Multi-source polar orbit satellite combined polar region daytime sea fog / low cloud detection method, equipment and medium

Abstract

The invention discloses a polar region daytime sea fog / low cloud detection method and device based on multi-source polar orbit satellites and a medium. The method comprises the steps that remote sensing image data collected by the multi-source polar orbit satellites are acquired and preprocessed; calculating a sea surface identification index by using data of two preset wave bands in the preprocessed remote sensing image data, and jointly removing areas related to seawater and sea ice / snow in the preprocessed remote sensing image data according to the sea surface identification index and near-infrared wave band reflectivity data in the remote sensing image data; for each pixel of a third preset wave band in the remote sensing image data from which the seawater, the sea ice and the snow are removed, calculating a gray-level co-occurrence matrix of the pixels, and counting a plurality of texture features based on the gray-level co-occurrence matrix; and inputting the statistical texture feature values corresponding to the remote sensing images corresponding to all polar orbit satellites into the trained random forest model, identifying and removing medium and high clouds in the random forest model, and obtaining sea fog / low clouds as the remainder. The influence of satellite difference and solar altitude angle change on the method is small.

Description

Technical Field

[0001] The present invention belongs to the field of image processing and meteorological services, and in particular relates to a polar daytime sea fog / low cloud detection method, equipment and medium combined with multi-source polar-orbiting satellites. Background Art

[0002] Trans-Arctic routes are significantly less time-consuming and labor-intensive than traditional routes. However, the melting of Arctic sea ice significantly contributes to the formation of sea fog and low clouds. The positive correlation between the occurrence of sea fog and low clouds and sea ice melting has been widely documented. For example, steam fog, formed when warm seawater comes into direct contact with cold air after melting sea ice, poses a serious potential threat to shipping safety. Therefore, conducting Arctic fog detection is crucial for navigation safety.

[0003] Polar-orbiting meteorological satellites offer high temporal and spatial resolution. A single satellite has a re-entry period of approximately 100 minutes and a spatial resolution of approximately 1 km. The combined presence of multiple medium-resolution multispectral polar-orbiting satellites allows for temporal resolution up to 5 minutes, providing a robust platform and data foundation for large-scale, accurate monitoring of Arctic sea fog and low clouds. However, differences in spectral bandwidth, ground response sensitivity, radiometric calibration, and solar zenith angle among multi-source sensors can still lead to significant errors in multi-source satellite sea fog and low cloud data.

[0004] Traditional Arctic sea fog monitoring relies primarily on ground-based observation stations, ships, and buoys. However, the harsh geographical environment of the Arctic, the sparse distribution of ground observation stations, and the limited distribution of ships and buoys result in a severe lack of monitoring data, making it difficult to effectively monitor sea fog over large areas of Arctic shipping routes.

[0005] Remote sensing monitoring algorithms based on visible-near-infrared reflectance and infrared radiation characteristics, widely used in mid- and low-latitude regions, face significant challenges in the Arctic. The Arctic's complex underlying surface, with interlaced sea ice and seawater, creates subtle differences in reflectivity between different landforms. These features are also susceptible to the influence of the solar zenith angle, making the reflectivity of sea ice / snow very similar to that of sea fog, making sea fog detection more difficult. Furthermore, the high latitude and weak solar radiation in the Arctic make it difficult for traditional remote sensing algorithms to accurately distinguish sea fog from other landforms. Summary of the Invention

[0006] The present invention provides a polar daytime sea fog / low cloud detection method, equipment and medium jointly using multiple polar-orbiting satellites, which are less affected by satellite differences and changes in solar altitude angles.

[0007] In order to achieve the above technical objectives, the present invention adopts the following technical solutions:

[0008] A polar daytime sea fog / low cloud detection method using a combination of multiple polar-orbiting satellites comprises:

[0009] Acquire remote sensing image data collected by multi-source polar-orbiting satellites and perform preprocessing;

[0010] Calculating a sea surface identification index using the first and second preset band data in the pre-processed remote sensing image data, and then removing areas related to seawater and sea ice / snow in the pre-processed remote sensing image data based on the sea surface identification index and near-infrared band reflectance data in the remote sensing image data;

[0011] For each pixel in the third preset band of the remote sensing image data with seawater and sea ice / snow removed, the gray level co-occurrence matrix is ​​calculated, and a number of texture features are statistically analyzed based on the gray level co-occurrence matrix;

[0012] The texture feature values ​​corresponding to the remote sensing images of all polar-orbiting satellites are input into the trained random forest model to identify the medium and high clouds. After removing the medium and high clouds, the remaining ones are sea fog / low clouds.

