Forest carbon reserve estimation method, device and equipment and storage medium
By combining multi-source data from air, space, and land, the problem of inaccurate forest carbon storage estimation was solved, high-precision quantification of forest structure and high-precision estimation of carbon storage were achieved, and the interpretability and estimation accuracy of the model were improved.
Patent Information
- Application Number
- CN202510920221.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-04
- Publication Date
- 2025-10-03
AI Technical Summary
Existing technologies for estimating forest carbon stocks are inaccurate. Traditional methods rely on ground surveys and fail to fully consider the complexity of forest canopy structure, resulting in insufficient accuracy and theoretical explanatory power.
A method based on multi-source data from air, space, and land is adopted. The regional grid canopy entropy is calculated by obtaining point cloud data, and spatial partitioning is performed by combining forest, grassland and wetland survey data. A physical model for carbon storage accounting of single-layer forests and multi-layer forests is constructed. Forest cover parameters are calculated using multispectral image data, and an allometric growth equation is constructed to calculate carbon storage, thereby achieving high-precision quantification of forest structure.
It has achieved highly precise estimation of forest carbon stocks, provided more reliable technical support, offered a scientific basis for forest management and carbon trading, and improved the estimation accuracy and interpretability of the model.
Smart Images
Figure CN120746045A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of forest carbon stock estimation, and in particular to a forest carbon stock estimation method, device, equipment and storage medium. Background Art
[0002] Forests are the largest carbon reservoirs in terrestrial ecosystems, and changes in their carbon storage have a significant impact on the global carbon cycle and climate change. Accurately estimating forest carbon stocks helps assess the role of forest ecosystems in the global carbon balance and provides a scientific basis for addressing climate change. Furthermore, understanding the current status and changing trends of forest carbon stocks supports the rational development, protection, and management of forest resources, promoting the sustainable development of forest ecosystems. Furthermore, in the carbon trading market, accurate forest carbon storage data is the basis for calculating emission reductions in carbon sink projects and is therefore crucial for promoting forest carbon sink trading. Summary of the Invention
[0003] The present application provides a method for estimating forest carbon reserves, which is used to solve the problem of inaccurate forest carbon reserve estimation in the prior art.
[0004] According to one aspect of the present application, a method for estimating forest carbon stocks is provided, the method comprising: Acquire multi-source data of the target area, including regional-scale point cloud data and forest, grassland and wetland survey data; The regional grid canopy entropy of the target area is calculated based on point cloud data; the regional grid canopy entropy is used to characterize the complexity of the forest structure in the target area; The forest extent data of the target area is obtained based on the forest, grassland and wetland survey data. The forest extent data is spatially partitioned based on the regional grid canopy entropy and the pre-generated canopy entropy classification threshold to obtain the spatial distribution range of single-layer forests and multi-layer forests in the target area. Within the spatial distribution range of single-layer forests, the carbon stock of single-layer forests is estimated based on the physical model for carbon stock accounting of single-layer forests. Within the spatial distribution range of multi-layer forests, the carbon stock of multi-layer forests is estimated based on the physical model for carbon stock accounting of multi-layer forests. The spatial distribution data of forest carbon storage in the target area is generated based on the merging of single-layer forest carbon storage and multi-layer forest carbon storage.
[0005] In one possible implementation, the calculation of the regional grid canopy entropy of the target area based on point cloud data includes: Divide the point cloud data into blocks based on the standard grid to generate grid point cloud data; The regional grid canopy entropy of the target area is calculated based on the grid point cloud data.
[0006] In one possible implementation, the canopy entropy classification threshold is generated based on the following method: Acquire sample plot image data of multiple forest sample plots within the target area; Based on the standard grid, multiple canopy plots were selected from the sample area image data, and the canopy type and sample canopy entropy of each canopy plot were determined; the canopy types included single-layer forest and multi-layer forest; The canopy entropy classification threshold is determined based on the canopy type and sample canopy entropy of each canopy plot; among them, the canopy entropy classification threshold is used to divide single-layer forests and multi-layer forests.
[0007] In another possible implementation, the multi-source data further includes regional-scale multispectral image data and forest resource inventory data; The physical models for carbon stock accounting in single-layer forests and multi-layer forests are constructed based on the following methods: Using multispectral image data, multiple forest sampling sites were selected from multiple sampling sites in the forest resource inventory data; Construct modeling samples and validation samples based on each forest sampling site and the canopy type corresponding to the forest sampling site; The model parameters were calculated based on multi-source data, and the physical models for carbon stock accounting of single-layer forest and multi-layer forest were constructed according to the model parameters, modeling samples and verification samples.
