A multi-source data fusion sea fog detection method

Abstract

The present application provides a kind of multi-source data fusion sea fog detection method, comprising the following steps: comprising the following steps: S1: obtaining the multi-source data for data fusion;S2: the quality control, spatial interpolation and deviation correction processing to the multi-source data obtained in S1, obtain sea fog range data;S3: check sea fog range data, identify the consistency of sea fog range data, and correct sea fog range data;S4: calculate the sea fog detection rate of sea fog range data, with sea fog detection rate as fusion weight, generate sea fog range fusion data;S5: using the data after S4 processing, divide fog grade for sea fog range fusion data, generate specified spatiotemporal resolution multi-source sea fog fusion live data.The present application has beneficial effects: solve the deficiencies that the existing sea fog detection technology cannot be considered in space coverage, data accuracy and multi-layer cloud fog condition applicability, etc., provide high-quality live data for the monitoring and early warning of sea fog, forecast verification, marine disaster prevention and mitigation.

Description

Technical Field

[0001] This invention belongs to the field of sea fog detection, and in particular relates to a multi-source data fusion method for sea fog detection. Background Technology

[0002] Sea fog is a frequent marine hazardous weather event along my country's coast. When sea fog occurs, it reduces horizontal visibility at sea and along the coast, severely affecting port operations, ship pilotage, and maritime shipping, and may even cause major maritime accidents. Therefore, real-time sea fog data has important application value in marine meteorological services.

[0003] While various methods exist for acquiring real-time sea fog data, these methods have limitations due to marine environmental or technological constraints. For instance, automatic weather stations offer relatively accurate data, but their low density at sea makes it difficult to reflect large-scale sea fog distribution. Satellite remote sensing can cover vast sea areas and has a clear advantage in monitoring offshore sea fog dynamics, but it still has limitations in separating sea fog from stratiform clouds, and its passive sensing method cannot penetrate multiple layers of clouds. Numerical models are unaffected by multi-layered cloud and fog conditions, but the model's boundary layer turbulence scheme, cloud microphysics scheme, and initial field accuracy directly impact simulation results. Lidar can also detect sea fog, but it can only achieve small-area sea fog observations in coastal ports and wharves. Summary of the Invention

[0004] In view of this, the present invention aims to propose a multi-source data fusion method for sea fog detection, in order to solve at least one of the above-mentioned technical problems.

[0005] To achieve the above objectives, the technical solution of the present invention is implemented as follows:

[0006] A multi-source data fusion method for sea fog detection includes the following steps:

[0007] S1: Acquire multi-source data for data fusion;

[0008] S2: Perform quality control, spatial interpolation, and bias correction on the multi-source data obtained in S1 to obtain sea fog range data;

[0009] S3: Check the consistency of multi-source data in identifying sea fog and correct the sea fog extent data;

[0010] S4: Calculate the sea fog detection rate of the sea fog range data, and use the sea fog detection rate as the fusion weight to generate sea fog range fused data;

[0011] S5: Using the data processed by S4, classify the fog levels for the sea fog range fusion data and generate multi-source sea fog fusion real-time data with a specified spatiotemporal resolution.

[0012] Furthermore, the multi-source data mentioned in step S1 includes visibility and relative humidity observation data from automatic weather stations and their quality control codes, L2-level fog monitoring data and cloud top temperature data from FY-4A geostationary satellites, visibility, relative humidity and surface temperature data from CLDAS, sea surface temperature data from CODAS, and L1-level 5km full-disk channel data and L2-level cloud attribute data from Himawari-9 geostationary satellites.

[0013] Furthermore, the process of processing the multi-source data in step S2 is as follows:

[0014] S201: Perform quality control on automatic weather station observation data according to quality control code and threshold check method, interpolate automatic weather station observation data to grid points using inverse distance weighting method, identify sea fog, and obtain automatic weather station interpolated sea fog data.

