Sea fog three-dimensional distribution detection method based on satellite-borne active and passive remote sensing fusion

Abstract

The invention discloses a sea fog three-dimensional distribution detection method based on satellite-borne active and passive remote sensing fusion. The high-precision vertical section detection capability of a satellite-borne active remote sensing laser radar and the large-range long-time continuous observation capability of passive remote sensing of a stationary satellite are fully exerted, an optimal band combination is selected through a reconstruction verification method, the multi-band spectrum similarity is evaluated, the most suitable active remote sensing profile is selected, and the multi-band spectrum similarity is evaluated. And expanding the sea fog to a certain distance outside an active remote sensing detection range, and constructing three-dimensional distribution of sea fog levels. Compared with a sea fog level detection result of an existing method, the method has the advantages that the satellite three-dimensional detection range is effectively expanded, the method has the advantages of being high in spatial resolution, capable of detecting vertical-direction levels and the like, is suitable for multiple scenes, facilitates understanding of sea fog formation and dissipation mechanisms and can also be used as a high-precision three-dimensional initial field for sea fog forecasting.

Description

Technical Field

[0001] The invention relates to the field of atmospheric remote sensing monitoring, and in particular to a method for detecting three-dimensional distribution of sea fog based on satellite-borne active and passive remote sensing fusion. Background Art

[0002] Sea fog is a phenomenon of condensation of water vapor in the lower atmosphere above the ocean. It is formed by the accumulation of a large amount of water droplets or ice crystals (or both), which can reduce horizontal visibility to less than 1 km. As a common meteorological disaster in marine or coastal areas, sea fog directly affects air, sea and land transportation, and also causes great inconvenience to communications, fishing, maritime military activities, etc. Therefore, the formation, diffusion mechanism and spatiotemporal distribution of sea fog have always been the research focus of scholars at home and abroad. In order to better clarify the formation, evolution and dissipation of sea fog, it is necessary to accurately grasp its three-dimensional distribution information. High-precision three-dimensional sea fog distribution grid data can be used as the initial field input for predicting the formation, diffusion and dissipation of sea fog, thereby improving the accuracy of sea fog prediction.

[0003] Since meteorological stations are sparsely distributed in the ocean, in order to meet the demand for obtaining large-scale, high-quality sea fog observation data, satellite-borne passive remote sensing has become an important means of sea fog monitoring. It has the advantages of large observation range, high real-time performance, and rich information. However, due to the extremely high similarity of radiation and brightness temperature in each channel between sea fog and stratus clouds, how to distinguish between the two has always been a difficult problem in sea fog monitoring. The only difference between the two is that the bottom of the sea fog layer is close to the ocean surface, while the bottom of the stratus cloud layer is a certain distance above the ocean surface.

[0004] It is difficult to obtain information about the vertical structure of clouds and fog by relying solely on passive remote sensing. Whether it is the obstruction of high clouds or the misjudgment between low-level clouds and sea fog, the accuracy of passive remote sensing detection of sea fog is restricted. Unlike passive remote sensing, active remote sensing such as space-borne lidar can obtain the vertical structure of the atmosphere and has the potential to distinguish sea fog from low-level clouds.

[0005] However, satellite-borne active remote sensing (lidar, cloud radar, etc.) also has obvious shortcomings. Its observation targets are arranged in a point-like pattern along the subsatellite track along the flight direction, with a very small width, and only three-dimensional observation is achieved within the coverage area of ​​its track. The range of sea fog is usually large, and using satellite-borne lidar to obtain only one profile of information is not enough to support three-dimensional monitoring of sea fog in the entire sea area.

[0006] The invention patent with authorization announcement number CN 114022782A discloses a sea fog detection method based on MODIS satellite data, including preprocessing MODIS satellite data and extracting features; selecting ground object samples in combination with CALIOP VFM satellite data and dividing them into training sets and test sets; using the information gain rate of node entropy as the impurity criterion for node splitting, and assigning weights to each typical correlation tree in the typical correlation forest, and training with the training set to obtain an improved typical correlation forest model; adjusting and updating the weight of each typical correlation tree in post-testing to obtain a continuously optimized improved typical correlation forest model; and using the model to identify satellite images to obtain sea fog recognition results.