[0008] In another possible implementation, the above-mentioned calculation of model parameters based on multi-source data includes: Based on the sampling range of the forest sampling site, the canopy entropy of the sampling site is calculated in combination with the point cloud data; Interpolation processing is performed based on point cloud data to obtain the digital surface model and digital elevation model of the target area; The average tree height of the forest sampling site was calculated based on the difference between the digital surface model and the digital elevation model; Calculate forest cover parameters based on multispectral image data; Calculate the carbon storage of forest sampling sites based on the spatial distribution range of single-layer forests and multi-layer forests; The canopy entropy of the sampling site, the average tree height of the forest sampling site, the forest cover parameter and the carbon storage of the forest sampling site were used as model parameters.
[0009] In another possible implementation, the calculation of the average tree height of the forest sampling site based on the difference between the digital surface model and the digital elevation model includes: Determine the height parameters of the target area based on the standard grid based on the difference between the digital surface model and the digital elevation model; According to the sampling range of the forest sampling site, the average height of trees in the forest sampling site was obtained based on the height parameter statistics of the standard grid.
[0010] In another possible implementation, the above-mentioned calculation of forest cover parameters based on multispectral image data includes: Calculate the normalized vegetation index of the target area using multispectral image data; The kernel normalized difference vegetation index was calculated based on the normalized difference vegetation index; The average kernel normalized difference vegetation index within the sampling range of the forest sampling site was calculated; Forest cover parameters were calculated based on the mean kernel normalized difference vegetation index.
[0011] In another possible implementation, the calculation of the carbon storage of the forest sampling site based on the spatial distribution range of the single-layer forest and the spatial distribution range of the multi-layer forest includes: Based on the spatial distribution range of single-layer forests and multi-layer forests, the canopy type of each forest sampling site was determined; Based on each forest sampling site and the canopy type corresponding to the forest sampling site, the carbon storage of the forest sampling site was calculated using the allometric growth equation.
[0012] According to another aspect of an embodiment of the present application, a device for estimating forest carbon stocks is provided, the device comprising: The acquisition module is used to obtain multi-source data of the target area; the multi-source data includes regional-scale point cloud data and forest, grassland and wetland survey data; A calculation module is used to calculate the regional grid canopy entropy of the target area based on the point cloud data; wherein the regional grid canopy entropy is used to characterize the complexity of the forest structure in the target area; A partitioning module is used to obtain forest extent data of the target area based on forest, grassland and wetland survey data, and spatially partition the forest extent data based on regional grid canopy entropy and pre-generated canopy entropy classification thresholds to obtain the spatial distribution range of single-layer forests and multi-layer forests in the target area; An estimation module is used to estimate the carbon stock of single-layer forests based on the physical model for carbon stock accounting of single-layer forests within the spatial distribution range of single-layer forests, and to estimate the carbon stock of multi-layer forests based on the physical model for carbon stock accounting of multi-layer forests within the spatial distribution range of multi-layer forests; The generation module is used to generate the spatial distribution data of forest carbon storage in the target area based on the merging of single-layer forest carbon storage and multi-layer forest carbon storage.
[0013] According to another aspect of the present application, an electronic device is provided, which includes: a memory, a processor and a computer program stored in the memory, and the processor executes the computer program to implement the steps of the method shown in the first aspect of the present application.
[0014] According to another aspect of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method shown in the first aspect of the present application are implemented.
[0015] The beneficial effects of the technical solution provided by this application are: The forest carbon stock estimation method provided in this application can spatially partition forest range data through regional grid canopy entropy and pre-generated canopy entropy classification thresholds to obtain the spatial distribution range of single-layer forests and the spatial distribution range of multi-layer forests in the target area; then, based on different forest carbon stock accounting physical models, the single-layer forest carbon stock and the multi-layer forest carbon stock are estimated separately within the spatial distribution range of forests of different canopy types; at the same time, by fusing multi-source data, effective quantification of the complexity of forest structure is achieved, and high-precision estimation of regional forest carbon stocks is achieved. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.
