A method for precision selection of standard plot of Chinese fir plantation based on airborne LiDAR

Abstract

The application discloses a kind of cedar plantation standard plot precision selection method based on airborne LiDAR, belong to the standard plot design method field of forestry field investigation.It is first carried out unmanned aerial vehicle image data collection of target area, and constructs stand DEM, DSM and CHM after three-dimensional reconstruction, carries out single tree segmentation of whole stand based on ResUnet convolutional neural network, extracts single tree spatial position, tree height, stand canopy density and other parameters, according to the overall shape of target area, DEM and standard plot size requirements, etc., grid target area cedar forest forms FG, eliminates the edge effect and does not include target species in all standard quadrats, forms SG, statistics secondary standard plot group all standard plot single tree, stand parameters, according to the overall deviation of target area stand parameters, select the most representative plot as RA.The application uses airborne LiDAR data acquisition means and computer-aided decision-making, and it is a more efficient and accurate standard plot selection scheme.

Description

Technical Field

[0001] This invention belongs to the field of standard plot design methods for forestry field surveys, specifically involving a method for selecting standard plots for Chinese fir plantations based on UAV-borne LiDAR. Background Technology

[0002] Standard plot surveys are a crucial technical method used in national forest resource surveys, such as continuous forest resource inventories and forest resource planning and design surveys, to investigate the overall growth of forest stands. Typically, experienced forestry technicians select fixed-size plots that represent the overall characteristics of the entire stand or compartment after conducting a field survey of the entire forest stand. Various stand factors (such as tree height, diameter at breast height, crown width, and density) within the standard plots are measured to investigate the forest resources and growth status of the entire stand or compartment. Standard plot surveys are a method of forest resource surveying that extrapolates from specific points to broader areas; the importance of the accuracy of stand representation is self-evident. For plain forests with very similar site conditions and stand growth and development, standard plots surveyed using traditional methods have relatively high representativeness of the target stand or compartment. However, the representativeness of standard plots in complex mountainous forests is limited by two factors: first, the technical personnel selecting the plots must have extensive survey experience, and the subjectivity of the surveyors can significantly impact the representativeness of the standard plots; second, in actual operations, it is impossible to measure all tree growth parameters of the target forest stand as a reference, and the surveyed area only accounts for a portion of the target forest stand, and the reference accuracy of the standard plots selected through on-site surveys cannot be accurately quantified. Furthermore, large-scale forest resource surveys require substantial human, material, and financial resources. To save costs and accelerate the survey process, surveyors often prioritize the convenience of setting up standard plots rather than focusing on the preliminary forest stand survey.

[0003] The aforementioned limitations primarily stem from the lack of an efficient and cost-effective technique for surveying standard plots and target forest stands to accurately acquire their growth parameters and quantify the differences in stand growth between the two scales. With technological advancements, UAV remote sensing technology is widely applied in forestry. Airborne LiDAR can rapidly extract stand growth parameters at different scales and conduct high-precision positioning and matching surveys. To address the current problem of rapidly selecting highly representative standard plots in forestry surveys, this invention fully considers the timeliness and comprehensiveness of stand growth status surveys at different scales, quantifies the differences in stand parameters between standard plots and target forest stands, and utilizes airborne LiDAR data acquisition and computer-aided decision-making to develop a more efficient and accurate standard plot selection scheme. Summary of the Invention

[0004] The purpose of this invention is to improve the representativeness accuracy of standard plots selected in forestry surveys, reduce the cost of standard plot surveys, and provide a method for accurately selecting standard plots for Chinese fir plantations based on airborne LiDAR. This invention mainly focuses on three key aspects of standard plot surveys: high representativeness accuracy of the overall forest stand, timeliness of obtaining overall forest stand parameters, and high efficiency of standard plot selection. It rapidly obtains the overall growth and development parameters of the target forest stand before the selection of Chinese fir plantation standard plots, comprehensively quantifies the deviation of each standard plot in the secondary standard plot group from the overall forest stand parameters, and selects the standard plots with high representativeness accuracy, thus providing a new method for standard plot selection.

[0005] The specific technical solution adopted in this invention is as follows:

[0006] This invention provides a method for accurately selecting standard plots for Chinese fir plantations based on airborne LiDAR, as detailed below:

[0007] S1. Collect lidar point cloud data of the target area;

[0008] S2. Perform 3D reconstruction and noise reduction on the point cloud data of the target area collected in S1 to obtain the lidar point cloud data of the target area.