[0007] In the current existing technology, the sea fog distribution detection algorithm mainly detects the sea fog coverage, relying on the multi-channel spectral data of satellite-borne passive remote sensing, but does not fully utilize the vertical structure detection capability of satellite-borne active remote sensing, resulting in the sea fog distribution detection results being limited to a two-dimensional plane. Therefore, it is necessary to construct a three-dimensional sea fog distribution detection method based on the fusion of satellite-borne active and passive remote sensing, which is conducive to the construction of a high-precision three-dimensional sea fog model, and thus better study and forecast the mechanism of sea fog. Summary of the invention

[0008] In order to solve the problem that most of the existing satellite-borne sea fog distribution detection algorithms can only obtain two-dimensional distribution and cannot perform large-scale sea fog layer height distribution, the present invention proposes a three-dimensional distribution detection method for sea fog based on satellite-borne active and passive remote sensing fusion. The algorithm uses the 532nm attenuated backscattering coefficient, scattering ratio and backscattering coefficient obtained by the satellite-borne laser radar and the multi-band albedo or brightness temperature data of the geostationary satellite as input to identify the sea fog layer and construct the three-dimensional structure of the sea fog, thereby obtaining continuous and high spatial resolution three-dimensional distribution information of the sea fog layer.

[0009] The specific technical solutions adopted are as follows:

[0010] A method for detecting three-dimensional distribution of sea fog based on satellite-borne active and passive remote sensing fusion, comprising the following steps:

[0011] (1) Obtaining satellite-borne laser radar remote sensing observation data as active remote sensing data, including obtaining the calibrated attenuated backscattering coefficient of level 1 (i.e., level 1), and simultaneously obtaining the corresponding backscattering coefficient and scattering ratio of level 2 (i.e., level 2) as the basis for subsequent identification of sea fog layers; obtaining geostationary satellite remote sensing observation data as passive remote sensing data, including multi-band spectral information, specifically including the albedo of each channel of the visible light and near-infrared bands and the brightness temperature of each channel of the infrared band; performing spatiotemporal matching between the satellite-borne laser radar remote sensing observation data and the geostationary satellite remote sensing observation data;

[0012] (2) Through the optical property criterion, the scattering ratio, attenuated backscattering coefficient and backscattering coefficient thresholds are set, and the cloud and aerosol layers in the detection profile are distinguished based on the thresholds; through the physical position criterion, the vertical detection advantage of the spaceborne lidar is brought into play, and the cloud layers within a certain threshold range above the sea surface are screened, and the pixels in the screened layers are identified as sea fog; on this basis, the layer continuity criterion is used to obtain a complete and continuous sea fog layer;

[0013] (3) Compare the sea fog identification results in step (2) with the actual observation data in the International Integrated Ocean-Atmosphere Dataset ICOADS, and analyze the accuracy of sea fog identification in step (2);

[0014] (4) Based on the time-space matching of the vertical detection profile of the space-borne laser radar and the multi-band spectral information of the geostationary satellite, the information of the measured vertical profile is expanded to a certain range outside the orbit by analyzing the multi-channel spectral similarity between the points outside the orbit of the space-borne laser radar and the points on the orbit;

[0015] (5) reconstructing the vertical detection profile of the spaceborne lidar using the method in step (4), and comparing the sea fog layer heights before and after reconstruction under different band conditions to select the band combination most suitable for detecting the sea fog layer height distribution;

[0016] (6) Apply the method of step (4) in combination with the band combination selected in step (5) to construct a three-dimensional distribution of sea fog within a certain range from the orbit of the satellite-borne laser radar.

[0017] The present invention combines the high-precision atmospheric vertical profile detection capability of space-borne lidar with the large-scale continuous observation capability of geostationary satellites, and realizes three-dimensional distribution detection of sea fog based on the active and passive fusion method. It can effectively expand the three-dimensional detection range of satellites and improve the ability to distinguish between sea fog and low clouds.

[0018] Preferably, in step (1), the channel attenuation backscattering coefficient, scattering ratio, and backscattering coefficient are obtained for better sea fog identification. Of course, optical characteristic parameters such as extinction coefficient, depolarization ratio, and radar ratio can also be obtained. These optical characteristic parameters can assist in sea fog identification. Generally speaking, compared with aerosols, sea fog has stronger backscattering, smaller laser radar ratio, stronger attenuation of laser radar signals, larger scattering ratio, and clearer layer edges.

