CN112395989A - Snow coverage mixed pixel decomposition method for multi-satellite sensor - Google Patents

Abstract

The invention relates to a snow coverage degree mixed pixel decomposition method for a multi-satellite sensor, wherein corresponding image data are obtained based on a plurality of preset remote sensing satellite sensors; extracting end members of a plurality of types of land on the basis of the multispectral reflectivity data and a preset image end member extraction rule to obtain an end member set of the land; when the extraction fails, calling a ground measurement spectrum curve of the ground object from the pre-acquired auxiliary data, calculating an end member matched with the spectrum response of the sensor wave band according to a pre-acquired spectrum response function of the remote sensing satellite sensor, and acquiring a ground measurement end member library; acquiring a typical ground class end member library based on the end member set of the ground class; performing multi-end member spectrum mixed analysis to obtain the coverage of the accumulated snow based on the typical end member library and the ground measurement end member library; and calculating the coverage of the satellite pixel accumulated snow.

Description

A snow cover mixed pixel decomposition method for multi-satellite sensors

technical field

The invention relates to the technical field of satellite sensor data processing and application, in particular to a snow coverage mixed pixel decomposition method for multi-satellite sensors.

Background technique

Snow cover contributes to the Earth's radiative energy balance and acts as an extensive aquifer that affects various climatic and hydrological processes. Conventional ground station monitoring cannot accurately obtain large-scale monitoring results, and whether the monitoring range of ground stations is representative directly affects the final monitoring accuracy. With the development of satellite technology, the abundant band information of sensors in optical remote sensing can provide accurate snow cover information and realize large-scale observation at the same time.

SUMMARY OF THE INVENTION

(1) Technical problems to be solved

In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides a method for decomposing snow cover mixed pixels for a multi-satellite sensor.

(2) Technical solutions

In order to achieve the above-mentioned purpose, the main technical scheme adopted in the present invention includes:

An embodiment of the present invention provides a snow cover mixed pixel decomposition method for a multi-satellite sensor, including:

S1. Obtain corresponding image data based on a plurality of preset remote sensing satellite sensors;

The image data includes: multispectral surface reflectance data, imaging geometry and mask data;

S2, based on the multi-spectral reflectance data and the preset image endmember extraction rule, perform endmember extraction of various land types, and obtain endmember sets of land types;

The end member set of the land type includes the end members of a plurality of land types;

When the extraction fails, the ground measurement spectral curve of the ground object is retrieved from the pre-acquired auxiliary data, and according to the pre-acquired spectral response function of the remote sensing satellite sensor, the endmember that matches the spectral response of the sensor band is calculated, and the ground measurement endmember library is obtained. ;

The pre-acquired auxiliary data includes snow cover, vegetation, soil spectral curves, and spectral response data;

S3, based on the endmember set of the land class, obtain a typical class endmember library;

S4, based on the typical endmember library of the ground type and the ground measurement endmember library, perform multi-endmember spectral hybrid analysis to obtain snow coverage;

S5. Calculate the satellite pixel snow coverage product.

preferably,

The image data includes: MODIS image data, Landsat TM/ETM+/OLI image data, Sentinel-2MSI image data, FY-4A AGRI image data and Himawari-8 AHI image data.

preferably,

The preset image terminal extraction rules in step S2 are:

The snow cover endmember extraction rules in the MODIS image data, Sentinel-2/MSI image data, and Landsat TM/ETM+/OLI image data are: NDVI<-0.035 and NDSI>0.75 and R _0.55 >0.7; vegetation endmember extraction The rules are: NDVI>0.7 and NDSI<-0.4; the extraction rules of soil and rock endmembers are: 0<NDVI<0.15 and NDSI<-0.4; the extraction rules of water body endmembers are: NDWI>0.2 and R _0.86 <0.2;

NDVI refers to the normalized vegetation index; NDSI refers to the normalized snow cover index; R _0.55 refers to the surface reflectance in the 0.55 micron band;

The snow cover extraction rules in Himawari-8/AHI image data and FY-4A/AGRI image data are: NDVI<-0.03 and NDSI>0.74 and R _0.55 >0.5; vegetation endmember extraction rules are: NDVI>0.65 and NDSI<-0.4; the extraction rules of soil and rock endmembers are: 0<NDVI<0.15 and NDSI<-0.4.