[0009] S3. The lidar point cloud data obtained in S2 is separated into ground points, and then digital elevation model, digital surface model and canopy height model are constructed respectively. Normalized point cloud data of the target Chinese fir forest is extracted according to the spatial vector range of the target Chinese fir forest.

[0010] S4. Using the canopy height model obtained in S3, employ a tree segmentation method based on a ResUnet convolutional neural network to obtain the spatial coordinates of individual trees throughout the entire forest stand. Visualize the spatial location of individual trees using ArcGIS software, and statistically calculate the total stand density D based on the target stand's vector spatial range. t And count the number of individual trees in the entire forest stand;

[0011] S5. Combining the canopy height model obtained in S3 and the individual tree spatial coordinates obtained in S4, the local maximum algorithm is used to obtain the individual tree height, and the average stand height H is statistically calculated. t ;

[0012] S6. Separate the tree layer, shrub layer, and ground layer point clouds from the normalized point cloud data obtained in S3. Calculate the canopy closure (CD) of the stand based on the proportion of the tree layer point cloud to the total point cloud within the target forest stand area. t ;

[0013] S7. Based on the individual tree height and number of individual trees in the entire stand obtained from S4 and S5, calculate the average height SH of the dominant trees in the entire Chinese fir plantation. t ;

[0014] S8. Based on the overall shape of the target Chinese fir forest stand, the size of the standard plot area, and the shape requirements of the standard plot, the target forest stand is gridded as the initial standard plot group FG. Standard plots with edge effects in FG are removed to obtain the secondary standard plot group SG. Following the processing methods of S4 to S7, the forest stand parameters of all standard plots within SG are obtained; the forest stand parameters of the standard plots include the standard plot tree density D. S The average height of standard forest stands, H S Standard forest stand canopy density (CD) S The average height of dominant trees in standard plots is SH S ;

[0015] S9. Based on the stand parameters of each standard plot obtained in S8, calculate the overall deviation coefficient P between it and the stand parameters of the whole stand. Combine this with the digital elevation model and digital surface model constructed in S3 to determine the operating conditions, and select the one with the smallest deviation coefficient and the best operating conditions as the final standard plot RA of the target stand. The stand parameters of the whole stand include the total stand density D. t Average stand height H t Forest stand canopy closure CD t And the average height of dominant trees is high SH t .

[0016] Preferably, in step S1, the acquisition of lidar point cloud data is achieved by using the DJI Matrice 300RTK flight platform equipped with the Zenmuse L1 lidar system.

[0017] As a preferred embodiment, the method for extracting the normalized point cloud data is as follows:

[0018] (1) Use the advanced version of DJI Terra software to perform three-dimensional reconstruction of LiDAR point clouds;

[0019] (2) Use LiDAR 360 to denoise and normalize the laser radar point cloud data to obtain normalized point cloud data.

[0020] Preferably, step S3 is as follows:

[0021] (1) An improved progressive encryption triangular mesh filtering algorithm is used for ground point separation;

[0022] (2) The digital elevation model is generated using the interpolation method of irregular triangular meshes, as shown in the following formula:

[0023] Z = b0 + b1x + b2y + b3xy

[0024] In the formula: x and y are plane coordinates, Z is elevation, and b0, b1, b2, and b3 are all unknown coefficients;

[0025] (3) The digital surface model is generated using the inverse distance weighting interpolation method, as shown in the following formula:

[0026]

[0027]

[0028] In the formula: Z j Z0 represents the elevation value of the interpolation point, Z0 represents the elevation value of the known laser foot point, and λ represents the elevation value of the interpolation point. j The weights of the laser feet represent the number of laser feet within a specified range, n represents the number of laser feet within a specified range, P represents the power exponent, and D represents the weights of the laser feet. j This represents the distance between the j-th interpolation point and the laser foot point;

[0029] (4) The difference between the spatial location models of the Digital Elevation Model (DEM) and the Digital Surface Model (DSM) is the canopy height model (CHM) of the Chinese fir forest, and the formula is as follows:

[0030] CHM = DSM-DEM.