[0013] Furthermore, the remote sensing image data collected by the multi-source polar-orbiting satellite includes: MODIS data, AVHRR data, VIRR data and MERSI-Ⅱ data.

[0014] Furthermore, the preprocessing includes: first calibrating each band to reflectance or brightness temperature, then converting the calibrated remote sensing image data from different polar-orbiting satellites into the same coordinate system, and then performing mask processing on the Arctic land area in each remote sensing image data in the same coordinate system.

[0015] Furthermore, the reflectivity data of the 0.8μm band and the 0.6μm band are used to calculate the sea surface identification index, which is expressed as:

[0016]

[0017] Where, Represents the sea surface identification index, R ~0.8μm and R ~0.6μm are the reflectivities of each polar-orbiting satellite in the ~0.6μm and ~0.8μm bands, respectively. The ~0.6μm band and ~0.8μm band refer to the spectral bands covered by the channels with central wavelengths of 0.6μm and 0.8μm, respectively.

[0018] Furthermore, the sea surface identification index and the near-infrared band reflectance data in the remote sensing image data are used to jointly remove the areas related to sea water, sea ice and snow in the pre-processed remote sensing image data, specifically:

[0019] If the remote sensing image data is MODIS data, VIRR data, or MERSI-Ⅱ data, the sea surface identification index is used to distinguish whether it is sea water, sea ice, or snow at dawn and dusk, and the near-infrared band reflectance is used to distinguish whether it is sea water, sea ice, or snow at daytime.

[0020] If the remote sensing image data is AVHRR data, the sea surface identification index is used to distinguish whether it belongs to sea water, sea ice / snow at any time during the day.

[0021] Furthermore, for MODIS data at dawn and dusk, if the sea surface identification index of a pixel is less than 0.9, the pixel belongs to sea water, sea ice / snow; otherwise, it does not belong to sea water, sea ice / snow. For MODIS data during the daytime, if the near-infrared band reflectance of a pixel is less than 0.02, the pixel belongs to sea water, sea ice / snow; otherwise, it does not belong to sea water, sea ice / snow.

[0022] For VIRR data at dawn and dusk, if the sea surface identification index of a pixel is less than 0.8, the pixel belongs to sea water, sea ice, or snow; otherwise, it does not belong to sea water, sea ice, or snow. For VIRR data at daytime, if the near-infrared band reflectance of a pixel is less than 0.04, the pixel belongs to sea water, sea ice, or snow; otherwise, it does not belong to sea water, sea ice, or snow.

[0023] For MERSI-II data at dawn and dusk, if the sea surface identification index of a pixel is less than 0.9, the pixel belongs to sea water, sea ice, or snow; otherwise, it does not belong to sea water, sea ice, or snow. For MERSI-II data during the daytime, if the near-infrared band reflectance of a pixel is less than 0.04, the pixel belongs to sea water, sea ice, or snow; otherwise, it does not belong to sea water, sea ice, or snow.

[0024] For AVHRR data at any time during the day, if the sea surface identification index of a pixel is less than 0.75, the pixel belongs to sea water, sea ice / snow, otherwise it does not belong to sea water, sea ice / snow.

[0025] Furthermore, the third preset band selects ~12μm band data.

[0026] Furthermore, when calculating the gray-level co-occurrence matrix, four directions of 0°, 45°, 90° and 135° are selected to calculate a gray-level co-occurrence matrix respectively, and then the mean of the four obtained gray-level co-occurrence matrices is taken as the final gray-level co-occurrence matrix of the pixel; several texture features based on the gray-level co-occurrence matrix statistics include: mean, variance and contrast.

[0027] An electronic device includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor implements the above-mentioned polar daytime sea fog / low cloud detection method.

[0028] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the above-mentioned polar daytime sea fog / low cloud detection method.

[0029] Compared with existing technologies, the present invention has the following advantages: Based on the analysis of visible light spectral characteristics, the present invention combines the solar zenith angle with a specific threshold to eliminate the influence of spectral bandwidth, ground response sensitivity, radiometric calibration differences, and solar altitude variations among multi-source polar satellite sensors, thereby removing elements such as seawater and sea ice / snow. Subsequently, based on the analysis of the texture differences between sea fog / low clouds and mid-to-high clouds in the thermal infrared water vapor absorption band, the gray-level co-occurrence matrix and random forest method are combined to extract sea fog / low clouds. Experimental results show that the proposed method has high consistency and stability across different satellite platforms, and the detection effect is minimally affected by satellite differences and solar altitude angle variations. The algorithm's average detection accuracy (POD), false positive rate (FAR), and stability (CSI) are 75.39%, 12.35%, and 68.17%, respectively. The accuracy, false positive rate, and stability differences between different sensors are all less than 6%. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 The figure is a flow chart of the polar daytime sea fog / low cloud detection method according to the present invention.

