Method and system for acquiring biomass of alpine grassland, and storage medium

Abstract

The invention provides an alpine grassland biomass obtaining method and system and a storage medium, and belongs to the technical field of data processing, and the method comprises the steps: obtaining the vegetation height, vegetation coverage and aboveground biomass of each sample region in each aerial image; according to the actually measured aboveground biomass and the spectral information of the pixel points in each sampling area, obtaining the aboveground biomass predicted value of each sampling area of all aerial images; and according to the above-ground biomass predicted values of all the sampling areas in each remote sensing image and the spectral information of the pixel points corresponding to the sampling areas, obtaining above-ground biomass of the areas corresponding to the pixel points in the remote sensing images. The objective of the invention is to solve the problem that the difference between a prediction result and an actual result is too large because the difference between the scale of a remote sensing satellite and the scale of a ground sampling quadrat is too large and a saturation phenomenon occurs when a single vegetation index is adopted to fit the above-ground biomass when an above-ground biomass upscaling estimation model is established at present.

Description

Technical Field

[0001] The present invention belongs to the technical field of data processing, and in particular relates to a method, system and storage medium for obtaining alpine grassland biomass. Background Art

[0002] As one of the most dominant vegetation types in the Third Pole, alpine grasslands are among the world's most unique grassland ecosystems, playing a vital role in carbon cycling, climate regulation, soil and water conservation, and maintaining biodiversity. However, affected by climate change and human activities, approximately 19-60% of alpine grasslands on the Qinghai-Tibet Plateau are experiencing varying degrees of degradation. Accurately monitoring the health of alpine grasslands is essential for maintaining their ecological and production functions. Aboveground biomass, a key indicator of grassland health and reflecting its productivity and carrying capacity, is crucial for assessing grassland health. Estimating aboveground biomass is crucial for grassland resource management and adaptive decision-making.

[0003] Currently, when using ground surveys combined with multi-source satellite remote sensing inversion to estimate aboveground biomass in alpine grasslands, satellite images are first taken of the surveyed area to generate satellite remote sensing images. Spectral information is then obtained for each pixel in the satellite remote sensing image. Several pixels in the satellite remote sensing image are then divided into several small sample areas, and the aboveground biomass within each sample area is manually estimated. Using a random forest algorithm, an upscaling model for estimating aboveground biomass from the small sample areas to the areas corresponding to the pixels in the satellite remote sensing image is constructed based on the aboveground biomass and spectral information of the pixels within the satellite remote sensing image. Due to the significant difference between the area of ​​the ground area corresponding to each pixel in the remote sensing image and the area of ​​each sample area, and the poor resolution of the remote sensing image, the aboveground biomass estimated through ground surveys combined with multi-source satellite remote sensing inversion deviates significantly from the measured value. Furthermore, the vegetation index calculated from vegetation spectral information tends to lose sensitivity in areas with high vegetation cover, leading to premature saturation of the biomass estimates. Improvements to the ground survey and multi-source satellite remote sensing inversion process are urgently needed. Summary of the Invention

[0004] In order to solve the problem that when obtaining the aboveground biomass of alpine grassland, the scale of the ground sampling area is too different from the scale of the satellite, resulting in a large difference between the biomass predicted by the existing method and the actual biomass, the present invention provides a method for obtaining the biomass of alpine grassland.

[0005] In order to achieve the above object, the present invention provides the following technical solutions: Acquire remote sensing images of alpine grassland areas, obtain several sampling areas from the remote sensing images, obtain several aerial photography areas from each sampling area, and obtain several sample areas from each aerial photography area; obtain vegetation height, vegetation cover and aboveground biomass of each sample area, vegetation cover of each aerial photography area, vegetation cover of each sampling area, and spectral information of each pixel in the remote sensing image; Based on the vegetation height, vegetation coverage and aboveground biomass of all sample areas in each aerial photography area, the predicted relationship between the aboveground biomass of each aerial photography area and the vegetation height is obtained; based on the predicted aboveground biomass values, vegetation height and vegetation coverage of all aerial photography areas in each sampling area, the predicted relationship between the aboveground biomass of each sampling area and the vegetation height is obtained; based on the spectral information of each pixel point in each remote sensing image and the predicted aboveground biomass value of the corresponding sampling area, a prediction model for the aboveground biomass of the area corresponding to the pixel point in each remote sensing image is obtained; Input the vegetation cover and vegetation height of each aerial photography area into the prediction relationship of the aboveground biomass of the aerial photography area to obtain the predicted value of the aboveground biomass of each aerial photography area; input the vegetation cover and vegetation height of each sampling area into the prediction relationship of the aboveground biomass of the sampling area to obtain the predicted value of the aboveground biomass of the sampling area; input the spectral information of each pixel point in each remote sensing image into the prediction model of the aboveground biomass of the area corresponding to the pixel point in the remote sensing image to obtain the predicted value of the aboveground biomass of the area corresponding to each pixel point in the remote sensing image. Obtain a remote sensing image of an alpine grassland area, obtain several sampling areas from the remote sensing image, and obtain several aerial photography areas from each sampling area; obtain several sample areas from each aerial photography area, and obtain the vegetation height, vegetation cover and aboveground biomass of each sample area, the vegetation cover of each aerial photography area, the vegetation cover of each sampling area, and the spectral information of each pixel point in the remote sensing image; According to the vegetation height, vegetation cover and aboveground biomass of all sample areas in each aerial photography area, the prediction relationship between aboveground biomass and vegetation height of each aerial photography area is obtained; combined with the vegetation cover and vegetation height of the aerial photography area, the predicted aboveground biomass of each aerial photography area is obtained. Based on the predicted aboveground biomass, vegetation height and vegetation cover of all aerial photographed areas in each sampling area, the predicted relationship between aboveground biomass and vegetation height in each sampling area was obtained; Input the vegetation coverage and vegetation height of each sampling area into the prediction relationship of the aboveground biomass of the sampling area to obtain the predicted aboveground biomass of the sampling area; According to the spectral information of each pixel point in each remote sensing image and the predicted aboveground biomass of the corresponding sampling area, a prediction model of the aboveground biomass of the area corresponding to the pixel point in each remote sensing image is obtained; The spectral information of each pixel point in each remote sensing image is input into the prediction model of the aboveground biomass of the area corresponding to the pixel point in the remote sensing image to obtain the predicted aboveground biomass of the area corresponding to each pixel point in the remote sensing image.