[0019] The difficulty in identifying sea fog lies mainly in distinguishing it from stratus clouds. Stratus clouds belong to the family of low clouds, and the cloud base is usually less than 2 km above the ground or ocean surface, which often seriously affects the safety and stability of aviation flights. The large amount of water vapor in the cloud will also interfere with communication electromagnetic wave signals. The main difference between sea fog and stratus clouds is that stratus clouds are not directly connected to the surface, while the bottom of sea fog is in direct contact with the ocean surface. There is no obvious difference in physical properties between the two. Due to the inability to obtain high-precision layer information, passive remote sensing often finds it difficult to accurately distinguish sea fog from stratus clouds. Therefore, in step (2), it is necessary to give full play to the high-resolution atmospheric vertical profile detection capability of the satellite-borne lidar, accurately obtain the layer height, and improve the ability to distinguish sea fog from stratus clouds.

[0020] Preferably, in step (2), the optical property judgment threshold is derived from a large number of historical observation results. According to the long-term observation data of ground-based laser radar, the optical property judgment formula for distinguishing cloud aerosol is:

[0021] SR>10

[0022] ATB≥0.02km -1 sr -1

[0023] β≥0.03km -1 sr -1

[0024] Where SR is the scattering ratio, ATB is the attenuated backscattering coefficient, and β is the backscattering coefficient. When the above SR, ATB, and β thresholds are met at the same time, it is considered to be a cloud, otherwise it is an aerosol.

[0025] When the extinction coefficient, depolarization ratio, and radar ratio are used to assist in sea fog identification, clouds can be considered to be clouds if the scattering ratio is less than 10, the depolarization ratio is greater than 0.15, and the radar ratio is less than 40. The extinction coefficient of clouds tends to increase with the thickness of clouds due to multiple scattering, so this can also be used to assist in sea fog identification.

[0026] Preferably, in step (2), the cloud layers within a certain threshold range above the sea surface are screened, and the accurate cloud base height is obtained by satellite-borne laser radar, and its distance from the sea surface is calculated as a basis for distinguishing sea fog from stratus clouds, and the pixels within the distance are identified as sea fog. The CALIOP-based method points out that CALIOP's VFM data may mistakenly identify sea fog as the sea surface. Since the satellite-borne laser radar is also equipped with an altimetry payload and has real-time measurement data of the surface elevation, this problem will not exist when using the satellite-borne laser radar to obtain the sea fog layers.

[0027] Preferably, in step (2), the method for obtaining a complete and continuous sea fog layer by means of a layer continuity criterion is as follows: for each pixel point identified as sea fog, if the pixel point directly connected to it satisfies the optical property criterion, then the directly connected pixel point is identified as sea fog.

[0028] Preferably, in step (3), when making comparison, the trajectory points in the sea fog identification results with a time interval of less than 2 hours and a spatial distance of less than 100 km are screened for comparison with the ICOADS data points. If both are identified as sea fog or both are identified as no sea fog, it is determined that the identification is accurate; if it is identified as sea fog but ICOADS does not record sea fog, it is determined to be a false identification; if it is not identified as sea fog but ICOADS records it as sea fog, it is determined to be missing, and all matching cases are counted, and the accuracy rate, missing rate and false identification rate are analyzed to determine the accuracy of the method.

[0029] Preferably, in step (4), the basic assumption of the method of fusing active and passive remote sensing detection data to expand the information of the measured vertical profile to a certain range outside the orbit is: when the geographical locations are close and the background conditions such as the surface type are similar, if the radiation amounts of two pixels in the passive remote sensing data in multiple spectral channels involving atmospheric features such as clouds and aerosols are similar, then it can be considered that the vertical structures of clouds and aerosols of the two pixels are also similar. In the area where active remote sensing and passive remote sensing are detected simultaneously, the vertical structure of clouds and aerosols and the spectral information of each band are known, and this information can be used as the "donor" in the expansion process. The pixels outside the orbit only have the spectral information of each band of passive remote sensing. Therefore, these pixels can be used as "receptors" and matched by selecting "donor" points that are sufficiently similar to their own spectral information.