[0031] Figure 2 It is a flow chart in an embodiment of the present invention.

[0032] Figure 3 This is the land and sea distribution, remote sensing imagery, and CALIOP transit map of the study area in the embodiments of the present invention.

[0033] Figure 4 This is a flowchart of the fusion of solar zenith angle data and multi-source polar-orbiting satellite images in an embodiment of the present invention.

[0034] Figure 5 This is a graph showing the time series statistics of the AVHRR reflectivity, brightness temperature, and SSRI of the observed object in an embodiment of the present invention.

[0035] Figure 6 These are the ~11μm and ~12μm channel images and texture comparison diagrams of the observed object in an embodiment of the present invention.

[0036] Figure 7 Schematic diagram of the four direction angles of the gray-level co-occurrence matrix.

[0037] Figure 8 The detection and CALIOP verification results of the method described in an embodiment of the present invention on August 5, 2021; (a) and (b) are the false color image and detection results at 12:10 on August 5, 2021, respectively. The blue line in (a) is the CALIOP motion trajectory, the red line in (b) represents the sea fog in the CALIOP VFM results, and (c) represents the CALIOP VFM observation results.

[0038] Figure 9 Figure 1 shows the detection results of the method described in an embodiment of the present invention from the morning to noon of July 15, 2016. (a) and (b) are the false color image and detection result at 03:30, respectively; (c) and (d) are the false color image and detection result at 05:35, respectively; (e) and (f) are the false color image and detection result at 07:31, respectively; (g) and (h) are the false color image and detection result at 10:05, respectively; (i) and (j) are the false color image and detection result at 11:26, respectively; the blue line in (i) is the CALIOP motion trajectory, and the red line in (j) represents the sea fog in the CALIOP VFM results.

[0039] Figure 10 Figure 1 shows the time series detection results from noon to dusk on July 15, 2016, using the method described in an embodiment of the present invention. (a) and (b) are the false color image and detection results at 13:07, (c) and (d) are the false color image and detection results at 15:05, (e) and (f) are the false color image and detection results at 16:40, (g) and (h) are the false color image and detection results at 17:42, and (i) and (j) are the false color image and detection results at 21:30.

[0040] Figures 11 to 15 The accuracy of sea fog detection results (UTC-11) for three different dates in 2016, 2018, 2019, 2020, and 2021 according to the embodiment of the present invention is shown. Figure 16 It is the overall average detection result of each satellite in these five years. DETAILED DESCRIPTION

[0041] The following is a detailed description of an embodiment of the present invention. This embodiment is based on the technical solution of the present invention, provides a detailed implementation method and a specific operation process, and further explains the technical solution of the present invention.

[0042] This embodiment provides a multi-source polar orbit satellite joint polar daytime sea fog / low cloud detection method, referring to Figure 1 、 Figure 2 As shown, the following steps are included:

[0043] Step 1: Obtain remote sensing image data collected by multi-source polar-orbiting satellites and perform preprocessing.

[0044] The remote sensing image data collected by the multi-source polar-orbiting satellites described in this embodiment include: MODIS data, AVHRR data, VIRR data, and MERSI-II data. MODIS data is remote sensing image data acquired by the Moderate-resolution Imaging Spectraradiometer (MRSI) aboard the polar-orbiting meteorological satellites Terra and Aqua, and is referred to as MODIS data. AVHRR data (Advanced Very High Resolution Radiometer) is remote sensing image data acquired by the Very High Resolution Radiometer (VHRR) aboard the polar-orbiting meteorological satellites NOAA-18, NOAA-19, METOP-A, and METOP-B, and is referred to as AVHRR data. VIRR data is remote sensing image data acquired by the Visible and Infra-Red Radiometer (VIRR) aboard the polar-orbiting meteorological satellites FY-3C and FY-3B, and is referred to as VIRR data. MERSI-II is remote sensing image data acquired by the Medium Resolution Spectral Imager-II aboard the polar-orbiting meteorological satellite FY-3D, and is referred to as MERSI-II data.

[0045] Step A1: Band calibration.

[0046] Step A1.1: There are 36 bands in the original MODIS data, of which bands 1-19 and 26 are calibrated as reflectance, and bands 20-25 and 27-36 are calibrated as brightness temperature.

