Dynamic threshold method for identifying snow cover based on high-resolution optical satellite

Abstract

The application discloses a dynamic threshold snow cover identification method for high-resolution series optical satellites. A plurality of high-resolution optical satellite sensors are preprocessed according to a preset condition, and high-resolution satellite reflectivity products are obtained. The high-resolution cloud and snow area is separated by fusing a cloud detection section product of a Fengyun meteorological satellite. Based on the optical characteristics and spatial distribution characteristics of snow, a blue wave band dynamic threshold algorithm is applied to different regions to realize adaptive identification threshold setting. Then, the spatial distribution of snow cover is obtained through a threshold extraction method, and snow cover monitoring is realized. The snow cover identification method for high-resolution optical satellites has good scalability, can adaptively set the snow identification threshold, improve the identification accuracy, and efficiently and accurately monitor the snow surface change.

Description

Technical Field

[0001] This invention relates to the field of quantitative remote sensing inversion technology for geographic information systems, and more particularly to a dynamic threshold snow cover identification method for Gaofen series optical satellites. Background Technology

[0002] Snow cover helps maintain the Earth's radiative energy balance and, as a vast water reservoir, influences various climate and hydrological processes. Conventional ground-based monitoring cannot accurately obtain large-scale monitoring results, and the representativeness of the monitoring range of ground stations directly affects the final monitoring accuracy. With the development of domestic high-resolution satellite technology, high spatial resolution optical sensor band information can provide accurate snow cover information and enable large-scale observation.

[0003] Currently, research on snow cover monitoring algorithms based on optical satellites has matured. However, snow cover identification algorithms still have limitations for high spatial resolution optical imagery. Taking GF-1 / 2 / 6 / WFV optical data as an example: current snow cover identification methods are hampered in application due to the lack of shortwave infrared channel information, which is crucial for monitoring snow cover; secondly, the limited number of high-resolution satellite bands makes it difficult to effectively distinguish between clouds and snow; furthermore, the large volume of high-resolution optical data necessitates the development of flexible and efficient snow cover identification algorithms to support operational snow cover identification. Therefore, it is necessary to improve this technology. Summary of the Invention

[0004] This invention addresses the shortcomings of existing technologies by providing a dynamic threshold snow cover identification method for Gaofen series optical satellites, thereby improving the snow cover identification capability of Gaofen satellites.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0006] A dynamic threshold snow cover identification method for Gaofen series optical satellites includes the following steps:

[0007] S1. Preprocessing is performed on multiple pre-set high-resolution optical satellite sensors to obtain high-resolution satellite reflectivity products;

[0008] S2. By integrating the cloud detection segment products of Fengyun meteorological satellites, high-resolution cloud and snow differentiation can be achieved;

[0009] S3. Based on the optical properties and spatial distribution characteristics of snow, a blue band dynamic threshold algorithm is applied to different regions to achieve adaptive identification threshold setting.

[0010] S4. Then, the spatial distribution of snow cover is obtained through threshold extraction method to realize snow cover monitoring;

[0011] Furthermore, the high-resolution optical satellite data preprocessing and quality control process in S1 includes:

[0012] S11, Radiation Calibration;

[0013] S12, Geographic Registration;

[0014] S13, orthorectification and atmospheric correction;

[0015] The data preprocessing and quality control methods are as follows: For the input GFL1 level standardized products, georegistration, radiometric calibration, orthorectification and atmospheric correction are achieved sequentially by using absolute radiometric calibration coefficients, favorable polynomial coefficient equations and 6S models.

[0016] Furthermore, the high-resolution optical satellite data pre-set in S1 includes remote sensing data from the wide-field camera of the Gaofen-1 satellite, remote sensing data from the panchromatic multispectral camera of the Gaofen-2 satellite, and remote sensing data from the wide-field camera of the Gaofen-6 satellite.

