Tree carbon reserve monitoring method and device based on laser radar and storage medium

Abstract

The invention provides a tree carbon reserve monitoring method and device based on a laser radar and a storage medium. The method comprises the following steps: scanning each sample plot in N sample plots sampled in a monitoring area corresponding to forest vegetation through a laser radar, and acquiring first point cloud data corresponding to a tree; preprocessing the first point cloud data to obtain second point cloud data of the trees in each sample plot; based on the second point cloud data of each sample plot, performing individual tree segmentation and parameter extraction to obtain a first individual tree number and a first feature parameter corresponding to the individual tree; correcting the first characteristic parameter based on a pre-established correction model to obtain a corrected first characteristic parameter; and based on the first individual tree number of each sample plot and the corrected first characteristic parameter, determining the tree carbon reserve of the monitoring area through a preset carbon measurement model. Therefore, the monitoring efficiency and the monitoring accuracy of the tree carbon reserves can be improved.

Description

Technical Field

[0001] This invention relates to the field of forest monitoring technology, and more specifically, to a method, device, and storage medium for monitoring tree carbon storage based on lidar. Background Technology

[0002] To accurately understand the current status and changing trends of forest vegetation types, distribution, biodiversity, carbon storage, and ecological functions, rapid monitoring of forest vegetation resources is essential. Forest vegetation carbon storage is typically assessed by monitoring tree carbon storage. For tree carbon storage monitoring, manual sampling surveys or remote sensing estimation are generally used. Currently, manual surveys for tree carbon storage monitoring are time-consuming, costly, and have long intervals, generally only conducted every three to five years, failing to reflect forest vegetation growth in a timely manner. Remote sensing estimation, on the other hand, often utilizes satellite remote sensing imagery and, through correction and other processing, assesses forest resources; however, image accuracy is limited, and it cannot obtain data on forest tree growth, relying only on forest cover for estimation, and is subject to significant errors depending on the season. Summary of the Invention

[0003] In view of this, the purpose of this application is to provide a method, device and storage medium for monitoring tree carbon storage based on lidar, which can improve the problems of inaccurate and inefficient monitoring of tree carbon storage.

[0004] To achieve the above technical objectives, the technical solution adopted in this application is as follows:

[0005] In a first aspect, embodiments of this application provide a method for monitoring tree carbon storage based on lidar, the method comprising:

[0006] S110, at the first time node, each of the N sample plots is scanned by lidar to collect the first point cloud data corresponding to the trees. The N sample plots are sample plots randomly sampled from M regions. The M regions are regions obtained by dividing the monitoring area corresponding to the forest vegetation. M is an integer greater than 2 and N is a positive integer less than M.

[0007] S120, preprocess the first point cloud data to obtain the second point cloud data of trees in each sample plot;

[0008] S130, based on the second point cloud data of each sample plot, perform single tree segmentation and parameter extraction to obtain the first number of single trees and the first feature parameters corresponding to the single trees. The first feature parameters include diameter at breast height and tree height.

[0009] S140, Based on the pre-established correction model, the first feature parameter is corrected to obtain the corrected first feature parameter;

[0010] S150, based on the number of first individual trees in each sample plot and the corrected first characteristic parameter, the carbon storage of trees in the monitoring area is determined through a preset carbon measurement model.

[0011] Secondly, embodiments of this application also provide an electronic device, which includes a processor and a memory coupled to each other. The memory stores a computer program, and when the computer program is executed by the processor, the electronic device performs the above-described method for monitoring tree carbon storage based on lidar.

[0012] Thirdly, embodiments of this application also provide a computer-readable storage medium storing a computer program that, when run on a computer, causes the computer to execute the aforementioned method for monitoring tree carbon storage based on lidar.

[0013] The invention employing the above technical solution has the following advantages:

[0014] In the technical solution provided in this application, scanning N random sample plots with lidar replaces manual on-site tree-by-tree monitoring, which helps to shorten data acquisition time, reduce manpower and time costs, and improve monitoring efficiency. Furthermore, by segmenting individual trees and extracting parameters, core characteristic parameters of individual trees such as diameter at breast height (DBH) and tree height are directly obtained, rather than the "regional coverage" obtained through remote sensing. This data is more aligned with the needs of carbon storage calculation. Setting up a "calibration model" to correct the extracted characteristic parameters helps the first characteristic parameter of the tree after correction to closely match the measured characteristic parameter, reducing the estimation error of carbon storage calculation and improving the accuracy of tree carbon storage monitoring. Attached Figure Description

[0015] This application can be further illustrated by the non-limiting embodiments given in the accompanying drawings. It should be understood that the following drawings only illustrate some embodiments of this application and should not be considered as limiting the scope. For those skilled in the art, other related drawings can be obtained from these drawings without any inventive effort.