Preferably, the step S3 includes:

Based on the endmember set of the land type, the image endmembers in the endmember set of the land type are screened to obtain a typical endmember library.

Preferably, the step S3 specifically includes:

S31, using formula (1) to calculate the vector length of each endmember spectrum;

where ||r|| is the vector length of the endmember spectrum; rk is the reflectivity of the _k -th band; N is the number of bands;

S32, arranging all vector lengths in each type of endmembers of snow, vegetation, and soil features in ascending order to obtain sequences corresponding to each type of endmembers of snow, vegetation, and soil features;

S33, for the sequence corresponding to each type of endmembers of the snow cover, vegetation, soil features, divide each type of endmembers into n subsets at equal intervals;

Wherein, the interval is ||R _i ||;

n is a preset value; ||R|| _max pixel maximum reflectivity; ||R|| _min pixel minimum reflectivity;

S34, taking the median or the mean value of all spectra in each subset as a typical end member of the subset;

S35, obtaining a typical class endmember library based on the typical endmembers of each subset;

The typical endmember library library includes: n typical endmembers of snow cover, n typical endmembers of vegetation, and n typical endmembers of soil.

Preferably, the step S4 includes:

S41. In the endmember library based on the typical endmember library and the ground measurement, mark the pixels with NDSI<0 or the reflectivity of the 1.6μm band greater than 0.3 as no snow, and mark the pixels with NDSI>0.7 as pure snow;

S42, marking the pixels that successfully extract snow endmembers as pure snow, and marking the pixels that are not snow-extracted as no snow;

S43. According to the snow, soil and vegetation spectra of the endmember library and the preset endmember area ratio constraint, establish a mixed pixel spectrum database;

S44. According to the mixed pixel spectral database, only the mixed pixels are processed one by one to determine a snow cover-soil dichotomy model and a snow cover-vegetation dichotomy model;

The mixed pixels are pixels outside the pixels marked as pure snow and no snow;

When the NDVI of the pixel exceeds 0.3, only the snow-vegetation model is used, otherwise, the snow-soil model and the snow-vegetation model are used in sequence;

S45. Calculate the difference between the mixed pixel spectrum and the mixed pixel spectral data corresponding to different snow cover degrees, and record the result with the smallest RMSE passing the residual test and the corresponding snow cover degree;

Among them, RMSE is the root mean square error (root mean square error);

If none of them meet the residual test and the current pixel NDVI does not exceed 0.3, record the RMSE that passes the residual test, repeat step S45 with the snow cover-vegetation model, and obtain a new RMSE for the calculated snow cover, and compare the two The snow cover corresponding to the smaller is selected as the final result.

Preferably, the preset end-member area ratio constraint in step S43 is:

The constraints are:

F _i ≥ 0;

where R _λ is the reflectivity of the mixed pixel at wavelength λ; R _iλ and F _i are the reflectivity and abundance of the ith endmember, respectively; _ελ is the fitting residual; M is the number of endmembers;

When solving, meet the preset first condition, second condition, third condition, fourth condition and fifth condition;

The first condition is: the area ratio of the end members complies with the constraints of "non-negative" and "sum is one", but an error of 1% is allowed;

The second condition is: when the NDVI of the pixel exceeds 0.3, use the snow-vegetation mixed model, otherwise use the snow-soil model and the snow-vegetation model in sequence;

The third condition is: during the residual test, the RMSE of the residuals of all bands shall not exceed 0.025;

The fourth condition is: during the residual error test, the residual error of half of the continuous bands in the number of bands does not exceed 0.025;

The fifth condition is: the result with the smallest RMSE is taken as the area ratio of each end member.

Preferably, the step S5 includes:

S51, mask the snow cover inversion result layer on the basis of the mask layer;

S52. Select a color table corresponding to the distribution range of the snow cover range, and use a Geotiff file of unsigned integer type and LZW compression method. The data file also saves the original projection information of the image, and the data file is the calculated snow cover. product.

