Method for calculating forest canopy structure complexity based on LiDAR and fractal dimension

Abstract

The invention relates to the field of forest ecological monitoring and operation management, and provides a method for quantifying the complexity of a three-dimensional structure of a forest canopy based on LiDAR and fractal dimension. Aiming at the key bottleneck that traditional canopy structure quantitative indexes are difficult to represent cross-scale three-dimensional space heterogeneity, high-precision three-dimensional point cloud data are acquired by using foundation and tower footing LiDAR, and precise individual tree segmentation is realized in combination with a machine learning clustering algorithm and tree structure features. According to the method, the three-dimensional void rate of the forest stand canopy is calculated, and the fractal dimension of a single tree is calculated by adopting a box counting method. The weight of a single tree is determined by comprehensively considering a single tree structure and a spatial distribution pattern of the single tree structure, the fractal dimension of a stand scale is calculated, a canopy structure complexity index is constructed in combination with a stand three-dimensional void rate, and scale-free quantification of the forest canopy three-dimensional structure complexity is achieved. According to the method, typical temperate zone secondary forest monitoring data is used for simulation and verification, and technical support is provided for forest accurate operation (cutting and complementary planting), carbon sink evaluation, biodiversity protection and the like.

Description

Technical Field

[0001] This invention relates to a method for calculating the complexity of forest canopy structure based on LiDAR and spatial distribution function, belonging to the fields of forest management and operation and soil and water conservation technology. Background Technology

[0002] Forests are the structural pillars and functional centers of terrestrial ecosystems, playing a vital role in regulating local and regional climates, preventing wind erosion and sandstorms, and mitigating natural disasters. The complex and diverse structure of forest canopies profoundly influences ecological processes such as carbon sequestration capacity, hydrological cycles, biodiversity, nutrient cycling, and fire risk. Accurately quantifying the complex structural characteristics of forest canopies is crucial for a deeper understanding of forest ecosystem functions and services.

[0003] Forest canopy structure complexity is a crucial parameter characterizing forest structure, reflecting the heterogeneity, diversity, and sophistication of the physical configuration of branches and leaves in three-dimensional space. Traditional field surveys often rely on manual measurements within limited sampling areas to obtain structural parameters such as canopy height, leaf area index, canopy width, and diameter at breast height (DBH). Classical ground measurement methods (such as hemispherical photography and leaf area indexers) often have limited sampling and struggle to capture the complexity of forest canopy structure. Furthermore, under field conditions, the accurate characterization of stand canopy structure complexity and the assessment and simulation of ecological functions are constrained by multiple factors, including measurement techniques and human measurement errors.

[0004] LiDAR technology can accurately and on a large scale quantify the three-dimensional structure of forests, making it the most powerful tool for monitoring the dynamic changes in forest canopy structure. Ground-based and tower-based LiDAR has been used to monitor forest canopy structure, acquiring high-precision, high-density, and high-resolution three-dimensional point cloud data of trees through scanning. This allows for the non-destructive and accurate extraction of information such as individual tree height, location, diameter at breast height (DBH), and crown width. However, the extraction of parameters such as forest branch and leaf distribution and branch structure from point cloud data is subject to significant uncertainties due to the influence of wind, shading, and the dynamic changes in tree branches and leaves during the leafy stage. Fractal dimension reflects the effectiveness of complex shapes in filling space, accurately describing the complexity of tree branch structures and their ability to fill and utilize space. However, research on the high integration of LiDAR and fractal dimension for calculating the complexity of forest canopy structure is still rare. To accurately quantify the structural complexity of forest canopy during the leafy / leafless stages, a method based on LiDAR and fractal dimension for quantifying forest canopy structural complexity is proposed. Summary of the Invention

[0005] This invention addresses the challenge of accurately quantifying forest stand structural complexity due to the dynamic changes in tree branches and leaves during the leafing stage. It proposes a method combining LiDAR and fractal dimension to quantify forest canopy structural complexity. The canopy structural complexity classification results can directly guide forest management (such as determining selective felling intensity) and identifying target trees for selective felling.