[0030] Preferably, in step (4), by analyzing the multi-channel spectral similarity between points outside the orbit of the satellite-borne laser radar and points on the orbit, the information of the measured vertical profile is expanded to a certain range outside the orbit:

[0031] First, record the position information of the donor pixel points on the ACHSRL (Aerosol-Cloud High-Spectral-Resolution Lidar) track and the receptor pixel points in the area to be expanded outside the ACHSRL track; the coordinates of the "donor" pixel point on the ACHSRL track are (a, 0), a∈[0, I], and the width range to be constructed is the range of J pixels on both sides of the active track, then the coordinates of the receptor pixel points to be constructed are (i, j), j∈[-J, 0) ∪(0, J];

[0032] Then, the loss function C(i, j; a) is calculated using the weighted minimum square error of the multi-channel albedo or brightness temperature data of the donor pixel and the acceptor pixel:

[0033]

[0034] Among them, Al m is the albedo of the mth channel selected by the geostationary satellite for spectral similarity comparison, BT n is the brightness temperature of the nth channel;

[0035] Next, the most suitable donor pixel is selected based on the background constraints and the distance index; the background constraints are: 1) the acceptor pixel and the donor pixel have the same surface type; 2) the acceptor pixel and the donor pixel have similar solar zenith angle and azimuth, that is, the difference between the solar zenith angle and azimuth is less than 10°. The best donor is to select a part of the pixel set with the smallest cost function with the acceptor pixel from all the candidate donor pixels under the above constraints, and then select the point with the closest distance among them;

[0036] Finally, in the target area, the vertical profile information of the donor pixel is used as the vertical profile information of each receptor pixel to match, so as to construct the three-dimensional structure of the sea fog.

[0037] Preferably, in step (5), a reconstruction algorithm is used to select a band combination. The following band combinations are tested by setting the blind area range to 30km, 60km, and 100km: Combination 1: The closest band to the band selected by MODIS, namely 0.64μm, 2.3μm, 8.6μm, and 12.4μm; Combination 2: The bands commonly used for daytime sea fog monitoring, namely 0.64μm, 0.86μm, and 1.6μm; Combination 3: The commonly used band combination for daytime and nighttime sea fog monitoring, namely 0.64μm, 0.86μm, 1.6μm, 3.9μm, and 11.2μm; Combination 4: All geostationary satellite bands. By comparing and analyzing the profiles before and after reconstruction, evaluating the overall matching rate TMR and the root mean square error of the sea fog height before and after reconstruction, the band combination that is most conducive to the expansion of the three-dimensional structure of sea fog is obtained. TMR is defined as:

[0038] TMR=(N cc +N aa +N ff +N 00 ) / N

[0039] Where N cc is the number of pixels identified as clouds before and after reconstruction, N aa is the number of pixels identified as aerosols before and after reconstruction, N ff is the number of pixels identified as sea fog before and after reconstruction, N00 It is the number of pixels identified as clear (i.e., cloudless aerosol) before and after reconstruction. The most suitable band combination for detecting the three-dimensional distribution of sea fog is 0.64μm, 2.3μm, 8.6μm and 12.4μm.

[0040] Preferably, in step (6), the method for constructing the three-dimensional distribution of sea fog within a certain range from the ACDL track is: select the profile of the known point with the highest spectral similarity to the point to be expanded, and move it outward to the point with expansion, so as to obtain the complete three-dimensional structure of the sea fog and obtain the distribution of the sea fog layer height.

[0041] Preferably, in step (6), the three-dimensional distribution of sea fog within a certain range from the ACDL track is constructed, and this distance is also evaluated in the reconstruction verification of step (5).

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] (1) The three-dimensional distribution detection method of sea fog based on satellite-borne active and passive remote sensing fusion combines the high-precision atmospheric vertical profile detection capability of ACDL with the large-scale continuous observation capability of Himawari-8 to achieve three-dimensional distribution detection of sea fog, which can effectively expand the three-dimensional detection range of satellites.

[0044] (2) The sea fog recognition results obtained by the sea fog three-dimensional distribution detection method based on satellite-borne active and passive remote sensing fusion of the present invention are compared with the ICOADS records matched with time and space, and the accuracy rate is more than 90%, the missing rate is about 1%, and the misidentification rate is about 9%.

[0045] (3) Compared with the existing satellite-based remote sensing sea fog layer distribution detection algorithm, the proposed method has the advantages of high spatial resolution and the ability to detect vertical layers. It is suitable for multiple scenarios and helps to understand the formation and dissipation mechanism of sea fog. It can also be used as a high-precision three-dimensional initial field for sea fog forecasting. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 Schematic diagram of the process of the three-dimensional distribution detection method of sea fog based on the fusion of satellite-borne active and passive remote sensing in this embodiment.

[0047] Figure 2 Schematic diagram of a case of sea fog recognition results in this embodiment.