[0015] S202: Correct the bias of CLDAS visibility and relative humidity data, identify sea fog, and obtain CLDAS sea fog data;

[0016] S203: Convert the FY-4A geostationary satellite fog monitoring data from full-disk nominal grid data to latitude and longitude grid data, and use the nearest neighbor interpolation method to interpolate the data to equal latitude and longitude grid points to obtain FY-4A satellite sea fog data;

[0017] S204: Process FY-4A geostationary satellite cloud top temperature data using the data processing methods in S203;

[0018] S205: Use bilinear interpolation to interpolate CODAS sea surface temperature data and CLDAS land surface temperature data to a grid of specified resolution.

[0019] Furthermore, the process of processing some of the multi-source data in step S2 is as follows:

[0020] For nighttime data, based on brightness temperature data from the Himawari-9 geostationary satellite, a dual-channel thresholding method was used to identify nighttime stratus and sea fog regions.

[0021] For daytime data, the stratocumulus and stratus cloud regions from the Himawari-9 geostationary satellite cloud attribute data are extracted as daytime stratocumulus and sea fog regions;

[0022] Using processed FY-4A geostationary satellite cloud top temperature data, CLDAS land surface temperature data, and CODAS sea surface temperature data, stratus cloud data from Himawari satellite stratus cloud and sea fog data were removed to obtain Himawari satellite sea fog data.

[0023] Furthermore, the specific implementation process of step S4 is as follows:

[0024] S401: Calculate the sea fog detection rate of automatic weather station interpolated sea fog data, CLDAS sea fog data, FY-4A satellite sea fog data, and Himawari satellite sea fog data respectively;

[0025] S402: At each grid point, the sea fog detection rate is used as a weight, and the weighted averaging method is used to process the automatic weather station interpolated sea fog data, CLDAS sea fog data, FY-4A satellite sea fog data, and Himawari satellite sea fog data to obtain fused data;

[0026] S403: Define a threshold value. When the fused data is greater than or equal to the threshold value, there is sea fog. When the fused data is less than the threshold value, there is no sea fog.

[0027] S404: Compare the sea fog range of the fused data with the identification results of the automatic weather station when the critical value takes different values ​​to obtain the optimal limit value;

[0028] S405: Mark regions where the fused data is greater than or equal to the optimal threshold as having sea fog, and regions where the fused data is less than the optimal threshold as not having sea fog, and generate sea fog range fused data.

[0029] Furthermore, the method for correcting the deviations in CLDAS visibility and relative humidity data in step S202 is as follows:

[0030] The CLDAS data is interpolated using a bilinear interpolation method to obtain CLDAS data interpolated to the location of the automatic weather station. The deviation of the CLDAS data relative to the observation data of the automatic weather station is calculated. The deviation of the CLDAS data is interpolated to the grid points using the inverse distance weighting method to obtain the CLDAS deviation field. The CLDAS data is then corrected using the CLDAS deviation field.

[0031] Furthermore, the method for checking the consistency of sea fog data in step S3 is as follows:

[0032] Set four conditions for determining whether there is no fog at a grid point in the multi-source data:

[0033] Condition 1: The relative humidity interpolated to the grid point by the automatic weather station is less than 80% or the visibility is greater than 1.5km;

[0034] Condition 2: The corrected CLDAS relative humidity is less than 80% or the visibility is greater than 1.5 km;

[0035] Condition 3: FY-4A satellite sea fog data detected clear skies;

[0036] Condition 4: The cloud attribute data of the Himawari-9 geostationary satellite detected clear skies or brightness temperature data, but no stratus clouds or sea fog were detected;

[0037] If an automatic weather station identifies sea fog at a certain grid point based on interpolated sea fog data, but conditions two, three, and four are met simultaneously, then the sea fog in the interpolated sea fog data from the automatic weather station is determined to be a false positive.

[0038] If CLDAS sea fog data identifies sea fog at a certain grid point, but conditions one, three, and four are met simultaneously, then the sea fog in the CLDAS sea fog data is determined to be a false alarm.

[0039] If the FY-4A satellite sea fog data identifies sea fog at a certain grid point, but conditions one, two, and four are met simultaneously, then the sea fog in the FY-4A satellite sea fog data is determined to be a false alarm.

[0040] If the Himawari satellite sea fog data identifies sea fog at a certain grid point, but conditions one, two, and three are met simultaneously, then the sea fog in the Himawari satellite sea fog data is determined to be a false positive.