[0017] Furthermore, the high-resolution satellite cloud and snow differentiation method in S2 resamples the cloud detection segment products of wind and cloud meteorological satellites within the same or similar transit time range, and then integrates multi-source remote sensing products to meet the product requirements for cloud and snow differentiation.

[0018] Furthermore, the high-resolution satellite cloud and snow differentiation method in S2 includes:

[0019] S21. Determine the spatial consistency between high-resolution optical data and Fengyun meteorological satellites;

[0020] S22. Set the transit time difference between the two types of data to ensure the time consistency of the Fengyun meteorological satellite data used;

[0021] S23. Based on the L1-level georegistration data of Fengyun meteorological satellites, perform georegistration and spatial resampling on the cloud detection segment products of Fengyun meteorological satellites.

[0022] S24. Achieve spatial registration between Fengyun data and Gaofen data, and implement cloud masking.

[0023] Furthermore, in the product requirement of integrating multi-source remote sensing products to achieve cloud and snow differentiation, when the spatial range or transit time difference between the wind and cloud meteorological satellites and high-resolution satellites acquired by the system exceeds a preset value, the system considers the wind and cloud detection segment product to be ineffective and will no longer provide cloud and snow differentiation function.

[0024] Furthermore, the dynamic threshold snow cover identification method for high-resolution series optical satellites in S4 is characterized in that the spatial distribution of snow cover is obtained through a threshold extraction method in S4, and the principle of this snow cover identification method is as shown in formula (1):

[0025]

[0026] In the formula, image(x, y) is the snow-identified pixel value at the image raster position (x, y), ρ blue denoted as blue light band reflectivity, and t is the dynamic threshold obtained by the algorithm.

[0027] Furthermore, the adaptive identification threshold setting achieved by applying the blue band dynamic threshold algorithm to different regions in S3 includes:

[0028] S31. Calculate the average value of the blue band in the current research sample area.

[0029] S32, when the current region If the reflectance is greater than 0.7, the reflectance of the area is considered to be dominated by snow cover, and 0.7 is used as the identification criterion (t) for snow cover;

[0030] S33, when When the reflectance is less than 0.7 and greater than 0.4, it is considered that the reflectance of this area is jointly dominated by snow and non-snow-covered ground objects, and the reflectance exhibits a bimodal distribution.

[0031] S34. If the reflectance does not exhibit a bimodal distribution, then... As a classification standard for snow accumulation and non-snow accumulation;

[0032] S35, and for If the threshold is less than 0.4, the area is considered to be mainly dominated by non-snow-covered objects, so 0.4 is used as the snow accumulation detection threshold.

[0033] S36. Apply the blue band dynamic threshold algorithm to identify snow cover.

[0034] Furthermore, in the process of snow cover identification using the blue band dynamic threshold algorithm, the system defaults to using the blue band dynamic threshold method as the priority strategy for snow cover identification. During the calculation, the snow cover identification strategy is changed when the pre-set algorithm boundary conditions are met. The algorithm boundary conditions are: when the slope corresponding to the current pixel is greater than 35°, the random forest algorithm is used. The system embeds ALOSDSM satellite digital elevation data and its derived slope data.

[0035] Furthermore, in step S33, the valley value in the reflectivity frequency distribution map is selected as the blue light threshold, which can effectively distinguish between snow and non-snow.

[0036] Beneficial effects: Compared with the prior art, the present invention has the following beneficial effects:

[0037] This invention presents a dynamic threshold snow cover identification method for Gaofen series optical satellites, applicable to domestic Gaofen series optical satellite imagery including GF-1 / 6WFV and GF-2PMS images. It offers good scalability and integrates Fengyun meteorological satellite cloud products, enhancing the cloud-snow differentiation capability of Gaofen satellites and providing a practical tool for high spatial resolution snow cover remote sensing mapping. The method provides a complete set of Gaofen optical data preprocessing and quality control functions, boasting high algorithm efficiency and strong operational capabilities. Furthermore, this invention utilizes a dynamic threshold identification method to achieve high spatial resolution and high-precision snow cover identification. Attached Figure Description