[0031] Furthermore, the method for separating ground points is as follows:

[0032] The system initially selects ground points from the LiDAR point cloud, then constructs a triangular network based on the initially selected ground points. It traverses all point clouds to determine whether the angle and distance from each laser point to the triangular face are within the threshold range for separating ground points. If they are, they are marked as new ground points; otherwise, they are not marked as new ground points, thus filtering out all ground points in the LiDAR point cloud.

[0033] Preferably, step S4 is as follows:

[0034] (1) The canopy height model generated by S3 is segmented into individual trees using a single tree canopy segmentation model based on ResUnet convolutional neural network to obtain the spatial coordinates of individual trees in the whole forest stand;

[0035] (2) Using ArcGIS software, the spatial range A of the target forest stand was vectorized based on the lidar point cloud data obtained in S2. The vectorized coordinates of individual trees in the entire forest stand were overlaid, and the tree density D of the entire forest stand was statistically calculated. t .

[0036] Preferably, in step S5, the individual trees in the entire forest are arranged by height, and the top 20% are selected as the dominant trees of the target species, and the average height H of the dominant trees is calculated. t .

[0037] Preferably, step S8 is as follows:

[0038] (1) Using ArcGIS software, based on the size of the standard plot area and the overall shape of the target Chinese fir forest stand, design the shape requirements of the standard plot, and grid the target Chinese fir forest range in S3 to obtain the initial sample plot group FG that includes the entire range of the target forest stand. Remove the standard plots in FG that have edge effects or incomplete distribution of target tree species to obtain the secondary standard plot group SG.

[0039] (2) Referring to the methods in S4 to S7, extract each standard plot in SG, and statistically calculate the stand density D of each standard plot in the secondary standard plot group. S Average stand height H S Forest stand canopy closure CD S And the average height of dominant trees is high SH S .

[0040] Preferably, step S9 is as follows:

[0041] (1) Based on the stand parameters of each standard plot in SG obtained by S8, calculate the overall deviation coefficient P between it and the stand parameters of the whole stand. The calculation formula is as follows:

[0042]

[0043] Where: P is the overall deviation coefficient, rounded to two decimal places; i is the number of forest stand parameter indicators; a i b represents the stand parameters for the standard plots in SG; i These are the corresponding stand parameters for the entire forest stand;

[0044] (2) Based on the deviation coefficient P of each standard plot in SG, select the three standard plots with the smallest P value, overlay the digital elevation model and digital surface model constructed in S3, analyze the working conditions, and arrange the selection priority according to the P value from small to large in combination with whether they are feasible to obtain the final standard plot RA of the target forest stand.

[0045] Furthermore, the operating conditions include the ease of access to the standard site, the changes in terrain undulation, the slope, and the density of shrubs and grasses.

[0046] Compared with the prior art, the present invention has the following advantages:

[0047] (1) Comprehensive and scientific. Based on airborne LiDAR data, the parameters of the entire forest stand in the target area are presented in a comprehensive and scientific manner, which makes up for the shortcomings of traditional manual survey methods, which can only obtain empirical values ​​based on the growth of forest stands within the field of view;

[0048] (2) Quantitative representation. The plot with the smallest deviation coefficient from the growth parameters of the whole stand in the first and second selection standard plots is selected as the standard plot. The accuracy of the standard plot in representing the whole stand is quantified, which solves the problem of the representativeness of the standard plot selected by traditional manual visual experience;

[0049] (3) High efficiency and cost-effectiveness. Overlaying forest land DEM for preliminary assessment of operational terrain conditions, from the selection of standard plots to the final survey operation, effectively shortens the time for field surveys and saves manpower and material costs. Attached Figure Description

[0050] Figure 1 A schematic diagram of the process of this invention;

[0051] Figure 2 A schematic diagram of the ground point separation process for airborne LiDAR point cloud data;

[0052] Figure 3 3D reconstruction of airborne LiDAR point cloud over the target forest stand area;

[0053] Figure 4 Digital elevation model (DEM) of the target forest stand area (DEM) (a); Digital surface model (DSM) (b); Canopy height model (DSM) (c); Normalized point cloud data (TNPC) (d);

[0054] Figure 5 Individual tree segmentation results and spatial distribution map of the target forest stand;

[0055] Figure 6 Example diagram of the method for selecting standard plots for Chinese fir plantations based on airborne LiDAR point clouds. Detailed Implementation

[0056] The present invention will be further described and illustrated below with reference to the accompanying drawings and specific embodiments. The technical features of each embodiment of the present invention can be combined accordingly, provided that there is no mutual conflict.