[0006] This invention provides a method for calculating the structural complexity of forest canopy during the leafy or leafless stages based on LiDAR and fractal dimension. Through steps such as radar point cloud data fusion, spatial point cloud normalization, stand porosity calculation, individual tree segmentation and structural feature extraction, tree trunk / branch / leaf reconstruction, multi-scale calculation of tree trunk / branch / leaf volume and surface area, and individual tree fractal dimension simulation, the method accurately measures the canopy structure and branch and leaf characteristics of forest stands, and then calculates the structural complexity of forest canopy.

[0007] The method includes the following:

[0008] (1) Point cloud data acquisition: LiDAR scanner Riegl was used. TM The VZ-400i and Faro Focus S 350 scanned the trees in the forest stand to obtain point cloud data of the trees, with scanning resolutions of 5mm and 1mm, respectively.

[0009] (2) Point cloud processing: After the point cloud data is spliced, cropped, denoised and filtered, ground points and non-ground points are separated; the point cloud elevation values ​​are converted into relative height (HeightAbove Ground) and normalized to eliminate the influence of terrain; the normalized non-ground point cloud data is subjected to spatial horizontal and vertical grid voxelization processing.

[0010] (3) Calculation of stand gap fraction: The gap fraction is defined as the probability (proportion) that a laser beam passes through a voxel on the forest floor without interception. The canopy gap fraction (P) is calculated using Formula 1.

[0011]

[0012] After voxelizing the point cloud, calculate the gap ratio p(θ, s) of the s-th layer at angle θ; where n single n is the number of returns per cycle. multiple For multiple return values, n total denoted as the total number of laser beams emitted by the s-th genus; μ is a constant in [0, 1], representing the proportion of gap information contained in the multiple returning laser beams.

[0013] (3) Individual tree extraction: Using normalized non-ground point clouds, relying on the differences in point cloud density, and combining geometric features (such as curvature, normal vector) and K-means algorithm to perform clustering to segment individual trees.

[0014] (4) Leaf Reconstruction: Adding leaves by combining the radiation propagation model and Beer's law can more realistically reflect the radiation characteristics, reflection characteristics, and intercalation rate of the canopy. Based on the point cloud density, adding leaves by combining the branch structure, the resulting canopy directional intercalation rate and leaf area index are close to the true values. Based on the measured data of leaf shape and structure of different tree types, the leaves are parametrically simplified and modeled. Combined with the voxel division of the leaf point cloud, the leaves are reconstructed according to the leaf point cloud density and branch nodes.

[0015] (5) Extraction of structural features of individual trees: LiDAR360 software was used to extract the diameter at breast height (DBH), tree height and crown width of individual trees, and the cross-sectional area at DBH was determined.

[0016] (6) Calculation of fractal dimension of a single tree: The forest canopy can be regarded as a fractal porous medium with statistical self-similarity. Based on the Hausdorff dimension, a 3D cube with a side length of d (> leaf length) is used to fill or cover the forest canopy. The number of cubes required, N(d), is counted to calculate the fractal dimension, denoted as D.

[0017] For a D-dimensional object, the relationship between N and d is as follows:

[0018]

[0019] Taking the logarithm of both sides yields the formula for calculating the fractal dimension:

[0020]

[0021] The numerical range of D is (1, 3).

[0022] (7) Calculate the weight (W) of the i-th tree in the stand. i ):

[0023]

[0024] Where C i Let d be the crown width of the i-th tree. ij S is the horizontal distance between the i-th tree and the j-th tree within the forest stand. j Let I be the cross-sectional area of ​​the chest height at the j-th position, and let I be the indicator function.

