Grassland species diversity evaluation method and device based on vegetation radiation transfer model

Abstract

The present application relates to grassland species evaluation technical field, disclose a kind of grassland species diversity evaluation method and device based on vegetation radiation transmission model, the method comprises: construct the three-dimensional model of grassland quadrat with different species composition and different crown layer structure complexity;Simulate grassland plant leaf reflectivity spectrum curve and the reflectivity spectrum curve of soil background;According to grassland plant leaf reflectivity spectrum curve and the reflectivity spectrum curve of soil background, simulate grassland crown layer hyperspectral image;Calculate the crown layer spectral diversity index and crown layer structure character of grassland quadrat three-dimensional model;Using single or combined crown layer spectral diversity index and crown layer structure character constructs grassland species diversity evaluation model and evaluates grassland species diversity.The present application can effectively control the possible influence factor such as parameter based on crown layer spectral measurement evaluation grassland species diversity in field experiment, eliminate interference factor for grassland species evaluation, reach variable control, accurate evaluation effect.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of grassland species evaluation, and particularly relates to a grassland species diversity evaluation method and device based on a vegetation radiation transfer model. BACKGROUND

[0002] Grassland ecosystems are an important component of terrestrial ecosystems, and their plant diversity is a key factor affecting the functioning of grassland ecosystems, such as productivity and community stability. In recent years, with the intensification of human activities and global climate change, grassland species diversity has been continuously lost, and the structure and function of grasslands have been increasingly degraded. The reduction of grassland species diversity and the simplification of vegetation structure have seriously threatened the ability of grassland ecosystems to provide products and regulate the climate. Therefore, accurate evaluation of grassland species diversity is crucial for the maintenance of grassland ecosystem function and the protection of grassland resources.

[0003] Traditional methods of obtaining grassland plant species diversity based on manual field surveys are time-consuming and labor-intensive, and are often limited by time and space. Remote sensing technology has become an alternative method for evaluating species diversity because it can efficiently collect large-scale data. Currently, multispectral remote sensing data have been gradually applied to the estimation of plant species diversity, but due to the limited and discontinuous bands of multispectral remote sensing data, it is difficult to extract rich species information from them. The rise of hyperspectral remote sensing technology provides a new perspective for solving this problem. By combining spectral characteristics with imaging technology, hyperspectral remote sensing can quickly obtain the response spectrum of ground objects in thousands of continuous bands. These continuous hyperspectral remote sensing response spectra can efficiently capture more abundant species information, making up for the limitations of field surveys and multispectral remote sensing in evaluating grassland species diversity.

[0004] However, due to the interference of biological and non-biological factors such as vegetation canopy structure and environmental factors, the hyperspectral image is prone to disturbance, resulting in significant uncertainty in the inversion of grassland species diversity using hyperspectral. In recent years, with the rapid development of quantitative remote sensing, vegetation radiation transfer models have been developed and applied to vegetation spectral feature simulation. The model simulates the absorption, scattering and transmission process of light energy in the vegetation canopy by controlling the input of multiple variables affecting the vegetation leaf and canopy spectrum, establishes the physical connection between the vegetation physicochemical parameters and the spectral response, and thus inverts the spectral characteristics of the vegetation. On the one hand, the leaf-scale radiation transfer model (such as PROSPECT, LIBERTY model) simulates the leaf reflectance and transmittance spectral characteristics by characterizing the leaf internal structure, pigment content and water content. On the other hand, the canopy-scale radiation transfer model (such as PROSAIL, LESS model) considers the influence of canopy structure and environmental factors (solar incident angle, soil background) on radiation transfer, and is commonly used for the simulation of optical properties of vegetation canopy. In summary, the vegetation radiation transfer model can effectively simulate the spectral characteristics of vegetation at different scales by controlling the input of leaf, canopy and environmental variables. However, there are few studies on evaluating the species diversity of grassland using radiation transfer model. Therefore, we invented a method and device for evaluating the species diversity of grassland based on vegetation radiation transfer model. SUMMARY

[0005] The present application provides a method and device for evaluating the species diversity of grassland based on vegetation radiation transfer model, which effectively controls the possible influencing factors and other parameters for evaluating the species diversity of grassland based on canopy spectral measurement in field experiments, eliminates the interference factors for evaluating the species diversity of grassland, and achieves the effects of variable control and accurate evaluation.

[0006] Therefore, the present application provides the following technical solutions:

[0007] A method for evaluating the species diversity of grassland based on vegetation radiation transfer model, the method comprising:

[0008] Step 1, constructing a three-dimensional model of grassland plot with different species composition and different canopy structure complexity;

[0009] Step 2, simulating the leaf reflectance spectral curve of grassland plants and the reflectance spectral curve of soil background;

[0010] Step 3, simulating the canopy hyperspectral image of the three-dimensional model of grassland plot according to the leaf reflectance spectral curve of grassland plants and the reflectance spectral curve of soil background;

[0011] Step 4, extracting canopy structural traits and calculating canopy spectral diversity index;

[0012] Step 5, using the canopy spectral diversity index and canopy structural traits to build a model for evaluating the grassland species diversity.

[0013] Optionally, in step 1, by scaling, rotating, arranging and combining one or several three-dimensional models of grassland plant species, a three-dimensional model of the grassland plot with different species composition and different canopy structure complexity is constructed along the species diversity gradient.

[0014] The size of the three-dimensional model of the grassland plant species is scaled according to the actual height range of the grassland plant species; the species diversity gradient is set at equal interval; the different species composition is randomly combined within a reasonable range; the quantification level of the different canopy structure complexity is divided into different aspects of canopy level, vertical, surface and whole; the size of the three-dimensional model of the grassland plot is scaled proportionally to the size of the actual plot in the field.

[0015] Optionally, in step 2, the leaf scale radiation transfer model is used to simulate the reflectance spectrum curve of the grassland plant leaf; the soil reflectance model is used to simulate the reflectance spectrum curve of the soil background.

[0016] The key parameters for simulating the plant leaf reflectance spectrum curve using the leaf scale radiation transfer model include chlorophyll content, leaf area index, leaf thickness, leaf dry matter content and leaf water content.