(3) Beneficial effects

The beneficial effects of the present invention are as follows: the snow cover mixed pixel decomposition method for multi-satellite sensors of the present invention is suitable for common remote sensing optical images including MODIS, Sentinel-2/MSI, Landsat TM/ETM+/OLI images , as well as the Himawari-8/AHI and FY-4A/AGRI images of the geostationary meteorological satellites, which have good scalability and provide practical tools for remote sensing mapping of snow cover, fusion and comparative analysis of multi-source snow cover results.

Based on the theory of multi-endmember spectral hybrid analysis and the automatic construction of image endmember library, the algorithm has a physical basis and high precision.

Using the image-based automatic endmember extraction and selection technology and the look-up table technology for the calculation of snow cover, the calculation efficiency is high, and the remote sensing real-time monitoring of snow cover can be realized according to this method.

Description of drawings

1 is a flowchart of a method for decomposing snow cover mixed pixels for multi-satellite sensors according to the present invention;

2 is a schematic diagram of a method for decomposing snow cover mixed pixels for a multi-satellite sensor in an embodiment of the present invention;

3 is an example diagram of the results of automatic endmember selection (a) and endmember extraction (b) of MODIS image snow types in an embodiment of the present invention;

4 is an example diagram of the results of automatic endmember selection (c) and endmember extraction (d) of MODIS image soil types in an embodiment of the present invention;

FIG. 5 is an example diagram of the results of automatic endmember selection (e) and endmember extraction (f) of vegetation types in MODIS images according to an embodiment of the present invention.

Detailed ways

In order to better explain the present invention and facilitate understanding, the present invention will be described in detail below with reference to the accompanying drawings and through specific embodiments.

For better understanding of the above technical solutions, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention may be embodied in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided so that the present invention will be more clearly and thoroughly understood, and will fully convey the scope of the present invention to those skilled in the art.

Referring to FIG. 1 , this embodiment provides a method for decomposing snow cover mixed pixels for multi-satellite sensors, including:

S1. Obtain corresponding image data based on a plurality of preset remote sensing satellite sensors;

The image data includes: multispectral surface reflectance data, imaging geometry and mask data;

S2, based on the multi-spectral reflectance data and the preset image endmember extraction rule, perform endmember extraction of multiple land types, and obtain endmember sets of land types;

The end member set of the land type includes the end members of a plurality of land types;

When the extraction fails, the ground measurement spectral curve of the ground object is retrieved from the pre-acquired auxiliary data, and according to the pre-acquired spectral response function of the remote sensing satellite sensor, the endmember that matches the spectral response of the sensor band is calculated, and the ground measurement endmember library is obtained. ;

The pre-acquired auxiliary data includes snow cover, vegetation, soil spectral curves, and spectral response data;

In this embodiment, the auxiliary data includes snow cover, vegetation, and soil spectra extracted from the Johns Hopkins University spectral library, and spectral response data officially obtained from each sensor image product.

S3, based on the endmember set of the land class, obtain a typical class endmember library;

S4, based on the typical endmember library of the ground type and the ground measurement endmember library, perform multi-endmember spectral hybrid analysis to obtain snow coverage;

S5. Calculate the satellite pixel snow coverage product and save it to a file.

preferably,

The image data includes: MODIS image data, Landsat TM/ETM+/OLI image data, Sentinel-2MSI image data, FY-4A AGRI image data and Himawari-8 AHI image data.

In this example, MODIS image data (satellite pixel surface reflectance data (MOD09GA and MYD09GA) provided by NASA, including imaging geometry and mask data), Landsat TM/ETM+/OLI image data (surface reflectance data provided by USGS) ), Sentinel-2 MSI image data (L2A surface reflectance product provided by ESA), FY-4AAGRI image data (AGRI China multi-channel imagery, cloud mask and imaging geometry products provided by China Meteorological Administration) and Himawari- 8 AHI image data (AHI multi-channel images produced by the Japan Meteorological Agency include imaging geometry, cloud mask and imaging geometry products).

preferably,

The preset image terminal extraction rules in step S2 are:

The snow cover endmember extraction rules in the MODIS image data, Sentinel-2/MSI image data, and Landsat TM/ETM+/OLI image data are: NDVI<-0.035 and NDSI>0.75 and R _0.55 >0.7; vegetation endmember extraction The rules are: NDVI>0.7 and NDSI<-0.4; the extraction rules of soil and rock endmembers are: 0<NDVI<0.15 and NDSI<-0.4; the extraction rules of water body endmembers are: NDWI>0.2 and R _0.86 <0.2.