[0047] Step A1.2: There are five bands in the original AVHRRR data, of which bands 1-2 are calibrated as reflectance and bands 3-5 are calibrated as brightness temperature.

[0048] Step A1.3: There are 10 bands in the original VIRR data, of which bands 1-8 are calibrated as reflectance and bands 9-10 are calibrated as brightness temperature.

[0049] Step A1.4: There are 25 bands in the original MERSI-II data, of which bands 1-19 are calibrated as reflectance and bands 20-25 are calibrated as brightness temperature.

[0050] Step A2: Geometric positioning. Convert the calibrated MODIS, AVHRR, VIRR, and MERSI-II data to the WGS-84 coordinate system, select the UTM projection, and resample the data spatial resolution to 1 km.

[0051] Step A3: Land masking: Use the standard Arctic land boundary SHP file to mask the MODIS, AVHRR, VIRR, and MERSI-II data, assigning the pixel values ​​of the land area to 0.

[0052] Step A4: CALIOP data is a laser sounding radar data jointly developed by CNES and NASA. The vertical feature layer distribution data (VFM) in the CALIOP data level 2 product is selected to verify the sea fog / low cloud detection results of this embodiment. The CALIOP VFM product has a total of eight feature layer classification identifiers, and sea fog, low cloud, and medium-high cloud are uniformly identified as Cloud. Clouds with a height of less than 400m are defined as sea fog. In addition, if the base height of the cloud is 0m, that is, it is close to the ocean surface, it is also defined as sea fog even if its top height is greater than 400m.

[0053] Figure 3 The land and sea distribution, remote sensing images, and CALIOP radar transit maps of the study area of ​​this embodiment are displayed.

[0054] Since the present embodiment uses the solar zenith angle data later, the solar zenith angle data is further fused with the pre-processed remote sensing image data in the data pre-processing operation. Figure 4 The specific process is as follows:

[0055] Step B1: Resampling the SZA data. The MODIS data provides SZA data, but the size of this dataset (271 × 406) differs from the size of the preprocessed MODIS data (1354 × 2030). Therefore, the SZA data needs to be resampled. Cubic convolution resampling is used to uniformly resample the SZA data to 1354 × 2030.

[0056] Step B2: Fuse the solar zenith angle data after step B1 with the MODIS, AVHRR, VIRR, and MERSI-II data after step A1. The specific steps are as follows:

[0057] Step B2.1: Geo-reference. Select the GCP ground point control method for geo-reference. Pixels in the solar zenith angle data are matched and mapped pixel by pixel with those in the MODIS, AVHRR, VIRR, and MERSI-II images.

[0058] Step B2.2: Land masking of solar zenith angle data. Use the standard Arctic land boundary SHP file to mask the solar zenith angle data and assign the pixel values ​​of the land area to 0.

[0059] Step B2.3: Data fusion: The solar zenith angle data obtained in step B2.2 are fused with the MODIS, AVHRR, VIRR, and MERSI-II images obtained in step A3 to obtain the MODIS, AVHRR, VIRR, and MERSI-II images with fused solar zenith angles.

[0060] Step 2: Removal of seawater, sea ice / snow.

[0061] The reflectivity of seawater and sea ice / snow decreases with increasing wavelength. The multi-source polar satellite sensors described in this embodiment are all equipped with ~0.6μm and ~0.8μm channels, for example Figure 5 (a) and Figure 5 (b) is the reflectivity curve of the daytime observation object on MODIS sensors B1 (0.630μm) and B2 (0.850μm). The reflectivity of seawater and sea ice / snow in the ~0.8μm channel is lower than that of ~0.6μm, while sea fog / low clouds do not have this feature. Figure 5 (c) is the time series statistical result of the sea surface identification index SSRI of the observed object AVHRR data. The SSRI values ​​of sea water, sea ice / snow, sea fog / low cloud, and medium and high cloud remain stable and have a low correlation with the solar zenith angle. Among them, the SSRI values ​​of sea fog / low cloud and medium and high cloud are all higher than or close to 0.8, while the SSRI values ​​of sea water and sea ice / snow are all lower than 0.8. Therefore, for the existing technology based on near-infrared reflectivity R NIR The problem that the removal of seawater, sea ice / snow based on the threshold is greatly affected by the solar altitude angle is solved in the present invention. The method for removing seawater, sea ice / snow based on the SSRI threshold is constructed: the sea surface recognition index is calculated using the preset band data in the pre-processed remote sensing image data, and then the areas related to seawater, sea ice and snow in the pre-processed remote sensing image data are jointly removed based on the sea surface recognition index and the near-infrared band reflectivity data in the remote sensing image data.