[0017] Figure 1 A schematic flow chart of a method for estimating forest carbon reserves provided in an embodiment of the present application; Figure 2-1 A schematic diagram of a single-layer forest scenario in a forest carbon storage estimation method provided in an embodiment of the present application; Figure 2-2 A schematic diagram of a multi-layer forest scenario in a method for estimating forest carbon reserves provided in an embodiment of the present application; Figure 3 A statistical diagram of canopy entropy classification threshold classification results in a forest carbon storage estimation method provided in an embodiment of the present application; Figure 4 A data sampling diagram for a forest carbon storage estimation method provided in an embodiment of the present application; Figure 5 A spatial distribution map of forest carbon reserves for an example of a forest carbon reserve estimation method provided in an embodiment of the present application; Figure 6 A schematic diagram of the structure of a forest carbon stock estimation device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0018] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are merely illustrative and are not intended to limit the scope of the present disclosure. In addition, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.
[0019] The accompanying drawings illustrate various schematic diagrams of structures according to embodiments of the present disclosure. These figures are not drawn to scale, and for the purpose of clarity, certain details are exaggerated and certain details may be omitted. The shapes of the various regions and layers shown in the figures, as well as their relative sizes and positional relationships, are merely exemplary and may deviate in practice due to manufacturing tolerances or technical limitations. Those skilled in the art may design regions / layers with different shapes, sizes, and relative positions as needed.
[0020] In the context of the present disclosure, when a layer / element is referred to as being "on" another layer / element, it can be directly on the other layer / element or an intervening layer / element may be present therebetween. In addition, if a layer / element is "on" another layer / element in one orientation, it may be "below" the other layer / element when the orientation is reversed.
[0021] Accurately assessing forest carbon stocks is a core challenge in global carbon cycle research, addressing climate change, and establishing a carbon trading system. The inventors discovered that traditional methods rely on ground surveys and sampling, but due to high data costs, simplified models, and a failure to fully consider the complexity of forest canopy structure, existing models have significant limitations in accuracy and theoretical explanatory power. The development of lidar technology has opened up new possibilities for forest carbon stock accounting. However, how to quantify the complexity of forest structure and improve the accuracy of carbon stock models remains a key issue that needs to be addressed.
[0022] The inventors also discovered that related technologies have explored methods for constructing forest biomass and carbon stock models using different parameters. These methods have improved the ability to estimate biomass and carbon stocks at the regional scale, but they generally ignore the complexity of forest structure. Their main limitations are: on the one hand, many studies use machine learning methods (such as random forests, support vector machines, and decision trees) to calculate biomass and carbon stocks, and these models lack interpretability; on the other hand, existing parameters do not adequately represent the complex structure of forests. Traditional models rely on parameters such as canopy height, height quantiles, texture characteristics, and cover, which can only reflect two-dimensional or macroscopic three-dimensional characteristics of the forest, and often fail to effectively account for the complex structure of the forest. This is especially true in multi-layered forests, where ignoring structural complexity can lead to large errors. Although some researchers have attempted to improve models in recent years by introducing more structural parameters such as leaf area index (LAI) and canopy roughness, these parameters are still insufficient to fully describe the spatial complexity of forest canopy structure.
[0023] Based on the above technical issues, some embodiments of this application propose a high-precision simulation and estimation method for forest carbon storage based on multi-source data from the air, space, and land. This method divides forests into single-layer forests and multi-layer forests. Compared with single-layer forests, multi-layer forests have a complex forest structure, and the canopy structure needs to be considered when modeling. By integrating multi-source data and quantifying the complexity of forest canopy structure, a segmented simulation and estimation model that is more interpretable than traditional linear regression and machine learning methods is constructed, thereby providing more reliable technical support for regional forest carbon storage monitoring.
[0024] The following describes in detail the technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.
[0025] The present invention provides a method for estimating forest carbon reserves. Figure 1 As shown, the method includes: S101, acquiring data of a target area.
[0026] Among them, multi-source data include regional-scale point cloud data and forest, grassland and wetland survey data.
[0027] S102, calculating the regional grid canopy entropy of the target area based on the point cloud data.
[0028] The regional grid canopy entropy is used to characterize the complexity of the forest structure in the target area. The above point cloud data can be obtained by preprocessing the original point cloud data obtained by lidar detection.