[0057] like Figure 1 As shown, this invention provides a method for accurately selecting standard plots for Chinese fir plantations based on airborne LiDAR. The method is as follows:

[0058] S1. Collect point cloud data of the target area using LiDAR.

[0059] In practical applications, LiDAR point cloud data acquisition can be achieved using the DJI Matrice 300RTK flight platform equipped with the Zenmuse L1 LiDAR system.

[0060] S2. Perform 3D reconstruction and denoising on the point cloud data of the target area collected in S1 to obtain the lidar point cloud data of the target area.

[0061] In practical applications, the specific steps are as follows:

[0062] 1) Use DJI Terra (Advanced) software to perform 3D reconstruction of LiDAR point clouds;

[0063] 2) Use LiDAR 360 to denoise and normalize LiDAR point cloud data to obtain normalized point cloud data (TNPC).

[0064] S3. The lidar point cloud data obtained in S2 is separated into ground points and then constructed into digital elevation models (DEM), digital surface models (DSM), and canopy height models (CHM), respectively. Normalized point cloud data (TNPC) of the target Chinese fir forest is extracted based on the spatial vector range of the target Chinese fir forest.

[0065] In practical applications, the canopy height model (CHM) is established based on the digital elevation model (DEM) and the digital surface model (DSM), as detailed below:

[0066] 1) An improved progressive encryption triangular mesh filtering algorithm is used for ground point separation. The specific process and method are as follows: Figure 2 As shown, a brief explanation follows:

[0067] The system initially selects ground points from the LiDAR point cloud, then constructs a triangular network based on the initially selected ground points. It traverses all point clouds to determine whether the angle and distance from each laser point to the triangular face are within the threshold range for separating ground points. If they are, they are marked as new ground points; otherwise, they are not marked as new ground points, thus filtering out all ground points in the LiDAR point cloud.

[0068] (2) The digital elevation model (DEM) is generated using the interpolation method of irregular triangular meshes, as shown in the following formula:

[0069] Z = b0 + b1x + b2y + b3xy

[0070] In the formula: x and y are plane coordinates, Z is elevation, and b0, b1, b2, and b3 are all unknown coefficients;

[0071] (3) Generate the digital surface model (DSM) using the inverse distance weighting interpolation method, as shown in the following formula:

[0072]

[0073]

[0074] In the formula: Z j Z0 represents the elevation value of the interpolation point, Z0 represents the elevation value of the known laser foot point, and λ represents the elevation value of the interpolation point. j The weights of the laser feet represent the number of laser feet within a specified range, n represents the number of laser feet within a specified range, P represents the power exponent, and D represents the weights of the laser feet. jThis represents the distance between the j-th interpolation point and the laser foot point;

[0075] (4) The difference between the spatial location models of the Digital Elevation Model (DEM) and the Digital Surface Model (DSM) is the canopy height model (CHM) of the Chinese fir forest, and the formula is as follows:

[0076] CHM = DSM-DEM.

[0077] S4. Using the canopy height model obtained in S3, a single-tree segmentation method based on ResUnet convolutional neural network is employed to obtain the spatial coordinates of individual trees throughout the entire stand. ArcGIS software is used to visualize the spatial location of individual trees across the entire stand. Based on the target stand's vector spatial range, the total stand density D is statistically calculated. t And count the number of individual trees in the entire forest stand.

[0078] In practical use, the steps are as follows:

[0079] (1) The canopy height model generated by S3 is segmented into individual trees using a single tree canopy segmentation model based on ResUnet convolutional neural network to obtain the spatial coordinates of individual trees in the whole forest stand;

[0080] (2) Using ArcGIS software, the spatial range A of the target forest stand was vectorized based on the lidar point cloud data obtained in S2. The vectorized coordinates of individual trees in the entire forest stand were overlaid, and the tree density D of the entire forest stand was statistically calculated. t .

[0081] S5. Combining the canopy height model obtained in S3 and the individual tree spatial coordinates obtained in S4, the local maximum algorithm is used to obtain the individual tree height, and the average stand height H is statistically calculated. t .