[0038] Figure 1 This is a schematic diagram of the dynamic threshold snow cover identification method for high-resolution series optical satellites according to the present invention;

[0039] Figure 2 This is a schematic diagram of high-resolution reflectance data obtained using data preprocessing and quality control methods;

[0040] Figure 3 This is a technical roadmap for the snow cover identification algorithm based on the dynamic threshold method involved in this invention;

[0041] Figure 4 This is a schematic diagram of snow cover obtained using the dynamic threshold snow cover identification method for the Gaofen series optical satellites. Detailed Implementation

[0042] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings and examples.

[0043] Please see Figure 1 As shown, the present invention discloses a dynamic threshold snow cover identification method for high-resolution optical satellites: First, preprocessing is performed on multiple pre-set high-resolution optical satellite sensors to obtain high-resolution satellite reflectivity products. Specifically, the high-resolution optical satellite data preprocessing and quality control includes radiometric calibration, georegistration, orthorectification, and atmospheric correction.

[0044] Specifically, the data preprocessing and quality control methods are as follows: For the input GF L1 level standardized products, georegistration, radiometric calibration, orthorectification, and atmospheric correction are achieved sequentially using the absolute radiometric calibration coefficient, the RPC model (Rational Polynomial Coefficient), and the 6S model (Second Simulation of the Satellite Signal in the Solar Spectrum).

[0045] Please refer to the following: Figure 2 As shown, Figure 2 This is a schematic diagram of high-resolution reflectance data obtained using the data preprocessing and quality control modules.

[0046] The pre-defined high-resolution optical satellite data includes remote sensing data from the wide-field camera of the Gaofen-1 satellite, remote sensing data from the panchromatic multispectral camera of the Gaofen-2 satellite, and remote sensing data from the wide-field camera of the Gaofen-6 satellite, namely, L1-level standardized products of wide-field camera (WFV) and high-resolution camera (PMS).

[0047] Secondly, by integrating the cloud detection segment products of Fengyun meteorological satellites, high-resolution satellite cloud and snow differentiation can be achieved.

[0048] The high-resolution satellite cloud-snow differentiation method based on Fengyun meteorological satellite cloud detection segment products resamples Fengyun meteorological satellite cloud detection segment products (FY-3CLM) within the same or similar transit time range as cloud mask information of high-resolution optical data, and then integrates multi-source remote sensing products to achieve the product requirements for cloud-snow differentiation.

[0049] When the time difference between the passage of the Fengyun meteorological satellite and the Gaofen satellite, obtained by the dynamic threshold snow cover identification method for the Gaofen series optical satellites, exceeds the preset value, the system considers the Fengyun cloud detection segment product to be invalid and will no longer provide the cloud-snow distinction function.

[0050] Specifically, the technical process for high-resolution satellite cloud and snow differentiation is as follows:

[0051] First, determine the spatial consistency between high-resolution optical data and Fengyun meteorological satellites; set the transit time difference between the two types of data to ensure the temporal consistency of the Fengyun meteorological satellite data used; perform georegistration and spatial resampling on the Fengyun meteorological satellite cloud detection segment products based on the Fengyun meteorological satellite L1-level georegistration data; realize the spatial registration of Fengyun data and high-resolution data to achieve cloud masking.

[0052] If the spatial range or transit time difference between the Fengyun meteorological satellite and the Gaofen satellite obtained by the dynamic threshold snow cover identification method for Gaofen series optical satellites exceeds the preset value, the method considers the Fengyun cloud detection segment product to be ineffective and will no longer provide cloud and snow differentiation function.

[0053] Based on the optical properties and spatial distribution characteristics of snow, a blue band dynamic threshold algorithm is applied to different regions to achieve adaptive identification threshold setting.

[0054] Finally, the spatial distribution of snow cover is obtained through threshold extraction, thus realizing snow cover monitoring.

