A method for extracting bare ice glaciers based on remote sensing cloud computing platform

Abstract

The present invention discloses a method for extracting bare ice glaciers based on a remote sensing cloud computing platform. The method comprises extracting optical remote sensing images of a study area over a certain period of time based on the remote sensing cloud computing platform, processing them to obtain a surface reflectance image, identifying ice and snow and cloud-covered pixels in the surface reflectance image, and performing mask calculation on these pixels in the surface reflectance image to obtain areas free of ice and snow and cloud cover, sorting the masked surface reflectance images according to cloud cover, using the surface reflectance image with the lowest cloud cover as the base image, and sequentially superimposing the remaining surface reflectance images on the base image to fill in the masked ice and snow and cloud-covered pixels in the base image to obtain a maximally filled image, and encoding and decoding the filled image to obtain a glacier distribution vector dataset. The method has the advantages of eliminating the interference problem of snow and cloud cover during glacier extraction and enabling automatic identification and extraction of glacier distribution.

Description

Technical Field

[0001] The present invention relates to the technical field of remote sensing image-oriented ground feature classification, and in particular to a method for extracting bare ice glaciers based on a remote sensing cloud computing platform. Background Art

[0002] Formed from the accumulation and evolution of snowfall or solid precipitation, glaciers are a vital component of the cryosphere and are considered sensitive indicators of climate change. In recent years, glaciers on the Qinghai-Tibet Plateau have been retreating, and glacial meltwater has increased. This has not only altered surface water reserves but also increased the risk of disasters such as floods, ice and snow avalanches, posing significant challenges to protecting public property and social stability. Existing glacier monitoring technologies primarily consist of two types: field surveys and remote sensing. Field surveys are relatively accurate, but they are time-consuming, labor-intensive, and difficult to monitor over large areas, hindering the assessment of overall glacier trends. Remote sensing overcomes this limitation by extracting and analyzing the unique spectral signatures of glaciers in optical images, facilitating large-scale glacier research. However, the image extraction process is susceptible to interference from snow and cloud cover, making accurate glacier identification difficult. Therefore, a rapid glacier interpretation technology that overcomes the interference of snow and cloud cover has become a top priority in glacier research. Summary of the Invention

[0003] The purpose of the present invention is to provide a method for extracting bare ice glaciers based on a remote sensing cloud computing platform, thereby solving the aforementioned problems existing in the prior art.

[0004] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0005] A method for extracting bare ice glaciers based on a remote sensing cloud computing platform includes the following steps:

[0006] S1. Calculation of surface reflectance image mask:

[0007] Identify ice, snow, and cloud-covered pixels in several surface reflectance images of the target area acquired via a remote sensing cloud computing platform, and perform mask calculations on these pixels in each surface reflectance image to obtain areas free of ice, snow, and cloud. Sorting the masked surface reflectance images according to cloud cover creates an image set.

[0008] S2, Image filling:

[0009] The surface reflectance image with the lowest cloud cover in the image set is used as the base image, and the remaining surface reflectance images are superimposed on the base image in sequence to fill in the masked ice, snow and cloud cover pixels in the base image, thereby obtaining a filled image with the greatest degree of filling.

[0010] S3. Obtain glacier distribution data:

[0011] The null pixels in the filled image are encoded, and the filled image is converted into a binary image using a raster calculator. The binary image is vectorized to obtain a glacier distribution vector dataset.

[0012] Preferably, step S1 specifically includes the following contents:

[0013] S11. Use the CFMask algorithm to identify the pixels covered by ice, snow, and clouds in each surface reflectance image;

[0014] S12. Select the "QA_PIXEL" band in the surface reflectance image and perform binary bit operations on the clouds, cloud shadows, and snow represented in binary. In the calculation result, pixels with a pixel value of 1 represent pixels covered by ice, snow, and clouds, and pixels with a pixel value of 0 represent actual ground objects. Mask out the pixels with a pixel value of 1 to obtain the non-ice, snow, and non-cloud covered areas in the surface reflectance image.

[0015] S13. Sort the surface reflectance images in the image set from high to low according to the cloud cover attribute field "CLOUD_COVER" in the surface reflectance image; the surface reflectance image at the back has the least cloud cover.

