A method for monitoring the growth of grassland vegetation

Abstract

The present invention provides a method for monitoring the growth change of grassland vegetation, comprising the following steps: Step 1: Obtain binary discrete point cloud data through in-situ LiDAR observation; Step 2: Perform standardization processing on the point cloud data obtained in Step 1, including: first, perform angle conversion and correction on the data obtained in Step 1 using the known observation angle information of LiDAR to obtain a daily discrete point cloud data set with unified angle reference, then perform noise filtering to eliminate non-vegetation ground objects and random noise, and finally perform surface elevation normalization in combination with temporal information to obtain standardized point cloud data; and Step 3: Invert the vertical structure parameters of grassland vegetation, and use the inversion results to represent the change trend of the vegetation growth process. The present invention can obtain high-resolution grassland vegetation point cloud data that is spatio-temporally continuous, and reveal the growth change of vegetation from a three-dimensional perspective.

Description

Technical Field

[0001] The present invention relates to a method for monitoring the growth of grassland vegetation, and in particular to a method for monitoring the growth of grassland vegetation based on in-situ lidar observation, belonging to the field of lidar vegetation remote sensing. Background Art

[0002] As a main type of terrestrial vegetation classification, grassland occupies most of the carbon storage in the terrestrial ecosystem and is the main object of study in natural resource surveys. Grassland vegetation natural resources are sensitive to changes in external parameters such as climate and precipitation, showing obvious interannual and seasonal changes. Therefore, realizing the monitoring of the growth process of grassland in different growth periods can accurately reveal the complete growth process of grassland vegetation, which is of great significance for realizing accurate and efficient grassland natural resource surveys.

[0003] Physical vegetation ecological parameters such as coverage, leaf area index, and vegetation height, as the main indicators in vegetation ecological remote sensing research, can objectively reflect the productivity level of the current grassland ecosystem to a certain extent, and different parameter inversion methods and their abilities to reflect the vegetation growth level are also different. The grassland ecosystem shows an obvious response to changes in external conditions in the time dimension, with significant changes in vegetation parameters such as coverage and vegetation height. Therefore, accurately estimating the above parameters can reflect the dynamic change process of grassland vegetation growth.

[0004] Compared with the method of using relevant remote sensing indices by satellite remote sensing to evaluate the growth status of large-scale grassland areas, near-ground remote sensing observation can more accurately reflect the growth changes of grassland vegetation at the plot scale. The existing near-ground remote sensing grassland vegetation parameter estimation methods can be divided into two observation methods based on passive optics and active optics according to the differences in equipment observation modes. The parameter inversion based on passive optical remote sensing, as a traditional method for monitoring the growth of grassland vegetation, generally uses digital cameras, etc. to set fixed sampling points in the target plot for digital photography. This method of inverting grassland parameters using two-dimensional images mainly estimates parameters such as coverage and leaf area index by classifying the vegetation and non-vegetation areas in the image, so as to achieve the purpose of evaluating the current vegetation growth status. However, since digital photography often relies on manual observation at fixed sample points, its time resolution is difficult to fully characterize the specific growth process of grassland vegetation. At present, although some studies obtain digital images containing multi-band information based on long-term site observations and calculate relevant remote sensing indices to represent the specific growth changes of vegetation, such methods can only represent the two-dimensional structural feature information of vegetation, and it is difficult to obtain the vertical structural features of vegetation. Moreover, the measurement accuracy of traditional optical images depends on the universality of the current light environment conditions, exposure parameter settings, and image processing algorithms, etc., which also causes more uncertainties in evaluating the growth status of grassland vegetation based on optical digital photography.