[0048] Figure 3 This is a comparison chart of the sea fog recognition results in this embodiment and the ICOADS data.

[0049] Figure 4 3 is a comparison diagram of the top height of the sea fog layer before and after reconstruction of different blind area ranges in this embodiment.

[0050] Figure 5 (a) is a three-dimensional map of the sea fog layers constructed in this embodiment, and (b) is a distribution map of the top height of the sea fog layer. DETAILED DESCRIPTION

[0051] To make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific implementation methods described herein are only used to explain the present invention and do not limit the scope of protection of the present invention.

[0052] This embodiment provides a method for detecting three-dimensional distribution of sea fog based on the fusion of satellite-borne active and passive remote sensing. Figure 1 As shown, the three-dimensional distribution detection method of sea fog includes five steps, and each step is described in detail below:

[0053] (1) Active and passive remote sensing data acquisition and time-space matching

[0054] The original remote sensing observation data of the space-borne lidar ACDL are obtained as active remote sensing data. The time range is from June 1, 2022 to December 31, 2023. The spatial range is Bohai Sea (116°E-124°E, 36°N-42°N), Yellow Sea (118°E-128°E, 32°N-40°N), East China Sea (120°E-130°E, 25°N-32°N), and South China Sea (108°E-124°E, 0°N-25°N). The main data products include the attenuated backscatter coefficient of the 532nm channel at the level 1 level, the 532nm backscatter coefficient, scattering ratio, radar ratio, depolarization ratio and extinction coefficient at the level 2 level.

[0055] The original remote sensing observation data of the geostationary satellite Himawari-8 are obtained as passive remote sensing data. The main data products include the albedo of the visible light and near-infrared channels with a time resolution of 10 minutes and a spatial resolution of 5 km, as well as the brightness temperature data of the infrared channel, which constitute multi-band spectral information;

[0056] For ACDL and Himawari-8, since Himawari-8 is a geostationary satellite, the R time-space match is the part of the ACDL trajectory that passes through the Himawari-8 detection area.

[0057] (2) ACDL sea fog recognition using spaceborne lidar

[0058] There are three main criteria for identifying sea fog based on the measured remote sensing observation data of the space-borne lidar ACDL:

[0059] Optical property criteria: ACDL detected layer SR>10, ATB≥0.02km-1 sr -1 , β ≥ 0.03 km -1 sr -1 ;

[0060] Physical location criterion: the bottom of the layer is directly connected to the ocean surface, or the distance from the bottom of the layer to the ocean surface is d≤2Δh, where Δh is the vertical resolution of the ACDL data product; and the surface type corresponding to the pixel point is ocean;

[0061] Hierarchical continuity criterion: For each pixel identified as sea fog, if the pixel directly connected to it meets the optical property criterion, it can be determined to be sea fog.

[0062] Figure 2 This is an example of sea fog identification based on the above criteria. The sea fog event was detected by ACDL around 17:40 (UTC time) on June 2, 2022. The ACDL optical property data product and sea fog identification results are shown.

[0063] (3) Verification of ACDL sea fog recognition results by spaceborne lidar

[0064] In the selected time range from June 1, 2022 to December 31, 2023, and in the spatial range of the four major sea areas of Bohai Sea, Yellow Sea, East China Sea, and South China Sea, there are 1,873 cases of spatiotemporal matching between ACDL trajectories and ICOADS records. Among them, there are 26 cases in Bohai Sea, 173 cases in the Yellow Sea, 420 cases in the East China Sea, and 1,254 cases in the South China Sea. Comparing the ACDL sea fog recognition results with the ICOADS records: there are 1,688 cases in which both results are foggy or fog-free, 16 cases recorded by ICOADS as foggy and ACDL as fog-free, and 169 cases recorded by ICOADS as fog-free and ACDL as foggy, that is, the accuracy rate is 90.12%, the missing rate is 0.85%, and the misidentification rate is 9.02%. Figure 3 The accuracy of the ACDL sea fog recognition algorithm of the spaceborne lidar is demonstrated.