[0016] Preferably, step S2 specifically includes the following contents:

[0017] S21. Selecting the surface reflectance image with the least cloud cover that is sorted last in the image set as the bottom image; the bottom image includes masked pixels and unmasked pixels, and the masked pixels are ice, snow, and cloud-obscured areas;

[0018] S22, starting from the second-to-last surface reflectance image in the image set, sequentially superimpose them on the underlying image, and fill the corresponding mask pixels in the underlying image to obtain a filled image with the maximum degree of filling;

[0019] M=M1+M2+M3+…M x …+M n

[0020]

[0021] Among them, M1, M2, M3, ..., M x ,…,M n are the 1st, 2nd, 3rd, ..., x, ..., nth surface reflectance images superimposed on the underlying image; x ij It is the pixel value in the i-th row and j-th column of the x-th surface reflectance image superimposed on the underlying image.

[0022] Preferably, step S3 specifically includes the following contents:

[0023] S31, assigning a value of -99 to the null-valued pixels in the filled image, and extracting the pixels with a value of -99 by decoding with a raster calculator to convert the filled image into a binary image of 0 and 1;

[0024] S32, performing median filtering on the binary image through a 3×3 moving window to eliminate individual interference values;

[0025] S33. Convert the filtered image into a vector, extract the vector elements whose field "Label" is 1, and merge the vectors to obtain a glacier distribution vector dataset, that is, the glacier distribution range.

[0026] Preferably, before step S1, step S0 is also included, obtaining a plurality of Landsat optical remote sensing images based on a remote sensing cloud computing platform, and obtaining a plurality of surface reflectance images after preprocessing them; step S0 specifically includes the following contents:

[0027] S01. Based on the Google Earth Engine remote sensing cloud computing platform, extract and select several Landsat optical remote sensing images according to the study area and the time range of image extraction;

[0028] S02, performing radiation correction, geometric correction, and image alignment on each of the screened optical remote sensing images to obtain a number of pre-processed images after row and column pixel alignment;

[0029] S03. Multiply all pixel values ​​of the optical bands in the plurality of pre-processed images by the scale coefficient, and then subtract the set value to obtain a plurality of surface reflectance images.

[0030] The beneficial effects of this invention are as follows: 1. The method solves the problem of interference from snow and clouds during glacier extraction, enabling automatic identification and extraction of glacier distribution. 2. The method effectively avoids interference from snow accumulation and cloud cover, enabling refined dynamic analysis of glaciers, which is crucial for studying the response mechanisms of glaciers to global climate change. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 1 is a flow chart of a method according to an embodiment of the present invention;

[0032] Figure 2 3 is a schematic diagram of the result of glacier identification using the method of the present invention in an embodiment of the present invention. DETAILED DESCRIPTION

[0033] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0034] Example 1

[0035] like Figure 1 As shown, this embodiment provides a method for extracting bare ice glaciers based on a remote sensing cloud computing platform. This method, based on the Google Earth Engine remote sensing cloud computing platform, acquires several Landsat remote sensing images of the target area over the past 1 to 2 years. The CFMask algorithm is used to identify pixels covered by clouds, snow, and glaciers in the images. Mask calculation is then used to extract areas of non-ice, snow, and non-cloud imagery. An image overlay algorithm is used to supplement the pixel values ​​masked out by cloud cover and snow cover, indirectly obtaining a glacier distribution map. The method of the present invention mainly includes the following steps:

[0036] S1. Calculation of surface reflectance image mask:

[0037] Identify ice, snow, and cloud-covered pixels in several surface reflectance images of the target area acquired via a remote sensing cloud computing platform, and perform mask calculations on these pixels in each surface reflectance image to obtain areas free of ice, snow, and cloud. Sorting the masked surface reflectance images according to cloud cover creates an image set.

[0038] S2, Image filling:

[0039] The surface reflectance image with the lowest cloud cover in the image set is used as the base image, and the remaining surface reflectance images are superimposed on the base image in sequence to fill in the masked ice, snow and cloud cover pixels in the base image, thereby obtaining a filled image with the greatest degree of filling.

[0040] S3. Obtain glacier distribution data:

[0041] The null pixels in the filled image are encoded, and the filled image is converted into a binary image using a raster calculator. The binary image is vectorized to obtain a glacier distribution vector dataset.