[0025] (8) Divide the stand point cloud into different height layers, and calculate the average stand gap ratio (P) based on the gap ratio of different height layers (Z-axis). Z ):

[0026]

[0027] p kThis is the gap ratio at the height of the i-th layer of the stand. Similarly, the gap ratios (P) along the X and Y axes can be calculated. X and P Y By combining the gap ratios along the X, Y, and Z axes, the fill ratios of the trunk, branches, and leaves in three-dimensional space are calculated:

[0028]

[0029] (9) Calculation of Stand Canopy Structure Complexity (SSC):

[0030]

[0031] in M represents the total number of individual trees in the stand, and the range of SSC is...

[0032] The calculation of diameter at breast height (DBH), crown width, fractal dimension, and gap ratio of individual trees enables the construction of stand canopy complexity and the accurate quantification of forest canopy structure complexity.

[0033] Select forest stands with a canopy coverage of ≥10%, an area of ​​≥0.5 hectares, a tree height of ≥500cm, and at least one species of normally growing tree.

[0034] Using LiDAR to scan forest stands from different angles while keeping the target position constant, the scanning angle can completely cover the target forest stand.

[0035] Determine the horizontal coordinates of a single tree, and then take the median of the corresponding X and Y axis data of the point cloud after dividing the point cloud by the 10th to 60th percentiles of the Z-axis.

[0036] The standard for three-dimensional voxel spatial discretization of the crown width of a single tree is used, with voxel side lengths ranging from 5 to 20 cm.

[0037] The fractal dimension is calculated based on the box counting method. The side length of the boxes divided by the cube scale is 5-100cm, and the number of scales is 4-6.

[0038] The weighting coefficient of a single tree is calculated based on its cross-sectional area at breast height and crown width, taking into account its geographical coordinates.

[0039] The canopy structure complexity index consists of two parts: the base and the index. It integrates two scales: individual tree and stand. The base is the weighted sum of the fractal dimensions of individual trees, and the index is the gap rate or canopy closure of the stand canopy.

[0040] The advantages of this invention are: The method calculates the three-dimensional gap ratio of the forest stand canopy and uses box counting to calculate the fractal dimension of individual trees. By comprehensively considering the structure and spatial distribution pattern of individual trees to determine their weights, the fractal dimension at the stand scale is calculated. Combined with the three-dimensional gap ratio of the stand, a canopy structure complexity index is constructed, achieving scale-free quantification of the complexity of the forest canopy's three-dimensional structure. This method utilizes monitoring data from typical temperate secondary forests for simulation and verification, providing technical support for precision forest management (harvesting and replanting), carbon sequestration assessment, and biodiversity conservation. Attached Figure Description

[0041] Figure 1 This is a schematic diagram illustrating the construction of canopy structure complexity in an embodiment of the present invention.

[0042] Figure 2 The tree coordinates and canopy spatial pattern of the forest stand are shown in the embodiments of the present invention.

[0043] Figure 3 This is an example of canopy voxelization in an embodiment of the present invention.

[0044] Figure 4 This invention simulates the variation in canopy structure complexity under different harvesting radii in an embodiment of the invention. Detailed Implementation

[0045] The following is in conjunction with the instruction manual appendix. Figure 1-4 The present invention will be described in further detail below.

[0046] A method for calculating forest canopy structure complexity based on LiDAR and fractal dimension is proposed. This method accurately measures the canopy structure and branch and leaf characteristics of forest stands through steps such as radar point cloud data fusion, spatial point cloud normalization, stand porosity calculation, individual tree segmentation and structural feature extraction, tree trunk / branch / leaf reconstruction, multi-scale calculation of tree trunk / branch / leaf volume and surface area, and individual tree fractal dimension simulation, thereby calculating the complexity of forest canopy structure.

[0047] Includes the following processes:

[0048] (1) Point cloud data acquisition: LiDAR scanner Riegl was used. TM The VZ-400i and Faro Focus S 350 scanned the trees in the forest stand to obtain point cloud data of the trees, with scanning resolutions of 5mm and 1mm respectively.