[0005] As the existing mainstream active optical remote sensing observation method, Light Detection and Ranging (LiDAR) has the potential to overcome the above-mentioned defects of passive optical remote sensing because the short-wavelength laser pulses can penetrate the vegetation canopy through the gaps between vegetation and branches, and can accurately describe the vertical structure inside the vegetation canopy. Although LiDAR can better reflect the three-dimensional structure of vegetation and effectively estimate relevant structure parameters, limited by the high hardware and data acquisition costs, the current data acquisition method shows discrete characteristics in time, and the data acquired at different times also vary in basic parameters such as scanning mode, point cloud density, and incident angle, making it difficult to directly and effectively comprehensively apply. Therefore, how to obtain the three-dimensional structure information of grassland vegetation with high spatio-temporal resolution still needs to be solved; in addition, grassland vegetation is densely distributed in space and mostly consists of low-growing vegetation, which responds significantly to changes in external conditions such as precipitation and temperature. This makes it difficult to capture the short-term growth change process of grassland vegetation using traditional LiDAR observation modes such as station-mounted or airborne. Therefore, based on the acquired high spatio-temporal resolution LiDAR grassland vegetation data, it is necessary to develop an effective method for inverting vertical structure parameters to accurately capture the growth change process of grassland. Summary of the Invention

[0006] A brief overview of the present disclosure will be given below to provide a basic understanding of certain aspects of the present disclosure. It should be understood that this overview is not an exhaustive overview of the present disclosure. It is not intended to identify the key or important parts of the present disclosure, nor is it intended to limit the scope of the present disclosure. Its purpose is merely to present some concepts in a simplified form as a prelude to a more detailed description to be presented later.

[0007] The object of the present invention is to quantitatively characterize the growth process of grassland vegetation by fully exploiting the advantages of the site observation network in view of the limitations of the existing monitoring means for the growth change of grassland vegetation, especially the deficiencies of the existing methods for observing the growth change of grassland vegetation using LiDAR. First, this method scans the target sample plot based on the established LiDAR observation site and obtains the daily grassland discrete point cloud data through a wireless transmission network; second, this method uses relevant laser point cloud processing algorithms and combines the temporal characteristics of the discrete point cloud to establish a daily LiDAR point cloud standardization process; finally, the LiDAR parameter inversion method is used to invert various grassland vertical structure parameters such as daily plant height, vertical coverage, and canopy volume, so as to quantitatively reveal the growth change process of grassland vegetation.

[0008] To solve the above technical problems, the present invention provides the following technical solutions:

[0009] The present invention provides a method for monitoring the growth change of grassland vegetation, comprising the following steps: Step 1: Obtain binary discrete point cloud data through LiDAR in-situ observation, including using a LiDAR with a fixed scanning field of view angle to scan a target quadrat at a certain height above the ground and at a certain inclination angle to obtain spatio-temporally continuous daily structured binary discrete point cloud data; Step 2: Perform standardization processing on the point cloud data obtained in Step 1, including: first, perform angle conversion and correction on the data obtained in Step 1 using the known observation angle information of the LiDAR to obtain a daily discrete point cloud data set with unified angle reference, then perform noise filtering to eliminate non-vegetation ground objects and random noise, and finally perform surface elevation normalization in combination with temporal information to obtain standardized point cloud data; and Step 3: Invert the vertical structure parameters of grassland vegetation, and use the inversion results to represent the change trend of the vegetation growth process. Among them, the inversion of the vertical structure parameters of grassland vegetation includes the inversion of the overall height of vegetation plants, the inversion of the vertical coverage of vegetation, and the inversion of the vegetation volume; the inversion of the overall height of the plants is characterized by quantitatively describing the vegetation height parameter, the inversion of the vertical coverage of the vegetation is characterized by constructing a LiDAR penetration index, and the inversion of the vegetation volume is characterized by quantitatively calculating the volume change of the canopy vegetation point cloud in the three-dimensional space in the quadrat area.

[0010] Further, before performing the standardization processing on the point cloud data, convert the binary discrete point cloud data in Step 1 into a single-echo las data format based on the LiDAR hardware conversion rule.

[0011] Further, the steps for performing angle conversion and correction in Step 2 include: establishing a spatial rotation matrix based on the principle of spatial rectangular coordinate transformation, performing angle correction on the point cloud data to obtain a discrete point cloud data set with unified angle reference. For the point cloud data with known observation angles and fixed observation frequencies, according to the rotation angles and sequences around the X, Y, and Z axes in the three-dimensional space, establish coordinate transformation formulas for the rotation around each axis in turn, as shown in the following formula:

[0012]

[0013] Among them, and are the three-dimensional coordinates of the point cloud data before and after angle conversion respectively; represents the angle values of the point cloud data rotating around the Y, X, and Z axes respectively; R represents the established angle rotation matrix, the Z-axis direction is the LiDAR laser pulse emission scanning direction, and the rotation around the Z-axis represents the self-rotation angle κ of the LiDAR along the pulse emission direction; the rotation angle around the Y-axis represents the inclination angle of the LiDAR pulse emission azimuth The rotation angle around the X-axis represents the observation azimuth angle ω of the LiDAR pulse emission direction.