[0006] Furthermore, the specific steps of obtaining a plurality of sampling areas from the remote sensing image are as follows: Use remote sensing satellites to obtain remote sensing images of the study area. Randomly select the areas corresponding to several pixels in a remote sensing image.

[0007] Furthermore, the specific steps of obtaining the predicted relationship between the aboveground biomass and the vegetation height of each aerial photography area based on the vegetation height, vegetation coverage and aboveground biomass of all sample areas within each aerial photography area are as follows: First The product of vegetation height and vegetation coverage of the sample area in each aerial photography area is used as the horizontal axis, and the aboveground biomass of the sample area is used as the vertical axis to construct a two-dimensional sample space. First In the aerial photography area The product of vegetation height and vegetation coverage in the sample area is used as the first The horizontal coordinate of the data point; In the aerial photography area The aboveground biomass of the sample area is taken as the The vertical coordinate of the data point; get the data points; The first All sample areas in the aerial photography area are mapped into a two-dimensional sample space to obtain a number of data points in the two-dimensional sample space; Use the random forest algorithm to fit the data points in the two-dimensional sample space, and record the obtained fitting relationship as The prediction relationship of aboveground biomass in the aerial photography area; The first The mean value of vegetation height of all sample areas in the aerial photography area is recorded as The vegetation height of the aerial photography area.

[0008] Furthermore, the specific steps of obtaining the aboveground biomass prediction value of each aerial photography area are as follows: The first The product of vegetation height and vegetation coverage in the aerial photography area is used as the The prediction relationship of aboveground biomass in the aerial photography area is input, and the output is recorded as The predicted aboveground biomass of the aerial photography area.

[0009] Furthermore, the specific steps of obtaining the predicted relationship between the aboveground biomass and the vegetation height in each sampling area are as follows: First The product of vegetation height and vegetation coverage in the aerial photography area within each sampling area is used as the horizontal axis, and the predicted value of aboveground biomass in the aerial photography area is used as the vertical axis to construct a two-dimensional coordinate system. First Within the sampling area The product of vegetation height and vegetation coverage in the aerial photography area is used as the first The horizontal coordinate of the sample point; Within the sampling area The aboveground biomass of the aerial photography area is used as the first The vertical coordinate of the sample point; get the samples; The first All aerial photography areas within a sampling area are mapped into a two-dimensional coordinate system to obtain a number of data points in the two-dimensional coordinate system; Use the random forest algorithm to fit the data points in the two-dimensional coordinate system, and record the obtained fitting relationship as The prediction relationship of aboveground biomass in each sampling area; The first The mean value of vegetation height in all aerial photography areas within the sampling area is recorded as Vegetation height in the sampling area.

[0010] Furthermore, the specific steps of obtaining the predicted aboveground biomass value of each sampling area are as follows: The first The mean value of vegetation height in all aerial photography areas within the sampling area is recorded as Vegetation height in each sampling area; The first The product of vegetation height and vegetation coverage in the sampling area is taken as the The output of the prediction relationship of aboveground biomass in the sampling area is recorded as The predicted aboveground biomass of the sampling area.

[0011] Furthermore, the specific steps of obtaining a prediction model for the aboveground biomass of an area corresponding to a pixel point in each remote sensing image based on the spectral information of each pixel point in each remote sensing image and the aboveground biomass prediction value of the corresponding sampling area are as follows: The first The remote sensing image corresponds to the alpine grassland area The aboveground biomass prediction values ​​for the sampling area, and The remote sensing image corresponds to the alpine grassland area The sampling area corresponds to the The spectral information of the pixels in the remote sensing image is respectively used as Remote sensing image The two-dimensional data values ​​of the target point are obtained All target points in a remote sensing image; Using the random forest algorithm, according to The two-dimensional data values ​​of all target points in the remote sensing image are used to train a model, which is recorded as A prediction model for aboveground biomass in the area corresponding to a pixel point in a remote sensing image.

[0012] The present invention also provides a system for obtaining alpine grassland biomass, comprising: The data acquisition module is used to obtain remote sensing images of alpine grassland areas, obtain several sampling areas from the remote sensing images, obtain several aerial photography areas from each sampling area, and obtain several sample areas from each aerial photography area; obtain the vegetation height, vegetation cover and aboveground biomass of each sample area, the vegetation cover of each aerial photography area, the vegetation cover of each sampling area, and the spectral information of each pixel point in the remote sensing image; The prediction model construction module is used to obtain the prediction relationship between the aboveground biomass and vegetation height of each aerial photography area based on the vegetation height, vegetation cover and aboveground biomass of all sample areas in each aerial photography area; Based on the predicted aboveground biomass values, vegetation height, and vegetation cover of all aerial photographed areas within each sampling area, the predicted relationship between aboveground biomass and vegetation height in each sampling area was obtained; According to the spectral information of each pixel point in each remote sensing image and the predicted value of the aboveground biomass of the corresponding sampling area, a prediction model of the aboveground biomass of the area corresponding to the pixel point in each remote sensing image is obtained; The prediction model uses a module for inputting the vegetation cover and vegetation height of each aerial photography area into a prediction relationship of the aboveground biomass of the aerial photography area to obtain a predicted value of the aboveground biomass of each aerial photography area; The vegetation coverage and vegetation height of each sampling area are input into the prediction relationship of the aboveground biomass of the sampling area to obtain the predicted value of the aboveground biomass of the sampling area; The spectral information of each pixel point in each remote sensing image is input into the prediction model of the aboveground biomass of the area corresponding to the pixel point in the remote sensing image to obtain the predicted value of the aboveground biomass of the area corresponding to each pixel point in the remote sensing image.

[0013] The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is loaded by a processor, it can execute any one of the steps in the method for obtaining alpine grassland biomass.