[0017] The PROSPECT-D model is selected as the leaf scale radiation transfer model to simulate the plant leaf reflectance spectrum curve, and the GSV model is selected as the soil reflectance model to simulate the reflectance spectrum curve of the soil background.

[0018] Optionally, in step 3, the plant leaf reflectance spectrum curve and the soil background reflectance spectrum curve are input into the canopy three-dimensional radiation transfer model, and the canopy hyperspectral image of the three-dimensional model of the grassland plot is simulated by setting environmental factors and image parameters.

[0019] The environmental factors and image parameters input into the canopy three-dimensional radiation transfer model include sensor type, sensor height, spectral band, spectral resolution, image size, image height, image width, image resolution, observation zenith angle, observation azimuth angle, solar zenith angle and solar azimuth angle.

[0020] Optionally, in step 4, the canopy structure traits are extracted from the grassland plot three-dimensional model, and the canopy spectral diversity index is calculated through the canopy hyperspectral image of the grassland plot three-dimensional model; the canopy spectral diversity index includes the spectral variation coefficient, the convex hull area, the convex hull volume, the spectral angle mapping, and the hyperspectral data clustering; the canopy structure trait refers to the quantifiable parameter of the three-dimensional morphological characteristics of the grassland canopy, including the canopy average height, the canopy coverage, the canopy surface roughness, and the canopy entropy, which represent the vertical, horizontal, surface, and overall canopy structure traits, respectively;

[0021] In the calculation of the canopy spectral diversity index, the spectral average variation coefficient is used as the canopy spectral diversity index; before calculating the spectral variation coefficient, in order to avoid the influence of background information on the calculation of the canopy spectral diversity, the soil background of the canopy hyperspectral image of each grassland plot is first removed; the NDVI threshold method is used to remove the soil background, that is, the pixels with NDVI≤0.3 are defined as the soil background, and the remaining pixels are determined as vegetation; the pixel points determined as vegetation are used to calculate the spectral average variation coefficient as the canopy spectral diversity index; the calculation formula of the spectral average variation coefficient is as follows:

[0022]

[0023] wherein, represents the average spectral variation coefficient of all bands participating in the calculation, represents the standard deviation of the reflectivity of the λ band image, represents the average value of the reflectivity of the λ band image, is the number of bands participating in the calculation;

[0024] In the calculation of the canopy structure trait, first, the grassland plot three-dimensional model is preprocessed, the grassland three-dimensional point cloud data is converted from the grassland plot three-dimensional model in the simulation experiment by using Poisson sampling, and based on the preprocessed grassland three-dimensional point cloud data, the canopy structure trait of each grassland plot is extracted, including the canopy average height, the canopy coverage, the canopy roughness, and the canopy entropy.

[0025] Optionally, in step 5, the correlation analysis of the canopy spectral diversity and the grassland species diversity is performed; and one or several indicators of the canopy spectral diversity index and the canopy structure trait are randomly combined to construct the grassland species diversity evaluation model to evaluate the species diversity; the types of the constructed grassland species diversity evaluation model include the unary regression model, the multivariate regression model, the stepwise regression model, the random forest, and the neural network.

[0026] Optionally, the method further comprises:

[0027] Step 6, select any natural grassland across scales as the research object, obtain the canopy spectral diversity index and canopy structure traits of the natural grassland, and use them as the real verification data set of the simulation experiment to construct a grassland species diversity evaluation model based on the measured data to evaluate the grassland species diversity of the natural grassland.

[0028] Optionally, step 6 comprises:

[0029] Step 61, select any natural grassland across scales as the research object, lay out sample plots and quadrats for field investigation, and record the grassland species diversity information and the accurate center of the quadrat position;

[0030] Step 62, obtain the three-dimensional point cloud data of the grassland quadrat and the canopy spectral data of the grassland quadrat of the natural grassland;

[0031] Step 63, pre-process the three-dimensional point cloud data of the grassland quadrat, including cropping, denoising, filtering, and normalization, and extract the canopy structure traits of the natural grassland;

[0032] Step 64, pre-process the canopy spectral data of the grassland quadrat, remove the soil background by using the NDVI threshold method, and calculate the canopy spectral diversity index of the natural grassland;

[0033] Step 65, according to the canopy spectral diversity index and the canopy structure traits of the natural grassland, construct the grassland species diversity evaluation model based on the measured data to evaluate the grassland species diversity of the natural grassland.

[0034] A grassland species diversity evaluation device based on a vegetation radiation transfer model, the device comprises:

[0035] A grassland quadrat three-dimensional model construction unit constructs a grassland quadrat three-dimensional model with different species composition and different canopy structure complexity;

[0036] A reflectance spectrum curve simulation unit simulates the reflectance spectrum curve of grass plant leaves and the reflectance spectrum curve of soil background;

[0037] A grassland canopy hyperspectral image simulation unit simulates a grassland canopy hyperspectral image according to the reflectance spectrum curve of grass plant leaves and the reflectance spectrum curve of soil background;

[0038] A grassland quadrat three-dimensional model parameter calculation unit calculates the canopy spectral diversity index and the canopy structure traits of the grassland quadrat three-dimensional model;

[0039] A grassland species diversity evaluation unit uses single or combined canopy spectral diversity index and canopy structure traits to construct a grassland species diversity evaluation model to evaluate the grassland species diversity.

[0040] A computer readable storage medium, having stored thereon a computer program, which, when executed by a processor, performs the steps of the method for evaluating grassland species diversity based on a vegetation radiation transfer model.

[0041] The application provides a method and device for evaluating grassland species diversity based on a vegetation radiation transfer model, which simulates the plant community state of a real grassland by constructing a three-dimensional model of a grassland plant quadrat with different species diversity gradients, simulates the canopy hyperspectral image of the grassland quadrat by controlling parameters such as vegetation leaves, soil and environmental factors, and further evaluates the grassland species diversity by calculating the spatial variation index of the spectrum and combining the canopy structure characteristics. The application establishes an experiment with only the change of species diversity and the change of canopy structure as a single factor, which can well verify the spectral variation hypothesis, that is, whether the spectrum can effectively evaluate the species diversity of the grassland. Compared with the prior art, the application has the following beneficial effects:

[0042] 1) The application realizes the construction of a three-dimensional model of a grassland quadrat and the computer simulation of canopy spectrum.