[0029] Specifically, the original point cloud data can be spliced and noise can be manually removed first, and then the original point cloud data after preprocessing can be filtered and classified to determine the classification types such as ground points, vegetation, buildings, vehicles, etc., to generate the above-mentioned point cloud data.
[0030] S103, based on the forest, grassland and wetland survey data, the forest range data of the target area is obtained, and the forest range data is spatially partitioned based on the regional grid canopy entropy and the pre-generated canopy entropy classification threshold to obtain the spatial distribution range of the single-layer forest and the multi-layer forest in the target area.
[0031] Among them, the canopy entropy classification threshold is used to partition the forest range of the target area based on the canopy entropy of each spatial region.
[0032] S104: within the spatial distribution range of single-layer forests, estimate the carbon stock of single-layer forests based on the physical model for carbon stock accounting of single-layer forests; within the spatial distribution range of multi-layer forests, estimate the carbon stock of multi-layer forests based on the physical model for carbon stock accounting of multi-layer forests.
[0033] S105, generating forest carbon storage spatial distribution data of the target area based on the merging of the single-layer forest carbon storage and the multi-layer forest carbon storage.
[0034] The embodiment of the present application spatially partitions the forest range data through regional grid canopy entropy and pre-generated canopy entropy classification thresholds to obtain the spatial distribution range of single-layer forests and the spatial distribution range of multi-layer forests in the target area; then, based on different physical models for forest carbon stock accounting, the carbon stocks of single-layer forests and multi-layer forests are estimated separately within the spatial distribution ranges of forests with different canopy types; at the same time, by fusing multi-source data, effective quantification of the complexity of forest structure is achieved, and a high-precision estimation of regional forest carbon stocks is achieved.
[0035] In an embodiment of the present application, a possible implementation method is provided, wherein the above-mentioned calculation of the regional grid canopy entropy of the target area based on point cloud data includes: S201, performing block processing on the point cloud data based on a standard grid to generate grid point cloud data.
[0036] S202, calculating the regional grid canopy entropy of the target area based on the grid point cloud data.
[0037] Specifically, for the target area, we can layer it at 1m intervals according to the vertical height, and calculate the regional grid canopy entropy E by counting the density of each layer. The calculation formula is as follows: (1) in It is The point ratio of the layer; n is the total number of layers, the minimum height of the forest is set to 2 meters, and both i and n are positive integers.
[0038] The embodiment of the present application introduces canopy entropy to quantitatively describe the complexity of forest structure, and based on this, the spatial partitioning model of the forest is performed, overcoming the problem that traditional methods cannot effectively capture the spatial structure of the forest.
[0039] The present application provides a possible implementation method, in which the canopy entropy classification threshold is generated based on the following method: S301, acquiring sample area image data of a plurality of forest sample areas within a target area.
[0040] The above-mentioned sample area image data may be high-resolution orthophoto image data of the forest sample area.
[0041] Specifically, based on aerial sampling, high-resolution orthophoto data of multiple forest sample areas within the target area can be obtained as sample area image data.
[0042] S302 , screening multiple canopy plots from the sample area image data based on a standard grid, and determining the canopy type and sample canopy entropy of each canopy plot.
[0043] Among them, canopy types include single-layer forest and multi-layer forest.
[0044] Specifically, based on the standard grid, canopy plots with typical forest canopy characteristics are randomly selected from the sample area image data; and the canopy type of each canopy plot is determined through image recognition or on-site investigation, that is, the canopy plots are divided into single-layer forest and multi-layer forest, such as Figure 2-1 As shown in the figure, it is a schematic diagram of a single-layer forest. Figure 2-2 Shown is a schematic diagram of a multi-layer forest.
[0045] S303: Determine a canopy entropy classification threshold based on the canopy type and sample canopy entropy of each canopy plot.
[0046] Among them, the canopy entropy classification threshold is used to divide single-layer forests and multi-layer forests.
[0047] Specifically, the sample canopy entropy and canopy type corresponding to each canopy plot can be used as samples, and the canopy entropy classification threshold for distinguishing single-layer forests from multi-layer forests can be calculated using the maximum F1-score (harmonic mean of precision and recall) criterion.
[0048] The F1-score criterion formula is as follows: Among them, Precision is the accuracy, Recall is the recall rate; TP is the number of samples predicted as positive when they are actually positive, FP is the number of samples predicted as positive when they are actually negative, and FN is the number of samples predicted as negative when they are actually positive.