[0014] The method for obtaining alpine grassland biomass provided by the present invention has the following beneficial effects: The present invention obtains the aboveground biomass of alpine grasslands through ground surveys combined with multi-source satellite remote sensing inversion. The adaptability of the scales of aerial images and sample areas, as well as the adaptability of the scales of aerial images and remote sensing satellites, are superior to the adaptability of the scales of remote sensing satellites and sample areas. When obtaining the aboveground biomass of the area corresponding to the remote sensing satellite, the aboveground biomass of the aerial image is first obtained through the aboveground biomass of multiple small sample areas in the aerial photography area, and the aboveground biomass of the area corresponding to the remote sensing satellite is obtained based on the aboveground biomass of the aerial photography area. This solves the problem of poor adaptability between the scales of the sample areas and the aerial photography areas, and reduces the difference between the predicted value of the aboveground biomass of an area and the actual value. Differences in aboveground biomass; Currently, the aboveground biomass of multiple sample plots in an aerial photography area is fitted with vegetation cover, and the aboveground biomass prediction value of the aerial photography area is obtained according to the vegetation cover of the aerial photography area; However, due to the complex vertical structure of grassland, the aboveground biomass of an area is not only related to vegetation cover but also to vegetation height. Therefore, this method obtains the aboveground biomass prediction value of an aerial photography area by multiplying the product of vegetation height and vegetation cover of multiple sample plots in an aerial photography area with the aboveground biomass, thereby enhancing the accuracy of the aboveground biomass of an aerial photography area and improving the accuracy of the aboveground biomass prediction value of the remote sensing satellite corresponding area. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] To more clearly illustrate the embodiments of the present invention and its design, the following briefly introduces the drawings required for this embodiment. The drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be derived from these drawings without inventive effort.

[0016] Figure 1 A flowchart of the steps of the method for obtaining alpine grassland biomass; Figure 2 The spatial distribution of grassland types and sampling areas in Qilian Mountains; Figure 3 Schematic diagram for scaling up Figure 4 Schematic diagram of the scaling process; Figure 5 It is a collection of aerial images; Figure 6 A collection of aerial photographs of the sample area; Figure 7 Schematic diagram of the process of building models for estimating aboveground biomass at the quadrat and aerial photo scales and resampling to the plot scale; Figure 8 To obtain sampling information of aboveground biomass, vegetation cover and vegetation height; Figure 9 The aboveground biomass and vegetation cover and vegetation cover × vegetation height regression fitting diagrams at the sample scale and aerial photo scale are shown in Figure 2. Figure 10 Maps showing the accuracy of aboveground biomass estimation at the quadrat scale and aerial photograph scale.

[0017] Figure 11 This is the distribution map of aboveground biomass in the Qilian Mountains region. DETAILED DESCRIPTION

[0018] In order to enable those skilled in the art to better understand the technical solution of the present invention and to be able to implement it, the present invention is described in detail below with reference to the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solution of the present invention and are not intended to limit the scope of protection of the present invention.

[0019] Example 1 The present invention provides a method for obtaining alpine grassland biomass, specifically as follows: Figure 1 As shown, the following steps are included: Step S001: Obtain a remote sensing image of an alpine grassland area, obtain several sampling areas from the remote sensing image, obtain several aerial photography areas from each sampling area, and obtain several sample areas from each aerial photography area; obtain the vegetation height, vegetation coverage and aboveground biomass of each sample area, the vegetation coverage of each aerial photography area, the vegetation coverage of each sampling area, and the spectral information of each pixel point in the remote sensing image.

[0020] It should be noted that, with the development of science and technology, small drones can quickly obtain high-resolution vegetation cover and other spectral information at each aerial scale, and the coverage of drone aerial images can effectively match the spatial scale of ground surveys and satellite pixels, making drone aerial images an intermediate bridge connecting the aboveground biomass from the sample area scale to the pixel point scale estimation in the remote sensing image. Therefore, in the process of upscaling, the present invention first randomly selects the areas corresponding to several pixel points in the remote sensing image as sampling areas, and then divides each sampling area into multiple small aerial areas, and divides each aerial area into multiple smaller sample areas. The upscaling process is as follows: Figure 3 shown. Figure 3 Figure (a) is an aerial image of the sampling area corresponding to a pixel point in the remote sensing image. Figure 3 The line connecting multiple points in Figure (b) is the flight path of the UAV in the sampling area, and the UAV takes an aerial photo at each point. Figure 3 Figure (c) in the figure is a schematic diagram of the sample area distribution in each aerial photography area. Figure 7 Schematic diagram of the process of building a model for estimating aboveground biomass at the quadrat and aerial photo scales and resampling to the plot scale. Figure 7 Figure (a) is a quadrat-scale image. Figure 7 Figure (b) shows the distribution of quadrats within the scale of aerial photos. Figure 7 Each small square in Figure (c) corresponds to an aerial photo scale.

[0021] It should be further explained that there are currently two methods for obtaining the aboveground biomass in an area. One is to obtain the aboveground biomass in an area through manual collection, and the other is to divide a large area into multiple small areas, and obtain the relationship between the aboveground biomass and vegetation coverage in the large area based on the aboveground biomass and vegetation coverage in the multiple small areas. The aboveground biomass of the large area is obtained based on the vegetation coverage of the large area. However, due to the complex vertical structural characteristics of grassland, the aboveground biomass of each area is not only related to the vegetation coverage but also to the vegetation height. Therefore, in the upscaling process, in addition to obtaining the vegetation coverage of each sample area, the vegetation height of each sample area is also obtained. Then, based on the acquired data, the aboveground biomass prediction value of the area corresponding to each pixel point in the remote sensing image is obtained through the idea of ​​upscaling. Among them Figure 4 It is a schematic diagram of the scaled process of the present invention, wherein Figure 4 Figure (a) shows the aboveground biomass, vegetation height, and vegetation coverage of multiple sample areas within each aerial photography area obtained using existing methods. Figure 4Figure (b) shows the relationship between the aboveground biomass of all sample areas within each aerial photography area and the product of vegetation height and vegetation cover, obtained using the random forest algorithm based on the data collected for each sample area. Figure 4 Figure (c) shows the predicted aboveground biomass value of the aerial photography area based on the vegetation height and vegetation coverage. Figure 4 Figure (d) continues to use the upscaling idea to obtain the predicted aboveground biomass value of the area corresponding to each pixel in the remote sensing image. Figure 8 Figure (a) shows the sampling information of aboveground biomass. Figure 8 Figure (b) shows the sampling information of vegetation height. Figure 8 Figure (c) shows the sampling information of vegetation height.