[0042] Before executing step S1, it is necessary to extract the image of the target area. Therefore, it can be seen that the method of the present invention mainly includes four parts: image extraction, mask calculation, image filling, and glacier acquisition; the following describes these four parts separately:

[0043] 1. Image Extraction

[0044] This part corresponds to step S0, that is, obtaining a number of Landsat optical remote sensing images based on the remote sensing cloud computing platform, and obtaining a number of surface reflectance images after preprocessing them;

[0045] Specifically, it includes the following contents:

[0046] S01. Based on the Google Earth Engine remote sensing cloud computing platform, several Landsat optical remote sensing images were extracted and screened according to the study area and the time range of image extraction.

[0047] S02, performing radiation correction, geometric correction, and image alignment on each of the screened optical remote sensing images to obtain a number of pre-processed images after row and column pixel alignment;

[0048] S03. Multiply all pixel values ​​of the optical bands in the plurality of pre-processed images by the scale coefficient, and then subtract the set value to obtain a plurality of surface reflectance images.

[0049] In this embodiment, the time range of image extraction is 1 to 2 years. For example, all Landsat images within the Shayok Basin in the upper reaches of the Indus River from 2019 to 2020 were screened using a remote sensing cloud computing platform to obtain 359 optical images. These 359 optical images were subjected to geometric correction, radiometric correction, and image alignment, and surface reflectance images were obtained through conversion calculation.

[0050] The proportional coefficient and the set value in step S03 can be set according to actual conditions to better meet actual needs. In this embodiment, the proportional coefficient is 0.0000275 and the set value is 0.2.

[0051] 2. Mask calculation

[0052] This part corresponds to step S1 and includes the following contents:

[0053] S11. Use the CFMask algorithm to identify the pixels covered by ice, snow, and clouds in each surface reflectance image;

[0054] S12. Select the "QA_PIXEL" band in the surface reflectance image and perform binary bit operations on clouds (binary representation: 00001000), cloud shadows (binary representation: 00010000), and snow (binary representation: 00100000). In the calculation results, pixels with a pixel value of 1 represent pixels covered by ice, snow, and clouds, and pixels with a pixel value of 0 represent actual ground objects. Mask out the pixels with a pixel value of 1 to obtain non-ice, snow, and non-cloud covered areas in the surface reflectance image.

[0055] S13. Sort the surface reflectance images in the image set from high to low according to the cloud cover attribute field "CLOUD_COVER" in the surface reflectance image; the surface reflectance image at the back has the least cloud cover.

[0056] 3. Image Filling

[0057] This part corresponds to step S2, and specifically includes the following contents:

[0058] S21. Selecting the surface reflectance image with the least cloud cover that is sorted last in the image set as the bottom image; the bottom image includes masked pixels and unmasked pixels, and the masked pixels are ice, snow, and cloud-obscured areas;

[0059] S22, starting from the second-to-last surface reflectance image in the image set, sequentially superimpose them on the underlying image, and fill the corresponding mask pixels in the underlying image to obtain a filled image with the maximum degree of filling;

[0060] M=M1+M2+M3+…M x …+M n

[0061]

[0062] Among them, M1, M2, M3, ..., M x ,…,M n are the 1st, 2nd, 3rd, ..., x, ..., nth surface reflectance images superimposed on the underlying image; x ij It is the pixel value in the i-th row and j-th column of the x-th surface reflectance image superimposed on the underlying image.

[0063] 4. Glacier Extraction

[0064] This part corresponds to step S3, and specifically includes the following contents:

[0065] S31, assigning a value of -99 to the null-valued pixels in the filled image, and extracting the pixels with a value of -99 by decoding with a raster calculator to convert the filled image into a binary image of 0 and 1;

[0066] S32, performing median filtering on the binary image through a 3×3 moving window to eliminate individual interference values;

[0067] S33. Convert the filtered image into a vector, extract the vector elements whose field "Label" is 1, and merge the vectors to obtain a glacier distribution vector dataset, that is, the glacier distribution range.

[0068] Example 2

[0069] In this example, in order to verify the effectiveness of the method of the present invention, the Shayok Basin in the upper reaches of the Indus River in 2019-2020 was used as the glacier identification test object. The satellite image was Landsat-8 optical image, and the cloud platform was Google EarthEngine remote sensing cloud computing platform. The identification results are as follows: Figure 2 As shown. Figure 2It can be seen that the textures such as the boundaries of the glaciers extracted by the method of the present invention are clear and complete, which shows that the present invention is accurate in identifying glaciers.