The snow cover extraction rules in Himawari-8/AHI image data and FY-4A/AGRI image data are: NDVI<-0.03 and NDSI>0.74 and R _0.55 >0.5; vegetation endmember extraction rules are: NDVI>0.65 and NDSI<-0.4; the extraction rules of soil and rock endmembers are: 0<NDVI<0.15 and NDSI<-0.4.

Preferably in this embodiment, the step S3 includes: based on the endmember set of the land type, performing a screening process on the image endmembers in the endmember set of the land type, and obtaining a typical endmember library (also known as the endmember library). It is the image automatic extraction and selection of endmember library in Figure 2).

Preferably in this embodiment, the step S3 specifically includes:

S31, using formula (1) to calculate the vector length of each endmember spectrum;

where ||r|| is the vector length of the endmember spectrum; rk is the reflectivity of the _k -th band; N is the number of bands.

S32 , arranging all vector lengths in each type of endmembers of snow cover, vegetation, and soil features in ascending order to obtain a sequence corresponding to each type of endmembers of snow cover, vegetation, and soil features.

S33, for the sequence corresponding to each type of endmembers of the snow cover, vegetation, soil features, divide each type of endmembers into n subsets at equal intervals;

Wherein, the interval is ||R _i ||;

n is a preset value; ||R|| _max pixel maximum reflectivity; ||R|| _min pixel minimum reflectivity.

Figure 3, Figure 4, and Figure 5 show the results of automatic endmember selection and endmember extraction of snow, soil, and vegetation types in a MODIS image. The number of vegetation endmembers directly extracted from the image is huge. A few well-represented endmembers that basically represent the spectral differences within the class can be extracted.

S34. Take the median or average value of all spectra in each subset as a typical end member of the subset.

S35. Obtain a typical endmember library based on the typical endmembers of each subset.

The typical endmember library library includes: n typical endmembers of snow cover, n typical endmembers of vegetation, and n typical endmembers of soil.

Preferably in this embodiment, the step S4 includes:

S41. In the endmember library based on the typical endmember library and the ground measurement, mark the pixels with NDSI<0 or the reflectivity of the 1.6μm band greater than 0.3 as no snow, and mark the pixels with NDSI>0.7 as pure snow.

S42 , marking the pixels that successfully extract snow endmembers as pure snow, and marking the pixels that are not snow-extracted as no snow.

S43 , according to the snow, soil and vegetation spectra of the endmember library and the preset endmember area ratio constraints, establish a mixed pixel spectrum database.

S44. According to the mixed pixel spectral database, only the mixed pixels are processed one by one to determine the snow cover-soil dichotomy model and the snow cover-vegetation dichotomy model.

The mixed cells are cells outside the cells marked as pure snow and no snow.

Only the snow-vegetation model is used when the NDVI of the cell exceeds 0.3, otherwise the snow-soil model and the snow-vegetation model are used in order.

S45. Calculate the difference between the mixed pixel spectrum and the mixed pixel spectral data corresponding to different snow cover degrees, and record the result with the smallest RMSE passing the residual test and the corresponding snow cover degree;

If none of them meet the residual test and the current pixel NDVI does not exceed 0.3, record the RMSE that passes the residual test, repeat step S45 with the snow cover-vegetation model, and obtain a new RMSE for the calculated snow cover, and compare the two The snow cover corresponding to the smaller is selected as the final result.

In the practical application of this implementation, the specific steps of multi-endmember spectral hybrid analysis are as follows:

In step 1, the pixels with NDSI < 0 or the reflectivity of the 1.6 μm band greater than 0.3 are marked as no snow, and the pixels with NDSI > 0.7 are marked as pure snow.

In step 2, the pixels that successfully extract snow endmembers are marked as pure snow, and the pixels that are not snowy, such as woodland and grassland, are marked as no snow.