[0082] S6. Separate the tree layer, shrub layer, and ground layer point clouds from the normalized point cloud data obtained in S3. Calculate the canopy closure (CD) of the stand based on the proportion of the tree layer point cloud to the total point cloud within the target stand area. t .

[0083] S7. Based on the individual tree height and number of individual trees in the entire stand obtained from S4 and S5, calculate the average height SH of the dominant trees in the entire Chinese fir plantation. t .

[0084] In practical use, the steps are as follows:

[0085] The individual trees in the entire stand were ranked by height, and the top 20% were selected as the dominant trees of the target species. The average height H of these dominant trees was calculated. t .

[0086] S8. Based on the overall shape of the target Chinese fir forest stand, the size of the standard plot area, and the shape requirements of the standard plot, the target forest stand is gridded as the initial standard plot group FG. Standard plots with edge effects in FG are removed to obtain the secondary standard plot group SG. Following the processing methods of S4 to S7, the forest stand parameters (i.e., the tree density D of the standard plot) of all standard plots within SG are obtained. S The average height of standard forest stands, H S Standard forest stand canopy density (CD) S The average height of dominant trees in standard plots is SH S ).

[0087] In practical use, the steps are as follows:

[0088] (1) Using ArcGIS software, based on the size of the standard plot area and the overall shape of the target Chinese fir forest stand, design the shape requirements of the standard plot, and grid the target Chinese fir forest range in S3 to obtain the initial sample plot group FG that includes the entire range of the target forest stand. Remove the standard plots in FG that have edge effects or incomplete distribution of target tree species to obtain the secondary standard plot group SG.

[0089] (2) Referring to the methods in S4 to S7, extract each standard plot in SG, and statistically calculate the stand density D of each standard plot in the secondary standard plot group. S Average stand height H S Forest stand canopy closure CD S And the average height of dominant trees is high SH S .

[0090] The above-mentioned references to S4 to S7 are as follows:

[0091] The canopy height model obtained in S3 was used to segment individual trees using a ResUnet convolutional neural network to obtain the spatial coordinates of individual trees throughout the entire stand. The spatial location of individual trees throughout the entire stand was visualized using ArcGIS software. The tree density D of the standard plot was calculated based on the vector spatial range of the standard plot. S Combining the canopy height model obtained in S3 and the individual tree spatial coordinates obtained in S4, the local maximum algorithm was used to obtain the individual tree height, and the average height H of the standard forest stand was statistically calculated. S From the normalized point cloud data obtained in S3, the tree layer, shrub and grass layer, and ground layer point clouds were separated. Based on the proportion of tree layer point clouds within the standard plot to the total point cloud area within the standard plot, the canopy closure (CD) of the standard plot was obtained. S Based on the individual tree height and the number of individual trees in the entire stand obtained above, the average height (SH) of the dominant trees in the standard plot of the Chinese fir plantation was statistically calculated. S .

[0092] S9. Based on the stand parameters of each standard plot obtained in S8, calculate their comparison with the stand parameters of the whole stand (including the total stand density D). t Average stand height H t Forest stand canopy closure CD t And the average height of dominant trees is high SH t The overall deviation coefficient P of the target forest stand is determined, and the operating conditions are judged by combining the digital elevation model and digital surface model constructed by S3. The stand with the smallest deviation coefficient and the best operating conditions is selected as the final standard site RA of the target forest stand.

[0093] In practical use, the steps are as follows:

[0094] (1) Based on the stand parameters of each standard plot in SG obtained in S8, calculate the overall deviation coefficient P between it and the stand parameters of the whole stand obtained in S4 to S7. The calculation formula is as follows:

[0095]

[0096] In the formula: P is the overall deviation coefficient (the smaller the coefficient, the stronger the representativeness of the standard plot), rounded to two decimal places; i is the number of stand parameter indicators (this invention selects 4 indicators); a i For the forest stand parameters of the standard plot in SG (in this invention, canopy closure D), S Average stand height H S Tree density CD S The average height of dominant trees is high. S );b i The corresponding stand parameters for the entire forest stand (in this invention, canopy closure D) are given. t Average stand height H t Tree density CD t The average height of dominant trees is high. t );

[0097] (2) Based on the deviation coefficient P of each standard plot in SG, select the three standard plots with the smallest P value, overlay the digital elevation model and digital surface model constructed in S3, analyze the working conditions such as the slope and the density of shrubs and grasses of the standard plots, and arrange them according to the priority of P value from small to large in combination with whether they are feasible to work, so as to obtain the final standard plot RA of the target forest stand.