[0022] Specifically, remote sensing satellites are used to obtain remote sensing images of the study area and obtain the spectral information of each pixel in each remote sensing image. The study area here is the Qilian Mountains area on the northeastern edge of the Qinghai-Tibet Plateau. Random selection within a remote sensing image The area corresponding to each pixel is recorded as the sampling area. And the area of ​​the ground area corresponding to one pixel in the remote sensing image is Among them, the number of sampling areas in the area corresponding to each remote sensing image preset in this embodiment is , the preset pixel scale , will be described using this as an example, and other implementations may be set to other values. Figure 2 The spatial distribution map of grassland types and sampling areas in Qilian Mountain is shown in FIG. The acquisition of spectral information of each pixel in a remote sensing image is a well-known technique and will not be described in detail in this embodiment.

[0023] Furthermore, in each sampling area, The size is The aerial photography area is taken by using a drone to take pictures of each aerial photography area, and the aerial image of each aerial photography area is obtained, and then the vegetation coverage of each aerial photography area is obtained. The aerial photography of each sampling area is taken by using a drone to obtain the vegetation coverage of each sampling area. Among them, obtaining the vegetation coverage of each area based on the aerial image of each area is a well-known technology, which will not be described in detail in this embodiment. Among them, the number of aerial photography areas in each sampling area preset in this embodiment is , the length of the aerial photography area , the width of the aerial photography area , is described as an example, and can be set to other values ​​in other implementations. Figure 3 Figure (a) is the drone's vertex photo taking interface. Figure 3Figure (b) shows the flight path of the drone within the sampling area, and the drone takes a photo at each digital node. Figure 5 Aerial image collection.

[0024] Furthermore, each aerial photography area is divided into Sample areas were used to obtain the vegetation height and aboveground biomass of each sample area using existing methods. The sample areas were photographed using drones to obtain the vegetation coverage of each sample area. The size of the sample area is Among them, the number of sample areas in each aerial photography area preset in this implementation is , the length of the preset sample area Meters, width This is described as an example, and other values ​​can be set in other implementations. Figure 4 Figure (c) shows the distribution of sample areas within the aerial photography area. Figure 6 A collection of aerial photos of the sample area.

[0025] At this point, the spectral characteristics of each pixel in the remote sensing image, the vegetation coverage of each sampling area, the vegetation coverage of each aerial photography area, the vegetation height, vegetation coverage and aboveground biomass of each sample area are obtained.

[0026] Step S002: Based on the vegetation height, vegetation coverage and aboveground biomass of all sample areas in each aerial photography area, obtain the predicted relationship between the aboveground biomass and vegetation height of each aerial photography area; input the vegetation coverage and vegetation height of each aerial photography area into the predicted relationship between the aboveground biomass and vegetation height of the aerial photography area to obtain the predicted value of the aboveground biomass of each aerial photography area.

[0027] It should be noted that due to the complex vertical structure of grassland, the aboveground biomass within the sample area with the same vegetation coverage may vary greatly. Therefore, the product of vegetation height and vegetation coverage in each sample area is fitted with the aboveground biomass of each sample area to obtain the prediction relationship of the aboveground biomass of each pixel in each aerial photography area. Figure 9 Figure (a) is the regression fitting diagram of aboveground biomass and vegetation cover at the sample scale and aerial photo scale. Figure 9 Figure (b) shows the regression fitting of aboveground biomass and vegetation cover × vegetation height at the sample scale and aerial photo scale. It can be found that when directly using vegetation cover and aboveground biomass for fitting, overfitting will occur earlier than when fitting by the product of vegetation height and vegetation cover. Figure 9 Figure (a) shows the accuracy of aboveground biomass estimation at the quadrat scale. Figure 9 Figure (b) shows the accuracy of aboveground biomass estimation based on aerial photographs. Figure 9N represents the aboveground biomass involved in the modeling, the black line is 1:1, the red line is the fitting line, the gray represents the 95% confidence interval, and the bar graph represents the scattered density distribution.

[0028] Specifically, The product of vegetation height and vegetation coverage of the sample area in each aerial photography area is used as the horizontal axis, and the aboveground biomass of the sample area is used as the vertical axis to construct a two-dimensional sample space.

[0029] Further, the In the aerial photography area The product of vegetation height and vegetation coverage in the sample area is used as the first The horizontal coordinate of the data point; In the aerial photography area The aboveground biomass of the sample area is taken as the The vertical coordinate of the data point; get the data points.

[0030] Further, the All sample areas in the aerial photography area are mapped into a two-dimensional sample space to obtain a number of data points in the two-dimensional sample space; Furthermore, the random forest algorithm is used to fit the data points in the two-dimensional sample space, and the obtained fitting relationship is recorded as The prediction relationship of the aboveground biomass of the aerial photography area is shown in FIG. The random forest algorithm is used to fit the data points in the two-dimensional sample space, which is a well-known technique and will not be described in detail in this embodiment.

[0031] Further, the The mean value of vegetation height of all sample areas in the aerial photography area is recorded as The vegetation height of the aerial photography area. The product of vegetation height and vegetation coverage in the aerial photography area is used as the The prediction relationship of aboveground biomass in the aerial photography area is input, and the output is recorded as The predicted aboveground biomass of the aerial photography area.

[0032] At this point, the predicted value of aboveground biomass in each aerial photography area is obtained.

[0033] Step S003: Based on the predicted aboveground biomass values, vegetation heights, and vegetation coverages of all aerial photography areas within each sampling area, obtain the predicted relationship between the aboveground biomass and vegetation height of each sampling area; input the vegetation coverage and vegetation height of each sampling area into the predicted relationship between the aboveground biomass of the sampling area to obtain the predicted aboveground biomass value of the sampling area.

[0034] Specifically, The product of vegetation height and vegetation coverage in the aerial photography area within each sampling area is used as the horizontal axis, and the predicted value of aboveground biomass in the aerial photography area is used as the vertical axis to construct a two-dimensional coordinate system.

[0035] Further, the Within the sampling area The product of vegetation height and vegetation coverage in the aerial photography area is used as the first The horizontal coordinate of the sample point; Within the sampling area The aboveground biomass of the aerial photography area is used as the first The vertical coordinate of the sample point; get the samples.

[0036] Further, the All aerial photography areas within a sampling area are mapped into a two-dimensional coordinate system to obtain several data points in the two-dimensional coordinate system.

[0037] Furthermore, the random forest algorithm is used to fit the data points in the two-dimensional coordinate system, and the obtained fitting relationship is recorded as The random forest algorithm is used to fit the data points in the two-dimensional sample space, which is a well-known technique and will not be described in detail in this embodiment.