[0049] In a specific embodiment, the above F1-score is used to characterize the classification performance of the canopy entropy classification threshold. An initial canopy entropy classification threshold can be set first, and the classification result of the initial canopy entropy classification threshold can be determined. Then, the corresponding real-time F1-score is determined by the classification result. The initial canopy entropy classification threshold is adjusted by the real-time F1-score, and the above adjustment steps are repeated until the best classification performance, i.e., the F1-score value, is obtained. The initial canopy entropy classification threshold after the last adjustment is the final canopy entropy classification threshold. In actual engineering calculations, when F1-score=0.94, the corresponding classification discrimination effect is as follows: Figure 3 As shown in Figure 3, the classification performance is the best when the canopy entropy classification threshold is 1.87.
[0050] The embodiment of the present application realizes data collection at the sample plot scale by acquiring sample plot image data of multiple forest sample plots within the target area. Combined with the regional scale data in the collected data, the present application can integrate multi-scale (including sample plot scale and regional scale) and multi-source data (including point cloud data, forest and grassland wetland survey data, sample plot image data, etc.) to conduct a comprehensive and comprehensive analysis of the target area, providing strong support for the subsequent carbon storage estimation accuracy; at the same time, the embodiment of the present application screens out multiple canopy sample plots from the sample plot image data, determines the canopy type of each canopy sample plot, and uses the above-mentioned canopy sample plots as samples to calculate the canopy entropy classification threshold, which can effectively improve the accuracy of the canopy entropy classification threshold and improve the classification accuracy of the canopy type.
[0051] A possible implementation method is provided in an embodiment of the present application, where the multi-source data further includes multispectral image data and forest resource inventory data.
[0052] The physical models for carbon stock accounting in single-layer forests and multi-layer forests are constructed based on the following methods: S401, screening multiple forest sampling sites from multiple sampling sites in forest resource inventory data using multispectral image data.
[0053] For example, based on the coordinate values of the 61 sampling sites in the target area, the sampling sites can be vectorized according to their size (25.8m×25.8m), relevant attribute information can be assigned to each sampling site, and the sampling sites located in the forest area, namely forest sampling sites, can be screened through multispectral image data, totaling 30.
[0054] S402: constructing modeling samples and verification samples according to each forest sampling site and the canopy type corresponding to the forest sampling site.
[0055] Specifically, the forest sampling sites include multiple sampling sites of single-layer forest types and multiple sampling sites of multi-layer forest types. A part of the sampling sites of the multiple single-layer forest types and a part of the sampling sites of the multiple multi-layer forest types are selected to obtain modeling samples; the remaining sampling sites in the forest sampling sites are used as verification samples, and the verification samples can also include multiple sampling sites of single-layer forest types and multiple sampling sites of multi-layer forest types.
[0056] Specifically, for each forest type, the number of modeling samples may be greater than the number of validation samples.
[0057] In some possible implementations, the ratio between the number of modeling samples and the number of validation samples may be in the range of 2:1.
[0058] Specifically, when there are 30 forest sampling sites, including 15 single-layer forest types and 15 multi-layer forest types, 10 single-layer forest sampling sites and 10 multi-layer forest sampling sites can be randomly selected to construct modeling samples; the remaining 5 single-layer forest sampling sites and 5 multi-layer forest sampling sites are used to construct verification samples.
[0059] S403, calculating model parameters based on multi-source data, and constructing a single-layer forest carbon stock accounting physical model and a multi-layer forest carbon stock accounting physical model according to the model parameters, modeling samples and verification samples.
[0060] Specifically, physical models for carbon storage accounting of regional single-layer forests and multi-layer forests can be constructed respectively; the physical model for carbon storage accounting of single-layer forests is as shown in formula (5), and the physical model for carbon storage accounting of multi-layer forests is as shown in formula (6): (5) (6) Among them, C1 and C2 are the carbon storage of single-layer forest and multi-layer forest, and the unit is kg; S represents the forest cover information, and the unit is m 2 ; H is the canopy height, representing the vertical information of the forest, and the unit is m; E is the canopy entropy, representing the complex structural information of the forest; a, b, c and d are the target parameters to be solved.