[0049] (2) Point cloud processing: After the point cloud data is spliced, cropped, denoised and filtered, ground points and non-ground points are separated; the point cloud elevation values ​​are converted into relative height (HeightAbove Ground) and normalized to eliminate the influence of terrain; the normalized non-ground point cloud data is subjected to spatial horizontal and vertical grid voxelization processing.

[0050] (3) Calculation of stand gap fraction: The gap fraction is defined as the probability (proportion) that a laser beam passes through a voxel on the forest floor without interception; the canopy gap fraction (P) is calculated using Formula 1.

[0051]

[0052] After voxelizing the point cloud, calculate the gap ratio p(θ, s) of the s-th layer at angle θ; where n single n is the number of returns per cycle. multiple For multiple return values, n total is the total number of laser beams emitted by the s-th voxel; μ is a constant in [0, 1], representing the proportion of gap information contained in the multiple returning laser beams;

[0053] (3) Single tree extraction: Using normalized non-ground point clouds, relying on the differences in point cloud density, and combining geometric features (such as curvature, normal vector) and K-means algorithm to perform clustering to segment single trees;

[0054] (4) Leaf reconstruction: Adding leaves by combining the radiation propagation model and Beer's law can more realistically reflect the radiation characteristics, reflection characteristics and gap ratio of the canopy; adding leaves based on the point cloud density and the branch structure can result in the canopy's directional gap ratio and leaf area index being closer to the true values; the leaves are parametrically simplified and modeled based on the measured data of leaf shape and structure of different tree types, and the leaves are reconstructed according to the leaf point cloud density and branch nodes by combining the voxel division of the leaf point cloud;

[0055] (5) Extraction of structural features of individual trees: LiDAR360 software was used to extract the diameter at breast height (DBH), tree height and crown width of individual trees, and the cross-sectional area at DBH was determined;

[0056] (6) Calculation of fractal dimension of single tree: The forest canopy can be regarded as a fractal porous medium with statistical self-similarity; Based on the Hausdorff dimension, a 3D cube with a side length of d (> leaf length) is used to fill or cover the forest canopy, and the number of cubes required N(d) is counted to calculate the fractal dimension, denoted as D;

[0057] For a D-dimensional object, the relationship between N and d is as follows:

[0058]

[0059] Taking the logarithm of both sides yields the formula for calculating the fractal dimension:

[0060]

[0061] The numerical range of D is (1,3);

[0062] (7) Calculate the weight (W) of the i-th tree in the stand. i ):

[0063]

[0064] Where C i Let d be the crown width of the i-th tree. ij S is the horizontal distance between the i-th tree and the j-th tree within the forest stand. j Let I be the cross-sectional area of ​​the breast height of the j-th digit, and let I be the indicator function.

[0065] (8) Divide the stand point cloud into different height layers, and calculate the average stand gap ratio (P) based on the gap ratio of different height layers (Z-axis). Z ):

[0066]

[0067] p k This is the gap ratio at the height of the i-th layer of the stand; similarly, the gap ratios (P) along the X and Y axes can be calculated. X and P Y By combining the gap ratios along the X, Y, and Z axes, the fill ratios of the trunk, branches, and leaves in three-dimensional space are calculated.

[0068]

[0069] (9) Calculation of Stand Canopy Structure Complexity (SSC):

[0070]

[0071] in M represents the total number of individual trees in the stand, and the range of SSC is...

[0072] The calculation of diameter at breast height (DBH), crown width, fractal dimension, and gap ratio of individual trees enables the construction of stand canopy complexity and the accurate quantification of forest canopy structure complexity.

[0073] Select forest stands with a canopy coverage of ≥10%, an area of ​​≥0.5 hectares, a tree height of ≥500cm, and at least one species of normally growing tree.

[0074] Using LiDAR to scan forest stands from different angles while keeping the target position constant, the scanning angle can completely cover the target forest stand.

[0075] Determine the horizontal coordinates of a single tree, and then take the median of the corresponding X and Y axis data of the point cloud after dividing the point cloud by the 10th to 60th percentiles of the Z-axis.