[0065] (4) Reconstruction verification for band combination selection and extended range evaluation

[0066] The reconstruction algorithm is used to select the band combination. The following band combinations are tested by setting the blind area range to 30km, 60km, and 100km: Band 1: The closest band to the band selected by MODIS, namely 0.64μm, 2.3μm, 8.6μm, and 12.4μm; Band 2: The band commonly used for daytime sea fog monitoring, namely 0.64μm, 0.86μm, and 1.6μm; Band 3: The band combination commonly used for daytime and nighttime sea fog monitoring, namely 0.64μm, 0.86μm, 1.6μm, 3.9μm, and 11.2μm; Band 4: All Himawari-8 bands. By comparing and analyzing the profiles before and after reconstruction, the overall matching rate TMR and the root mean square error of the sea fog height before and after reconstruction are evaluated. When the blind area is 30km, the TMRs of the four band combinations are 89.08%, 87.27%, 89.11%, and 89.07% respectively; when the blind area is 60km, the TMRs of the four band combinations are 86.82%, 86.17%, 86.84%, and 86.80% respectively; when the blind area is 100km, the TMRs of the four band combinations are 84.41%, 82.45%, 84.42%, and 84.39% respectively. Therefore, Band 3 is selected as the best band combination. Figure 4 The comparison of the top height of the sea fog layer before and after reconstruction under different blind area ranges is shown.

[0067] (4) Construction of active and passive fusion sea fog three-dimensional structure

[0068] This step depends on the band combination selection result obtained by reconstruction and verification in step (5) and the evaluation of the extended range. Band3 is selected as the band combination for evaluating spectral similarity, and the three-dimensional structure of sea fog within a certain distance of the active remote sensing detection range is constructed through active and passive fusion. Figure 5 It shows a stereoscopic image of the three-dimensional structure of sea fog and a distribution map of the top height of the sea fog layer.

Claims

1. A method for detecting three-dimensional distribution of sea fog based on satellite-borne active and passive remote sensing fusion, characterized in that: The following steps are involved: (1) Obtaining remote sensing observation data from spaceborne laser radar as active remote sensing data, including obtaining the calibrated attenuated backscatter coefficient of level 1, and simultaneously obtaining the corresponding backscatter coefficient and scattering ratio of level 2 as the basis for subsequent identification of sea fog layers; obtaining remote sensing observation data from geostationary satellites as passive remote sensing data, including multi-band spectral information, specifically including the albedo of each channel in the visible and near-infrared bands and the brightness temperature of each channel in the infrared band; performing spatiotemporal matching between the remote sensing observation data from spaceborne laser radar and the remote sensing observation data from geostationary satellites; (2) By using the optical property criterion, the scattering ratio, attenuated backscattering coefficient, and the threshold of the backscattering coefficient are set, and the cloud and aerosol layers in the detection profile of the spaceborne lidar are distinguished based on the threshold. By using the physical position criterion, the cloud layers within a certain threshold range above the sea surface are screened, and the distance between the cloud base height and the sea surface is calculated, and the pixels within this distance are identified as sea fog. On this basis, the layer continuity criterion is used to obtain a complete and continuous sea fog layer. (3) Compare the sea fog identification results in step (2) with the actual observation data in the International Integrated Ocean-Atmosphere Dataset ICOADS, and analyze the accuracy of sea fog identification in step (2); (4) Based on the time-space matching of the vertical detection profile of the space-borne laser radar and the multi-band spectral information of the geostationary satellite, the information of the measured vertical profile is expanded to a certain range outside the orbit by analyzing the multi-channel spectral similarity between the points outside the orbit of the space-borne laser radar and the points on the orbit; (5) reconstructing the vertical detection profile of the spaceborne lidar using the method in step (4), and comparing the sea fog layer heights before and after reconstruction under different band conditions to select the band combination most suitable for detecting the sea fog layer height distribution; (6) Apply the method of step (4) in combination with the band combination selected in step (5) to construct a three-dimensional distribution of sea fog within a certain range from the orbit of the satellite-borne laser radar.

2. The method for detecting three-dimensional distribution of sea fog based on satellite-borne active and passive remote sensing fusion according to claim 1 is characterized in that: In step (2), the thresholds for cloud and aerosol classification using optical characteristics are: scattering ratio greater than 10, attenuated backscatter coefficient greater than or equal to 0.02 km -1 sr -1 , backscatter coefficient is greater than or equal to 0.03km -1 sr -1 When the above scattering ratio, attenuated backscattering coefficient, and backscattering coefficient threshold are met at the same time, it is considered to be a cloud, otherwise it is an aerosol.