[0014] Further, in step two, the steps of noise filtering to eliminate non-vegetation ground objects and random noise include: determining the location of non-vegetation ground object areas in the single-day point cloud data through analysis, obtaining the fixed coordinate interval of the area to be filtered, and removing non-vegetation ground objects from the time-series point cloud data; and using the statistical outlier removal algorithm to remove random noise.

[0015] Further, the random noise includes isolated and drifting noise points, and the isolated and drifting noise points are removed by setting the neighborhood search area and the standard deviation multiple.

[0016] Further, in step two, the steps of surface elevation normalization by combining time-series information include: first performing ground filtering on the vegetation litter period data to accurately extract ground points and interpolating to generate a reference digital elevation model; and directly subtracting the pixel values of the corresponding reference digital elevation model from the point cloud data.

[0017] Further, in step three, the inversion of the vegetation height parameter includes: constructing a two-dimensional grid unit based on the standardized point cloud data, using the grid unit as the basic processing unit for parameter calculation to calculate the plant height, and calculating the average value H mean and the maximum value H max of all points in each grid unit to represent the height characteristic distribution of plants in the current grid, and finally synthesizing the height characteristic parameters of all grids to calculate the height characteristic distribution of plants in the current quadrat area.

[0018] Further, in step three, the inversion of the construction of the LiDAR penetration index includes: setting an empirical height threshold for the standardized point cloud data to distinguish vegetation from ground points, and realizing the inversion of the vertical vegetation coverage. The specific calculation formula is as follows:

[0019]

[0020] where ACI is the all-echo coverage index; ∑ALL is the number of all points in the current point cloud; Single canopy 、First canopy 、Intermediate canopy and Last canopy represent the number of single-echo points, first-echo points, intermediate-echo points, and last-echo points in the vegetation canopy area respectively. For the monitoring of grassland vegetation, all echo points are regarded as single-echo points, and a suitable height empirical threshold is set to distinguish vegetation from non-vegetation points, and thus the vertical vegetation coverage at the current time is calculated.

[0021] Further, in step three, the inversion of the vegetation volume includes: First, the point cloud in the sample area is divided into regular three-dimensional grids with voxels as the basic units. The height eigenvalue of the point cloud in each grid is counted as the voxel value of the grid, and the product of the height statistic and the bottom area of the grid is used as the volume of the grid. The overall volume estimation is obtained by summing up the volumes of all three-dimensional grids:

[0022]

[0023] where V all is the grassland vegetation volume of the current point cloud; V i is the vegetation volume of the i-th three-dimensional grid; h i is the height statistic of the vegetation in the i-th three-dimensional grid; n is the total number of sub-grids in the established three-dimensional grid set; S represents the bottom area of the established three-dimensional grid, and this parameter is determined by the grid step size set when the grid is established.

[0024] Further, the two-dimensional grid unit is 30×30 cm in size.

[0025] The present invention provides a method for monitoring the growth of grassland vegetation. Different from the traditional multi-temporal LiDAR vegetation growth change monitoring method, this method can obtain time-continuous high spatio-temporal resolution discrete point cloud data through LiDAR in-situ observation, and effectively combine the temporal feature information of the point cloud to accurately invert a variety of vertical structure parameters, quantitatively revealing the growth change process of grassland vegetation. This method has the following advantages: 1) It can obtain spatio-temporally continuous high-resolution grassland vegetation point cloud data; 2) It effectively combines the temporal information of the point cloud to invert the vertical structure parameters of the vegetation, and accurately reveals the growth change process of grassland vegetation from a three-dimensional perspective. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The specific content of the present disclosure will be described below with reference to the accompanying drawings, which will help to more easily understand the above and other objects, features and advantages of the present disclosure. The drawings are only for showing the principle of the present disclosure. The dimensions and relative positions of the units do not have to be drawn to scale in the drawings.