[0043] 2) The application effectively controls the possible influencing factors such as parameters for evaluating the species diversity of the grassland based on the canopy spectrum measurement in the field experiment, achieves the purpose of controlling variables, and excludes interference factors for the evaluation of the species diversity of the grassland based on the spectral diversity.

[0044] 3) The application invents a new method for evaluating the species diversity of the grassland by combining the spectral diversity and the canopy structure characteristics through computer simulation, which can effectively improve the evaluation accuracy of the species diversity of the grassland. BRIEF DESCRIPTION OF DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the following will briefly introduce the drawings needed in the embodiments. Obviously, the drawings in the following description only constitute some embodiments of the application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0046] Figure 1 A flowchart of the method for evaluating grassland species diversity based on a vegetation radiation transfer model is disclosed.

[0047] Figure 2 A three-dimensional model of a grassland plant species used in the embodiments of the application is shown.

[0048] Figure 3 A three-dimensional model of a grassland quadrat constructed along a species diversity gradient in the embodiments of the application is shown.

[0049] Figure 4 A data distribution violin plot of the grass canopy structure complexity in the embodiment of the present application;

[0050] Figure 5 A reflectance spectrum curve of leaves of different species and a soil reflectance spectrum curve simulated by PROSPECT-D and GSV in the embodiment of the present application;

[0051] Figure 6 A canopy spectral normalized difference vegetation index map of the three-dimensional model of the grass plot simulated by LESS in the embodiment of the present application;

[0052] Figure 7 A correlation between the canopy spectral diversity index and the grass species diversity in the embodiment of the present application;

[0053] Figure 8 A plot of evaluating the grass species diversity by using the canopy spectral diversity and the canopy structure traits in the embodiment of the present application;

[0054] Figure 9 A three-dimensional point cloud data collection schematic diagram of the grass plot in the embodiment of the present application;

[0055] Figure 10 A canopy structure trait extraction schematic diagram in the embodiment of the present application;

[0056] Figure 11 A correlation between the canopy spectral diversity index and the grass species diversity in the field verification experiment in the embodiment of the present application;

[0057] Figure 12 A plot of evaluating the grass species diversity by using the canopy spectral diversity and the canopy structure traits in the field verification experiment in the embodiment of the present application;

[0058] Figure 13 A structure schematic diagram of the grass species diversity evaluation device based on the vegetation radiation transfer model in the embodiment of the present application. DETAILED DESCRIPTION

[0059] The specific embodiments of the present application are described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely intended to illustrate and explain the present application, and are not intended to limit the present application.

[0060] In order to make the person skilled in the art better understand the technical scheme of the present application, the technical scheme in the embodiments of the present application will be described clearly and completely below in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by the person skilled in the art without creative labor should belong to the protection scope of the present application.

[0061] As Figure 1 shown is a flow chart of the grassland species diversity evaluation method based on the vegetation radiation transfer model in the embodiments of the present application, including the following steps:

[0062] Step 101, constructing a grassland quadrat three-dimensional model with different species composition and different canopy structure complexity along the species diversity gradient by scaling, rotating, arranging and combining one or several three-dimensional models of grassland plant species in proportion.

[0063] The size of the three-dimensional model of the grassland plant species can be scaled according to the real height range of the species; the species diversity gradient is set at equal difference or equal ratio steps; the species composition is randomly combined within a reasonable range; and the quantification level of the canopy structure complexity is divided into different aspects such as canopy level, vertical, surface and whole.

[0064] The grassland quadrat is a standardized sampling area, usually a small area of grassland unit delimited by a square or rectangular frame. The size of the three-dimensional model of the grassland quadrat is scaled in proportion to the size of the actual quadrat in the field (generally the grassland quadrat is ).

[0065] Step 102, simulating the plant leaf reflectance spectrum curve by using the leaf scale radiation transfer model; simulating the reflectance spectrum curve of the soil background by using the soil reflectance model;

[0066] The key parameters for simulating the plant leaf reflectance spectrum curve by using the leaf scale radiation transfer model usually include: chlorophyll content (Cab), leaf area index (LAI), leaf thickness (Cw), leaf dry matter content (Cm), leaf water content (LCC) and the like. These parameters can be obtained by measuring the sampled grassland plant leaves in the field or by accumulating experience.

[0067] Step 103, inputting the plant leaf reflectance spectrum curve and the spectrum curve of the soil background into the canopy three-dimensional radiation transfer model, and simulating the grassland canopy hyperspectral image by setting the environmental factors (such as the solar incident angle) and the image parameters (such as the number of wave bands, image resolution and the like);

[0068] The environmental factors and image parameters input to the canopy radiative transfer model include sensor type, sensor height, spectral band, spectral resolution, image size, image height, image width, image resolution, observation zenith angle, observation azimuth angle, solar zenith angle, and solar azimuth angle. By controlling these input variables, the hyperspectral image of the grassland canopy can be simulated.

[0069] Step 104: Extract canopy structural traits and calculate the canopy spectral diversity index;

[0070] Canopy structural traits are extracted from the three-dimensional model of the grassland quadrat, and the canopy spectral diversity index is calculated using the hyperspectral image of the canopy from the three-dimensional model of the grassland quadrat. The canopy spectral diversity index includes spectral variation coefficient, convex hull area, convex hull volume, spectral angle mapping, hyperspectral data clustering, etc., among which the spectral variation coefficient is the most commonly used spectral diversity measure. Canopy structural traits refer to all quantifiable parameters of the three-dimensional morphological characteristics of the grassland canopy. In the embodiments of this invention, the main canopy structural traits include four types: average canopy height, canopy coverage, canopy surface roughness, and canopy entropy, which represent vertical, horizontal, surface, and overall canopy structural traits, respectively.

[0071] Step 105: Construct grassland species diversity assessment models using single or combined canopy spectral diversity indices and canopy structural traits to assess grassland species diversity.