[0041] Furthermore, the calculation method for the sea fog detection rate of different data in step S401 is as follows:

[0042] For the sea fog detection rate of the interpolated sea fog data from automatic weather stations, the sea fog detection rate under sea fog conditions is marked as 1.0, and the sea fog detection rate under non-sea fog conditions is marked as 0.0. The inverse distance weighting method is used to interpolate to the grid points.

[0043] The detection rate of sea fog data from CLDAS, FY-4A, and Himawari satellites is calculated by comparing the observation data from automatic weather stations with the CLDAS, FY-4A, and Himawari satellite sea fog data at their respective grid points, calculating the sea fog detection rate of different sea fog data in the target area, and applying the results to all grid points in the target area.

[0044] Compared with existing technologies, the multi-source data fusion method for sea fog detection described in this invention has the following advantages:

[0045] The multi-source data fusion sea fog detection method described in this invention leverages the advantages of multi-source data fusion technology to absorb various real-time data, thereby obtaining high-quality, spatiotemporally continuous real-time sea fog data. This method can compensate for the shortcomings of other sea fog detection methods such as automatic weather stations, numerical models, and satellite remote sensing in terms of spatial coverage, data accuracy, and identification of fog under clouds, thus improving the accuracy of sea fog monitoring and early warning.

[0046] Meanwhile, the multi-source sea fog fusion data was used to classify sea fog levels based on visibility, making it easier for meteorological service departments to provide more accurate services for the differentiated needs of port operations, ship pilotage, and maritime shipping. Attached Figure Description

[0047] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings:

[0048] Figure 1 This is a schematic diagram of a multi-source data fusion method for detecting sea fog according to an embodiment of the present invention;

[0049] Figure 2 This is a schematic diagram of real-time data from multi-source sea fog fusion as described in an embodiment of the present invention. Detailed Implementation

[0050] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0051] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0052] A multi-source data fusion method for sea fog detection includes the following steps:

[0053] S1: Acquire multi-source data for data fusion; multi-source data includes automatic weather station observation data (visibility and relative humidity and their quality control codes), FY-4A geostationary satellite L2 level data (fog monitoring data and cloud top temperature data), CLDAS data (visibility, relative humidity and surface temperature), CODAS sea surface temperature data, Himawari-9 geostationary satellite L1 level 5km full disk channel data and L2 level cloud attribute data.

[0054] S2: Perform quality control, spatial interpolation, and bias correction on the multi-source data obtained in S1; the process of processing the multi-source data is as follows:

[0055] S201: Automatic weather station observation data marked as "correct" are filtered out according to the quality control code. Automatic weather station observation data marked as "not quality controlled" (data from some offshore platform stations) are filtered out using the threshold check method (abnormal data are removed according to the element observation thresholds: visibility is 0~35000m, relative humidity is 0~100%). The data is interpolated to a 0.05°×0.05° grid using the inverse distance weighting method (weights are assigned according to the distance between the data points and the location to be interpolated, and a weighted average is performed, with the weights being the square of the inverse of the distance). Sea fog is identified based on the relative humidity and visibility at the interpolated grid points, and the automatic weather station interpolated sea fog data FOG1 is obtained.

[0056] In automatic weather stations, the Quality Control Code is a coding system used to identify and record the quality of observation data. The Quality Control Code is usually used to check and screen the quality of observation data collected from meteorological sensors or instruments. The Quality Control Code is built into the automatic weather station. This invention only uses the automatic weather station and does not discuss the working principle of the automatic weather station.

[0057] S202: Correct the CLDAS visibility and relative humidity using the automatic weather station observation data after quality control, identify sea fog based on the corrected visibility and relative humidity, and obtain CLDAS sea fog data FOG2.

[0058] S203: The FY-4A geostationary satellite fog monitoring data is converted from full-disk nominal grid data to latitude and longitude grid data using a lookup table method, and then interpolated to a 0.05°×0.05° grid using the nearest neighbor interpolation method (selecting the grid data closest to the target point and inserting it into the target point) to obtain FY-4A satellite sea fog data FOG3;

[0059] S204: Process the FY-4A geostationary satellite cloud top temperature data using the data processing method in S203 to obtain FY-4A satellite cloud top temperature data of 0.05°×0.05°;

[0060] S205: Use bilinear interpolation (perform linear interpolation once in the radial and once in the latitudinal directions for the data of four adjacent grid points to obtain the attribute value of the interpolation position) to interpolate the 0.25°×0.25° CODAS sea surface temperature data and the 0.0625°×0.0625° CLDAS land surface temperature data to a 0.05°×0.05° grid.