[0027] Figure 1 is a flow chart of the present invention;

[0028] Figure 2 is a schematic diagram of vegetation height calculation;

[0029] Figure 3 is the calculation result of daily vegetation height;

[0030] Figure 4 is the calculation result of vegetation vertical coverage and quadrat volume. DETAILED DESCRIPTION OF THE INVENTION

[0031] In the following, the exemplary disclosure of the present disclosure will be described in conjunction with the accompanying drawings. For clarity and conciseness, not all features for implementing the present disclosure are described in the specification. However, it should be understood that many decisions specific to the present disclosure may be made during the development of any such implementation of the present disclosure in order to achieve the specific goals of the developer, and these decisions may vary depending on the different aspects of the present disclosure.

[0032] Here, it should also be noted that in order to avoid obscuring the present disclosure due to unnecessary details, only the pipeline network structure closely related to the solution according to the present disclosure is shown in the drawings, while other details less related to the present disclosure are omitted.

[0033] It should be understood that the present disclosure is not limited to the described embodiments only due to the following description with reference to the accompanying drawings. Herein, where feasible, features between different embodiments may be replaced or borrowed, and one or more features may be omitted in one embodiment.

[0034] The technical solution of the present invention will be elaborated in detail below in conjunction with the accompanying drawings.

[0035] The present invention realizes a method for monitoring the growth changes of grassland vegetation based on in-situ lidar observations. First, the method continuously scans the target quadrat based on in-situ observations at the site, and automatically wirelessly transmits to obtain daily grassland vegetation discrete point cloud data. Secondly, using spatial coordinate transformation and point cloud filtering algorithms and combining with the temporal characteristics of the point cloud, a standardized processing process for temporal LiDAR data is established. Finally, by accurately retrieving various grassland vertical structure parameters in the daily point cloud, such as plant height, vertical coverage, and vegetation volume, the growth change process of grassland vegetation is quantitatively revealed in three dimensions. The following takes the in-situ LiDAR monitoring of grassland growth changes as a specific example for illustration, and its process is as Figure 1 shown, and the specific implementation steps are as follows:

[0036] Step 1: Obtain binary discrete point cloud data through in-situ lidar observations.

[0037] Specifically, a LiDAR with a fixed scanning field of view angle is installed on the top of the observation tower at a certain height above the ground. Preferably, the height above the ground is 5m, and the target quadrat is scanned at a certain inclination angle to obtain spatio-temporally continuous daily structured binary discrete point cloud data. Preferably, the inclination angle is 0° to 12.5°, and the scanning time resolution is 8 hours. In this way, spatio-temporally continuous high-resolution grassland vegetation point cloud data can be obtained, which is convenient for revealing the growth changes of vegetation from a three-dimensional perspective in the subsequent process.

[0038] Step 2: Standardize the point cloud data.

[0039] Specifically, refer toFigure 1 , the LiDAR point cloud data standardization process can be divided into three main parts: coordinate rotation transformation, noise filtering, and ground elevation normalization. Specifically, for the daily structured binary discrete point cloud data obtained in Step 1, it is converted into a single-echo las data format based on the hardware conversion rules given by the LiDAR device manufacturer. Considering the LiDAR scanning angle, point cloud noise, and point cloud time series feature information comprehensively, a standardized data processing process is established. Specifically, the original data is corrected for angle conversion using the known observation angle information when the LiDAR is installed at the observation site. Based on the principle of spatial rectangular coordinate transformation in photogrammetry theory, a spatial rotation matrix is established to correct the angle of the original point cloud data, and finally, a daily discrete point cloud data set with unified angle reference is obtained. For point cloud data with known observation angles and fixed observation frequencies, according to the rotation angles and sequences around the X, Y, and Z axes in three-dimensional space, coordinate transformation equations are established for the rotation around each axis in turn. The above process can be expressed by the following formula:

[0040]