[0098] The methods and effects of the present invention will be specifically illustrated below through examples.

[0099] Example

[0100] This embodiment provides a method for selecting standard plots for Chinese fir plantations based on UAV-borne LiDAR. The specific steps are as follows:

[0101] S1. Collection of LiDAR point cloud data for the target area: The study area is located in Maoshan Forest Area, Changhua Forest Farm, Longgang Town, Lin'an District, Hangzhou City, Zhejiang Province, which is a 32-year-old pure plantation of Chinese fir. The study area covers 2.166 hm². 2 On March 29, 2022, a DJI Matrice 300RTK equipped with the Zenmuse L1 lidar system was used to conduct field operations in the forest stand area under study to obtain lidar point cloud data. The flight parameters were: terrain-following flight mode, relative flight altitude of 80m, lidar forward overlap rate of 30%, visible light lateral overlap rate of 80%, and three-echo non-repeating scan mode.

[0102] S2. Using DJI Terra software (advanced version), perform 3D reconstruction of the target area point cloud data collected in S1 to obtain LiDAR point cloud data (e.g., Figure 3 ).

[0103] S3. Denoise the point cloud using LiDAR 360 software. Separate ground points from the denoised point cloud using a progressively denser triangular mesh filtering algorithm. Generate digital elevation models (DEMs) from the separated point clouds using irregular triangular mesh interpolation methods. Figure 4 a) and inverse distance weighting interpolation methods generate digital surface models (DSMs, such as...) Figure 4 b) Based on the generated DEM and DSM, the canopy height model (CHM) of the sample plot is obtained by performing interpolation calculation. Figure 4 c) Normalize the LiDAR point cloud based on the generated DEM using TNPC (e.g., Figure 4 d) The spatial range A of the vectorized target Chinese fir forest stand.

[0104] S4. Individual Tree Location Identification in Chinese Fir Forest: Using the canopy height model obtained in S3, a tree segmentation method based on ResUnet convolutional neural network is employed to obtain the spatial coordinates of individual trees throughout the entire forest stand. The spatial locations of individual trees throughout the forest are then visualized using ArcGIS software. Figure 5 The spatial locations of individual trees in the entire forest are marked as points, and the total number of trees (D) of the forest stand is calculated statistically based on the target stand vector spatial range. t And count the number of individual trees in the entire forest stand.

[0105] S5. Combining the CHM obtained in S3 and the individual tree spatial coordinates obtained in S4, the local maximum algorithm is used to obtain the individual tree height, and the average stand height H is calculated statistically. t The method for extracting tree height based on local maximum values ​​of CHM is as follows: superimpose CHM onto the spatial coordinate vector data of a single tree, and extract the tree height H based on the spatial location correspondence method.

[0106] S6. Based on the TNPC obtained in S3, separate the point clouds of the tree layer, shrub layer, and ground layer. The canopy closure CD of the forest stand is calculated by the proportion of the tree layer point cloud to the total point cloud within the target forest stand area. t .

[0107] S7. Based on the whole stand individual tree height dataset obtained in S5, sort the individual trees in the whole stand by height; according to the number of individual trees in the whole stand, select the top 20% as the dominant trees of the target tree species, and calculate the average height H of the dominant trees. t .

[0108] S8. Based on the overall shape of the target area, the size of the standard plots, and the shape requirements of the standard plots, the gridded target forest stands are used as the initial standard plot group FG. Standard plots with edge effects in FG are removed to obtain the secondary standard plot group SG. Following the processing methods of S4, S5, S6, and S7, the standard plot density D of all standard plots within SG is obtained. S The average height of standard forest stands, H S Standard forest stand canopy density (CD) S Standard plot of dominant trees average height SH S stand parameters (such as...) Figure 6 ):

[0109] 1) Using ArcGIS software, set up square standard plots of 25m×25m, and vectorize the target forest stands in S3 into the initial sample plot group FG, which contains 31 standard plots; remove standard plots in FG that have edge effects or incomplete distribution of target tree species to obtain the secondary standard plot group SG, which contains 10 standard plots.