[0072] The assessment of grassland species diversity using single or combined canopy spectral diversity indices and canopy structural traits specifically includes: conducting correlation analysis between canopy spectral diversity and grassland species diversity; and constructing a model to assess species diversity using one or more random combinations of canopy spectral diversity indices and canopy structural traits. The types of models constructed include univariate regression models, multivariate regression models, stepwise regression models, random forests, and neural networks, etc., selected according to the actual situation.

[0073] In one embodiment, the method further includes: step 106 (not shown in the figure), selecting any cross-scale natural grassland as the research object, obtaining the quadrat canopy spectral diversity index and canopy structural traits of the natural grassland as the real verification dataset for the simulation experiment, and constructing a grassland species diversity assessment model based on the measured data to assess the grassland species diversity of the natural grassland.

[0074] Building upon the simulation experiments demonstrating the need to use canopy spectral diversity indices and canopy structural traits to assess grassland species diversity, this step further validates this through field experiments, confirming the necessity of using canopy spectral diversity indices and canopy structural traits to assess grassland species diversity in natural grasslands. Specifically, this includes:

[0075] Step 61: Select natural grasslands of any scale as the research object, set up sample plots and quadrats for field surveys, and record grassland species diversity information and precise quadrat location centers through satellite imagery.

[0076] Step 62: Obtain the three-dimensional point cloud data and canopy spectral data of the grassland quadrats. The three-dimensional point cloud data of the grassland quadrats can be obtained by using ground-based lidar or by using mobile photogrammetry. The canopy spectral data covering the grassland quadrats should be selected from satellite images with high spatiotemporal resolution and spectral resolution as much as possible.

[0077] The specific method for acquiring 3D point cloud data of grassland quadrats using ground-based lidar is as follows: high-reflectivity targets are placed at the four corners of the grassland quadrats, and 3D point cloud data of the grassland quadrats is collected at the center of the four sides of the quadrats using ground-based lidar. Riegl RiScan Pro software is then used to identify the pre-placed high-reflectivity targets in the quadrats and stitch the data together. The stitched data is then cropped to retain only the 3D point cloud data containing the grassland quadrats.

[0078] The specific method for acquiring 3D point cloud data of grassland quadrats using mobile photogrammetry is as follows: High-reflectivity targets are placed within the four corners of the grassland quadrats. Two 360° three-dimensional surround photographs are taken around the grassland quadrats using a standard digital camera. During the first surround photograph, the camera angle is set to tilt downwards, and the top angle of the camera lens is controlled within the range of 0-90°, performing a 360° surround tilt photograph of the grassland quadrats. During the second surround photograph, the camera angle is set to tilt upwards, and the top angle of the camera lens is controlled within the range of 90°-180°, also performing a 360° surround tilt photograph of the grassland quadrats. It is important to note that the overlap rate between photographs must be maintained at no less than 80% during the mobile photographing process (for example, assuming a field of view of 30°, the number of photographs taken in one 3D surround photograph should be no less than 60). Setting the overlap rate to no less than 80% is significant for improving the accuracy and precision of the 3D model reconstruction of the sugar beet plants. The collected 3D surround photograph data of the grassland quadrats is imported into Agisoft Metashape Professional software, and 3D reconstruction of the grassland quadrats is performed using SFM technology. First, image features are extracted from the acquired photographs to estimate the position of the plants in three-dimensional spatial coordinates. The photographs are then registered and stitched together using these three-dimensional positions to generate a dense point cloud. The RGB information and XYZ coordinates of the dense point cloud are preserved and exported as a PLY format point cloud. A high-reflectivity target is used to correct the size of the reconstructed grassland quadrat three-dimensional point cloud to match the actual size of the grassland quadrat. Background information and markers are removed, retaining only the grassland quadrat three-dimensional point cloud with accurate dimensions.

[0079] Step 63: Preprocess the three-dimensional point cloud data of the grassland quadrat and extract the canopy structure of the natural grassland;

[0080] Step 64: Preprocess the canopy spectral data of grassland quadrats based on satellite imagery, remove soil background using the NDVI thresholding method, and calculate the canopy spectral diversity index of natural grassland.

[0081] Step 65: Based on the canopy spectral diversity index and canopy structural characteristics of natural grasslands, construct a grassland species diversity assessment model based on measured data to assess the grassland species diversity of natural grasslands.

[0082] In a specific embodiment of the present invention, the grassland species diversity assessment based on the vegetation radiative transfer model includes:

[0083] S1. Combine three-dimensional models of arbitrary grassland plant species to construct three-dimensional models of grassland quadrats with different species compositions and different canopy structure complexities along the species diversity gradient.

[0084] Specifically, Figure 2 Three-dimensional models of 16 grassland plant species collected in this embodiment are provided, and specific grassland species information is shown in Table 1. Table 1 clarifies the family, genus, and species to which the species belong, as well as the actual height range of the species. Based on this, the three-dimensional models of the plant species are scaled proportionally to further construct three-dimensional models of grassland quadrats. Figure 3 Three-dimensional models of grassland quadrats with different species compositions and canopy structure complexities, constructed along a species diversity gradient, are presented. In this embodiment, the species diversity gradient is set to 1, 2, 4, 6, 8, 10, 12, 14, and 16 according to species richness; the species composition is randomly combined with 16 species according to the composition rules of natural grasslands, that is, each quadrat has one species as the dominant species to form the dominant community composition. Figure 4 The complexity of the canopy structure is defined as low, medium, and high, based on four levels: vertical, horizontal, surface, and overall. In this embodiment, the size of the grassland quadrat 3D model is set to... These grassland quadrat 3D models consist of one control group and five community quadrats with different dominant species. The main difference between the control and experimental groups lies in whether the plants exhibit significant differences in canopy height. These grassland quadrat 3D models will be further used to simulate grassland canopy spectra.

[0085] Table 1 Information on the 3D model of grassland plants

[0086]

[0087] S2. Simulate the reflectance spectrum curve of grassland plant leaves using a leaf-scale radiative transfer model; simulate the reflectance spectrum curve of soil background using a soil reflectance model.