[0061] The method for correcting CLDAS data bias in step S202 is as follows:

[0062] The CLDAS visibility and relative humidity were interpolated to the location of the automatic weather station using a bilinear interpolation method. The visibility and relative humidity deviations of the CLDAS data relative to the automatic weather station were calculated (CLDAS data interpolation result minus the automatic weather station observation value). Then, the inverse distance weighting method was used to interpolate the CLDAS data deviation to a 0.05°×0.05° grid to obtain the CLDAS data deviation field. Finally, the CLDAS data was corrected using the CLDAS data deviation field (CLDAS data minus CLDAS data deviation field).

[0063] The process of processing Himawari-9 geostationary satellite data in S2 is as follows:

[0064] For nighttime data (solar altitude angle ≤ 5°), the brightness temperature data of the full disk of the Himawari-9 geostationary satellite with 14 channels (11.2 μm) and 7 channels (3.9 μm) were used to identify the stratus cloud and sea fog regions at night using the dual-channel threshold method (the difference between the 3.9 μm and 11.2 μm brightness temperature data is less than 2 K, which is considered to be stratus cloud or sea fog).

[0065] For daytime data (solar altitude angle greater than 5°), since stratus and sea fog have similar radiation characteristics, the stratus and stratocumulus regions in the Himawari-9 geostationary satellite cloud attribute data are extracted as the daytime stratus and sea fog regions.

[0066] To distinguish between stratus clouds and sea fog, based on the characteristic that fog top temperatures are usually close to or higher than surface / sea surface temperatures, while cloud top temperatures are lower than surface / sea surface temperatures, stratus clouds in Himawari-9 geostationary satellite stratus cloud and sea fog data were removed using cloud top temperature data from FY-4A satellite, as well as surface temperature data from CLDAS and sea surface temperatures from CODAS, to obtain Himawari-9 satellite sea fog data FOG4.

[0067] S3: Check whether the sea fog at each grid point of the FOG1, FOG2, FOG3 and FOG4 data is a misjudgment according to the consistency check method. If it is a misjudgment, correct it to no sea fog.

[0068] The method for identifying erroneous sea fog data based on consistency checks is as follows:

[0069] Four criteria were set: 1. The relative humidity interpolated to the grid point by the automatic weather station is less than 80% or the visibility is greater than 1.5km; 2. The relative humidity after correction by CLDAS is less than 80% or the visibility is greater than 1.5km; 3. Clear sky was detected by FY-4A satellite sea fog data; 4. Clear sky was detected by Himawari-9 geostationary satellite cloud attribute data or no stratus clouds and sea fog were detected by brightness temperature data.

[0070] If FOG1 data identifies sea fog at a certain grid point, but conditions two, three, and four are met simultaneously, then the sea fog in the FOG1 data is determined to be a false alarm.

[0071] If FOG2 data identifies sea fog at a certain grid point, but conditions one, three, and four are met simultaneously, then the sea fog in the FOG2 data is determined to be a false alarm.

[0072] If FOG3 data identifies sea fog at a certain grid point, but conditions one, two, and four are met simultaneously, then the sea fog in the FOG3 data is determined to be a false alarm.

[0073] If FOG4 data identifies sea fog at a certain grid point, but conditions one, two, and three are met simultaneously, then the sea fog in the FOG4 data is determined to be a false positive.

[0074] S4: Calculate the sea fog detection rate of FOG1, FOG2, FOG3 and FOG4 data, and use the sea fog detection rate as the fusion weight to generate sea fog range fusion data;

[0075] S401: Calculate the sea fog detection rate for FOG1, FOG2, FOG3 and FOG4 data respectively;

[0076] S402: At each grid point, the sea fog detection rate is used as a weight, and the four types of sea fog data are fused using the weighted average method (when fusing, grid points with fog are marked as 1, and grid points without fog are marked as 0) to obtain the fused data Merge_FOG1, which can be regarded as the "probability of sea fog occurrence".