[0062] Step C1: Calculate the sea surface identification index using the reflectivity data of the 0.8μm and 0.6μm wavelength bands. The specific calculation formula is:

[0063] (1)

[0064] Where R ~0.8μm and R ~0.6μm They are the reflectivity of each polar-orbiting satellite near the 0.6μm and 0.8μm bands, respectively. The specific bands are B1 (0.645μm) and B2 (0.859μm) in MODIS, B1 (0.650μm) and B2 (0.865μm) in MERSI-Ⅱ, B1 (0.630μm) and B2 (0.865μm) in VIRR, and B1 (0.630μm) and B2 (0.862μm) in AVHRR.

[0065] Step C2: Remove seawater, sea ice / snow.

[0066] During periods of high solar altitude (0º≤SZA<75º), the reflectivity of seawater and sea ice / snow in the near-infrared band is low (almost close to 0), while sea fog / low clouds do not have this characteristic (RNIR is between 7% and 15%). However, during the dawn and dusk periods when the solar altitude is low (75º≤SZA≤90º), the reflectivity of the three is relatively close, and the sea fog detection algorithm based on the near-infrared reflectivity RNIR threshold has a large error. This is used to construct the R threshold for different sensors at different solar zenith angles. NIR And the segmentation threshold lookup table of SSRI is shown in Table 1:

[0067]

[0068] According to the segmentation threshold lookup table constructed above, sea water, sea ice / snow are identified and removed:

[0069] For MODIS data collected during the twilight hours (75° ≤ solar altitude angle SZA ≤ 90°), if the sea surface identification index of a pixel is less than 0.9, the pixel belongs to seawater, sea ice, or snow; otherwise, it does not belong to seawater, sea ice, or snow. For MODIS data collected during the daytime hours (0° ≤ solar altitude angle < 75°), if the near-infrared reflectance of a pixel is less than 0.02, the pixel belongs to seawater, sea ice, or snow; otherwise, it does not belong to seawater, sea ice, or snow.

[0070] For VIRR data at dawn and dusk, if the sea surface identification index of a pixel is less than 0.8, the pixel belongs to sea water, sea ice, or snow; otherwise, it does not belong to sea water, sea ice, or snow. For VIRR data at daytime, if the near-infrared band reflectance of a pixel is less than 0.04, the pixel belongs to sea water, sea ice, or snow; otherwise, it does not belong to sea water, sea ice, or snow.

[0071] For MERSI-II data at dawn and dusk, if the sea surface identification index of a pixel is less than 0.9, the pixel belongs to sea water, sea ice, or snow; otherwise, it does not belong to sea water, sea ice, or snow. For MERSI-II data during the daytime, if the near-infrared band reflectance of a pixel is less than 0.04, the pixel belongs to sea water, sea ice, or snow; otherwise, it does not belong to sea water, sea ice, or snow.

[0072] For AVHRR data at any time during the day, if the sea surface identification index of a pixel is less than 0.75, the pixel belongs to sea water, sea ice or snow; otherwise, it does not belong to sea water, sea ice or snow.

[0073] Step 3: Calculate the gray level co-occurrence matrix for each pixel in the third preset band of the remote sensing image data with sea water, sea ice and snow removed, and count several texture features based on the gray level co-occurrence matrix.

[0074] On the basis of removing surface elements such as seawater, sea ice / snow, it is necessary to further distinguish and extract sea fog / low clouds from medium and high clouds. The tops of sea fog / low clouds are smooth, while the tops of medium and high clouds are undulating, and the former is lower than the latter. These characteristics make sea fog / low clouds have uniform texture and consistent brightness in the thermal infrared channel, while the medium and high cloud areas have broken texture, strong brightness changes, and are accompanied by shadows. Figure 6 As shown in Figure 2, it constitutes the physical basis for remote sensing extraction of sea fog / low clouds and medium and high clouds.

[0075] Due to water vapor sensitivity, the texture of the objects observed in the ~12μm channel is more different than that in the ~11μm channel, especially in the mid-to-high clouds, where the texture fluctuations are more dramatic ( Figure 6 The blue circle frame) further counted the brightness temperature time series characteristics of the objects observed by the AVHRR sensor Band5 (12.00μm) during the day ( Figure 5 (d)) Except for seawater, the brightness temperature of the observed object is basically unaffected by changes in the solar altitude. Therefore, this embodiment of the present invention selects the brightness temperature in the 12 μm band and uses the gray-level co-occurrence matrix to calculate the texture features.