[0016] Figure 1 This is a schematic flowchart of a tree carbon storage monitoring method based on lidar provided in an embodiment of this application. Detailed Implementation

[0017] The present application will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that similar or identical parts are referred to by the same reference numerals in the drawings or description. Implementations not shown or described in the drawings are forms known to those skilled in the art. In the description of this application, terms such as "first" and "second" are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0018] This application provides an electronic device that may include a processor and a memory. The memory stores a computer program, which, when executed by the processor, enables the electronic device to perform corresponding steps in the following lidar-based tree carbon storage monitoring method. The electronic device may be, but is not limited to, a personal computer, a server, or similar equipment.

[0019] Please refer to Figure 1 This application provides a method for monitoring tree carbon storage based on lidar (hereinafter referred to as "this method"), which can be applied to the aforementioned electronic device, and the electronic device executes or implements the steps of the method. The electronic device can interact with the lidar and acquire point cloud data scanned by the lidar. Specifically, this method may include the following steps:

[0020] S110, at the first time node, each of the N sample plots is scanned by lidar to collect the first point cloud data corresponding to the trees. The N sample plots are sample plots randomly sampled from M regions. The M regions are regions obtained by dividing the monitoring area corresponding to the forest vegetation. M is an integer greater than 2 and N is a positive integer less than M.

[0021] S120, preprocess the first point cloud data to obtain the second point cloud data of trees in each sample plot;

[0022] S130, based on the second point cloud data of each sample plot, perform single tree segmentation and parameter extraction to obtain the first number of single trees and the first feature parameters corresponding to the single trees. The first feature parameters include diameter at breast height and tree height.

[0023] S140, Based on the pre-established correction model, the first feature parameter is corrected to obtain the corrected first feature parameter;

[0024] S150, based on the number of first individual trees in each sample plot and the corrected first characteristic parameter, the carbon storage of trees in the monitoring area is determined through a preset carbon measurement model.

[0025] In this embodiment, "random sampling + multi-stage modeling" ensures a more reliable final carbon storage result. This approach, through the aforementioned steps, offers the following advantages:

[0026] Rich in data dimensions: Through "single-tree segmentation + parameter extraction", directly obtain the core feature parameters of single trees such as diameter at breast height and tree height, rather than the "regional coverage rate" of remote sensing. The data is more in line with the requirements of carbon storage calculation.

[0027] More precise error control: Set a "correction model" to correct the extracted feature parameters, and at the same time avoid the problem that remote sensing is affected by the seasonal vegetation state, significantly reducing the estimation error.

[0028] Strong representativeness of plots: N plots are randomly selected from M divided areas (M > 2, N < M), covering the diversity of the monitoring area and avoiding the "local bias" of artificial or remote sensing sampling.

[0029] Rigorous calculation logic: Based on the "corrected single-tree feature parameters + number of single trees", combined with a preset carbon measurement model, replacing traditional rough estimation, the calculation process is more in line with the scientific logic of carbon storage assessment.

[0030] The following will elaborate on each step of this method as follows:

[0031] In step 110, the first time node can be understood as the time range from when the lidar starts scanning N plots to when it finishes scanning. This time range is usually within 10 days and can be flexibly determined according to the actual situation.

[0032] In this embodiment, the monitoring area can be flexibly planned according to the actual situation of the forest coverage range to be detected. The monitoring area can be divided into M areas of equal area on the electronic map. The number of M can be flexibly determined according to the actual situation, such as 100, 1000, or other numbers. When randomly sampling N areas from M areas, the value of N can be 5 - 20% of M. For example, when M is 100, N is 10.

[0033] The lidar can be a backpack lidar, and the lidar can be equipped with a positioning module. It can achieve the positioning of the lidar through the RTK (Real Time Kinematic) positioning algorithm. The position of the lidar is used to cooperate with the reference position coordinates of the plot to identify the plot boundary and facilitate filtering out the point cloud data outside the plot.

[0034] Each of the N plots is set with reference position coordinates. The reference position coordinates can be the longitude and latitude coordinates of the center point of the plot or other geometric points (such as points on the plot boundary, corner points of the plot, etc.). The selection method of the reference position is not specifically limited here.

[0035] In step S120, the first point cloud data is preprocessed to obtain the second point cloud data of the trees in each sample plot, which may include:

[0036] For each of the N sample plots, the position coordinates of the first point cloud data in the corresponding sample plot are determined based on the reference position coordinates;

[0037] Based on the location coordinates of the first point cloud data in the corresponding sample plot, the point cloud data that is not within the boundary line in the first point cloud data is filtered out to obtain the first point cloud data after filtering.