[0077] S403: Define a critical value F, where there is sea fog when Merge_FOG1≥F, and no sea fog when Merge_FOG1<F;

[0078] S404: Compare the sea fog results of Merge_FOG1 with the identification results of automatic weather station observation data when F takes different values, and take the F value when "the sum of the number of automatic weather stations that falsely report sea fog and the number of automatic weather stations that fail to report sea fog is the smallest" (F). t ) as the optimal limit value;

[0079] S405: Set Merge_FOG1≥F t The area is marked as having sea fog, and Merge_FOG1 is less than F. t The area was marked as having no sea fog, and sea fog fusion data Merge_FOG2 was generated.

[0080] The calculation method for the sea fog detection rate of different data in step S401 is as follows:

[0081] For the sea fog detection rate of the interpolated sea fog data from automatic weather stations, the sea fog detection rate under sea fog conditions is marked as 1.0, and the sea fog detection rate under non-sea fog conditions is marked as 0.0. The inverse distance weighting method is used to interpolate to the grid points.

[0082] The sea fog detection rate for CLDAS sea fog data, FY-4A satellite sea fog data, and Himawari satellite sea fog data is calculated by comparing the automatic weather station observation data with the CLDAS, FY-4A, and Himawari satellite sea fog data at the same grid point, and calculating the sea fog detection rate of different sea fog data in the target area (sea fog detection rate = a / (a+b), where a is the number of stations where both the tested data and the automatic weather station observation data identify sea fog, and b is the number of stations where the tested data identifies sea fog but the automatic weather station observation data identifies no sea fog). The results are then applied to all grid points in the target area.

[0083] S5: Using the CLDAS visibility data corrected for bias in step S4, the sea fog area in Merge_FOG2 is further divided into four regions: dense fog, heavy fog, very heavy fog, and extremely heavy fog, generating multi-source sea fog fusion real-time data with a spatiotemporal resolution of 1h / 0.05°×0.05°.

[0084] Work process:

[0085] Data was obtained from the meteorological big data cloud platform of the meteorological department, including automatic weather station observation data (visibility and relative humidity and their quality control codes), FY-4A geostationary satellite L2 data (fog monitoring data and cloud top temperature data), CLDAS data (visibility, relative humidity and surface temperature), and CODAS sea surface temperature data for the target area. L1 level 5km full-disk channel data and L2 level cloud attribute data from the Himawari-9 geostationary satellite were obtained via FTP provided by P-TreeSystem.

[0086] Quality control was performed on the automatic weather station observation data, retaining only data whose data quality control code meant "correct". Automatic weather station observation data marked "not quality controlled" (including some offshore oil platform station observation data) were removed based on element observation thresholds, eliminating abnormal data with visibility exceeding 0-35000m or relative humidity exceeding 0-100%. Then, an inverse distance weighting method (assigning weights based on the distance between data points and the interpolation location, performing a weighted average with the weights being the square of the inverse of the distance) was used to interpolate the data to a 0.05°×0.05° grid. Fog was identified based on the relative humidity and visibility at the interpolated grid points (visibility less than 1km and relative humidity greater than 90% was considered foggy; otherwise, no fog was identified), resulting in the automatic weather station interpolated fog data FOG1. The visibility and relative humidity of CLDAS data were corrected using quality-controlled automatic weather station observation data. Specifically, bilinear interpolation was used (performing linear interpolation once in the radial and once in the latitudinal directions for data from four adjacent grid points to obtain the attribute values ​​of the interpolation location) to interpolate the CLDAS visibility and relative humidity to the location of the automatic weather station. The deviation of the CLDAS data relative to the automatic weather station data was calculated (the CLDAS data interpolation result minus the automatic weather station observation value). Then, the inverse distance weighting method was used to interpolate the CLDAS data deviation to a 0.05°×0.05° grid to obtain the CLDAS data deviation field. The CLDAS data deviation field was then used to correct the CLDAS data (CLDAS data minus the CLDAS data deviation field). Finally, fog was identified based on the corrected CLDAS visibility and relative humidity (visibility less than 1km and relative humidity greater than 90% were considered foggy; otherwise, no fog was identified), resulting in CLDAS fog data FOG2. The FY-4A geostationary satellite fog monitoring data was converted from full-disk nominal grid data to latitude and longitude grid data using a lookup table method. Nearest neighbor interpolation (selecting the grid data closest to the target point and inserting it into the target point) was then used to interpolate to a 0.05°×0.05° grid to obtain FY-4A geostationary satellite fog data FOG3. The same method was used to process the FY-4A geostationary satellite cloud top temperature data to obtain 0.05°×0.05° FY-4A geostationary satellite cloud top temperature data. Bilinear interpolation was used to interpolate the 0.25°×0.25° CODAS sea surface temperature data and the 0.0625°×0.0625° CLDAS land surface temperature data to a 0.05°×0.05° grid.