[0088] Reference Figure 5 In this embodiment, the Prospect-D leaf radiative transfer model is used. The input parameters of this model include leaf internal structure parameters (N), chlorophyll content (Ct), carotenoid content (Cxc), anthocyanin content (Ant), brown pigment content (Brown), dry matter content (Cm), and water content (Cw). These key parameters from the 3D models of 16 grassland plant species are input into the Prospect-D leaf radiative transfer model to simulate the leaf spectral reflectance curve for each species. A soil reflectance spectral model is used to simulate the spectral reflectance curve of the soil background. The reflectance spectral bands of both plant leaves and soil are 400nm-2500nm.

[0089] Specifically, in this embodiment, the PROSPECT-D model is used as the leaf-scale radiative transfer model to simulate the reflectance spectrum curve of plant leaves, and the GSV model is used as the soil reflectance spectrum model to simulate the reflectance spectrum curve of the soil background. The PROSPECT-D model is a leaf optical property model based on radiative transfer theory, used to simulate the spectral reflectance and transmittance of plant leaves. The GSV model is a model specifically for quantifying the three-dimensional structure of grassland vegetation, mainly used to estimate grassland biomass, vegetation cover, and spatial distribution characteristics. Table 2 shows the key parameter inputs for the reflectance spectrum curves of grassland plant leaves in this embodiment:

[0090] Table 2 Key parameters of leaf reflectance spectra of grassland species

[0091]

[0092] S3. Input the reflectance spectral curves of plant leaves and the spectral curves of soil background into the three-dimensional radiative transfer model of the canopy, and simulate the hyperspectral image of the grassland canopy by setting environmental factors and image parameters.

[0093] Reference Figure 6 In this embodiment, Figure 6The normalized vegetation index (NDI) of canopy spectral reflectance for each grassland quadrat 3D model simulated using the canopy radiative transfer model in LESS (Large Scale Remote Sensing Data and Image Simulation Framework) is presented. Specifically, the leaf and soil spectral reflectance curves simulated in S2 were used as input parameters for LESS to further simulate the canopy spectral image of the grassland quadrat 3D model. To avoid the influence of vegetation shadows on the canopy spectral image, the solar altitude angle was set to 0° and the solar azimuth angle to 90°. The sensor used orthogonal images for simulation, with the image height and width set to 100×100, i.e., a resolution of 0.01 m×0.01 m. The observed zenith was 0°, and the observed azimuth angle was 180°. The simulation type was hyperspectral reflectance data, with the spectral band set to 400 nm–2500 nm and a spectral resolution of 1 nm. Table 3 shows the input information for each parameter simulated using the canopy radiative transfer model.

[0094] Table 3 Parameters of the three-dimensional canopy radiative transfer model in LESS

[0095]

[0096] S4. Extract canopy structural traits and calculate the canopy spectral diversity index.

[0097] Canopy structural traits are extracted from the 3D model of the grassland quadrat, and the canopy spectral diversity index is calculated using the hyperspectral image of the canopy from the 3D model of the grassland quadrat. In this embodiment, the coefficient of variation (COP) is used as the canopy spectral diversity index. Before calculating this index, to avoid the influence of background information on the calculation of canopy spectral diversity, the soil background of the canopy spectral image of each quadrat is first removed. The method for removing the soil background is the NDVI thresholding method, that is, pixels with NDVI ≤ 0.3 are defined as soil background, and the remaining pixels are determined as vegetation. The pixels determined as vegetation are used to calculate the average COP as the spectral diversity index. The formula for calculating the average COP is as follows:

[0098]

[0099] in, This represents the average spectral variation coefficient across all bands involved in the calculation. The standard deviation of the reflectance of a λ-band image is represented by... This represents the average reflectance of the λ-band image. It is the number of bands involved in the calculation.

[0100] Specifically, the method for extracting grassland canopy structural traits in this embodiment is as follows: First, the simulated grassland quadrat 3D model is preprocessed. Poisson sampling is used to convert the simulated grassland quadrat 3D model (.obj format) into 3D point cloud data (.las format). The point density of the converted point cloud data is 10000 pts / m². 2 Based on the preprocessed grassland 3D point cloud data, the canopy structure characteristics of each quadrat were extracted, namely the average canopy height ( H mean ), canopy coverage ( CC ), canopy roughness ( H sd ) and canopy entropy ( CE These canopy structural traits are considered important indicators of grassland canopy structural complexity in vertical, horizontal, surface, and overall dimensions, and are highly sensitive to grassland species diversity. The specific extraction process involves voxelizing each quadrat (5 cm × 5 cm) and calculating the average height and canopy roughness based on the voxels. The average height is calculated as the average height of points on the canopy surface, and the canopy roughness is calculated as the standard deviation of the height of these points. Canopy coverage is calculated based on the canopy spectral image of each grassland quadrat in the simulation experiment, i.e., the ratio of vegetation pixels to the total number of pixels. Canopy entropy is the degree of distribution of canopy leaves in three-dimensional space within the grassland quadrat; the more diffuse the distribution, the higher the overall complexity of the grassland canopy.

[0101] S5. Use single or combined canopy spectral diversity indices and canopy structural traits to construct models to assess grassland species diversity.

[0102] In this embodiment, Figure 7 The correlation analysis results between canopy spectral diversity and grassland species diversity are presented, and the analysis is carried out at different levels of canopy structural complexity. Figure 7 This study utilizes a traditional method: assessing the accuracy of grassland species diversity solely using canopy spectral diversity, denoted by the correlation coefficient r. It also demonstrates that the correlation differs across low, medium, and high canopy structure complexity gradients, proving that grassland species diversity is influenced by changes in canopy structure, affecting the accuracy of canopy spectral diversity assessment. The results show a significant correlation between canopy spectral diversity and species diversity, exhibiting a significant positive correlation at different levels of canopy structure complexity. This indicates that the method of calculating canopy spectral diversity based on vegetation radiative transfer models can effectively assess grassland species diversity. Figure 8 A model combining canopy spectral diversity and canopy structural traits, using multiple linear regression, is presented to assess and predict grassland species diversity. Figure 8This assesses the accuracy of species diversity by combining canopy spectral diversity and canopy structural diversity indices. The results show that compared to... Figure 7 The accuracy of grassland species diversity assessment is significantly improved by combining canopy structural traits and spectral diversity, rather than relying solely on canopy spectral diversity. Figure 7 and Figure 8 The results show that canopy structure is an indispensable factor in grassland species diversity assessment. Simultaneously considering canopy structure and canopy spectral diversity in grassland species diversity assessment is a new method to improve the accuracy of remote sensing assessment of grassland species diversity.