[0038] The filtered first point cloud data is then subjected to point cloud denoising and point cloud normalization to obtain the second point cloud data representing the trees in each sample plot.

[0039] In this embodiment, the lidar can achieve self-localization based on the RTK positioning algorithm, obtaining its own latitude and longitude information. The point cloud data obtained by the lidar scan carries three-dimensional coordinate information. This three-dimensional coordinate system is typically a three-dimensional coordinate system with the lidar as the origin (called the device coordinate system). By combining the reference position coordinates of the sample plot, the position of the point cloud in the device coordinate system can be converted to its position in the sample plot coordinate system. The sample plot coordinate system can be a coordinate system established based on the reference position coordinates of each sample plot as the origin, or a coordinate system established based on an electronic map of the detection area. After obtaining the position of the point cloud in the sample plot coordinate system, by combining the latitude and longitude information of the sample plot's boundary line on the electronic map, it can be determined whether the point cloud is within the sample plot's boundary line. Point clouds not within the boundary line are filtered out, thereby ensuring that the collected point cloud data does not include point clouds outside the sample plot's boundary line, improving the effectiveness of the collected data.

[0040] In this embodiment, point cloud denoising and point cloud normalization are both conventional methods. For example, when a lidar scanner scans a forest, in addition to the effective points of trees (trunks, branches, and leaves), various "noise points" are mixed in, such as ground weeds, airborne debris, equipment error points, fog / bird interference points, etc. Point cloud denoising is based on the "spatial distribution difference between effective points and noise points" to filter out point clouds that can represent the tree structure.

[0041] The purpose of point cloud normalization is to standardize the "height benchmark" and accurately calculate the actual height of trees. The original point cloud height from lidar scanning is based on the absolute height of the lidar device's own coordinate system (such as relative to sea level or the device's takeoff point). However, the ground in forest plots often has undulations (such as slopes and depressions), and directly using absolute height to calculate tree height will lead to errors (for example, tree height in higher areas is underestimated, and tree height in lower areas is overestimated). The core of normalization is to "convert the height of all points to relative heights relative to the plot ground." After normalization, the height values ​​of all points on the ground approach 0, the height of points at the base of trees approaches 0, and the height of points at the treetops is the actual height of the trees, providing an accurate benchmark for subsequent extraction of "tree height" parameters.

[0042] By denoising point clouds to remove invalid interference points and normalizing to unify the height benchmark, the point cloud data can truly reflect the three-dimensional structure of trees.

[0043] In step 130, individual tree segmentation is performed to segment the point cloud of each tree to distinguish each individual tree. Parameter extraction calculates the diameter at breast height (DBH) and tree height from the point cloud clusters of individual trees. Of course, other parameters can also be obtained through parameter extraction, such as feature parameters (which can refer to the first, second, or third feature parameters). In addition to DBH and tree height, these can also include parameters such as branch height and crown width.

[0044] The principle of single-tree segmentation is to utilize the spatial structural characteristics of a single tree (such as an independent crown, an upright trunk, and spatial spacing between adjacent trees) to break down a mixed point cloud into non-overlapping clusters of single-tree point clouds. Single-tree segmentation can be performed as follows:

[0045] Method 1, based on "distance clustering" segmentation (suitable for scenarios with large tree spacing and little canopy overlap): Utilizing the characteristic that "point clouds within the same tree are close together, while point clouds in different trees are far apart," a "distance threshold" (e.g., 0.5 meters) is set. Points within a distance of less than the threshold are grouped into the same cluster, while points within a distance of more than the threshold are grouped into different clusters. Ultimately, each cluster corresponds to an independent tree. That is, first, a point cloud is randomly selected as a "seed point" from the point cloud data. Then, surrounding point clouds that meet the distance requirement are gradually included until no more point clouds can be included. Then, the next unclassified seed point is selected, and the operation is repeated.

[0046] Method Two, based on "canopy structure" segmentation (suitable for dense forests with many overlapping canopies): First, find the "canopy apex" of each tree (the highest point in the single-tree point cloud). Then, using the apex as the center, segment downwards or outwards according to the "natural shape of canopy growth" (such as conical or ellipsoidal). Simultaneously, combine this with point cloud density to determine boundaries (in overlapping areas, points with higher density belong to the closer canopy apex). That is, starting with the point cloud of the canopy apex, gradually incorporate points with "continuous height and sufficient density" into the region until the point cloud boundaries of other trees are reached.