Step 3, build a lookup table (a lookup table is constructed for an image, according to the pixel mixture spectrum, find the "mixed spectrum" with the smallest difference in the lookup table, and the calculated snow cover that satisfies its conditions is the estimated target result. ). Taking 1% as the step size of snow cover, considering the estimation error of 1%, according to the snow cover, soil and vegetation spectra of the endmember pool, and the aforementioned endmember area ratio constraints, the snow cover-soil bipartite model and the snow cover are established. - Mixed pixel spectral database corresponding to different snow cover under the vegetation dichotomous model.

Step 4, start the calculation, and only process the pixels of the mixed pixels (that is, outside the pixels marked as pure snow and no snow) one by one, and first determine the mixed pixel model. Only the snow-vegetation model is used when the NDVI of the cell exceeds 0.3, otherwise the snow-soil model and the snow-vegetation model are used in order.

Step 5: Calculate the difference between the mixed pixel spectrum and the mixed pixel spectral data corresponding to different snow cover degrees, and record the result with the smallest RMSE passing the residual test and the corresponding snow cover degree.

Step 6, if none of them meet the residual test and the current pixel NDVI does not exceed 0.3, record the RMSE of step 5, repeat step 5 with the snow cover-vegetation model, and obtain a new RMSE for the calculation of snow cover. The user chooses the snow cover corresponding to the smaller one as the final result.

Preferably, the preset end-member area ratio constraint in step S43 is:

The constraints are:

F _i ≥ 0;

Among them, R _λ is the reflectivity of the mixed pixel at wavelength λ; R _iλ and F _i are the reflectivity and abundance of the ith endmember, respectively; _ελ is the fitting residual; M is the number of endmembers.

When solving, the preset first condition, second condition, third condition, fourth condition, and fifth condition are satisfied.

The first condition is: the area ratio of the end members complies with the "non-negative" and "sum is one" constraints, but an error of 1% is allowed.

The second condition is: when the NDVI of a pixel exceeds 0.3, use the snow-vegetation mixed model, otherwise use the snow-soil model and the snow-vegetation model in sequence.

The third condition is: during the residual test, the RMS of the residuals of all bands shall not exceed 0.025.

The fourth condition is: during the residual error test, the residual error of half of the continuous bands of the number of bands does not exceed 0.025.

The fifth condition is: the result with the smallest RMSE is taken as the area ratio of each end member.

Preferably, the step S5 includes:

S51 , mask the snow cover inversion result layer on the basis of the mask layer.

S52. Select a color table corresponding to the distribution range of the snow cover range (snow cover is blue-cyan-yellow gradient), use a Geotiff file of unsigned integer type and LZW compression method, and save the image at the same time as the data file Raw projection information, this data file is the calculated snow cover product.

The snow cover estimation error and the imaging geometry layer are retained in the output file in this embodiment to provide a reference for the quality assessment of snow cover.

In the snow cover mixed pixel decomposition method for multi-satellite sensors in this embodiment, the endmember extraction rules are determined according to the spectral band settings of different satellite optical sensors; endmembers such as snow, soil, and vegetation are automatically extracted according to multispectral images, A typical endmember library is established, and the typical endmembers are selected according to the endmember spectral vector; the mixed pixels are calculated according to the typical endmembers and the least squares method, and the optimal solution is obtained to realize the snow coverage inversion. The snow cover mixed pixel decomposition method of the multi-satellite sensor provided by the embodiment of the invention is suitable for various satellite images including MODIS, Sentinel-2/MSI, Landsat TM/ETM+/OLI images, and the stationary meteorological satellite Himawari-8/ AHI, FY-4A/AGRI images, through multi-endmember spectral hybrid analysis theory and image-based endmember automatic extraction and selection technology, realize automatic calculation of mixed pixel snow coverage under complex surface conditions, and the results can be used in mountainous areas Hydrological modelling, water resource management, numerical weather forecasting, land surface models, and climate change research.

The snow cover mixed pixel decomposition method for multi-satellite sensors in this embodiment is suitable for common remote sensing optical images including MODIS, Sentinel-2/MSI, Landsat TM/ETM+/OLI images, and the stationary meteorological satellite Himawari -8/AHI and FY-4A/AGRI images have good scalability and provide practical tools for remote sensing mapping of snow cover, fusion and comparative analysis of multi-source snow cover results. Based on the theory of multi-endmember spectral hybrid analysis and the automatic construction of image endmember library, the algorithm has a physical basis and high precision. Using the image-based automatic endmember extraction and selection technology and the look-up table technology for the calculation of snow cover, the calculation efficiency is high, and the remote sensing real-time monitoring of snow cover can be realized according to this method.