[0076] In view of the significant texture differences between sea fog / low clouds and medium and high clouds, this embodiment reflects the frequency and variation characteristics of the image in grayscale combination by calculating the joint distribution of pixel pairs in different directions and distances. Starting from the pixel point, when the distance is d and the direction is θ, the pixel value is The normalized probability of , the definition formula is:

[0077] (2)

[0078] In the formula For two pixels in The distance difference value in the direction, Represents the grayscale level. In order to improve the computational efficiency, this embodiment compresses the image grayscale to 3 bits and uses a 5×5 window with a distance of 1 to traverse the image; the direction θ is as follows Figure 7 As shown in Figure 1, four directions of 0°, 45°, 90° and 135° are selected to fully reflect the joint distribution characteristics of grayscale, and the mean of the grayscale co-occurrence matrix in the four directions is selected as the eigenvalue of the central pixel.

[0079] Eight texture statistics are defined on the gray-level co-occurrence matrix to describe the gray-level correlation and spatial distribution of the image. Considering the information redundancy caused by the connectivity between the statistics, this example selects the mean, variance, and contrast statistics to distinguish sea fog / low clouds from medium and high clouds. Their calculation formulas are as follows:

[0080] (3)

[0081] The mean describes the regularity of the texture of the object. The higher the mean value, the more uniform the image is. Indicates the number of gray levels (0~7), is the probability value corresponding to the pixel pair sum i in the gray-level co-occurrence matrix.

[0082] (4)

[0083] Variance describes the periodicity of the object texture. The higher the Variance value, the more obvious the repetition and regularity of the texture features in the image. are two index variables, used to traverse the rows and columns in the gray-level co-occurrence matrix, corresponding to different gray levels. Represents the mean, which refers to the mean of the pixel grayscale values ​​in the selected area. The element value at position (i, j) in the gray-level co-occurrence matrix represents the joint probability that a pixel with gray level i and a pixel with gray level j appear at the same time.

[0084] (5)

[0085] Contrast describes the grayscale difference between image pixels. The higher the Contrast value, the rougher the texture in the image.

[0086] Step 4: Input the statistical texture feature values ​​corresponding to the ~12μm band of all polar-orbiting satellite remote sensing images into the trained random forest model to identify the medium and high clouds. After removing the medium and high clouds, the remaining ones are sea fog and low clouds.

[0087] The texture feature values ​​corresponding to the ~12μm band of polar-orbiting satellite remote sensing images are a set of texture feature statistics at multiple scales, including directionality, uniformity, and complexity. The combination of these texture feature statistics gives the images high-dimensional characteristics. Random Forest (RF) is not susceptible to the "curse of dimensionality" and can effectively avoid the computational and storage problems caused by high-dimensional data. At the same time, it can achieve good classification results without excessive parameter tuning.

[0088] In this embodiment of the present invention, the texture statistics (mean, variance, and contrast) obtained by extracting the remote sensing data corresponding to multi-source polar satellites are input into a trained random forest model to identify medium and high clouds. After removing the medium and high clouds, the remaining parts are sea fog and low clouds.

[0089] The gray level co-occurrence matrix GLCM and random forest RF parameter settings in this embodiment are shown in Table 2:

[0090]

[0091] Next, we verify the accuracy of the onion model test results of the present invention:

[0092] Fifteen fog cases from different dates, including 2016 and the summer of 2018-2021 (June-August), were selected to quantitatively verify the universality of the algorithm. Figures 11 to 16 The results show that the proposed method has an average correct detection rate (POD) of 75.39%, an average false alarm rate (FAR) of 12.35%, and a stability index (CSI) of 68.17%. The three indicators vary less than 6% between different sensors, indicating that the algorithm's detection effect is basically unaffected by sensor changes.

[0093] Figure 8 is the detection result of MODIS / Aqua during this period, such as Figure 8 (a) Figure 8 As shown in (b), a large number of missed detections occurred in the sea fog detection results. Compared with the detection results of CALIOP ( Figure 8 (white dashed box in (c)) and Figure 8 (a) (white solid line circle), the missed sea fog mainly appears in the thin sea fog area with obvious texture shadows, thin thickness and complex underlying surface distribution. The thin sea fog limits its shielding of the underlying surface's radiant energy, which causes the sea fog in this part to mix with the underlying surface's energy at the brightness temperature value of ~12um, that is, spectral mixing occurs. Due to the obvious difference in brightness temperature between sea ice / snow and seawater in the underlying surface, and their staggered distribution, the temperature fluctuation of this part of the sea fog is relatively large. At the same time, due to the movement of sea air, compared with normal sea fog, the texture of this part of the thin sea fog is relatively rough, with a "shadow" phenomenon, that is, the fog top has high and low fluctuations within a short distance. The pixel mixing, the underlying surface temperature fluctuations and the thin fog shadow make the texture characteristics of this part of the thin sea fog rough and change more dramatically, so it is misclassified as cloud.