[0047] Method 3, segmentation based on "trunk detection": First, select "low-height, vertically continuous" trunk points (such as dense points with a height of 0.5-3 meters) from the normalized point cloud. Then, determine the trunk position and radius of each tree by fitting a cylinder (the trunk is approximately cylindrical). Finally, with the trunk as the center, classify the upward-extending canopy point clouds into the same tree. This helps to solve the problem of overlapping canopies.

[0048] After individual tree segmentation, each tree corresponds to an independent point cloud cluster. The core principle of parameter extraction is based on the "three-dimensional geometric morphology of trees". Key parameters that reflect the biomass of trees are extracted from the point cloud cluster. Among them, tree height extraction is relatively direct, while diameter at breast height (DBH) extraction requires the definition and geometric fitting of "standard DBH (e.g., 1.3 meters)".

[0049] As an example, tree height extraction: The point cloud has been normalized (the height reference is the ground of the sample plot, ground height ≈ 0). The "tree height" of a single tree is the "maximum normalized height of all points in the point cloud cluster of that tree"—because the highest point corresponds to the treetop, and the lowest point (bottom of the trunk) is close to the ground, the difference between the two is the actual height of the tree from the ground to the treetop. By iterating through the "normalized height values" (Z-axis coordinates) of all points in the single tree's point cloud cluster, the maximum value is selected, and this value is the "tree height" of that tree.

[0050] Diameter at Breast Height (DBH) Extraction: Fitting the "diameter of the tree trunk cross-section at a height of 1.3 meters". DBH is defined as the diameter of the tree trunk at a height of 1.3 meters above the ground. However, actual tree trunks are not perfect cylinders, and the point cloud at a height of 1.3 meters may be affected by branch interference. Therefore, extraction is achieved through "screening trunk points + geometric fitting". Specifically, the first step is to screen the "trunk points at the 1.3-meter height layer". From the single-tree point cloud clusters, points with a "normalized height between 1.2 and 1.4 meters" (allowing ±0.1 meter error to avoid missing points) are extracted. Simultaneously, branch points "far from the trunk center" in this layer are removed (screened using a "distance threshold from the trunk center", e.g., only points within a 0.3-meter radius of the center are retained). The second step is to fit a "circle of the tree trunk cross-section". For the screened trunk points at the 1.3-meter height layer, a circle is fitted using the "least squares method". That is, a circle is found that minimizes the sum of the squares of the distances from all trunk points to the circle; the diameter of this circle is the "DBH". For anomaly handling, if there is a fork at a height of 1.3 meters (without a complete trunk), take the "nearest complete height layer below the fork" (such as 1.2 meters, 1.1 meters) and repeat the above fitting steps to ensure that the diameter at breast height (DBH) data is valid.

[0051] In step 140, the calibration model is a relational function used to convert the characteristic parameters such as diameter at breast height (DBH) and tree height represented by radar-acquired point clouds into characteristic parameters that approximate the measured DBH and tree height. The calibration model is obtained through calibration prior to step 110. The method for generating / creating the calibration model can be found in steps S010 to S070 below.

[0052] The correction model can be a linear correction model or a nonlinear correction model.

[0053] As an example, the linear correction model can be: y = k0·x + b1; where y refers to the corrected characteristic parameter (e.g., diameter at breast height); x refers to the first characteristic parameter acquired by the radar (e.g., diameter at breast height); k0 refers to the slope parameter, used to correct the proportional error between the radar-acquired characteristic parameter and the measured characteristic parameter; b1 is the intercept parameter, correcting system fixed errors, such as the inherent bias of lidar. The parameters k0 and b1 can be obtained by fitting multiple sets of measured characteristic parameters and the first characteristic parameter acquired by the radar using the least squares method.

[0054] The nonlinear correction model can be: y = a·x 2 +b2·x+c; where y refers to the corrected characteristic parameter (e.g., diameter at breast height); x refers to the first characteristic parameter acquired by the radar (e.g., diameter at breast height); a, b2, and c are the coefficients of a quadratic polynomial, which can be obtained by fitting multiple sets of measured characteristic parameters and the first characteristic parameter acquired by the radar using the least squares method.

[0055] Step S150, based on the first single tree quantity in each sample plot and the corrected first characteristic parameter, determines the tree carbon storage in the monitoring area using a preset carbon measurement model, including:

[0056] Based on the number of individual trees in each sample plot and the corrected first characteristic parameter, the carbon storage of a single tree is determined using the calculation formula corresponding to the carbon measurement model. The carbon measurement model may include the following formulas (1)-(3), expressed as:

[0057] (1)

[0058] In the formula, Carbon storage of a single tree in the i-th sample plot (unit: tons of carbon). The relationship between the total biomass of the j-th tree in the i-th plot and its diameter at breast height and height (unit: kg per tree); The diameter at breast height (DBH) of the j-th tree in the i-th sample plot (unit: cm); The height of the j-th tree in the i-th sample plot (unit: meters); Carbon content in tree species biomass (unit: tons of carbon per ton).