[0076] The standard for three-dimensional voxel spatial discretization of the crown width of a single tree is used, with voxel side lengths ranging from 5 to 20 cm.

[0077] The fractal dimension is calculated based on the box counting method. The side length of the boxes divided by the cube scale is 5-100cm, and the number of scales is 4-6.

[0078] The weighting coefficient of a single tree is calculated based on its cross-sectional area at breast height and crown width, taking into account its geographical coordinates.

[0079] The canopy structure complexity index consists of two parts: the base and the index. It integrates two scales: individual tree and stand. The base is the weighted sum of the fractal dimensions of individual trees, and the index is the gap rate or canopy closure of the stand canopy.

[0080] Example: A forest survey plot (30m×30m) is used as an example.

[0081] The canopy structure complexity of the sample plot is calculated according to the procedure of this invention and used as a reference standard. A tree is then removed sequentially, and the canopy structure complexity of the sample plot is recalculated. By comparing the changes in canopy structure complexity before and after removal, the contribution and ranking of each tree in the sample plot to the canopy structure complexity are obtained.

[0082] Once the tree density in the sample plot exceeds the optimal density, the number of thinning operations can be determined. The harvesting targets are then determined by ranking each tree according to its contribution to the structural complexity of the sample plot.

[0083] When the tree density in the sample plot is less than the optimal density, the number of replanted trees can be determined. The canopy structure complexity of the sample plot is recalculated based on the coordinates of the replanted saplings and the target structural parameters. By comparing the changes in canopy structure complexity before and after replanting at different coordinates, the coordinates of the replanted saplings are determined.

[0084] After simulating logging within a defined logging area, the canopy structure complexity of the sample plots is recalculated. By comparing this value with that of unlogged sample plots, the target logging area is determined.

[0085] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. The method provided by the present invention also provides valuable reference for quantification methods of canopy structure complexity in other forest types. Any modifications, equivalent substitutions, and improvements 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 calculating the structural complexity of forest canopy based on LiDAR and fractal dimension, characterized in that: This method accurately measures the canopy structure and branch and leaf characteristics of forest stands through steps such as radar point cloud data fusion, spatial point cloud normalization, stand porosity calculation, individual tree segmentation and structural feature extraction, tree trunk / branch / leaf reconstruction, multi-scale tree trunk / branch / leaf volume and surface area calculation, and individual tree fractal dimension simulation, and then calculates the complexity of the forest canopy structure.