[0041] Wherein, and are the three-dimensional coordinates of the point cloud data before and after angle conversion respectively; represents the angle values of the point cloud data rotating around the Y, X, and Z axes respectively; R represents the established angle rotation matrix. In actual LiDAR observations, the Z-axis direction is the LiDAR laser pulse emission scanning direction, and the rotation around the Z-axis represents the self-rotation angle κ of the LiDAR along the pulse emission direction; the rotation angle around the Y-axis represents the tilt angle of the LiDAR pulse emission azimuth; the rotation angle around the X-axis represents the observation azimuth angle ω of the LiDAR pulse emission direction. Generally, during actual observations, only the emission tilt angle and the observation azimuth angle ω of the LiDAR are adjusted, and the device itself does not rotate around the observation direction, so κ is generally 0, and only the numerical values of and ω are considered.

[0042] Based on the discrete point cloud obtained after coordinate rotation transformation, noise filtering is carried out to eliminate the influence of non-vegetation ground objects and random noise on data processing. Noise filtering is divided into two parts according to the classification process: for the daily discrete point cloud data obtained from continuous site observations, LiDAR scanning generally samples non-vegetation ground objects such as the observation tower itself and the quadrat fence at the same time, resulting in the appearance of a fixed noise area. By analyzing and determining the position of the non-vegetation ground object area in the single-day point cloud data, the fixed coordinate interval of the area to be filtered is obtained, and the non-vegetation ground objects in the time-series point cloud data are removed; for the noise points formed in the daily point cloud due to the influence of the surrounding environment and scanning characteristics, the statistical outlier removal algorithm is used to remove such random noise, and the isolated and drifting noise points and other noises are removed by setting the neighborhood search area and the standard deviation multiple.

[0043] Carry out surface elevation normalization by combining time-series information. For grassland vegetation without an obvious vertical stratification structure, first perform ground filtering on the vegetation litter period data to accurately extract ground points and interpolate to generate a Digital Elevation Model (DEM). Considering that it is difficult for LiDAR pulses to penetrate the grassland vegetation canopy to reach the ground during the peak vegetation growth season, resulting in the difficulty of the ground filtering algorithm to effectively classify ground points and non-ground points, therefore, by combining time-series information, the DEM accurately extracted during the litter period is used to process the daily point cloud data of the entire growth period. When performing ground filtering on the litter period point cloud, the progressive densification triangular mesh filtering algorithm is used to classify ground and non-ground points. Therefore, by effectively applying the reference DEM extracted from the litter period in the time-series data, surface elevation normalization of the data of the entire growth period is carried out. Considering that the terrain in the grassland area is generally relatively flat and the ground object size is small, generally set a small ground object search size, terrain slope, and a more refined iteration angle and distance to obtain ground point information. When performing elevation normalization on the time-series data using the reference DEM, in order to retain the point cloud information of plants with different heights in the vegetation canopy to the greatest extent, the reference DEM is generated by irregular triangular mesh interpolation to ensure the high resolution and continuity of the ground area. The elevation normalization is achieved by directly subtracting the corresponding DEM pixel value from the point cloud to avoid the loss of part of the plant height information during normalization and obtain standardized point cloud data.

[0044] Step 3: Invert the vertical structure parameters of the vegetation and use the inversion results to represent the change trend of the vegetation growth process.

[0045] See Figure 1, the overall growth process change trend of grassland vegetation is characterized by inverting the vertical structure parameters of vegetation in different growth periods. Among them, the overall height characteristics of plants in the region are characterized by quantitatively describing the changes in vegetation height parameters such as average plant height and maximum plant height to represent vegetation growth; the vertical coverage of the grassland is represented by constructing the LiDAR penetration index (LPI) to show the change trend of the vegetation canopy over growth time; the grassland volume is represented by quantitatively calculating the volume change of the canopy vegetation point cloud in the three-dimensional space in the quadrat area to represent the canopy growth change.

[0046] Such as Figure 2 shown, for the inversion of the height parameters of grassland plants in the quadrat area, a two-dimensional grid cell is constructed on the basis of the processed elevation-normalized discrete point cloud, and the grid cell is used as the basic processing unit for parameter calculation to calculate the plant height. By statistically calculating the average value H mean and the maximum value H max of all points in each grid cell to represent the height characteristic distribution of plants in the current grid, and finally comprehensively integrating the height characteristic parameters of all grids to calculate the height characteristic distribution of plants in the current quadrat area. Specifically, the arithmetic mean of the height characteristic parameters of all grids is calculated to obtain the height characteristic distribution of plants in the current quadrat area. When setting the basic unit size of the two-dimensional grid, the grid is generally set to a size of 30×30 cm to avoid repeated statistics of the same plant for smaller grids.