[0059] Based on the number of individual trees and their carbon storage in each sample plot, the carbon storage of each sample plot is obtained. (Unit: tons of carbon), expressed as:

[0060] (2)

[0061] In the formula, J refers to the number of the first single tree in the sample plot;

[0062] Based on the tree carbon storage in each sample plot, the estimated tree carbon storage C in the monitored area is expressed as:

[0063] (3)

[0064] In the formula, A refers to the total area of ​​the N sample plots (unit: hectares); A refers to the total area of ​​the monitoring area (unit: hectares); N refers to the number of sample plots.

[0065] In step 150, the carbon storage of each tree in the sample plot can be calculated using a carbon econometric model, as shown in formula (1); by summing the carbon storage of all trees in a single sample plot, the carbon storage of a single sample plot can be obtained, as shown in formula (2); by summing the carbon storage of all trees in all sample plots, the carbon storage of all trees in all sample plots can be obtained, i.e. Combined with the area ratio of the entire monitoring area and all sample plots The carbon storage of trees in the monitoring area can then be obtained, as shown in formula (3).

[0066] In this embodiment, the biomass and carbon storage of a single tree are calculated using the characteristic parameters of a single tree. The biomass equations and carbon content coefficients for different tree species are based on GB / T43648—2024, "Biomass Models and Carbon Measurement Parameters of Standing Trees of Major Tree Species".

[0067] Prior to step 110, the method may include the step of establishing a calibration model. That is, prior to step 110, the method may further include:

[0068] S010, obtain the number of second individual trees and the second characteristic parameters corresponding to each individual tree in P sample plots that are manually measured at the second time node. The P sample plots are sample plots randomly sampled from M regions. The second characteristic parameters include diameter at breast height and tree height.

[0069] S020, for each of the P sample plots, at the third time node, the corresponding sample plot is scanned by the lidar to collect the third point cloud data corresponding to the trees. The time difference between the third time node and the second time node is within a preset time difference range. The preset time difference range can be flexibly determined according to the actual situation, for example, not exceeding 10 days.

[0070] S030, preprocess the third point cloud data to obtain the fourth point cloud data of trees in each sample plot;

[0071] S040, Based on the fourth point cloud data, perform single-tree segmentation to obtain the third single-tree quantity;

[0072] S050, if the difference between the number of third individual trees and the number of second individual trees in any of the P sample plots is not within the first preset threshold range (the preset threshold range can be flexibly set according to the actual situation, for example, 10% of the number of second individual trees), then for any sample plot, the laser radar is used to rescan to obtain new third point cloud data, and steps S030 and S040 are repeated until the difference between the number of third individual trees and the number of second individual trees in each sample plot is within the first preset threshold range;

[0073] S060, when the difference between the number of third individual trees and the number of second individual trees in each sample plot is within the first preset threshold range, parameters are extracted from the fourth point cloud data to obtain the third feature parameters corresponding to the individual trees. The third feature parameters include diameter at breast height (DBH) and tree height.

[0074] S070, Based on the third feature parameter and the second feature parameter, establish a correction model for correcting the third feature parameter to the second feature parameter.

[0075] In step S010, to ensure the accuracy of the second characteristic parameter, multiple workers can measure each of the P sample plots, and different workers can obtain their own characteristic parameters. Then, outlier filtering is performed on the data from multiple workers. The average of the filtered outlier data is then calculated to obtain the accurate and effective characteristic parameter. The outlier filtering method is conventional; for example, for the same tree, if the dispersion of characteristic parameters (such as diameter at breast height) measured by different people exceeds a certain threshold, the characteristic parameters exceeding the threshold are filtered out.

[0076] In other implementations, if the overall ecological environment of each area within the monitoring region is similar, and the species and growth of trees in each area are also similar, then for each sample plot, only one manual measurement of the tree's characteristic parameters is required, eliminating the need for repeated measurements by multiple people at each sample plot as described above. The data obtained from the sample plot measurements need to be verified for accuracy. This verification can be performed in the following ways: if the accuracy requirements are not met, additional sample plots need to be added for measurement, and the accuracy verification repeated until the requirements are met. The validity verification method can be as follows:

[0077] Step 1: For the tree characteristic parameters of each sample plot, taking diameter at breast height (DBH) as an example, calculate the average DBH for each sample plot.

[0078] (4)

[0079] : Average diameter at breast height (DBH) of the i-th sample plot (unit: cm);

[0080] : Number of trees in the i-th sample plot (unit: trees);

[0081] : Diameter at breast height (DBH) of the j-th tree in the i-th sample plot (unit: cm).