[0103] S6. Select natural grasslands of arbitrary scales as research objects, obtain the canopy spectral diversity index and canopy structural traits of the natural grasslands, and use them as the real validation dataset for the simulation experiment. Construct a grassland species diversity assessment model based on measured data to evaluate the grassland species diversity of natural grasslands. Specifically, this includes:

[0104] Step 6a: Select natural grasslands of any scale as the research object, establish plots and quadrats for field surveys, and record grassland species diversity information and precise quadrat location centers; specifically including: plot size of... 3-5 samples were randomly selected from each plot. The quadrats were identified; the species composition within the quadrats was identified, and information such as their names, frequencies, and coverage was recorded; the location of the center of each quadrat was recorded using a CHCNAV Global Navigation Satellite System (GNSS) receiver to match satellite imagery, which is connected to the Qianxun Continuously Operating Reference Stations (CORS) network and provides centimeter-level positioning accuracy.

[0105] Using natural grasslands in northern China as the research object, sample plots of 30 m × 30 m were selected from east to west. Each sample plot had 3-5 quadrats of 1 m × 1 m. Species composition and frequency were recorded to obtain species diversity and to construct a field observation dataset that spans climate, topography and grassland types.

[0106] Step 6b: Obtain the 3D point cloud data and canopy spectral data of the grassland quadrats; refer to... Figure 9 Specifically, this includes: acquiring the three-dimensional point cloud data of grassland quadrats by using ground-based lidar or mobile photogrammetry; and selecting satellite imagery with high spatiotemporal and spectral resolution as much as possible for the canopy spectral data covering the grassland quadrats.

[0107] Specifically, in this embodiment, ground-based lidar is selected as the method for acquiring 3D point cloud data of grassland. The specific method for acquiring 3D point cloud data of grassland quadrats using ground-based lidar is as follows: high-reflectivity targets are deployed at the four corners of the grassland quadrat, and 3D point cloud data of the grassland quadrat is collected at the center of the four sides of the quadrat using ground-based lidar. Using RieglRiScan Pro software, the high-reflectivity targets pre-deployed in the quadrat are identified and the data is stitched together. The stitched data is then cropped, retaining only the 3D point cloud containing the grassland quadrats.

[0108] Specifically, in this embodiment, the canopy spectral data of grassland quadrats is obtained as follows: A buffer zone with a diameter of 20 m is drawn in ArcGIS 10.2, centered on the location coordinates of each quadrat. Sentinel-2 imagery covering each buffer zone is downloaded using Google Earth Engine (GEE). From the multiple spectral bands of the Sentinel-2 imagery, 10 bands (R, G, B, NIR, RE1, RE2, RE3, RE4, SWIR1, and SWIR2) are selected to calculate the spectral diversity index.

[0109] Step 6c: Preprocess the 3D point cloud data of the grassland quadrat and extract the canopy structure characteristics of the natural grassland; specifically, this includes: preprocessing steps for the 3D point cloud data of the grassland quadrat obtained using ground-based lidar or 3D surround photography, including noise reduction, filtering, and normalization. Canopy structure characteristics include the grassland canopy structure characteristics at different levels, such as horizontal, vertical, surface, and overall.

[0110] Specifically, the noise reduction step in the preprocessing process uses the k-nearest noise reduction algorithm to remove noise points caused by wind, flying objects (such as birds and insects), etc. Filtering uses a local minimum filtering algorithm to identify ground points. Normalization aims to eliminate the influence of terrain on the height measurement of ground-based lidar data by subtracting the height of the corresponding ground point from the height of vegetation points.

[0111] Specifically, Figure 10 A method for extracting grassland canopy structural traits is presented. Based on preprocessed grassland quadrat 3D point cloud data, this embodiment extracts the canopy structural traits of each quadrat, namely the average canopy height (…). H mean ), canopy coverage ( CC ), canopy roughness ( H sd ) and canopy entropy ( CEThese canopy structural traits are used as quantitative indicators of grassland canopy structure. Table 4 shows the specific calculation formulas for each canopy structural trait. These canopy structural traits are considered important indicators of grassland canopy structure complexity in vertical, horizontal, surface, and overall dimensions, and are highly sensitive to grassland species diversity. Each quadrat was voxelized, with a voxel size of 5 cm × 5 cm. Average height and canopy roughness were calculated based on the voxels. Average height was calculated as the average height of points on the canopy surface, and canopy roughness was calculated as the standard deviation of the height of points on the canopy surface. Canopy coverage was calculated based on the canopy spectral image of each grassland quadrat in the simulation experiment, i.e., the ratio of vegetation pixels to the total number of pixels. Canopy entropy is the degree of distribution of canopy leaves in three-dimensional space within the grassland quadrat; the more diffuse the distribution, the higher the overall complexity of the grassland canopy. The specific calculation formulas for grassland canopy structural traits are shown in Table 4.

[0112] Table 4. Description of grassland canopy structure

[0113]

[0114] Note: Representing the i The height of each canopy surface grid; Represents the total number of grid cells on the canopy surface; The number of grid points representing vegetation points; Represents the number of grids on all canopy surfaces; , and These represent the canopy entropy in the xy, xz, and yz planes, respectively.

[0115] Step 6d: Preprocess the canopy spectral data of grassland quadrats based on satellite imagery, remove soil background using the NDVI thresholding method, and calculate the canopy spectral diversity index of natural grassland.