[0087] At night (solar altitude angle ≤ 5°), brightness temperature data from the full disk of the Himawari-9 geostationary satellite (14 channels, 11.2 μm and 7 channels, 3.9 μm) are used to identify stratus / fog regions using a dual-channel thresholding method (a difference of less than 2 K between the 3.9 μm and 11.2 μm brightness temperature data is considered to indicate the presence of stratus or sea fog). During the day (solar altitude angle > 5°), due to the similar radiation characteristics of stratus and fog, stratus and stratocumulus regions from the Himawari-9 geostationary satellite cloud attribute data are extracted as daytime stratus / fog regions. To distinguish between fog and stratus, based on the characteristic that fog top temperatures are usually close to or higher than surface / sea surface temperatures, while cloud top temperatures are lower, the difference between cloud top temperature and surface / sea surface temperature is calculated using cloud top temperature data from the FY-4A geostationary satellite, as well as CLDAS surface temperature and CODAS sea surface temperature data. This difference is then compared with the fog range identified by automatic weather stations to determine the optimal threshold value T for the temperature difference. d When the temperature difference is ≥ T d If the fog is identified as fog, it is identified as stratus cloud, thus obtaining the fog data FOG4 from the Himawari-9 geostationary satellite;

[0088] Check the consistency of fog identification data from FOG1, FOG2, FOG3, and FOG4, and correct any deviations. The specific checking method is as follows:

[0089] Four criteria are set: ① The relative humidity interpolated to the grid point by the automatic weather station is less than 80% or the visibility is greater than 1.5km (considering the possible bias in the data, the relative humidity and visibility thresholds for determining no fog are set to 80% and 1.5km, instead of 90% and 1km); ② The corrected CLDAS relative humidity is less than 80% or the visibility is greater than 1.5km; ③ The fog monitoring data from the FY-4A geostationary satellite detects clear skies; ④ The cloud attribute data from the Himawari-9 geostationary satellite detects clear skies or the brightness temperature data does not detect stratus / fog.

[0090] If FOG1 data identifies fog at a certain grid point, but conditions ②, ③, and ④ are met simultaneously, then the fog in the FOG1 data is determined to be a false positive.

[0091] If FOG2 data identifies a grid point as having fog, but conditions ①, ③, and ④ are met simultaneously, then the fog in the FOG2 data is determined to be a false positive.

[0092] If FOG3 data identifies a grid point as having fog, but conditions ①, ②, and ④ are met simultaneously, then the fog in the FOG3 data is determined to be a false positive.

[0093] If FOG4 data identifies a grid point as having fog, but conditions ①, ②, and ③ are met simultaneously, then the fog in the FOG4 data is determined to be a false positive.

[0094] The "fog detection rate" of FOG1, FOG2, FOG3, and FOG4 data were calculated separately. The "fog detection rate" of FOG1 data was obtained by marking the "fog detection rate" of automatic weather stations with fog as 1.0 and the "fog detection rate" of automatic weather stations without fog as 0.0, and then interpolating to a 0.05°×0.05° grid using the inverse distance weighting method. The FOG2, FOG3, and FOG4 data (referred to as the "tested data") were obtained by comparing the automatic weather station observation data with the tested data at their respective grid points. The "fog detection rate" of FOG2, FOG3, and FOG4 data within the target area was calculated (fog detection rate = a / (a+b), where a is the number of stations where both the tested data and the automatic weather station observation data identified fog, and b is the number of stations where the tested data identified fog but the automatic weather station observation data identified no fog). The results were then applied to all grid points in the target area.