[0082] Step 2: Calculate the average diameter at breast height (DBH) of the total number of plots (P plots).

[0083] (5)

[0084] : Average diameter at breast height (DBH) of the sample plot (unit: cm);

[0085] Number of sample plots.

[0086] Step 3: Calculate the variance of the average diameter at breast height (DBH) of the total sample plots.

[0087] (6)

[0088] : Variance of average diameter at breast height (DBH) of the sample plots.

[0089] Step 4: Calculate the uncertainty

[0090] (7)

[0091] The uncertainty of the average value of the monitoring indicators in the sample plot, i.e., the relative error, requires that the sampling accuracy be no less than 90%, i.e., the relative error be no greater than 10%.

[0092] The reliability index has N-1 degrees of freedom (where N is the number of sample plots). The monitoring requires a confidence level of 90%, which means that the two-sided quantile values ​​of the t-distribution are required at a confidence level of 90% and with n-1 degrees of freedom.

[0093] Understandably, if the monitoring requires a sampling accuracy u of no less than 90%, then the manually measured tree characteristic parameters are considered valid. If the monitoring requires a sampling accuracy u of less than 90%, then the manually measured tree characteristic parameters are considered invalid, and additional plots need to be added for measurement, and the accuracy check needs to be repeated until the accuracy requirement is met. It should be noted that when the number of newly added plots reaches the maximum threshold, the characteristic parameters are considered valid / accurate data by default. The accuracy threshold can be other values, not limited to 90%.

[0094] In this embodiment, the data preprocessing in step S030, the single-tree segmentation in step S040, and the parameter extraction in step S060 can be referred to the foregoing description of the data preprocessing in step S120, the single-tree segmentation in step S130, and the parameter extraction, and will not be repeated here.

[0095] In step S070, the calibration model can be either the aforementioned linear model or a nonlinear model, and the corresponding parameters in the model can be obtained by least squares fitting based on the third characteristic parameter and the second characteristic parameter.

[0096] In other embodiments, the calibration model can be a fitting model, such as a fitting model between diameter at breast height (DBH) and tree height. This fitting model is used to convert the tree height acquired by the lidar into the corresponding DBH. The method of establishing the fitting model is similar to that of linear / nonlinear models. When selecting a calibration model, models with small errors (the error between the characteristic parameters measured by the lidar and the actual measured characteristic parameters) and high accuracy can be given priority. The method for judging model accuracy can be as follows:

[0097] (8)

[0098] Model accuracy;

[0099] Single tree number;

[0100] : Measured characteristic parameters of the kth individual tree;

[0101] The calibration model calculates the characteristic parameters of the kth individual tree.

[0102] : The average value of the characteristic parameters measured in the sample plot.

[0103] If the model accuracy is lower than the accuracy threshold, it is necessary to remodel or select another model. The accuracy threshold can be 0.8 or other values; no specific limitation is made here.

[0104] As an optional implementation, the method may further include:

[0105] S160, based on the pre-established correspondence between vegetation carbon storage and tree carbon storage, determine the vegetation carbon storage corresponding to the tree carbon storage in the monitoring area, and use it as the vegetation carbon storage in the monitoring area.

[0106] In this embodiment, the carbon storage of trees in a forest can indirectly reflect the carbon storage of the entire forest vegetation. The carbon storage of trees in forests of the same ecological species typically has a fixed correspondence / mapping relationship with the carbon storage of forest vegetation. By pre-calibrating the mapping function between the carbon storage of trees and vegetation in forests of different ecological species, and combining this with the ecological species of the monitoring area, the appropriate mapping function can be selected to obtain the vegetation carbon storage of the entire monitoring area. The creation method of this mapping function is similar to that of the aforementioned linear / nonlinear correction model, and will not be elaborated here.

[0107] Ecological categories can be classified based on factors such as climate zone, dominant tree species, and forest stand structure. As an example, ecological categories can be classified according to climate zone and vegetation type, including:

[0108] Tropical rainforests are characterized by high temperatures and humidity, abundant biodiversity, and well-developed understory vegetation (vines and epiphytes).

[0109] Subtropical evergreen broad-leaved forest: warm and humid, with dominant tree species being evergreen broad-leaved trees (such as camphor trees and chestnut trees), and shrubs and herbs under the forest canopy;

[0110] Temperate deciduous broad-leaved forest: distinct seasons, with leaves falling in winter, and the understory mainly composed of herbaceous plants and leaf litter;

[0111] Temperate coniferous forest: cold and dry, with dominant tree species such as pine, cypress, and spruce, and sparse understory vegetation;

[0112] Temperate coniferous forest (taiga): Low temperature and little rain, the trees are mainly larch and spruce, and there is very little understory vegetation;

[0113] Subtropical coniferous forests: such as Southern Chinese fir forests and Masson pine forests, with a small number of shrubs and herbs under the forest canopy;

[0114] Tropical monsoon forests: distinct dry and wet seasons, with understory vegetation changing with the seasons.