[0047] The inversion of the vertical coverage of grassland vegetation using LiDAR is realized based on the method of point cloud quantity statistics. The response of the laser pulse emitted by LiDAR to the canopy is not only a function of the canopy height but also a function of the coverage. Therefore, the vertical coverage can be calculated by constructing the LPI by calculating the ratio of the canopy echo to the overall echo using the echo information of the point cloud. In the construction of the LPI of grassland vegetation, an empirical height threshold is set for all echo point clouds with normalized elevation to distinguish vegetation from ground points, and the inversion of the vertical coverage of vegetation is realized. The specific calculation formula is as follows:

[0048]

[0049] Among them, ACI is the all echo coverage index; ∑ALL is the number of all points in the current point cloud; Single canopy 、First canopy 、Intermediate canopy and Last canopyrespectively represent the number of single echo points, first echo points, middle echo points, and last echo points in the vegetation canopy area. When applying Equation 2 to the monitoring of grassland vegetation growth trends, multiple pulse echoes were not distinguished. Therefore, all echo points were regarded as single echo points, and a suitable height empirical threshold was set to distinguish vegetation and non-vegetation points, and thus the vertical vegetation coverage at the current time was calculated.

[0050] The calculation of the grassland vegetation volume in the quadrat area is realized by using the volume surface difference method. First, the point cloud in the quadrat is divided into regular three-dimensional grids with voxels as the basic unit, and the height characteristic value of the point cloud in each grid is statistically used as the voxel value of the grid. Then, the product of the height statistic and the bottom area of the grid is used as the volume of the grid, and the overall volume estimation is obtained by accumulating all the three-dimensional grid volumes:

[0051]

[0052] Among them, V all is the grassland vegetation volume of the current point cloud; V i is the vegetation volume of the i-th three-dimensional grid; h i is the height statistic of the vegetation in the i-th three-dimensional grid; n is the total number of sub-grids in the established three-dimensional grid set; S represents the bottom area of the established three-dimensional grid, and this parameter is determined by the grid step size set during grid establishment. After inverting the daily vegetation vertical structure parameters using the above method, the growth changes of grassland vegetation during the entire growing season can be quantitatively revealed through the daily numerical changes of the three parameters, and the growth rate of the vegetation can be intuitively obtained.

[0053] Specific measurement example:

[0054] A LiDAR with a fixed observation field of view angle was installed at a height of 5 m above the ground in a certain grassland area, and the quadrat area was continuously observed at a certain inclination angle at different time periods. The observation time started from May 21, 2021 and ended on September 15, 2021. During the preliminary processing of the acquired data, 33 days of invalid data caused by the clock disorder and maintenance of the LiDAR itself were excluded, and finally 88 days of effective observations were carried out. In the acquired daily point cloud data, the observation angle changed twice due to maintenance needs, which were 12.5° inclination and vertical downward observation respectively. After the angle rotation transformation, they were all unified into the same coordinate system. The generation of the reference DEM was carried out using the progressive densification triangular network filtering algorithm. The maximum feature size of the ground object and the terrain slope were set to 1 m and 1° respectively, and the minimum iteration angle and iteration distance were 1° and 0.5 m respectively. Finally, the accurate ground points in the litter period were extracted, and a high-resolution reference DEM was generated based on the irregular triangular network interpolation method to realize the normalization of the daily point cloud elevation. The grassland vegetation height characteristics were obtained by statistically calculating the daily mean and maximum values of the normalized time series point cloud data. See Figure 3As shown, by using the maximum height value of the grassland vegetation in the quadrat area to characterize the best growth level of the current forage grass, and the average height value represents the average growth level of the current vegetation. Thus, based on the height statistical method in this paper, the growth change characteristics of the grassland vegetation in the vertical direction can be intuitively described. Figure 3 The results of the height characteristic changes show that the best growth level of the grassland vegetation area is always much greater than the average level in the quadrat, but the fluctuation range of the best growth level is relatively large. However, the overall average height change level tends to be stable and conforms to the general understanding of the grassland growing slowly first, then growing rapidly, and finally withering rapidly. The estimation of the vertical coverage is based on the normalized time-series point cloud dataset, and an empirical height threshold (5 cm) is set to calculate the proportion of vegetation points in the overall point cloud data, so as to comprehensively characterize the density of the grassland vegetation in the current quadrat area and the daily change of the grassland canopy coverage. Figure 4 The daily change broken line of the coverage in [reference] reveals that although there is a certain fluctuation range during the growth of the vegetation canopy, it always maintains continuous growth and then gradually withers. The vegetation volume estimation is carried out based on the volume surface difference method. By dividing the point cloud in the sample area into three-dimensional grids, the height characteristic value of the point cloud in each grid is counted as the voxel value of the grid, and the product of the height statistic and the bottom area of the grid is used as the volume of the grid. The overall volume estimation is obtained by summing up all the three-dimensional grid volumes. Here, the voxel grid size is set to 5 cm, and the height characteristic parameter of the grid scale is calculated using the mean value ([reference]) Figure 4 ), so as to characterize the three-dimensional spatial distribution change of the grassland vegetation from the three-dimensional structure perspective, comprehensively represent the canopy distribution level and vertical structure information of the vegetation. And from Figure 4 The vegetation volume change broken line graph shown in [reference] can also show that the overall change trend of the vegetation canopy growth is consistent with the coverage, indicating that the horizontal distribution and vertical growth level of the vegetation canopy in the area are unified. In summary, from Figure 3 and Figure 4 The changes of the vertical parameters of the grassland vegetation in the quadrat shown, the grassland vegetation has experienced the process of slow growth in the initial stage, rapid growth in the middle stage, and finally stable withering, indicating that this method can effectively monitor the specific growth changes of the grassland vegetation.

[0055] The above specific implementation manners are only for explaining the technical concept and structural characteristics of the present invention, aiming to enable those skilled in the art to implement it accordingly. However, the above content does not limit the protection scope of the present invention. Any equivalent changes or modifications made according to the technical characteristics of the present invention shall fall within the protection scope of the present invention.

[0056] The above description has been made in combination with specific implementation scenarios for the present disclosure. However, those skilled in the art should clearly understand that these descriptions are exemplary and do not limit the scope of protection of the present disclosure. Those skilled in the art can make various variations and modifications to the present disclosure according to the spirit and principles of the present disclosure, and these variations and modifications are also within the scope of the present disclosure.

Claims

1. A method for monitoring the growth change of grassland vegetation, comprising the following steps: Step 1: Obtain binary discrete point cloud data through in-situ LiDAR observation, including using a LiDAR with a fixed scanning field of view angle to scan a target sample plot at a certain height above the ground and at a certain inclination angle to obtain spatio-temporally continuous daily structured binary discrete point cloud data; Step 2: Standardize the point cloud data obtained in Step 1, including: first, perform angle conversion correction on the data obtained in Step 1 using the known observation angle information of the LiDAR to obtain a daily discrete point cloud data set with a unified angle reference, then perform noise filtering to eliminate non-vegetation ground objects and random noise, and finally perform surface elevation normalization in combination with temporal information to obtain standardized point cloud data; Step 3: Invert the vertical structure parameters of grassland vegetation, and use the inversion results to represent the change trend of the vegetation growth process. Among them, the inversion of the vertical structure parameters of grassland vegetation includes the inversion of the overall height of vegetation plants, the inversion of the vertical coverage of vegetation, and the inversion of the vegetation volume; the inversion of the overall height of the plants is characterized by quantitatively describing the vegetation height parameter, the inversion of the vertical coverage of the vegetation is characterized by constructing a LiDAR penetration index, and the inversion of the vegetation volume is characterized by quantitatively calculating the volume change of the canopy vegetation point cloud in the three-dimensional space in the sample plot area; Among them, the inversion of the vegetation height parameter in Step 3 includes: constructing a two-dimensional grid unit on the basis of the standardized point cloud data, using the grid unit as the basic processing unit for parameter calculation to calculate the plant height, and representing the height characteristic distribution of the plants in the current grid by statistically calculating the average value H_mean and the maximum value H_max of all points in each grid unit. Finally, comprehensively calculate the height characteristic parameters of all grids to calculate the height characteristic distribution of the plants in the current sample plot area.