[0095] At each grid point, using the "fog detection rate" as a weight, four types of fog data are fused using a weighted averaging method (during fusion, grid points with fog are marked as 1, and those without fog are marked as 0) to obtain the fused data Merge_FOG1, which can be considered as the "probability of fog occurrence". A critical value F is defined: fog is present when Merge_FOG1 ≥ F, and fog is absent when Merge_FOG1 is less than F. The fog range of Merge_FOG1 at different values ​​of F is compared with the automatic weather station identification results. The value of F that minimizes the sum of the number of automatic weather stations reporting fog and the number of automatic weather stations that missed reporting fog is selected. t () is taken as the optimal bounding value. Merge_FOG1≥F t The area is marked as foggy, and Merge_FOG1 is less than F. t The area was marked as fog-free, and sea fog fusion data Merge_FOG2 was generated.

[0096] Using the bias-corrected CLDAS visibility data, the foggy areas in the Merge_FOG2 data were further divided into four categories: dense fog (500m ≤ visibility < 1000m), heavy fog (200m ≤ visibility < 500m), very heavy fog (50m ≤ visibility < 200m), and extremely heavy fog (visibility < 50m). Specifically, when the Merge_FOG2 data identified fog and visibility ≥ 1000m, it was classified as a dense fog area. Finally, the fused data, classified by fog level, was stored in a netCDF file, generating multi-source sea fog fusion real-time data with a spatiotemporal resolution of 1h / 0.05°×0.05°.

[0097] Those skilled in the art will recognize that the units and method steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0098] In the several embodiments provided in this application, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the division of units described above is merely a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. The aforementioned units may or may not be physically separated. The components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the embodiments of the present invention according to actual needs.

[0099] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

[0100] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for detecting sea fog by multi-source data fusion, comprising the following steps: S1: obtaining multi-source data for data fusion; S2: performing quality control, spatial interpolation and bias correction on the multi-source data obtained in S1 to obtain sea fog range data; S3: checking the sea fog range data to identify the consistency of the sea fog and correcting the sea fog range data; S4: calculating the sea fog detection rate of the sea fog range data, taking the sea fog detection rate as the fusion weight to generate sea fog range fusion data; S5: using the data processed in S2 to divide the sea fog range fusion data into fog grades to generate multi-source sea fog fusion real-time data with specified spatial and temporal resolution; and the specific implementation process of step S4 is as follows: S401: respectively calculating the sea fog detection rates of the automatic weather station interpolated sea fog data, the CLDAS sea fog data, the FY-4A satellite sea fog data and the GOES satellite sea fog data; S402: using the weight average method to process the sea fog range data by taking the sea fog detection rates obtained in step S401 as the weight at each grid point to obtain fusion data; S403: defining a critical value, when the fusion data is greater than or equal to the critical value, there is sea fog, and when the fusion data is less than the critical value, there is no sea fog; S404: comparing the sea fog range of the fusion data at different critical value with the identification result of the automatic weather station to obtain the optimal limit value; and S405: marking the region with fusion data greater than or equal to the optimal limit value as having sea fog and marking the region with fusion data less than the optimal limit value as not having sea fog to generate sea fog range fusion data. 2.The method according to claim 1, wherein the multi-source data in step S1 comprises the visibility and relative humidity observation data of the automatic weather station, the L2 fog monitoring data and the cloud top temperature data of the FY-4A geostationary satellite, the CLDAS visibility, relative humidity and surface temperature data, the CODAS sea surface temperature data, and the L1 5km full-disk channel data and the L2 cloud property data of the GOES-9 geostationary satellite. 3.The method according to claim 2, wherein the process of spatial interpolation and bias correction on the multi-source data in step S2 is as follows: S201: performing quality control on the automatic weather station observation data according to the quality control code and threshold check method, interpolating the automatic weather station observation data to the grid points by using the inverse distance weighting method, identifying the sea fog to obtain the automatic weather station interpolated sea fog data; S202: performing bias correction on the CLDAS visibility and relative humidity data and identifying the sea fog to obtain the CLDAS sea fog data; S203: converting the FY-4A geostationary satellite fog monitoring data from the full-disk nominal grid data to the latitude and longitude grid data, interpolating the data to the equal latitude and longitude grid by using the inverse distance weighting method to obtain the FY-4A satellite sea fog data; S204: processing the FY-4A geostationary satellite cloud top temperature data by using the data processing method in S203; and S205: interpolating the CODAS sea surface temperature data and the CLDAS surface temperature data to the grid with specified resolution by using the bilinear interpolation method. ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ 4. The multi-source data fusion sea fog detection method according to claim 3, characterized in that: The quality control process of the sea fog range data in step S2 is as follows: For night data, according to the brightness temperature data of the GOES-9 stationary satellite, a double-channel threshold method is used to identify the night layer cloud and sea fog area; For daytime data, the layer cloud and stratocumulus cloud area of the GOES-9 stationary satellite cloud property data are extracted as the daytime layer cloud and sea fog area; The processed FY-4A stationary satellite cloud top temperature data and the CLDAS ground temperature and CODAS sea surface temperature data are used to remove the layer cloud data in the GOES-9 stationary satellite layer cloud and sea fog data, and the GOES satellite sea fog data is obtained.