In the description of the present invention, it should be understood that the terms "first" and "second" are only used for description purposes, and cannot be interpreted as indicating or implying relative importance or the number of indicated technical features. Thus, a feature defined as "first" or "second" may expressly or implicitly include one or more of that feature. In the description of the present invention, "plurality" means two or more, unless otherwise expressly and specifically defined.

In the present invention, unless otherwise expressly specified and limited, the terms "installed", "connected", "connected", "fixed" and other terms should be understood in a broad sense, for example, it may be a fixed connection or a detachable connection , or integrated; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium; it can be the internal communication of the two elements or the interaction relationship between the two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

In the present invention, unless otherwise expressly specified and limited, a first feature is "on" or "under" a second feature, which may be in direct contact between the first and second features, or through an intermediary between the first and second features indirect contact. Also, the first feature is "above", "above" and "above" the second feature, which may mean that the first feature is directly above or obliquely above the second feature, or simply means that the first feature is level higher than the second feature . The first feature is "below", "below" and "below" the second feature, which may mean that the first feature is directly below or diagonally below the second feature, or simply means that the first feature is level below the second feature.

In the description of this specification, the terms "one embodiment", "some embodiments", "embodiments", "example", "specific example" or "some examples" and the like are described in conjunction with the embodiment or example A particular feature, structure, material, or characteristic described is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms are not necessarily directed to the same embodiment or example. Furthermore, the particular features, structures, materials or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, those skilled in the art may combine and combine the different embodiments or examples described in this specification, as well as the features of the different embodiments or examples, without conflicting each other.

Although the embodiments of the present invention have been shown and described above, it should be understood that the above-mentioned embodiments are exemplary and should not be construed as limiting the present invention. Embodiments are subject to alterations, modifications, substitutions and variations.

Claims (8)

1. a snow cover mixed pixel decomposition method for multi-satellite sensors, is characterized in that, comprises: S1. Obtain corresponding image data based on a plurality of preset remote sensing satellite sensors; The image data includes: multispectral surface reflectance data, imaging geometry and mask data; S2, based on the multi-spectral reflectance data and the preset image endmember extraction rule, perform endmember extraction of various land types, and obtain endmember sets of land types; The end member set of the land type includes the end members of a plurality of land types; When the extraction fails, the ground measurement spectral curve of the ground object is retrieved from the pre-acquired auxiliary data, and according to the pre-acquired spectral response function of the remote sensing satellite sensor, the endmember that matches the spectral response of the sensor band is calculated, and the ground measurement endmember library is obtained. ; The pre-acquired auxiliary data includes snow cover, vegetation, soil spectral curves, and spectral response data; S3, based on the endmember set of the land class, obtain a typical class endmember library; S4, based on the typical endmember library of ground type and the ground measurement endmember library, perform multi-endmember spectral hybrid analysis to obtain snow coverage; S5. Calculate the satellite pixel snow coverage product. 2. The method according to claim 1, wherein The image data includes: MODIS image data, Landsat TM/ETM+/OLI image data, Sentinel-2 MSI image data, FY-4A AGRI image data and Himawari-8 AHI image data. 3. The method of claim 2, wherein The preset image terminal extraction rules in step S2 are: The snow cover endmember extraction rules in the MODIS image data, Sentinel-2/MSI image data, and Landsat TM/ETM+/OLI image data are: NDVI<-0.035 and NDSI>0.75 and R _0.55 >0.7; vegetation endmember extraction The rules are: NDVI>0.7 and NDSI<-0.4; the extraction rules of soil and rock endmembers are: 0<NDVI<0.15 and NDSI<-0.4; the extraction rules of water body endmembers are: NDWI>0.2 and R _0.86 <0.2; The snow cover extraction rules in Himawari-8/AHI image data and FY-4A/AGRI image data are: NDVI<-0.03 and NDSI>0.74 and R _0.55 >0.5; vegetation endmember extraction rules are: NDVI>0.65 and NDSI<-0.4; the extraction rules of soil and rock endmembers are: 0<NDVI<0.15 and NDSI<-0.4. 4. The method according to claim 3, wherein the step S3 comprises: Based on the endmember set of the land type, the image endmembers in the endmember set of the land type are screened to obtain a typical endmember library. 5. The method according to claim 4, wherein the step S3 specifically comprises: S31, using formula (1) to calculate the vector length of each endmember spectrum;