2. The monitoring method according to claim 1, wherein, Before standardizing the point cloud data, convert the binary discrete point cloud data in Step 1 into a single-echo las data format based on the LiDAR hardware conversion rule.

3. The monitoring method according to claim 2, wherein, The steps for performing angle conversion correction in Step 2 include: Establish a spatial rotation matrix based on the principle of spatial rectangular coordinate transformation, perform angle correction on the point cloud data to obtain a discrete point cloud data set with a unified angle reference. For point cloud data with known observation angles and fixed observation frequencies, according to the rotation angles and sequences around the X, Y, and Z axes in the three-dimensional space, establish coordinate transformation formulas for the rotation around each axis in turn, as shown in the following formula: Among them, and are the three-dimensional coordinates of the point cloud data before and after angle conversion, respectively; ω and κ represent the angle values by which the point cloud data rotates around the Y, X, and Z axes respectively; R represents the established angle rotation matrix. The Z-axis direction is the LiDAR laser pulse emission scanning direction, and the rotation around the Z-axis represents the self-rotation angle κ of the LiDAR along the pulse emission direction; the rotation angle around the Y-axis represents the tilt angle of the LiDAR pulse emission azimuth The rotation angle around the X-axis represents the observation azimuth angle ω of the LiDAR pulse emission direction.

4. The monitoring method according to claim 2, wherein, The steps for performing noise filtering to eliminate non-vegetation ground objects and random noise in Step 2 include: analyzing and determining the position of the non-vegetation ground object area in the single-day point cloud data, obtaining the fixed coordinate interval of the area to be filtered, and removing the non-vegetation ground objects in the temporal point cloud data; and using the statistical outlier removal algorithm to remove random noise.

5. The monitoring method according to claim 4, wherein, The random noise includes isolated and drifting noise points, and the isolated and drifting noise points are removed by setting a neighborhood search area and a standard deviation multiple.

6. The monitoring method according to claim 2, wherein, The steps of performing surface elevation normalization by combining temporal information in step two include: first, performing ground filtering on the vegetation litter fall period data to accurately extract ground points and interpolating to generate a reference digital elevation model; and directly subtracting the pixel values of the corresponding reference digital elevation model from the point cloud data.

7. The monitoring method according to claim 1, wherein, The inversion of constructing the LiDAR penetration index in step three includes: setting an empirical height threshold for the standardized point cloud data to distinguish vegetation from ground points, and realizing the inversion of vegetation vertical coverage. The specific calculation formula is as follows: Among them, ACI is the all echo coverage index; ∑ALL is the number of all points in the current point cloud; Single canopy , First canopy , Intermediate canopy and Last canopy respectively represent the number of single echo points, first echo points, intermediate echo points and last echo points in the vegetation canopy area. For the monitoring of grassland vegetation, all echo points are regarded as single echo points, and a suitable height empirical threshold is set to distinguish vegetation points from non-vegetation points, so as to calculate the vertical vegetation coverage at the current time.

8. The monitoring method according to claim 1, wherein, The inversion of vegetation volume in step three includes: first, dividing the point cloud in the sample area into regular three-dimensional grids with voxels as the basic unit, statistically calculating the height characteristic value of the point cloud in each grid as the voxel value of the grid, and taking the product of the height statistic and the bottom area of the grid as the volume of the grid. The overall volume estimation is obtained by accumulating all three-dimensional grid volumes: Among them, V all is the grassland vegetation volume of the current point cloud; V i is the vegetation volume of the i-th three-dimensional grid; h i is the height statistic of the vegetation in the i-th three-dimensional grid; n is the total number of sub-grids in the established three-dimensional grid set; S represents the bottom area of the established three-dimensional grid, and this parameter is determined by the grid step size set during grid establishment.

9. The monitoring method according to claim 1, wherein, The two-dimensional grid unit is 30×30 cm in size.

Images