[0115] To further improve mapping accuracy, the classification can be further subdivided based on dominant tree species, building upon the climate zone classification.

[0116] Coniferous forests: spruce forests, fir forests, pine forests, larch forests, etc.;

[0117] Broadleaf forests: oak forests, birch forests, poplar forests, pure camphor forests, and Castanopsis forests, etc.

[0118] The corresponding parameters of the mapping function are determined through "actual measurement calibration": select typical sample plots, and simultaneously measure the carbon storage of trees and the total carbon storage of vegetation (including shrubs, herbs, etc.). Statistical fitting is used to obtain the parameters of the corresponding ecological species to ensure the accuracy of the mapping relationship.

[0119] As an optional implementation, the method may further include:

[0120] S170, replace the first time node with multiple time nodes after the first time node, and repeat steps S110 to 160 at each of the multiple time nodes to obtain the vegetation carbon storage of the monitoring area at the corresponding time node.

[0121] S180, Based on the vegetation carbon storage at all time points, the relationship between carbon storage in the monitored area and time is obtained.

[0122] Understandably, the sample plot is scanned with lidar at regular intervals, and the trees in the monitoring area are scanned by lidar at multiple time points to obtain the tree characteristic parameters corresponding to each time point. For the tree characteristic parameters at each time point, by repeating steps S110 to 160, the vegetation carbon storage in the monitoring area at each time point can be obtained, forming the relationship between forest vegetation carbon storage and time.

[0123] Changes in carbon storage are closely related to ecosystem health, and carbon storage data obtained from monitoring at multiple time points provide crucial evidence for assessing the stability, resilience, and sustainability of ecosystems.

[0124] As an optional implementation, the method may further include:

[0125] S190, Based on the described relationship of change, perform visualization processing to obtain a trend graph.

[0126] During the generation of the trend chart, core data from the change relationship can be extracted, including time nodes (such as January 2023, July 2023, January 2024, etc.), vegetation carbon storage values ​​at each time node, and optional subdivision data (such as carbon storage in M ​​regional divisions). The trend chart can be a line graph, with the horizontal axis representing time nodes and the vertical axis representing vegetation carbon storage (unit: tons), visually presenting the increase or decrease trend of carbon storage through continuous lines.

[0127] Visualization transforms the abstract "time-carbon storage" relationship into intuitive graphics, significantly improving the data's understandability, dissemination efficiency, and application value.

[0128] In this embodiment, the processor can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor. For example, it can be a Central Processing Unit (CPU), a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components, capable of implementing or executing the methods, steps, and logic block diagrams disclosed in the embodiments of this application.

[0129] The memory can be, but is not limited to, random access memory, read-only memory, programmable read-only memory, erasable programmable read-only memory, electrically erasable programmable read-only memory, etc. In this embodiment, the memory can be used to store point cloud data, calibration models, carbon metrology models, tree carbon storage, etc. Of course, the memory can also be used to store programs, which the processor executes after receiving execution instructions.

[0130] It should be noted that those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the electronic device described above can be referred to the corresponding steps in the aforementioned method, and will not be elaborated further here.

[0131] This application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program that, when run on a computer, causes the computer to perform the lidar-based tree carbon storage monitoring method as described in the above embodiments.

[0132] Based on the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by hardware or by using software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solution of this application can be embodied in the form of a software product. This software product can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, mobile hard drive, etc.) and includes several instructions to cause a computer device (such as a personal computer, electronic device, or network device, etc.) to execute the methods described in the various implementation scenarios of this application.

[0133] In the embodiments provided in this application, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device and method embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of devices, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code, which includes one or more executable instructions for implementing a specified logical function. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions. Furthermore, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0134] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for monitoring carbon stock of trees based on a laser radar, characterized by, The method comprises: S110, at a first time node, scanning each sample plot in N sample plots by a laser radar to collect first point cloud data corresponding to trees, wherein the N sample plots are sample plots randomly sampled from M regions, the M regions are regions obtained by dividing a monitoring region corresponding to forest vegetation, M is an integer greater than 2, and N is a positive integer less than M; S120, preprocessing the first point cloud data to obtain second point cloud data of trees in each sample plot; S130, based on the second point cloud data of each sample plot, performing single-tree segmentation and parameter extraction to obtain a first number of single trees and first feature parameters corresponding to the single trees, the first feature parameters comprising diameter at breast height and tree height; S140, based on the first feature parameters, correcting the first feature parameters based on a pre-established correction model to obtain corrected first feature parameters; S150, based on the first number of single trees of each sample plot and the corrected first feature parameters, determining tree carbon storage of the monitoring region by a preset carbon measurement model.