where ||r|| is the vector length of the endmember spectrum; rk is the reflectivity of the _k -th band; N is the number of bands; S32, arranging all vector lengths in each type of endmembers of snow, vegetation, and soil features in ascending order to obtain the sequence corresponding to each type of endmembers of snow, vegetation, and soil features; S33, for the sequence corresponding to each type of endmembers of the snow cover, vegetation, soil features, divide each type of endmembers into n subsets at equal intervals; Wherein, the interval is ||R _i ||;

n is a preset value; ||R|| _max pixel maximum reflectivity; ||R|| _min pixel minimum reflectivity; S34, taking the median or the mean value of all spectra in each subset as a typical end member of the subset; S35, obtaining a typical class endmember library based on the typical endmembers of each subset; The typical endmember library library includes: n typical endmembers of snow cover, n typical endmembers of vegetation, and n typical endmembers of soil. 6. The method according to claim 5, wherein the step S4 comprises: S41. In the endmember library based on the typical endmember library and the ground measurement, mark the pixels with NDSI<0 or the reflectance greater than 0.3 in the 1.6 μm band as no snow, and mark the pixels with NDSI>0.7 as pure snow; S42, marking the pixels that successfully extract snow endmembers as pure snow, and marking the pixels that are not snow-extracted as no snow; S43. According to the snow, soil and vegetation spectra of the endmember library and the preset endmember area ratio constraints, establish a mixed pixel spectrum database; S44. According to the mixed pixel spectral database, the mixed pixels are processed one by one to determine a snow cover-soil dichotomy model and a snow cover-vegetation dichotomy model; The mixed pixels are pixels outside the pixels marked as pure snow and no snow; When the NDVI of the pixel exceeds 0.3, only the bipartite model is used, that is, the snow-vegetation model is used, otherwise, the snow-soil model and the snow-vegetation model are used in sequence; S45. Calculate the difference between the mixed pixel spectrum and the mixed pixel spectral data corresponding to different snow cover degrees, and record the result with the smallest RMSE that passes the residual test and the corresponding snow cover degree; If none of them meet the residual test and the current pixel NDVI does not exceed 0.3, record the RMSE that passes the residual test, repeat step S45 with the snow cover-vegetation model, and obtain a new RMSE for the calculated snow cover, and compare the two The snow cover corresponding to the smaller is selected as the final result. 7. The method according to claim 6, wherein the preset end member area ratio constraint in the step S43 is: The constraints are:

F _i ≥ 0; where R _λ is the reflectivity of the mixed pixel at wavelength λ; R _iλ and F _i are the reflectivity and abundance of the ith endmember, respectively; _ελ is the fitting residual; M is the number of endmembers; When solving, meet the preset first condition, second condition, third condition, fourth condition and fifth condition; The first condition is: the area ratio of the end members complies with the constraints of "non-negative" and "sum is one", but an error of 1% is allowed; The second condition is: when the NDVI of the pixel exceeds 0.3, use the snow-vegetation mixed model, otherwise use the snow-soil model and the snow-vegetation model in sequence; The third condition is: during the residual test, the RMS of the residuals of all bands shall not exceed 0.025; The fourth condition is: during the residual error test, the residual error of half of the continuous bands in the number of bands does not exceed 0.025; The fifth condition is: the result with the smallest RMSE is taken as the area ratio of each end member. 8. The method according to claim 7, wherein the step S5 comprises: S51, mask the snow cover inversion result layer on the basis of the mask layer; S52. Select a color table corresponding to the distribution range of the snow coverage value range, and use a Geotiff file of unsigned integer type and LZW compression method. The data file also saves the original projection information of the image, and the data file is the calculated snow coverage. product.

Images