[0061] Specifically, a is the density parameter, the unit is kg / m 3 , b, c and d are the correction parameters of S, H and E. The parameters a, b, c and d can be solved by the least square method based on the extracted model parameters C, S, H and E; at the same time, the coefficient of determination (R 2), the root mean square error (RMSE) and the root mean square error percentage (RMSE%) describe the accuracy of the model and the validation results. The formula of the indicator is as follows: (7) (8) (9) in, It is Actual values; It is predicted values; is the mean of all actual values; n: the number of samples. The closer R² is to 1, the better the fit. and The smaller it is, the smaller the degree of discreteness between the true value and the model predicted value is, and the higher the model accuracy is.
[0062] In an embodiment of the present application, a possible implementation method is provided, wherein the above-mentioned multi-source data-based model parameter calculation method includes: S501, for the sampling range of the forest sampling site, combined with the point cloud data, calculate the canopy entropy of the sampling site.
[0063] Specifically, based on the collected continuous forest resource inventory data (the sampling site size is one acre, 25.8m×25.8m), the sampling sites located in the forest area can be selected as modeling data, and the canopy entropy E of the sampling sites can be calculated based on the preprocessed lidar data to describe the forest canopy structure information.
[0064] S502 , performing interpolation processing based on the point cloud data to obtain a digital surface model and a digital elevation model of the target area; and calculating the average height of trees in the forest sampling site based on the difference between the digital surface model and the digital elevation model.
[0065] S503: Calculate forest coverage parameters based on the multispectral image data.
[0066] S504: Calculate the carbon storage of the forest sampling site based on the spatial distribution range of the single-layer forest and the spatial distribution range of the multi-layer forest.
[0067] S505, taking the canopy entropy of the sampling site, the average tree height of the forest sampling site, the forest cover parameter and the carbon storage of the forest sampling site as model parameters.
[0068] In an embodiment of the present application, a possible implementation method is provided, wherein the above-mentioned calculation of the average tree height of a forest sampling site based on the difference between a digital surface model and a digital elevation model includes: S601, determining a height parameter of a target area based on a standard grid based on a difference between a digital surface model and a digital elevation model; S602: For the sampling range of the forest sampling site, the average height of trees in the forest sampling site is obtained based on the height parameter statistics of the standard grid.
[0069] Specifically, we can interpolate the preprocessed point cloud data of the target area to obtain a Digital Surface Model (DSM) and a Digital Elevation Model (DEM). The height parameter H is obtained by subtracting the DEM from the DSM. Furthermore, after resampling and calculating the mean height of the sampling site (consistent with forest inventory data, 25.8m × 25.8m), the canopy height H can describe the forest height information. The formula is as follows: (10) Among them, H is the canopy height; DSM and DEM are digital surface model and digital elevation model.
[0070] In an embodiment of the present application, a possible implementation method is provided, wherein the forest cover parameters are calculated based on multispectral image data, including: S701, calculating the normalized vegetation index of the target area using multispectral image data; S702, calculating a kernel normalized difference vegetation index based on the normalized difference vegetation index; S703, calculate the average kernel normalized difference vegetation index within the sampling range of the forest sampling site; S704, calculating forest cover parameters based on the mean kernel normalized difference vegetation index.
[0071] Specifically, the Normalized Difference Vegetation Index (NDVI) can be calculated using multispectral image data, followed by the Kernel Normalized Difference Vegetation Index (kNDVI). At the same time, after resampling and statistically analyzing the mean of the sampling site (consistent with the forest inventory data, 25.8 m × 25.8 m), the forest cover parameter S is further calculated using the following formula: (11) (12) (13) in, and It is the reflectance of the near-infrared and red light bands; NDVI is the normalized vegetation index; is the hyperbolic tangent function; is the grid area (consistent with field survey data, 25.8m×25.8m).
[0072] In the embodiment of the present application, kNDVI can cope with problems such as saturation effect, complex phenological cycles and seasonal changes compared to NDVI. The present invention uses the forest coverage parameter S calculated by this index to estimate forest carbon storage. S is used to describe the planar information of the forest, which can better describe the forest coverage information and improve the model accuracy.
[0073] The present application provides a possible implementation method for calculating the carbon storage of a forest sampling site based on the spatial distribution range of a single-layer forest and the spatial distribution range of a multi-layer forest, including: S801: Determine the canopy type of each forest sampling site based on the spatial distribution range of single-layer forests and multi-layer forests; S802: Calculate the carbon storage of the forest sampling site using an allometric growth equation based on each forest sampling site and the canopy type corresponding to the forest sampling site.