[0110] 2) Following the methods in S4, S5, S6, and S7, extract the individual tree and stand parameters for each standard plot in SG, and statistically calculate the stand density D of the 10 standard plots in the secondary standard plot group. S Average stand height H S Forest stand canopy closure CD S The average height of dominant trees is high. S Forest stand parameters.

[0111] S9. Selection of Final Standard Plots: Based on the stand parameters of each standard plot obtained in S8, calculate the overall deviation coefficient P between these parameters and the stand parameters of the entire stand in S4, S5, S6, and S7, and combine this with the DEM and DSM constructed in S3 to determine operational conditions (such as...). Figure 4 a, b), select the one with the smallest deviation coefficient and the best operating conditions as the final standard site RA (e.g., Figure 6 ).

[0112] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the invention. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the invention. Therefore, all technical solutions obtained through equivalent substitution or transformation fall within the protection scope of the present invention.

Claims

1. A method for accurately selecting standard plots for Chinese fir plantations based on airborne LiDAR, characterized in that, Specifically as follows: S1. Collect lidar point cloud data of the target area; S2. Perform 3D reconstruction and noise reduction on the point cloud data of the target area collected in S1 to obtain the lidar point cloud data of the target area. S3. The lidar point cloud data obtained in S2 is separated into ground points, and then digital elevation model, digital surface model and canopy height model are constructed respectively. Normalized point cloud data of the target Chinese fir forest is extracted according to the spatial vector range of the target Chinese fir forest. S4. Using the canopy height model obtained in S3, employ a tree segmentation method based on a ResUnet convolutional neural network to obtain the spatial coordinates of individual trees throughout the entire forest stand. Visualize the spatial location of individual trees using ArcGIS software, and statistically calculate the total stand density D based on the target stand's vector spatial range. t And count the number of individual trees in the entire stand; S5. Combining the canopy height model obtained in S3 and the individual tree spatial coordinates obtained in S4, the local maximum algorithm is used to obtain the individual tree height, and the average stand height H is statistically calculated. t ; S6. Separate the tree layer, shrub layer, and ground layer point clouds from the normalized point cloud data obtained in S3. Calculate the canopy closure (CD) of the stand based on the proportion of the tree layer point cloud to the total point cloud within the target forest stand area. t ; S7. Based on the individual tree height and number of individual trees in the entire stand obtained from S4 and S5, calculate the average height SH of the dominant trees in the entire Chinese fir plantation. t ; S8. Based on the overall shape of the target Chinese fir forest stand, the size of the standard plot area, and the shape requirements of the standard plot, the target forest stand is gridded as the initial standard plot group FG. Standard plots with edge effects in FG are removed to obtain the secondary standard plot group SG. Following the processing methods of S4 to S7, the forest stand parameters of all standard plots within SG are obtained; the forest stand parameters of the standard plots include the standard plot tree density D. S The average height of standard forest stands, H S Standard forest stand canopy density (CD) S The average height of dominant trees in standard plots is SH S ; S9. Based on the stand parameters of each standard plot obtained in S8, calculate the overall deviation coefficient P between it and the stand parameters of the whole stand. Combine this with the digital elevation model and digital surface model constructed in S3 to determine the operating conditions, and select the one with the smallest deviation coefficient and the best operating conditions as the final standard plot RA of the target stand. The stand parameters of the whole stand include the total stand density D. t Average stand height H t Forest stand canopy closure CD t And the average height of dominant trees is high SH t .

2. The method for precise selection of standard plots for Chinese fir plantations based on airborne LiDAR according to claim 1, characterized in that, In step S1, the acquisition of lidar point cloud data is achieved by using the DJI Matrice 300 RTK flight platform equipped with the Zenmuse L1 lidar system.

3. The method for precise selection of standard plots for Chinese fir plantations based on airborne LiDAR according to claim 1, characterized in that, The specific method for extracting the normalized point cloud data is as follows: (1) Use the advanced version of DJI Terra software to perform three-dimensional reconstruction of LiDAR point clouds; (2) Use LiDAR 360 to denoise and normalize the laser radar point cloud data, and then obtain normalized point cloud data.