[0094] like Figure 9 and Figure 10 As shown in Figure 2, two fog areas with different thematic forms appear in the time series detection results during the day. Figure 9The middle green circle marks an independent leaf-shaped fog area. The algorithm relatively completely detected the body changes and movement process of this leaf-shaped fog area in the study area. It was generated in the floating ice area outside the study area, in the northwest of the Chukchi Sea, and entered the study area at 03:30 in the morning. After that, it gradually moved southward under the influence of wind and showed signs of main body breaking up and dissipating at 17:42 in the afternoon. By 21:30 at dusk, the leaf-shaped fog area had disappeared from the study area.

[0095] Satellite time-series imagery also reveals that the density and morphology of this leaf-shaped fog area correlate with the intensity of solar radiation, indicating that it is affected by diurnal variations. From 03:30 to 05:35, the leaf-shaped fog area was dense and extensive, nearly completely obscuring radiation information from the underlying surface. As solar radiation increased, the main body of the fog area began to shrink inward, gradually becoming narrower and thinner, while its density gradually thinned. At 13:07, the fog area reached its lowest density, exposing the underlying sea surface. From 13:07 to 15:05, as the sun's altitude decreased and radiation weakened, the fog area gradually expanded, and its density increased again. Subsequently, influenced by wind, the leaf-shaped fog area began to break up from the main veins, and its density decreased. By 15:05, the leaf-shaped fog area had broken up into a circular shape, and radiation information from the underlying sea surface was visible. By 21:30, the fog area had disappeared from the study area.

[0096] In addition to the leaf-shaped sea fog / low cloud area, there is also a large area of ​​surface advection fog in the study area. This advection fog occurred in the ice-edge zone in the southern part of the study area. Figure 9 The time series detection results shown are warm and humid air masses ( Figure 9 The central white box (center) entered the study area around 03:30 AM and, after encountering cold, began to form advection fog. Subsequently, the warm, moist air mass, influenced by winds, gradually moved southeastward, transforming into sea fog / low cloud. By 3:05 PM, the warm, moist air mass had completely transformed into sea fog / low cloud. Thereafter, the main body of the sea fog / low cloud area remained relatively stable, gradually moving eastward under the influence of winds. By approximately 21:30 PM, the entire area had drifted toward the coast of North America.

[0097] CALIOP data was used to quantitatively verify the detection results of the proposed method. The detection evaluation indicators include the correctness rate (POD), the false alarm rate (FAR), and the reliability index (CSI). POD represents the detection accuracy; a higher POD value indicates a higher algorithm accuracy. CSI represents the reliability index; a higher CSI value indicates a more reliable algorithm. The calculation method is as follows:

[0098]

[0099]

[0100]

[0101] Where: N H The number of pixels indicating that both CALIOP and MODIS detection results are sea fog / low cloud; N F represents the number of falsely detected pixels (i.e., the number of pixels where CALIOP is not sea fog / low cloud but MODIS is sea fog / low cloud). Otherwise, N M Represents the number of missed pixels, that is, the number of pixels that CALIOP detected as sea fog / low cloud but MODIS detected as non-sea fog / low cloud.

[0102] The present invention combines nine polar-orbiting meteorological satellites to detect Arctic sea fog and low clouds, demonstrating high consistency and stability across different satellite platforms. The detection results are minimally affected by satellite differences and variations in solar altitude. The fog detection results demonstrate that the present method comprehensively captures the dynamics of sea fog generation and dissipation, and its detection effectiveness is unaffected by factors such as variations in the solar zenith angle and differences in satellite sensors, further confirming its reliability.

[0103] The above embodiments are preferred embodiments of the present application. Ordinary technicians in this field can also make various changes or improvements on this basis. Without departing from the overall concept of the present application, these changes or improvements should fall within the scope of protection required by the present application.