2. The method for calculating the complexity of forest canopy structure based on LiDAR and fractal dimension as described in claim 1, characterized in that, The calculation process includes the following: (1) Point cloud data acquisition: The trees in the forest stand are scanned using a scanner to acquire point cloud data of the trees; (2) Point cloud processing: After the point cloud data is spliced, cropped, denoised and filtered, ground points and non-ground points are separated; the point cloud elevation values ​​are converted into relative heights and normalized to eliminate the influence of terrain; the normalized non-ground point cloud data is subjected to spatial horizontal and vertical grid voxelization processing. (3) Calculation of stand gap ratio: The gap ratio is the probability that a laser beam on the forest floor will pass through a voxel without interception; the canopy gap ratio (P) is calculated using Formula 1. After voxelizing the point cloud, calculate the gap ratio p(θ,s) of the s-th layer at angle θ; where n single n is the number of returns per cycle. multiple For multiple return values, n total is the total number of laser beams emitted by the s-th voxel; μ is a constant in [0, 1], representing the proportion of gap information contained in the multiple returning laser beams; (3) Individual tree extraction: Using normalized non-ground point clouds, relying on the differences in point cloud density, and combining geometric features and K-means algorithm to perform clustering to segment individual trees; (4) Reconstruction of branches and leaves: Adding leaves by combining the radiation propagation model and Beer's law can more realistically reflect the radiation characteristics, reflection characteristics and porosity of the canopy; Based on point cloud density, leaves were added in combination with branch structure, and the resulting canopy directional gap ratio and leaf area index were close to the true values. Based on the measured data of leaf shape and structure of different tree types, the leaves were parametrically simplified and modeled. Combined with the voxel division of leaf point cloud, the leaves were reconstructed according to leaf point cloud density and branch nodes. (5) Extraction of structural features of individual trees: Extract the diameter at breast height (DBH), tree height and crown width of individual trees, and determine the cross-sectional area at DBH; (6) Calculation of fractal dimension of single tree: The forest canopy can be regarded as a fractal porous medium with statistical self-similarity; Based on the Hausdorff dimension, a 3D cube with a side length of d (> leaf length) is used to fill or cover the forest canopy, and the number of cubes required N(d) is counted to calculate the fractal dimension, denoted as D; For a D-dimensional object, the relationship between N and d is as follows: Taking the logarithm of both sides yields the formula for calculating the fractal dimension: The numerical range of D is (1,3); (7) Calculate the weight (W) of the i-th tree in the stand. i ): Where C i Let d be the crown width of the i-th tree. ij S is the horizontal distance between the i-th tree and the j-th tree within the forest stand. j Let I be the cross-sectional area of ​​the breast height of the j-th digit, and let I be the indicator function. (8) Divide the stand point cloud into different height layers, and calculate the average stand gap ratio (P) based on the gap ratio of different height layers (Z-axis). Z ): p k This is the gap ratio at the height of the i-th layer of the stand; similarly, the gap ratios (P) along the X and Y axes can be calculated. X and P Y By combining the gap ratios along the X, Y, and Z axes, the fill ratios of the trunk, branches, and leaves in three-dimensional space are calculated. (9) Calculation of Stand Canopy Structure Complexity (SSC): in M represents the total number of individual trees in the stand, and the range of SSC is... The calculation of diameter at breast height (DBH), crown width, fractal dimension, and gap ratio of individual trees enables the construction of stand canopy complexity and the accurate quantification of forest canopy structure complexity.

3. The method for calculating the complexity of forest canopy structure based on LiDAR and fractal dimension as described in claim 2, characterized in that, Select forest stands with a canopy coverage of ≥10%, an area of ​​≥0.5 hectares, a tree height of ≥500 cm, and at least one species of normally growing tree.

4. The method for calculating the complexity of forest canopy structure based on LiDAR and fractal dimension as described in claim 2, characterized in that, Using LiDAR to scan forest stands from different angles while keeping the target position constant, the scanning angle can completely cover the target forest stand.

5. The method for calculating the complexity of forest canopy structure based on LiDAR and fractal dimension as described in claim 2, characterized in that, Determine the horizontal coordinates of a single tree, and then take the median of the corresponding X and Y axis data of the point cloud after dividing the point cloud by the 10th to 60th percentiles of the Z-axis.

6. The method for calculating the complexity of forest canopy structure based on LiDAR and fractal dimension as described in claim 2, characterized in that, The standard for three-dimensional voxel spatial discretization of the crown width of a single tree is used, with voxel side lengths ranging from 5 to 20 cm.

7. The method for calculating the complexity of forest canopy structure based on LiDAR and fractal dimension as described in claim 2, characterized in that, The fractal dimension is calculated based on the box counting method. The side length of the boxes divided by the cube scale is 5~100 cm, and the number of scales is 4~6.

8. The method for calculating the complexity of forest canopy structure based on LiDAR and fractal dimension as described in claim 2, characterized in that, The weighting coefficient of a single tree is calculated based on its cross-sectional area at breast height and crown width, taking into account its geographical coordinates.

9. The method for calculating the complexity of forest canopy structure based on LiDAR and fractal dimension as described in claim 2, characterized in that, The canopy structure complexity index consists of two parts: the base and the index. It integrates two scales: individual tree and stand. The base is the weighted sum of the fractal dimensions of individual trees, and the index is the gap rate or canopy closure of the stand canopy.

Images