[0038] Further, the The mean value of vegetation height in all aerial photography areas within the sampling area is recorded as The vegetation height of each sampling area. The product of vegetation height and vegetation coverage in the sampling area is taken as the The output of the prediction relationship of aboveground biomass in the sampling area is recorded as The predicted aboveground biomass of the sampling area.

[0039] At this point, the predicted value of aboveground biomass in each sampling area was obtained.

[0040] Step S004: Based on the spectral information of each pixel point in each remote sensing image and the predicted value of the aboveground biomass of the corresponding sampling area, a prediction model of the aboveground biomass of the area corresponding to the pixel point in each remote sensing image is obtained; the spectral information of each pixel point in each remote sensing image is input into the prediction model of the aboveground biomass of the area corresponding to the pixel point in the remote sensing image to obtain the predicted value of the aboveground biomass of the area corresponding to each pixel point in the remote sensing image.

[0041] Specifically, the The remote sensing image corresponds to the alpine grassland area The aboveground biomass prediction values ​​for the sampling area, and The remote sensing image corresponds to the alpine grassland area The sampling area corresponds to the The spectral information of the pixels in the remote sensing image is respectively used as Remote sensing image The two-dimensional data values ​​of the target point are obtained All target points in a remote sensing image.

[0042] Furthermore, using the random forest algorithm, according to The two-dimensional data values ​​of all target points in the remote sensing image are used to train a model, which is recorded as A prediction model for aboveground biomass in an area corresponding to a pixel point in a remote sensing image is provided. A random forest algorithm is used to train the model based on data, which is a well-known technique and will not be described in detail in this embodiment.

[0043] Further, the In the remote sensing image The spectral information of each pixel is used as the The input of the prediction model of aboveground biomass in the area corresponding to the pixel point in the remote sensing image is recorded as the output In the remote sensing image The aboveground biomass prediction value of the area corresponding to the pixel point is obtained. Figure 7 This is the distribution map of aboveground biomass in the Qilian Mountains region.

[0044] At this point, this embodiment is completed.

[0045] Example 2: Next, the method proposed in the present invention is further described by taking the Qilian Mountains region on the northeastern edge of the Qinghai-Tibet Plateau as an example.

[0046] Step 1: Study area and data acquisition The present invention area is located in the Qilian Mountains on the northeastern edge of the Qinghai-Tibet Plateau, which is an important ecological security barrier and water conservation area in northwest China. 5767m, the terrain gradually rises from southeast to northwest. The area has a plateau continental climate, with the characteristics of high altitude and high latitude mountain climate, with obvious temperature and precipitation zonation, the annual average temperature is about -3.24℃, and the annual precipitation is about 300mm. The unique geographical location and climatic conditions have formed a variety of typical grassland types ( Figure 2 ), including alpine meadows, alpine steppes, temperate steppes and alpine deserts, and common plant species are Kentucky bluegrass ( Poa pratensis )、Stipa purpurogena( Stipa purpurea )、Elymus nutans ( Elymus nutans ), Alpine Songgrass ( Carex parvula )、Moss tibetica( Carex moorcroftii )、Potentilla bifidum( Sibbaldianthe bifurca )wait.

[0047] Based on the UAV aerial photography-ground collaborative survey fixed observation plots deployed by the present invention team in the Qilian Mountains ( Figure 2 ), 65 of the sample plots were selected based on the drone aerial photography system (FragMap) in 2023 A field survey was conducted in the Qilian Mountains from July to August 2024. At each 200m×200m fixed observation plot, a new GRID flight route was created or an existing one was called based on FragMap. Figure 3 (b) in the figure), each route can obtain 16 aerial photos. After the route is added, the Mavic 2 Pro (DJI Innovations, China) drone is used to set the flight altitude (20m for this project), flight speed (4m / s for this project) and camera parameters (white balance, sensitivity, etc.) according to the aerial photography requirements. The drone will then start autonomous flight and shooting according to the preset parameters ( Figure 3 Each aerial photograph covers a surface area of ​​25m×36m with a resolution of 1cm. At the same time, a ground collaborative survey was carried out at the sample site that matched the scale of the drone aerial photography. Six 0.5m×0.5m sample boxes were randomly arranged in each sample site ( Figure 3 (c) in the figure) to survey vegetation height, cover, and aboveground biomass. The specific steps for the vegetation survey were: 1) Measure the height of vegetation within the sample plot (10 replicates each for the height of reproductive branches and leaf tips at ground level); 2) Take vertical photographs of the vegetation within the sample plot; 3) Cut the aboveground vegetation within the sample plot at ground level (for shrubs, cut green branches) and place it in a paper bag.

[0048] Based on the GEE (Google Earth Engine) platform, the 2023 The MOD13Q1 (61) product with a resolution of 30 m × 30 m in July-August 2024 was declouded and vegetation indices were calculated and extracted (Equations 1 and 2), including the normalized vegetation index mean (NDVI_mean), the normalized vegetation index maximum (NDVI_max), the normalized vegetation index minimum (NDVI_min), and the enhanced vegetation index (EVI). ASTER GDEM 30 m resolution elevation data (DEM) was downloaded from the geospatial data cloud platform, and the slope and aspect were extracted using surface analysis tools.

[0049] Where NIR is the near infrared band, R is the red light band, and B is the blue light band.

[0050] Step 2: Aerial Photo and Vegetation Sample Analysis The quadrat-scale photos were converted to double-precision type using Matlab software and normalized to the range of 0-1. The threshold was calculated based on the Excess Green Index (EGI=2G-RB, R, G, and B are red, green, and blue bands, respectively). The accuracy of the threshold was adjusted using GIS10.8 software. The percentage of pixels smaller than the threshold in the total number of pixels was calculated to obtain the vegetation cover at the quadrat-scale ( Figure 5 and Figure 6 The vegetation coverage at the aerial photo scale is based on the grassland vegetation coverage analysis software (Pixel Classifier Manual) independently developed by the author team. The OpenCV package is used to outline green vegetation pixels or bare ground pixels according to the threshold value, and the original photo is manually corrected. The proportion of the vegetation patch area to the entire photo area is the vegetation coverage ( Figure 5 and Figure 6 The aboveground part of the vegetation needs to be placed in an oven at 65°C for 24 h and air-dried to a constant weight (dry weight), and the aboveground biomass is weighed.