[0070] By adopting the above technical solution disclosed in the present invention, the following beneficial effects are obtained:

[0071] This paper provides a method for extracting bare ice glaciers based on a remote sensing cloud computing platform. This method addresses the interference caused by snow and cloud cover during glacier extraction, enabling automatic identification and extraction of glacier distribution. This method effectively avoids interference from snow accumulation and cloud cover, enabling detailed dynamic analysis of glaciers, which is crucial for studying the response mechanisms of glaciers to global climate change.

[0072] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for extracting bare ice glaciers based on a remote sensing cloud computing platform, characterized by: The following steps are included: S1. Calculation of surface reflectance image mask: Identify ice, snow, and cloud-covered pixels in several surface reflectance images of the target area acquired via a remote sensing cloud computing platform, and perform mask calculations on these pixels in each surface reflectance image to obtain areas free of ice, snow, and cloud. Sorting the masked surface reflectance images according to cloud cover creates an image set. S2, Image filling: The surface reflectance image with the lowest cloud cover in the image set is used as the base image, and the remaining surface reflectance images are superimposed on the base image in sequence to fill in the masked ice, snow and cloud cover pixels in the base image, thereby obtaining a filled image with the greatest degree of filling. S3. Obtain glacier distribution data: Encode the null pixels in the filled image, convert the filled image into a binary image using a raster calculator, and vectorize the binary image to obtain a glacier distribution vector dataset. Step S3 specifically includes the following contents: S31, assigning a value of -99 to the null-valued pixels in the filled image, and extracting the pixels with a value of -99 by decoding with a raster calculator to convert the filled image into a binary image of 0 and 1; S32, performing median filtering on the binary image through a 3×3 moving window to eliminate individual interference values; S33. Convert the filtered image into a vector, extract the vector elements whose "Label" field is 1, and merge the vector elements to obtain a glacier distribution vector dataset, that is, the glacier distribution range.

2. The method for extracting bare ice glaciers based on a remote sensing cloud computing platform according to claim 1, characterized in that: Step S1 specifically includes the following contents: S11. Use the CFMask algorithm to identify the pixels covered by ice, snow, and clouds in each surface reflectance image; S12. Select the "QA_PIXEL" band in the surface reflectance image and perform binary bitwise operations on the clouds, cloud shadows, and snow represented in binary. In the calculation result, pixels with a value of 1 represent pixels covered by ice, snow, and clouds, while pixels with a value of 0 represent actual ground objects. Mask out the pixels with a value of 1 to obtain areas not covered by ice, snow, or clouds in the surface reflectance image. S13. Sort the surface reflectance images in the image set from high to low according to the cloud cover attribute field "CLOUD_COVER" in the surface reflectance image; the surface reflectance image at the back has the least cloud cover.

3. The method for extracting bare ice glaciers based on a remote sensing cloud computing platform according to claim 2, characterized in that: Step S2 specifically includes the following contents: S21. Selecting the surface reflectance image with the least cloud cover that is ranked last in the image set as the bottom image; the bottom image includes masked pixels and unmasked pixels, and the masked pixels are ice, snow, and cloud-obscured areas; S22, starting from the second-to-last surface reflectance image in the image set, sequentially superimpose them on the underlying image, and fill the corresponding mask pixels in the underlying image to obtain a filled image with the maximum degree of filling; M=M1+M2+M3+L M x L+M n Among them, M1, M2, M3, L, M x ,L,M n These are the 1st, 2nd, 3rd, L, x, L, nth surface reflectance images superimposed on the underlying image; x ij It is the pixel value in the i-th row and j-th column of the x-th surface reflectance image superimposed on the underlying image.

4. The method for extracting bare ice glaciers based on a remote sensing cloud computing platform according to any one of claims 1 to 3, characterized in that: Before step S1, step S0 is also included, which is to obtain a number of Landsat optical remote sensing images based on the remote sensing cloud computing platform, and obtain a number of surface reflectance images after preprocessing them; step S0 specifically includes the following contents: S01. Based on the Google Earth Engine remote sensing cloud computing platform, extract and select several Landsat optical remote sensing images according to the study area and the time range of image extraction; S02, performing radiation correction, geometric correction, and image alignment on each of the screened optical remote sensing images to obtain a number of pre-processed images after row and column pixel alignment; S03. Multiply all pixel values ​​of the optical bands in the plurality of pre-processed images by the scale coefficient, and then subtract the set value to obtain a plurality of surface reflectance images.

Images