck function; ε ULST is the preset ULST test threshold; M ch2 represents the short-wave infrared observation brightness temperature, which is obtained by observing the short-wave infrared channel of the low-light imager in the target observation area; M ch6 It represents the infrared window observation brightness temperature, which is obtained by observing the infrared window channel of the low-light imager in the target observation area; S ch2 S represents the short-wave infrared clear sky simulated brightness temperature, which is obtained by simulating the observation of the short-wave infrared channel of the low-light imager in the target observation area through the radiation transfer model; S ch6 It represents the simulated brightness temperature of clear sky in infrared window area, which is obtained by simulating the observation of infrared window channel of low-light imager in the target observation area through radiation transmission mode; ε SG_DLS Indicates the preset SG_DLS test threshold.

3. The cloud detection method applicable to the FY-3E low-light imager according to claim 2, characterized in that: The input of the radiation transfer model is the surface infrared emissivity, atmospheric temperature and humidity, and surface elements within the observation range extracted based on the forecast field data.

4. The cloud detection method applicable to the FY-3E low-light imager according to claim 2, characterized in that: When judging each pixel in the target observation area, the cloud detection criterion is that any pixel in the target observation area satisfies: Mch7-Mch6>0 and (M ch6 -M ch2 +5) / 10-(M ch7 -M ch6 +4) / 6<0.16, it indicates that the corresponding pixel is a low cloud pixel; Where: M ch7 It represents the infrared dirty window observation brightness temperature, which is obtained by observing the infrared dirty window channel of the low-light imager in the target observation area.

5. The cloud detection method applicable to the FY-3E low-light imager according to claim 4, characterized in that: The cloud detection criterion is used to judge each pixel in the target observation area. When any pixel in the target observation area satisfies: or ρ(M ch6 ,M ch4 )>0.7 or M ch2 -M ch7 >ε N-OTC When , it indicates that the corresponding pixel is a cirrus cloud pixel; Where: is the preset first PFMFT test threshold; S ch7 represents the infrared dirty window clear sky simulated brightness temperature, which is obtained by simulating the observation of the infrared dirty window channel of the low-light imager in the target observation area through the radiation transfer model; ρ(M ch6 ,M ch4 ) represents the observed brightness temperature M of the infrared window channel of any pixel ch6 and the observed brightness temperature M of the water vapor infrared channel ch4 The correlation coefficient of M ch4 represents the brightness temperature of water vapor infrared observation, which is obtained by observing the water vapor infrared channel of the low-light imager in the target observation area; ε N-OTC It is the preset N-OTC test threshold.

6. The cloud detection method applicable to the FY-3E low-light imager according to claim 5, characterized in that: The cloud detection criterion is used to judge each pixel in the target observation area. When any pixel in the target observation area satisfies: or When , it indicates that the corresponding pixel is a cloud pixel; where: is the maximum observed brightness temperature of any pixel in the observation area observed through the infrared window channel within the preset neighborhood window B; z is the standard deviation of terrain height within the neighborhood window B, which is calculated based on the geographic information data read by the low-light imager; γ represents the temperature decrease rate; ε RTCT is the preset RTCT test threshold; R2(T) represents the radiance of the infrared window channel calculated from the temperature T according to Planck’s law; It is the radiation emitted by the top of the troposphere and the atmosphere above it, calculated based on the atmospheric temperature and humidity profiles and the optical thickness of the atmospheric layers in the forecast field; ε ETROP is the preset ETROP test threshold.

7. The cloud detection method applicable to the FY-3E low-light imager according to claim 6, characterized in that: The cloud detection criterion is used to judge each pixel in the target observation area. When any pixel in the target observation area satisfies: or When , it indicates that the corresponding pixel is a cloud pixel; Where: They represent the brightness temperature observed by the pixel at the NWC position through the infrared window channel and the infrared dirty window channel respectively; the NWC position represents the neighboring warm center of any pixel in the observation area, and represents the warmest pixel of any pixel in its preset neighborhood window A; is the preset second RFMFT test threshold; It represents the correction value of the low-light channel observation, obtained by correcting the solar altitude angle; It represents the minimum value of the correction value of the low-light channel observation of any pixel in the preset neighborhood window D.

8. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the computer program runs to execute the cloud detection method applicable to the FY-3E satellite low-light imager as described in any one of claims 1 to 7.

Images