[0055] Please refer to the following: Figure 3 As shown, Figure 3This is a technical roadmap for the dynamic threshold snow cover identification method involved in this invention. Snow cover identification is achieved using a blue band dynamic threshold algorithm. The principle of the snow cover identification method is shown in formula (1).

[0056]

[0057] In the formula, image(x, y) is the snow-identified pixel value at the image raster position (x, y), ρ blue denoted as blue light band reflectivity, and t is the dynamic threshold obtained by the algorithm.

[0058] First, calculate the average value of the blue band in the current study area. When the current area If the reflectance is greater than 0.7, the region is considered to have snow cover as the dominant factor, and 0.7 is used as the criterion for snow cover (t); when... When the reflectance is less than 0.7 and greater than 0.4, the reflectance in this area is considered to be jointly dominated by snow and non-snowy ground features, exhibiting a bimodal distribution. In this case, using the valley value in the reflectance frequency distribution map as the selected blue light threshold can effectively distinguish between snow and non-snowy areas. If the reflectance does not exhibit a bimodal distribution, then... As a classification standard for snow accumulation and non-snow accumulation; while for If the threshold is less than 0.4, the area is considered to be mainly dominated by non-snow-covered objects, so 0.4 is used as the snow accumulation detection threshold.

[0059] In the process of snow cover identification using the blue band dynamic threshold algorithm, the system defaults to using the blue band dynamic threshold method as the priority strategy for snow cover identification. During the calculation, the snow cover identification strategy is changed when the pre-set algorithm boundary conditions are met. The algorithm boundary conditions are: when the slope corresponding to the pixel is greater than 35°, the random forest algorithm is used. The system embeds ALOSDSM satellite digital elevation data and its derived slope data.

[0060] Please refer to the following: Figure 4 The image shows a high-resolution snow cover diagram obtained using the high-resolution satellite dynamic threshold snow cover identification method.

[0061] In summary, the dynamic threshold snow cover identification method for the Gaofen series optical satellites of this invention is applicable to domestic Gaofen series optical satellite imagery, including GF-1 / 6WFV and GF-2PMS images. It exhibits good scalability and integrates Fengyun meteorological satellite cloud products, enhancing the cloud-snow differentiation capability of Gaofen satellites and providing a practical tool for high spatial resolution snow cover remote sensing mapping. This method also provides a complete set of Gaofen optical data preprocessing and quality control methods, with high algorithm efficiency and strong operational capabilities. By employing a dynamic threshold selection method targeting the blue light band, this invention achieves high spatial resolution and high-precision snow cover identification. Furthermore, this dynamic threshold snow cover identification method for the Gaofen series optical satellites exhibits good scalability, adaptively setting the snow cover identification threshold to improve identification accuracy, and enabling efficient and accurate operational monitoring of snow cover surface changes.

[0062] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the concept and scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the design concept of the present invention should fall within the protection scope of the present invention. All technical contents for which protection is sought in this invention have been fully described in the claims.

Claims

1. A dynamic threshold snow cover discrimination method for high-resolution series optical satellites, characterized by, Includes the following steps: S1. Preprocessing is performed on multiple pre-set high-resolution optical satellite sensors to obtain high-resolution satellite reflectivity products; S2. By integrating the cloud detection segment products of Fengyun meteorological satellites, high-resolution satellite cloud-snow differentiation can be achieved; S3. Based on the optical properties and spatial distribution characteristics of snow, a blue band dynamic threshold algorithm is applied to different regions to achieve adaptive identification threshold setting. S4. Then, the spatial distribution of snow cover is obtained through threshold extraction method to realize snow cover monitoring; In S4, the spatial distribution of snow cover is obtained through a threshold extraction method. The principle of the blue band dynamic threshold snow cover identification method is shown in formula (1): ， In the formula, To identify the pixel value of snow at the image raster position (x,y), t represents the reflectivity of the blue light band, and t is the dynamic threshold obtained by the algorithm. The adaptive identification threshold setting in S3, which applies the blue band dynamic threshold algorithm to different regions, includes: S31. Calculate the average value of the blue band in the current research sample area. ; S32, when the current region If the reflectance is greater than 0.7, the reflectance of the area is considered to be dominated by snow cover, and 0.7 is used as the identification criterion (t) for snow cover; S33, when When the reflectance is less than 0.7 and greater than 0.4, it is considered that the reflectance of this area is jointly dominated by snow and non-snow-covered ground objects, and the reflectance exhibits a bimodal distribution. S34. If the reflectance does not exhibit a bimodal distribution, then... As a classification standard for snow accumulation and non-snow accumulation; S35, and for If the threshold is less than 0.4, the area is considered to be mainly dominated by non-snow-covered objects, so 0.4 is used as the snow accumulation detection threshold. S36. Apply the blue band dynamic threshold algorithm to identify snow cover.