5. The multi-source data fusion sea fog detection method according to claim 3, characterized in that: The method for correcting the CLDAS visibility and relative humidity data bias in step S202 is as follows: The CLDAS data is interpolated into the automatic weather station observation data using the bilinear interpolation method, the bias of the CLDAS data relative to the automatic weather station observation data is calculated, the bias of the CLDAS data is interpolated into the grid using the inverse distance weighting method, and the CLDAS bias field is obtained. The CLDAS data is corrected using the CLDAS bias field.

6. The multi-source data fusion sea fog detection method according to claim 4, characterized in that: The method for checking the consistency of sea fog range data identification in step S3 is as follows: Set four sea fog range data judgment grid points without fog conditions: Condition one, the relative humidity of the automatic weather station interpolated into the grid is less than 80% or the visibility is greater than 1.5 km; Condition two, the corrected CLDAS relative humidity is less than 80% or the visibility is greater than 1.5 km; Condition three, the FY-4A satellite sea fog data detects clear sky; Condition four, the cloud property data of the GOES-9 stationary satellite detects clear sky or the brightness temperature data does not detect layer cloud and sea fog; When the automatic weather station interpolated sea fog data identifies that a certain grid point has sea fog, but at the same time meets conditions two, three and four, it is determined that the sea fog of the automatic weather station interpolated sea fog data is misjudged; When the CLDAS sea fog data identifies that a certain grid point has sea fog, but at the same time meets conditions one, three and four, it is determined that the sea fog of the CLDAS sea fog data is misjudged; When the FY-4A satellite sea fog data identifies that a certain grid point has sea fog, but at the same time meets conditions one, two and four, it is determined that the sea fog of the FY-4A satellite sea fog data is misjudged; When the GOES satellite sea fog data identifies that a certain grid point has sea fog, but at the same time meets conditions one, two and three, it is determined that the sea fog of the GOES satellite sea fog data is misjudged.

7. The multi-source data fusion sea fog detection method according to claim 1, characterized in that: The calculation method of the sea fog detection rate of different data in step S401 is as follows: For the sea fog detection rate of the automatic weather station interpolated sea fog data, the sea fog detection rate of the automatic weather station with sea fog state is marked as 1.0, the sea fog detection rate of the automatic weather station without sea fog state is marked as 0.0, and the inverse distance weighting method is used for interpolation to the grid. For the sea fog detection rate of CLDAS sea fog data, FY-4A satellite sea fog data and HJ-1 satellite sea fog data, the automatic weather station observation data is compared with the CLDAS sea fog data, FY-4A satellite sea fog data and HJ-1 satellite sea fog data in the grid, and the sea fog detection rates of different sea fog data in the target area are calculated respectively. The results are applied to all grid points in the target area.

Images