Claims

1. A method for detecting polar daytime sea fog / low cloud using a combination of multiple polar-orbiting satellites, characterized in that: include: Acquire remote sensing image data collected by multi-source polar-orbiting satellites and perform preprocessing; Calculating a sea surface identification index using the first and second preset band data in the pre-processed remote sensing image data, and then removing areas related to seawater and sea ice / snow in the pre-processed remote sensing image data based on the sea surface identification index and near-infrared band reflectance data in the remote sensing image data; For each pixel in the third preset band of the remote sensing image data with seawater and sea ice / snow removed, the gray level co-occurrence matrix is ​​calculated, and a number of texture features are statistically analyzed based on the gray level co-occurrence matrix; The texture feature values ​​corresponding to the remote sensing images of all polar-orbiting satellites are input into the trained random forest model to identify the medium and high clouds. After removing the medium and high clouds, the remaining ones are sea fog / low clouds.

2. The polar daytime sea fog / low cloud detection method according to claim 1, characterized in that: The remote sensing image data collected by the multi-source polar-orbiting satellite includes: MODIS data, AVHRR data, VIRR data and MERSI-Ⅱ data.

3. The polar daytime sea fog / low cloud detection method according to claim 1, characterized in that: The preprocessing includes: first calibrating each band to reflectivity or brightness temperature, then converting the calibrated remote sensing image data from different polar-orbiting satellites into the same coordinate system, and then performing masking on the Arctic land area in each remote sensing image data in the same coordinate system.

4. The polar daytime sea fog / low cloud detection method according to claim 1, characterized in that: The sea surface identification index is calculated using the reflectivity data of the 0.8μm band and the 0.6μm band, which is expressed as: ； Where, Represents the sea surface identification index, R ~0.8μm and R ~0.6μm are the reflectivities of each polar-orbiting satellite in the ~0.6μm and ~0.8μm bands, respectively. The ~0.6μm band and ~0.8μm band refer to the spectral bands covered by the channels with central wavelengths of 0.6μm and 0.8μm, respectively.

5. The polar daytime sea fog / low cloud detection method according to claim 1, characterized in that: The method of removing the areas related to sea water, sea ice and snow in the pre-processed remote sensing image data based on the sea surface identification index and the near-infrared band reflectance data in the remote sensing image data is as follows: If the remote sensing image data is MODIS data, VIRR data, or MERSI-Ⅱ data, the sea surface identification index is used to distinguish whether it is sea water, sea ice, or snow at dawn and dusk, and the near-infrared band reflectance is used to distinguish whether it is sea water, sea ice, or snow at daytime. If the remote sensing image data is AVHRR data, the sea surface identification index is used to distinguish whether it belongs to sea water, sea ice / snow at any time during the day.

6. The polar daytime sea fog / low cloud detection method according to claim 5, characterized in that: For MODIS data at dawn and dusk, if the sea surface identification index of a pixel is less than 0.9, the pixel belongs to sea water, sea ice, or snow; otherwise, it does not belong to sea water, sea ice, or snow. For MODIS data during the daytime, if the near-infrared band reflectance of a pixel is less than 0.02, the pixel belongs to sea water, sea ice, or snow; otherwise, it does not belong to sea water, sea ice, or snow. For VIRR data at dawn and dusk, if the sea surface identification index of a pixel is less than 0.8, the pixel belongs to sea water, sea ice, or snow; otherwise, it does not belong to sea water, sea ice, or snow. For VIRR data at daytime, if the near-infrared band reflectance of a pixel is less than 0.04, the pixel belongs to sea water, sea ice, or snow; otherwise, it does not belong to sea water, sea ice, or snow. For MERSI-II data at dawn and dusk, if the sea surface identification index of a pixel is less than 0.9, the pixel belongs to sea water, sea ice, or snow; otherwise, it does not belong to sea water, sea ice, or snow. For MERSI-II data during the daytime, if the near-infrared band reflectance of a pixel is less than 0.04, the pixel belongs to sea water, sea ice, or snow; otherwise, it does not belong to sea water, sea ice, or snow. For AVHRR data at any time during the day, if the sea surface identification index of a pixel is less than 0.75, the pixel belongs to sea water, sea ice / snow, otherwise it does not belong to sea water, sea ice / snow.

7. The polar daytime sea fog / low cloud detection method according to claim 1, characterized in that: The third preset band selects ~12μm band data.

8. The polar daytime sea fog / low cloud detection method according to claim 1, characterized in that: When calculating the gray-level co-occurrence matrix, select four directions of 0°, 45°, 90° and 135° to calculate a gray-level co-occurrence matrix respectively, and then take the average of the four gray-level co-occurrence matrices obtained as the final gray-level co-occurrence matrix of the pixel; Several texture features based on gray-level co-occurrence matrix statistics, including mean, variance and contrast.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the computer program is executed by the processor, the processor is caused to implement the method according to any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images