2. The method of claim 1, wherein, Before step S110, the method further comprises: S010, obtaining a second number of single trees and second feature parameters corresponding to the single trees of each sample plot measured by a human being at a second time node, the P sample plots being sample plots randomly sampled from the M regions, and the second feature parameters comprising diameter at breast height and tree height; S020, for each sample plot in the P sample plots, scanning the corresponding sample plot by the laser radar at a third time node to collect third point cloud data corresponding to trees, a time difference between the third time node and the second time node being within a preset time difference range; S030, preprocessing the third point cloud data to obtain fourth point cloud data of trees in each sample plot; S040, based on the fourth point cloud data, performing single-tree segmentation to obtain a third number of single trees; S050, if a difference between the third number of single trees and the second number of single trees of any sample plot in the P sample plots is not within a first preset threshold range, then for the any sample plot, rescan by the laser radar to obtain new third point cloud data, and repeat steps S030 and S040 until the difference between the third number of single trees and the second number of single trees of each sample plot is within the first preset threshold range; S060, when the difference between the third number of single trees and the second number of single trees of each sample plot is within the first preset threshold range, performing parameter extraction from the fourth point cloud data to obtain third feature parameters corresponding to the single trees, the third feature parameters comprising diameter at breast height and tree height; S070, based on the third feature parameters and the second feature parameters, establishing a correction model for correcting the third feature parameters into the second feature parameters.

3. The method of claim 1, wherein, Each sample plot in the N sample plots is provided with a reference position coordinate; Step S120, preprocessing the first point cloud data to obtain second point cloud data of trees in each sample plot, comprises: For each of the N sample plots, based on the reference position coordinates, determine position coordinates of the first point cloud data in the corresponding sample plot; Based on the position coordinates of the first point cloud data in the corresponding sample plot, filter out the point cloud data in the first point cloud data that is not within the boundary line of the corresponding sample plot, to obtain filtered first point cloud data; For the filtered first point cloud data, perform point cloud denoising and point cloud normalization to obtain the second point cloud data representing the trees in each sample plot.

4. The method of claim 1, wherein, Step S150, based on the first number of individual trees of each sample plot and the corrected first feature parameter, determine the tree carbon storage of the monitoring area through a preset carbon measurement model, including: Based on the first number of individual trees of each sample plot and the corrected first feature parameter, determine the tree carbon storage of individual trees through the calculation formula corresponding to the carbon measurement model, represented as: ; wherein, denotes the tree carbon storage of the jth individual tree in the ith plot; denotes the function of the relationship between the whole-plant biomass and the diameter at breast height and the tree height of the jth individual tree in the ith plot; denotes the diameter at breast height of the jth individual tree in the ith plot; denotes the tree height of the jth individual tree in the ith plot; denotes the tree species-specific biomass carbon content. Based on the first number of individual trees of each sample plot and the tree carbon storage of individual trees, the tree carbon storage of each sample plot is obtained is expressed as: ; In the formula, J refers to the first number of individual trees in the sample plot; Based on the tree carbon storage of each sample plot, estimate the tree carbon storage C of the monitoring area, represented as: ; wherein denotes the total area of the N sample plots; A denotes the total area of the monitoring region; N denotes the number of sample plots.

5. The method of claim 1, wherein, The method further includes: S160, based on the pre-established corresponding relationship between vegetation carbon storage and tree carbon storage, determine the vegetation carbon storage corresponding to the tree carbon storage of the monitoring area as the vegetation carbon storage of the monitoring area.

6. The method of claim 5, wherein, The method further includes: S170, replace the first time node with multiple time nodes after the first time node, and at each time node of the multiple time nodes, repeat steps S110 to S160 to obtain the vegetation carbon storage of the monitoring area at the corresponding time node; S180, based on the vegetation carbon storage of all time nodes, obtain the change relationship of the carbon storage in the monitoring area over time.

7. The method of claim 6, wherein, The method further includes: S190, according to the change relationship, perform visual processing to obtain a change trend chart.

8. An electronic device, comprising: The electronic device includes a processor and a memory coupled to each other, and the memory stores a computer program. When the computer program is executed by the processor, the electronic device executes the method of any one of claims 1-7.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program. When the computer program runs on the computer, the computer executes the method of any one of claims 1-7.

Images