4. The method for precise selection of standard plots for Chinese fir plantations based on airborne LiDAR according to claim 1, characterized in that, Step S3 is as follows: (1) Ground point separation is performed using an improved progressive encryption triangular mesh filtering algorithm; (2) The digital elevation model is generated using the interpolation method of irregular triangular meshes, as shown in the following formula: ； In the formula: x and y are plane coordinates, and Z is the elevation. b1, b2, and b3 are all unknown coefficients; (3) The digital surface model is generated using the inverse distance weighting interpolation method, as shown in the following formula: ； ； In the formula: 𝑍 j The value represents the elevation of the interpolation point, 𝑍0 represents the elevation of the known laser foot point, and 𝜆 j The weights of the laser feet represent the number of laser feet within a specified range, n represents the number of laser feet within a specified range, P represents the power exponent, and 𝐷 j This represents the distance between the j-th interpolation point and the laser foot point; (4) The difference between the spatial location models of the Digital Elevation Model (DEM) and the Digital Surface Model (DSM) is the canopy height model (CHM) of the Chinese fir forest, and the formula is as follows: CHM = DSM-DEM.

5. The method for precise selection of standard plots for Chinese fir plantations based on airborne LiDAR according to claim 4, characterized in that, The specific method for separating ground points is as follows: The system initially selects ground points from the LiDAR point cloud, then constructs a triangular network based on the initially selected ground points. It traverses all point clouds to determine whether the angle and distance from each laser point to the triangular face are within the threshold range for separating ground points. If they are, they are marked as new ground points; otherwise, they are not marked as new ground points, thus filtering out all ground points in the LiDAR point cloud.

6. The method for precise selection of standard plots for Chinese fir plantations based on airborne LiDAR according to claim 1, characterized in that, Step S4 is as follows: (1) The canopy height model generated by S3 is segmented into individual trees using a single tree canopy segmentation model based on ResUnet convolutional neural network to obtain the spatial coordinates of individual trees in the whole stand; (2) Using ArcGIS software, the spatial range A of the target forest stand was vectorized based on the lidar point cloud data obtained in S2. The vectorized coordinates of individual trees in the entire forest stand were overlaid, and the tree density D of the entire forest stand was statistically calculated. t .

7. The method for precise selection of standard plots for Chinese fir plantations based on airborne LiDAR according to claim 1, characterized in that, In step S5, the individual trees in the entire forest are arranged by height, and the top 20% are selected as the dominant trees of the target species. The average height H of these dominant trees is then calculated. t .

8. The method for precise selection of standard plots for Chinese fir plantations based on airborne LiDAR according to claim 1, characterized in that, Step S8 is as follows: (1) Using ArcGIS software, design the shape requirements of the standard plots based on the size of the standard plot area and the overall shape of the target Chinese fir forest stand. Grid the target Chinese fir forest area in S3 and obtain the initial sample plot group FG that includes the entire range of the target forest stand. Remove the standard plots in FG that have edge effects or incomplete distribution of target tree species to obtain the secondary standard plot group SG. (2) Referring to the methods in S4~S7, extract each standard plot in SG, and statistically calculate the stand density D of each standard plot in the secondary standard plot group. S Average stand height H S Forest stand canopy closure CD S And the average height of dominant trees is high SH S .

9. The method for precise selection of standard plots for Chinese fir plantations based on airborne LiDAR according to claim 1, characterized in that, Step S9 is as follows: (1) Based on the stand parameters of each standard plot in SG obtained by S8, calculate the overall deviation coefficient P between it and the stand parameters of the whole stand. The calculation formula is as follows: ； Where: P is the overall deviation coefficient, rounded to two decimal places; i is the number of forest stand parameter indicators; a i b represents the stand parameters for the standard plots in SG; i These are the corresponding stand parameters for the entire forest stand; (2) Based on the deviation coefficient P of each standard plot in SG, select the three standard plots with the smallest P value, overlay the digital elevation model and digital surface model constructed in S3, analyze the working conditions, and arrange them according to the priority of P value from small to large to obtain the final standard plot RA of the target forest stand.

10. A method for precise selection of standard plots for Chinese fir plantations based on airborne LiDAR, as described in claim 9, is characterized in that... The operating conditions include the ease of access to the standard site, the terrain undulations, the slope, and the density of shrubs and grasses.

Images