[0116] Preprocessing of grassland canopy spectral data based on satellite imagery was performed: To avoid the influence of background information on the calculation of canopy spectral diversity, soil background was first removed from the canopy spectral image of each plot. The NDVI thresholding method was used to remove the soil background; pixels with NDVI ≤ 0.3 were defined as soil background, and the remaining pixels were identified as vegetation. Pixels identified as vegetation were used to calculate the spectral mean coefficient of variation as a spectral diversity index. Canopy spectral diversity indices include spectral coefficient of variation, convex hull area, convex hull volume, spectral angle mapping, and hyperspectral data clustering, selected according to the actual situation. Among these, the spectral coefficient of variation is the most commonly used spectral diversity measure.

[0117] Specifically, in this embodiment, the spectral mean coefficient of variation is selected as the measure of spectral diversity. The specific formula for calculating the spectral mean coefficient of variation is as follows:

[0118]

[0119] in, This represents the average spectral variation coefficient across all bands involved in the calculation. The standard deviation of the reflectance of a λ-band image is represented by... This represents the average reflectance of the λ-band image. It is the number of bands involved in the calculation.

[0120] Step 6e: Based on the canopy spectral diversity index and canopy structural traits of natural grasslands, construct a grassland species diversity assessment model based on measured data to assess the grassland species diversity of natural grasslands.

[0121] Figure 11 A correlation diagram showing the relationship between canopy spectral diversity index and grassland species diversity in a field validation experiment; Figure 12 Maps for assessing grassland species diversity using canopy spectral diversity and canopy structural traits, used in field validation experiments; and Figure 7 and Figure 8 The simulation results are different. Figure 11 and Figure 12 These are results from a field experiment. Figure 7 and Figure 11 As can be seen, using only the canopy spectral diversity index to assess grassland species diversity is inaccurate. And as... Figure 8 and 13 As shown, combining canopy structure and spectral diversity in species diversity assessment significantly improves accuracy. Therefore, both simulation and field experiments validate that, compared to using spectral diversity alone, combining canopy structure and spectral diversity not only improves assessment accuracy but also reverses the positive or negative direction of the correlation.

[0122] Accordingly, embodiments of the present invention also provide a grassland species diversity assessment device based on a vegetation radiative transfer model, such as... Figure 13 The image shows a schematic diagram of the device. This grassland species diversity assessment device based on a vegetation radiative transfer model includes the following modules:

[0123] Grassland quadrat 3D model building unit 901 is used to construct grassland quadrat 3D models with different species compositions and different canopy structure complexities.

[0124] Unit 902 simulates the reflectance spectrum curves of grassland plant leaves and the reflectance spectrum curves of soil background.

[0125] The grassland canopy hyperspectral image simulation unit 903 simulates a grassland canopy hyperspectral image based on the reflectance spectral curves of the grassland plant leaves and the reflectance spectral curves of the soil background.

[0126] The grassland quadrat three-dimensional model parameter calculation unit 904 calculates the canopy spectral diversity index and canopy structural characteristics of the grassland quadrat three-dimensional model;

[0127] Grassland species diversity assessment unit 905 uses single or combined canopy spectral diversity indices and canopy structural traits to construct a grassland species diversity assessment model to assess grassland species diversity. The grassland species diversity assessment model is a model established based on simulation experimental data.

[0128] In one embodiment, the device further includes a grassland species diversity assessment unit based on measured data. It selects natural grasslands of any scale as research objects, obtains the quadratic canopy spectral diversity index and canopy structural traits of the natural grasslands, uses them as a real verification dataset for simulation experiments, and constructs a grassland species diversity assessment model based on measured data to assess the grassland species diversity of natural grasslands.

[0129] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, because according to the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0130] The present invention also provides a storage medium, which is a computer-readable storage medium storing a computer program thereon, the computer program being executable when it runs. Figure 1 The method shown may include some or all of the steps. The storage medium may include read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, etc. The storage medium may also include non-volatile memory or non-transitory memory, etc.

[0131] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data provider to another website, computer, server, or data provider via wired or wireless means.

[0132] The embodiments of the present invention have been described in detail above. Specific implementation methods have been used to illustrate the present invention. The descriptions of the embodiments above are only for the purpose of helping to understand the methods and systems of the present invention, and are merely some, not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention, and the content of this specification should not be construed as a limitation of the present invention. Therefore, any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for assessing grassland species diversity based on a vegetation radiative transfer model, characterized in that, The method includes: Step 1: Construct three-dimensional models of grassland quadrats with different species compositions and different canopy structure complexities; Step 2: Simulate the reflectance spectrum curves of grassland plant leaves and soil background; Step 3: Based on the reflectance spectral curves of the grassland plant leaves and the reflectance spectral curves of the soil background, simulate the canopy hyperspectral image of the three-dimensional model of the grassland quadrat; Step 4: Extract canopy structural traits and calculate the canopy spectral diversity index; Step 5: Construct grassland species diversity assessment models using single or combined canopy spectral diversity indices and canopy structural traits to assess grassland species diversity; The simulation of the reflectance spectral curves of the grassland plant leaves and the soil background includes: simulating the reflectance spectral curves of the grassland plant leaves using a leaf-scale radiative transfer model; and simulating the reflectance spectral curves of the soil background using a soil reflectance model. The key parameters used in simulating the reflectance spectrum curve of plant leaves using the leaf-scale radiative transfer model include: chlorophyll content, leaf area index, leaf thickness, leaf dry matter content, and leaf moisture content. The PROSPECT-D model was selected as the radiative transfer model at the leaf scale to simulate the reflectance spectrum of the plant leaves, and the GSV model was selected as the soil reflectance model to simulate the reflectance spectrum of the soil background.

2. The grassland species diversity assessment method based on a vegetation radiative transfer model according to claim 1, characterized in that, In step 1, by scaling, rotating, arranging and combining one or more grassland plant species three-dimensional models at the same scale, a grassland quadrat three-dimensional model with the different species composition and the different canopy structure complexity is constructed along the species diversity gradient. The size of the 3D model of grassland plant species is scaled and adjusted according to the actual height range of grassland plant species; the species diversity gradient is set with arithmetic or geometric step sizes; the different species are randomly combined within a reasonable range; the quantification levels of different canopy structure complexity are divided into different aspects of canopy horizontal, vertical, surface and overall; the size of the 3D model of grassland quadrats is scaled proportionally to the size of the actual quadrats in the field.