2. The dynamic threshold snow cover identification method for Gaofen series optical satellites according to claim 1, characterized in that, The high-resolution optical satellite data preprocessing and quality control process in S1 includes: S11, Radiation Calibration; S12, Geographic Registration; S13, orthorectification and atmospheric correction; The data preprocessing and quality control methods are as follows: For the input GF L1 level standardized products, georegistration, radiometric calibration, orthorectification and atmospheric correction are achieved sequentially by using absolute radiometric calibration coefficients, favorable polynomial coefficient equations and 6S models.

3. The dynamic threshold snow cover identification method for Gaofen series optical satellites according to claim 2, characterized in that, The pre-set Gaofen optical satellite data includes: remote sensing data from the wide-field camera of Gaofen-1 satellite, remote sensing data from the panchromatic multispectral camera of Gaofen-2 satellite, and remote sensing data from the wide-field camera of Gaofen-6 satellite.

4. The dynamic threshold snow cover identification method for Gaofen series optical satellites according to claim 1, characterized in that, The method for distinguishing between clouds and snow using high-resolution satellites in S2 involves resampling the cloud detection segment products of meteorological satellites within the same or similar transit time range, and then fusing multi-source remote sensing products to achieve the product requirements for distinguishing between clouds and snow.

5. The dynamic threshold snow cover identification method for Gaofen series optical satellites according to claim 4, characterized in that, The method for distinguishing clouds and snow using high-resolution satellites in S2 includes: S21. Determine the spatial consistency between high-resolution optical data and Fengyun meteorological satellites; S22. Set the transit time difference between the two types of data to ensure the time consistency of the Fengyun meteorological satellite data used; S23. Based on the L1-level georegistration data of Fengyun meteorological satellites, perform georegistration and spatial resampling on the cloud detection segment products of Fengyun meteorological satellites. S24. Achieve spatial registration between Fengyun data and Gaofen data, and implement cloud masking.

6. The dynamic threshold snow cover identification method for Gaofen series optical satellites according to claim 5, characterized in that, In the product requirement of integrating multi-source remote sensing products to achieve cloud and snow differentiation, when the spatial range or transit time difference between the wind and cloud meteorological satellites and high-resolution satellites acquired by the system exceeds the preset value, the system considers the wind and cloud detection segment product to be ineffective and will no longer provide cloud and snow differentiation function.

7. The dynamic threshold snow cover identification method for Gaofen series optical satellites according to claim 1, characterized in that, In the process of snow cover identification using the blue band dynamic threshold algorithm, the system defaults to using the blue band dynamic threshold method as the priority strategy for snow cover identification. During the calculation, the snow cover identification strategy is changed when the pre-set algorithm boundary conditions are met. The algorithm boundary conditions are: when the slope corresponding to the pixel is greater than 35°, the random forest algorithm is used. The system embeds ALOS DSM satellite digital elevation data and its derived slope data.

8. The dynamic threshold snow cover identification method for Gaofen series optical satellites according to claim 1, characterized in that, In S33, the valley value in the reflectivity frequency distribution map is selected as the blue light threshold, which can effectively distinguish between snow and non-snow.

Images