[0051] Step 3: Modeling and estimation of aboveground biomass at different scales Random Forest is a machine learning algorithm widely used in simulation and prediction processes. It improves the predictive power and accuracy of the model by combining multiple decision trees and is widely used in estimating aboveground biomass. For model construction, the number of random forest regression trees (ntree) is first set to 500 by default and adjusted through 10-fold cross-validation (cv). The ranger package of the regression model is used to train and predict the optimized parameters. 70% of the data is randomly selected for model training and 30% for validation. The coefficient of determination (R) is then calculated. 2 ) and root mean square error (RMSE) to evaluate the accuracy of the model (Equations 3 and 4).

[0052] Where n is the number of samples, and represent the measured aboveground biomass and the predicted aboveground biomass, respectively. is the average of the measured aboveground biomass.

[0053] A random forest algorithm was used to combine measured aboveground biomass, drone images, and MODIS pixel data to construct an upscaling estimation model for aboveground biomass. The specific steps were as follows: 1) A random forest model with a sample size of 0.5 m × 0.5 m was established using vegetation cover, vegetation height, and measured aboveground biomass. Figure 7 (a) in Figure 1) 1) Estimating aerial photo-scale aboveground biomass (25 m × 36 m) based on vegetation cover and vegetation height; 2) Building an aerial photo-scale random forest model based on the estimated aerial photo-scale aboveground biomass and MODIS pixel data ( Figure 7 (b) in Figure 3) estimated the aboveground biomass at the pixel scale (30 m × 30 m) based on MODIS pixel data; 3) resampled the estimated aboveground biomass at the pixel scale to the plot scale (250 m × 250 m) to obtain the spatial distribution of aboveground biomass in the entire Qilian Mountain grassland ( Figure 7 (c) in the figure).

[0054] Step 4: Statistical analysis Linear and nonlinear regression were used to fit the relationships between aboveground biomass and vegetation cover, and between aboveground biomass and vegetation height × vegetation cover, and explanatory coefficients were calculated. MODIS image data were preprocessed and thematic maps were created using GIS 10.8 software. Permutation tests were used to test the significance of the aboveground biomass upscaling estimation model. All other statistical analyses and modeling were performed using R 4.3.1 software, and all plots were created using Origin and ggplot2.

[0055] The calculation results are analyzed below.

[0056] 1. Vegetation parameter characteristics and their fitting relationship with aboveground biomass Aboveground biomass is concentrated in range 0 275g·m -2 , showing an approximately normal distribution, with an average value of approximately 110 g·m -2 and has the highest frequency ( Figure 8 There are large differences in vegetation cover, with both low and high coverage areas having a large number of samples, while the number of samples in the intermediate coverage area (about 60%-80%) is relatively small ( Figure 8 (b) in the figure). The vegetation height shows a clear right skewness, and the main distribution range is 0 7cm, and the number of vegetation heights exceeding 15cm is relatively small ( Figure 8 (c) in Figure 1). At the sample and aerial photograph scales, vegetation cover and aboveground biomass showed a power function relationship, with vegetation cover explaining 39% and 61% of the variation in aboveground biomass, respectively ( Figure 9(a) in Figure 1). It is noteworthy that aboveground biomass reaches a saturated state with increasing vegetation cover. In contrast, vegetation height × vegetation cover has a linear relationship with aboveground biomass, with vegetation height × vegetation cover explaining 47% and 74% of the variation in aboveground biomass, respectively ( Figure 9 (b) in the figure).

[0057] 2. Assessment of the accuracy of aboveground biomass estimation The estimation model constructed at the sample and aerial photo scale can better estimate the aboveground biomass ( P <0.001, Figure 10 At the quadrat scale, the training set of the random forest model (R 2 =0.85, RMSE=27.97 g·m -2 ) and the test set (R 2 =0.72, RMSE=38.77 g·m -2 ) has a relatively high accuracy ( Figure 10 Compared with the quadrat scale, the training set of the random forest model at the aerial photo scale (R 2 =0.93, RMSE=13.5 g·m -2 ) and the test set (R 2 =0.78, RMSE=25.41 g·m -2 ) has higher accuracy and smaller error ( Figure 10 Secondly, the number of samples participating in model training (N=283) and testing (N=122) at the aerial photo scale is also higher than the number of samples participating in model training (N=646) and testing (N=278) at the quadrat scale.

[0058] 3. Spatial distribution of aboveground biomass The prediction results show that the estimated range of aboveground biomass of grassland in Qilian Mountains is 0 250g·m -2 , with an average value of about 110 g·m -2 The estimated results are close to the measured range. From a spatial scale perspective, the areas with higher aboveground biomass in the Qilian Mountains are mainly distributed in the southeast, while the areas with lower aboveground biomass are mainly concentrated in the northwest of the Qilian Mountains. The overall trend is a gradual decrease from the southeast to the northwest ( Figure 11 ).

[0059] The importance of vegetation height in estimating aboveground biomass is explained as follows: The results of the present invention show that the aboveground biomass shows a phenomenon of early saturation with the increase of vegetation cover at the sample plot and aerial photograph scales. Combined with the vegetation height parameter (vegetation cover × vegetation height), the aboveground biomass can be objectively estimated. Vegetation height reflects the structural characteristics of spatial heterogeneity of different grassland types in the vertical direction. Compared with the model constructed with vegetation cover as a single indicator, adding vegetation height can effectively eliminate the premature saturation phenomenon in estimating aboveground biomass and improve the estimation accuracy. Among the methods for extracting vegetation height (including three-dimensional reconstruction technology, canopy height model, etc.), existing studies have shown that drone aerial photography has a high accuracy in identifying forest ground height or high vegetation areas. However, for alpine grasslands with short plants, its prediction effect is poor (R 2 = 0.33). The results of this invention show that most of the vegetation in alpine grasslands is less than 7 cm in height ( Figure 8 Even aerial photographs with centimeter-level resolution can produce significant errors when capturing vegetation height. Therefore, the present invention measures vegetation height to ensure objectivity at the sample scale. However, compared with the measured biomass, the present invention's results show that the maximum value of aboveground biomass estimated at the aerial photograph scale is underestimated ( Figure 10 (b) in Figure 1). This is mainly because on the one hand, the -2 The number of aboveground biomass samples is small; on the other hand, the average value method is used to unify the vegetation height of aerial photos in the same flight route, which leads to saturation in areas with the same vegetation coverage but higher vegetation height, resulting in the higher aboveground biomass being underestimated.