3. The grassland species diversity assessment method based on a vegetation radiative transfer model according to claim 1, characterized in that, In step 3, the reflectance spectrum curves of the plant leaves and the reflectance spectrum curves of the soil background are input into the three-dimensional radiative transfer model of the canopy, and the hyperspectral image of the canopy of the grassland quadrat three-dimensional model is simulated by setting environmental factors and image parameters. The environmental factors and image parameters input to the canopy three-dimensional radiative transfer model include sensor type, sensor height, spectral band, spectral resolution, image size, image height, image width, image resolution, observation zenith angle, observation azimuth angle, solar zenith angle, and solar azimuth angle.

4. The grassland species diversity assessment method based on a vegetation radiative transfer model according to claim 1, characterized in that, In step 4, canopy structural traits are extracted from the three-dimensional model of the grassland quadrat, and the canopy spectral diversity index is calculated using the hyperspectral image of the canopy from the three-dimensional model of the grassland quadrat. The canopy spectral diversity index includes the spectral variation coefficient, convex hull area, convex hull volume, spectral angle mapping, and hyperspectral data clustering. Canopy structural traits refer to quantifiable parameters of the three-dimensional morphological characteristics of the grassland canopy, including four canopy structural traits representing vertical, horizontal, surface, and overall canopy structure, respectively: average canopy height, canopy coverage, canopy surface roughness, and canopy entropy. When calculating the canopy spectral diversity index, the spectral mean coefficient of variation is used as the canopy spectral diversity index. Before calculating the spectral mean coefficient of variation, to avoid the influence of background information on the calculation of canopy spectral diversity, the soil background of the canopy hyperspectral image of each grassland plot is first removed. The NDVI thresholding method is used to remove the soil background, that is, pixels with NDVI ≤ 0.3 are defined as soil background, and the remaining pixels are determined as vegetation. The pixels determined as vegetation are used to calculate the spectral mean coefficient of variation as the canopy spectral diversity index. The formula for calculating the spectral mean coefficient of variation is as follows: in, This represents the average spectral variation coefficient across all bands involved in the calculation. The standard deviation of the reflectance of a λ-band image is represented by... This represents the average reflectance of the λ-band image. This refers to the number of bands involved in the calculation; When calculating the canopy structure characteristics, the grassland quadrat three-dimensional model is first preprocessed. Poisson sampling is used to convert the grassland quadrat three-dimensional model in the simulation experiment into grassland three-dimensional point cloud data. Based on the preprocessed grassland three-dimensional point cloud data, the canopy structure characteristics of each grassland quadrat are extracted, including the average canopy height, canopy coverage, canopy roughness and canopy entropy.

5. The grassland species diversity assessment method based on a vegetation radiative transfer model according to claim 1, characterized in that, In step 5, a correlation analysis is performed on canopy spectral diversity and grassland species diversity; and the grassland species diversity assessment model is constructed by randomly combining one or more indicators of canopy spectral diversity index and canopy structural traits to assess species diversity. The types of grassland species diversity assessment models constructed include univariate regression model, multivariate regression model, stepwise regression model, random forest and neural network.

6. The grassland species diversity assessment method based on a vegetation radiative transfer model according to claim 1, characterized in that, The method further includes: Step 6: Select natural grasslands of any scale as the research object, obtain the canopy spectral diversity index and canopy structural traits of the natural grassland quadrats, and use them as the real verification dataset for the simulation experiment. Construct a grassland species diversity assessment model based on the measured data to assess the grassland species diversity of natural grasslands.

7. The grassland species diversity assessment method based on a vegetation radiative transfer model according to claim 6, characterized in that, Step 6 includes: Step 61: Select natural grasslands of any scale as the research object, set up sample plots and quadrats for field investigation, and record grassland species diversity information and precise quadrat location centers. Step 62: Obtain the three-dimensional point cloud data of the grassland quadrats and the canopy spectral data of the grassland quadrats; Step 63: Perform preprocessing such as cropping, denoising, filtering and normalization on the three-dimensional point cloud data of the grassland quadrat, and extract the canopy structure characteristics of the natural grassland. Step 64: Preprocess the canopy spectral data of the grassland quadrats, remove the soil background using the NDVI thresholding method, and calculate the canopy spectral diversity index of the natural grassland. Step 65: Based on the canopy spectral diversity index and canopy structural characteristics of natural grassland, construct the grassland species diversity assessment model based on measured data to assess the grassland species diversity of natural grassland.

8. A grassland species diversity assessment device based on a vegetation radiative transfer model, characterized in that, The device includes: Grassland quadrat 3D model building unit, constructing grassland quadrat 3D models with different species compositions and different canopy structure complexities; The reflectance spectrum curve simulation unit simulates the reflectance spectrum curves of grassland plant leaves and the reflectance spectrum curves of the soil background. The grassland canopy hyperspectral image simulation unit simulates a grassland canopy hyperspectral image based on the reflectance spectral curves of the grassland plant leaves and the reflectance spectral curves of the soil background. The grassland quadrat three-dimensional model parameter calculation unit calculates the canopy spectral diversity index and canopy structural characteristics of the grassland quadrat three-dimensional model; The grassland species diversity assessment unit uses single or combined canopy spectral diversity indices and canopy structural traits to construct a grassland species diversity assessment model to assess grassland species diversity; The reflectance spectral curve simulation unit simulates the reflectance spectral curves of the grassland plant leaves and the soil background by: simulating the reflectance spectral curves of the grassland plant leaves using a leaf-scale radiative transfer model; and simulating the reflectance spectral curves of the soil background using a soil reflectance model. The key parameters used in simulating the reflectance spectrum curve of plant leaves using the leaf-scale radiative transfer model include: chlorophyll content, leaf area index, leaf thickness, leaf dry matter content, and leaf moisture content. The PROSPECT-D model was selected as the radiative transfer model at the leaf scale to simulate the reflectance spectrum of the plant leaves, and the GSV model was selected as the soil reflectance model to simulate the reflectance spectrum of the soil background.

Images