3. The method for detecting three-dimensional distribution of sea fog based on satellite-borne active and passive remote sensing fusion according to claim 1 is characterized in that: In step (2), the method for obtaining a complete and continuous sea fog hierarchy through the hierarchy continuity criterion is as follows: for each pixel point identified as sea fog, if the pixel point directly connected to it meets the optical characteristic criterion, then the directly connected pixel point is identified as sea fog.

4. The method for detecting three-dimensional distribution of sea fog based on satellite-borne active and passive remote sensing fusion according to claim 1 is characterized in that: In step (3), when making comparisons, the trajectory points in the sea fog identification results with a time interval of less than 2 hours and a spatial distance of less than 100 km are screened for comparison with the actual observation data points in ICOADS. If both are identified as sea fog or both are identified as no sea fog, the identification is determined to be accurate; if it is identified as sea fog but ICOADS does not record sea fog, it is determined to be a false identification; if it is not identified as sea fog but ICOADS records it as sea fog, it is determined to be missing. All matching cases are counted, and the accuracy rate, missing rate and false identification rate are analyzed to determine the accuracy of the method.

5. The method for detecting three-dimensional distribution of sea fog based on satellite-borne active and passive remote sensing fusion according to claim 1 is characterized in that: In step (4), the multi-channel spectral similarity analyzed is a loss function of the albedo or brightness temperature of the selected channel between the pixels outside the track and the points on the ACDL track.

6. The method for detecting three-dimensional distribution of sea fog based on satellite-borne active and passive remote sensing fusion according to claim 1 is characterized in that: By analyzing the multi-channel spectral similarity between points outside the orbit of the satellite-borne lidar and points on the orbit, the information of the measured vertical profile is extended to a certain range outside the orbit, including: First, record the position information of the donor pixel points on the satellite laser radar track and the receptor pixel points in the area to be expanded outside the track; the coordinates of the donor pixel points on the satellite laser radar track are (a, 0), a∈[0, I], and the width range to be constructed is the range of J pixels on both sides of the active track, then the coordinates of the receptor pixel points to be constructed are (i, j), j∈[-J, 0) ∪(0, J]; Then, the loss function C(i, j; a) is calculated using the weighted minimum square error of the multi-channel albedo or brightness temperature data of the donor pixel and the acceptor pixel: Among them, Al m is the albedo of the mth channel selected by the geostationary satellite for spectral similarity comparison, BT n is the brightness temperature of the nth channel; Next, the most suitable donor pixel is selected based on the background constraints and the distance index; the background constraints are: 1) the acceptor pixel and the donor pixel have the same surface type; 2) the acceptor pixel and the donor pixel have similar solar zenith angle and azimuth, that is, the difference between the solar zenith angle and azimuth is less than 10°. The best donor pixel is to select a part of the pixel set with the smallest cost function with the acceptor pixel from all the candidate donor pixels under the above constraints, and then select the point with the closest distance among them; Finally, in the target area, the vertical profile information of the donor pixel is used as the vertical profile information of each receptor pixel to match, so as to construct the three-dimensional structure of the sea fog.

7. The method for detecting three-dimensional distribution of sea fog based on satellite-borne active and passive remote sensing fusion according to claim 1 is characterized in that: In step (5), the method for reconstructing the vertical detection profile of the satellite-borne laser radar is as follows: a certain blind area is set around each position, assuming that the vertical profile in the blind area is unknown, and the corresponding position is reconstructed using the part outside the blind area.

8. The method for detecting three-dimensional distribution of sea fog based on satellite-borne active and passive remote sensing fusion according to claim 1 is characterized in that: In step (5), the goodness of fit and root mean square error under different band combinations are calculated based on the results of comparative analysis of the sea fog layer height before and after reconstruction, so as to select the band combination most suitable for reconstructing the sea fog layer height.

9. The method for detecting three-dimensional distribution of sea fog based on satellite-borne active and passive remote sensing fusion according to claim 1 is characterized in that: In step (5), the band combination most suitable for detecting the three-dimensional distribution of sea fog is 0.64μm, 2.3μm, 8.6μm and 12.4μm.

10. The method for detecting three-dimensional distribution of sea fog based on satellite-borne active and passive remote sensing fusion according to claim 1, characterized in that: In step (6), the method for constructing the three-dimensional distribution of sea fog within a certain range from the satellite-borne laser radar track is: select the profile of the known point with the highest spectral similarity to the point to be expanded, move it outward to the point with expansion, thereby obtaining the complete three-dimensional structure of the sea fog and obtaining the distribution of the sea fog layer height.

Images