[0060] Explanation of the estimation accuracy of aboveground biomass: Satellite remote sensing inversion typically captures periodic remote sensing images at fixed time intervals, but it has a lag in monitoring real-time changes in aboveground biomass. In contrast, drone aerial photography offers greater flexibility, is less affected by seasons and climate, and can quickly and efficiently obtain high-precision vegetation information over a wide area, making it advantageous for estimating aboveground biomass in alpine grasslands. The results of this study demonstrate that models constructed at both the plot and aerial photograph scales can accurately estimate aboveground biomass, demonstrating that drones can bridge the spatial scale mismatch between ground samples and satellite pixels and further improve the accuracy of aboveground biomass estimation. Notably, the aboveground biomass estimation model constructed at the aerial photograph scale achieved higher accuracy and sample size than the aboveground biomass estimation model constructed at the plot scale, with the increase in scale and sample size significantly improving estimation accuracy. Although current regression models based on drone aerial photography at small scales can effectively estimate aboveground biomass, their applicability and scalability across different grassland types are limited by scale limitations and spatial heterogeneity, and they lack validation. The aboveground biomass estimation model constructed in this study incorporates the aboveground biomass of multiple grassland types, effectively avoiding the influence of spatial heterogeneity. At the same time, the model takes into account the horizontal and vertical information of vegetation, and has high estimation accuracy at the sample plot and aerial photograph scales. The estimated values ​​of large-scale grassland aboveground biomass and its spatial distribution are close to the measured results.

[0061] Based on ground surveys, drone aerial photography, and remote sensing images, this paper constructs a model for estimating aboveground biomass from plots to aerial photos and then to pixels. The results show that the model estimation has high accuracy. Vegetation height is an important parameter for estimating aboveground biomass. Considering vegetation height can solve the problem of oversaturation of aboveground biomass when fitting a single vegetation cover. The aboveground biomass variation estimated based on the model is 0 250g·m -2 , the measured aboveground biomass variation is 0 275g·m -2 The estimated value is close to the measured result. Higher grassland aboveground biomass is mainly distributed in the southeastern region of Qilian Mountain and gradually decreases toward the northwest. The results of this paper show that the aboveground biomass upscaling estimation model constructed based on drone aerial photography can be used to accurately invert the aboveground biomass of large-scale alpine grasslands, providing a theoretical basis and practical reference for grassland resource management and adaptive decision-making.

[0062] Another embodiment of the present invention provides a system for obtaining alpine grassland biomass, the system comprising: The data acquisition module is used to obtain remote sensing images of alpine grassland areas, obtain several sampling areas from the remote sensing images, obtain several aerial photography areas from each sampling area, and obtain several sample areas from each aerial photography area; obtain the vegetation height, vegetation cover and aboveground biomass of each sample area, the vegetation cover of each aerial photography area, the vegetation cover of each sampling area and the spectral information of each pixel point in the remote sensing image.

[0063] The prediction model construction module is used to obtain the prediction relationship between the aboveground biomass and vegetation height of each aerial photography area based on the vegetation height, vegetation cover and aboveground biomass of all sample areas in each aerial photography area; Based on the predicted aboveground biomass values, vegetation height, and vegetation cover of all aerial photographed areas within each sampling area, the predicted relationship between aboveground biomass and vegetation height in each sampling area was obtained; According to the spectral information of each pixel point in each remote sensing image and the predicted value of the aboveground biomass of the corresponding sampling area, a prediction model of the aboveground biomass of the area corresponding to the pixel point in each remote sensing image is obtained.

[0064] The prediction model uses a module for inputting the vegetation cover and vegetation height of each aerial photography area into a prediction relationship of the aboveground biomass of the aerial photography area to obtain a predicted value of the aboveground biomass of each aerial photography area; The vegetation coverage and vegetation height of each sampling area are input into the prediction relationship of the aboveground biomass of the sampling area to obtain the predicted value of the aboveground biomass of the sampling area; The spectral information of each pixel point in each remote sensing image is input into the prediction model of the aboveground biomass of the area corresponding to the pixel point in the remote sensing image to obtain the predicted value of the aboveground biomass of the area corresponding to each pixel point in the remote sensing image.

[0065] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0066] The present invention is described with reference to flowcharts and / or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0067] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0068] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0069] It should be noted that the specific embodiments described above can enable those skilled in the art to more fully understand the present invention, but do not limit the present invention in any way. Therefore, although this specification and examples have described the present invention in detail, those skilled in the art should understand that the present invention can still be modified or replaced with equivalents; and all technical solutions and improvements that do not depart from the spirit and scope of the present invention are included in the scope of protection of the patent for the present invention. Any reference signs in the claims should not be construed as limiting the claims involved.

Claims

1. A method for obtaining alpine grassland biomass, characterized in that: include: Acquire remote sensing images of alpine grassland areas, obtain several sampling areas from the remote sensing images, obtain several aerial photography areas from each sampling area, and obtain several sample areas from each aerial photography area; obtain vegetation height, vegetation cover and aboveground biomass of each sample area, vegetation cover of each aerial photography area, vegetation cover of each sampling area, and spectral information of each pixel in the remote sensing image; Based on the vegetation height, vegetation coverage and aboveground biomass of all sample areas in each aerial photography area, the predicted relationship between the aboveground biomass of each aerial photography area and the vegetation height is obtained; based on the predicted aboveground biomass values, vegetation height and vegetation coverage of all aerial photography areas in each sampling area, the predicted relationship between the aboveground biomass of each sampling area and the vegetation height is obtained; based on the spectral information of each pixel point in each remote sensing image and the predicted aboveground biomass value of the corresponding sampling area, a prediction model for the aboveground biomass of the area corresponding to the pixel point in each remote sensing image is obtained; The vegetation coverage and vegetation height of each aerial photography area are input into the prediction relationship of the aboveground biomass of the aerial photography area to obtain the aboveground biomass prediction value of each aerial photography area; the vegetation coverage and vegetation height of each sampling area are input into the prediction relationship of the aboveground biomass of the sampling area to obtain the aboveground biomass prediction value of the sampling area; The spectral information of each pixel point in each remote sensing image is input into the prediction model of the aboveground biomass of the area corresponding to the pixel point in the remote sensing image to obtain the predicted value of the aboveground biomass of the area corresponding to each pixel point in the remote sensing image.

2. The method for obtaining alpine grassland biomass according to claim 1, characterized in that: The specific steps of obtaining a plurality of sampling areas from the remote sensing image are as follows: Use remote sensing satellites to obtain remote sensing images of the study area. Randomly select the areas corresponding to several pixels in a remote sensing image.

3. The method for obtaining alpine grassland biomass according to claim 1, characterized in that: The specific steps of obtaining the predicted relationship between the aboveground biomass and the vegetation height of each aerial photography area based on the vegetation height, vegetation coverage and aboveground biomass of all sample areas within each aerial photography area are as follows: First The product of vegetation height and vegetation coverage of the sample area in each aerial photography area is used as the horizontal axis, and the aboveground biomass of the sample area is used as the vertical axis to construct a two-dimensional sample space. First In the aerial photography area The product of vegetation height and vegetation coverage in the sample area is used as the first The horizontal coordinate of the data point; In the aerial photography area The aboveground biomass of the sample area is taken as the The vertical coordinate of the data point; get the data points; The first All sample areas in the aerial photography area are mapped into a two-dimensional sample space to obtain a number of data points in the two-dimensional sample space; Use the random forest algorithm to fit the data points in the two-dimensional sample space, and record the obtained fitting relationship as The prediction relationship of aboveground biomass in the aerial photography area; The first The mean value of vegetation height of all sample areas in the aerial photography area is recorded as The vegetation height of the aerial photography area.

4. The method for obtaining alpine grassland biomass according to claim 1, characterized in that: The specific steps of obtaining the predicted aboveground biomass value of each aerial photography area are as follows: The first The product of vegetation height and vegetation coverage in the aerial photography area is used as the The prediction relationship of aboveground biomass in the aerial photography area is input, and the output is recorded as The predicted aboveground biomass of the aerial photography area.

5. The method for obtaining alpine grassland biomass according to claim 1, characterized in that: The specific steps of obtaining the predicted relationship between the aboveground biomass and the vegetation height in each sampling area are as follows: First The product of vegetation height and vegetation coverage in the aerial photography area within each sampling area is used as the horizontal axis, and the predicted value of aboveground biomass in the aerial photography area is used as the vertical axis to construct a two-dimensional coordinate system. First Within the sampling area The product of vegetation height and vegetation coverage in the aerial photography area is used as the first The horizontal coordinate of the sample point; Within the sampling area The aboveground biomass of the aerial photography area is used as the first The vertical coordinate of the sample point; get the samples; The first All aerial photography areas within a sampling area are mapped into a two-dimensional coordinate system to obtain a number of data points in the two-dimensional coordinate system; Use the random forest algorithm to fit the data points in the two-dimensional coordinate system, and record the obtained fitting relationship as The prediction relationship of aboveground biomass in each sampling area; The first The mean value of vegetation height in all aerial photography areas within the sampling area is recorded as Vegetation height in the sampling area.

6. The method for obtaining alpine grassland biomass according to claim 1, characterized in that: The specific steps of obtaining the predicted aboveground biomass value of each sampling area are as follows: The first The mean value of vegetation height in all aerial photography areas within the sampling area is recorded as Vegetation height in each sampling area; The first The product of vegetation height and vegetation coverage in the sampling area is taken as the The output of the prediction relationship of aboveground biomass in the sampling area is recorded as The predicted aboveground biomass of the sampling area.

7. The method for obtaining alpine grassland biomass according to claim 1, characterized in that: The specific steps of obtaining the prediction model of the aboveground biomass of the area corresponding to each pixel point in each remote sensing image based on the spectral information of each pixel point in each remote sensing image and the aboveground biomass prediction value of the corresponding sampling area are as follows: The first The remote sensing image corresponds to the alpine grassland area The aboveground biomass prediction values ​​for the sampling area, and The remote sensing image corresponds to the alpine grassland area The sampling area corresponds to the The spectral information of the pixels in the remote sensing image is respectively used as Remote sensing image The two-dimensional data values ​​of the target point are obtained All target points in a remote sensing image; Using the random forest algorithm, according to The two-dimensional data values ​​of all target points in the remote sensing image are used to train a model, which is recorded as A prediction model for aboveground biomass in the area corresponding to a pixel point in a remote sensing image.

8. A system for obtaining alpine grassland biomass, characterized in that: include: The data acquisition module is used to obtain remote sensing images of alpine grassland areas, obtain several sampling areas from the remote sensing images, obtain several aerial photography areas from each sampling area, and obtain several sample areas from each aerial photography area; obtain the vegetation height, vegetation cover and aboveground biomass of each sample area, the vegetation cover of each aerial photography area, the vegetation cover of each sampling area, and the spectral information of each pixel point in the remote sensing image; The prediction model construction module is used to obtain the prediction relationship between the aboveground biomass and vegetation height of each aerial photography area based on the vegetation height, vegetation cover and aboveground biomass of all sample areas in each aerial photography area; Based on the predicted aboveground biomass values, vegetation height, and vegetation cover of all aerial photographed areas within each sampling area, the predicted relationship between aboveground biomass and vegetation height in each sampling area was obtained; According to the spectral information of each pixel point in each remote sensing image and the predicted value of the aboveground biomass of the corresponding sampling area, a prediction model of the aboveground biomass of the area corresponding to the pixel point in each remote sensing image is obtained; The prediction model uses a module for inputting the vegetation cover and vegetation height of each aerial photography area into a prediction relationship of the aboveground biomass of the aerial photography area to obtain a predicted value of the aboveground biomass of each aerial photography area; The vegetation coverage and vegetation height of each sampling area are input into the prediction relationship of the aboveground biomass of the sampling area to obtain the predicted value of the aboveground biomass of the sampling area; The spectral information of each pixel point in each remote sensing image is input into the prediction model of the aboveground biomass of the area corresponding to the pixel point in the remote sensing image to obtain the predicted value of the aboveground biomass of the area corresponding to each pixel point in the remote sensing image.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is loaded into a processor, it is capable of executing the steps of the method for obtaining alpine grassland biomass according to any one of claims 1 to 7.

Images