[0074] Specifically, the carbon storage of individual trees can be calculated using the "allometric equation" based on the dominant tree species, number of trees, diameter at breast height and other parameters from the continuous forest resource survey data, and the carbon storage of the sampling site can be statistically obtained. The formula for the allometric equation is as follows: (14) Where C is the carbon storage of a single tree; is the carbon density of the tree species; D is the diameter at breast height; is the parameter of the tree species. See Table 1 for details.
[0075] Table 1 Allometric equations and parameters In order to better understand the above-mentioned forest carbon storage estimation method, Figure 4 and Figure 5 An example of a forest carbon stock estimation method according to the present application is described in detail. The method includes the following steps: S1. Data Collection: Using a mountainous area as the research area, collect the data required for the model. This includes: Regional-scale data: ① Forest, grassland and wetland survey data to obtain the forest distribution range in the study area; ② Forest resource inventory data from 2014 to 2018 (scale is one mu, 25.8m×25.8m), with a total of 61 sampling sites, including attributes such as tree species, number of trees, diameter at breast height, and coordinates of each sampling site; ③ Sentinel-2 multispectral imagery (August-November 2023, 10m resolution) for extracting vegetation indices; ④ Airborne LiDAR point cloud data covering the entire study area (density: 1 point / m²) for extracting forest three-dimensional structural information.
[0076] Sample plot scale data: Five aerial sample plots (area ≥ 1 km2) with typical forest structure characteristics were selected in the study area. 2 ), collect high-resolution orthophotos of the sample area using drones (October 2023, with a resolution of 2.5 cm × 2.5 cm, e.g. Figure 4 (shown in area (a)).
[0077] S2. Data Preprocessing Point cloud data processing: The original point cloud file is a plurality of strip-shaped point cloud files, and the point cloud needs to be preprocessed, including: splicing the strip-shaped original point cloud into a complete sample area point cloud, manually removing noise points, and then filtering to complete the classification of point clouds such as ground points, vegetation, buildings, vehicles, etc., and correcting misclassified points; at the same time, a standard grid (10m×10m) of the entire study area is generated, and the digital surface model (DSM) and digital elevation model (DEM) of the study area are obtained by interpolating the preprocessed point cloud.
[0078] Spatial vectorization of forestry inventory sampling sites: Based on the coordinates of the 61 sampling sites within the study area and the size of the sampling sites (25.8m×25.8m), the sampling sites were vectorized as follows: Figure 4 As shown in the middle area (b), relevant attribute information was assigned, and multispectral images were used to screen sampling sites located in forest areas, totaling 30.
[0079] S3. Calculate the canopy entropy threshold of the forest partition: Sample site selection: Based on high-resolution orthophotos, sample sites (10m×10m) with typical forest canopy characteristics were randomly selected in standard grid units within the five existing aerial sampling areas. 124 single-layer forest samples and 54 multi-layer forest samples were selected. The selected sample sites were divided into single-layer forest and multi-layer forest through manual interpretation of high-resolution orthophotos and field surveys. Figure 4 As shown in areas (c) and (d).
[0080] The canopy entropy values of the standard grid of the entire area were calculated using the preprocessed lidar point cloud. The canopy entropies of the obtained single-layer forest and multi-layer forest plots were used as samples. The optimal canopy entropy threshold for distinguishing single-layer forest and multi-layer forest was determined by the maximum F1-score (harmonic mean of precision and recall) criterion.
[0081] S4. Forest spatial partitioning based on canopy entropy threshold: At the regional scale, based on the forest range of the study area obtained from the forest, grassland and wetland monitoring data, the forest range of the study area was spatially divided based on the regional grid canopy entropy E and canopy entropy threshold calculated by S3, and the spatial distribution range of single-layer forests and multi-layer forests in the study area was obtained.
[0082] S5. Carbon stock accounting model parameter extraction: 1. Canopy entropy E of forestry inventory sampling sites: Based on the 30 sampling map patches selected after vectorization in S2 and combined with point cloud data, the canopy entropy of the sampling map patches is calculated using Formula 1; 2. Average height parameter H: Based on the pre-processed DSM and DEM in S2, the height parameter H on the standard grid is obtained by subtracting DEM from DSM. The average height of trees in the sampling area is calculated using the sampling map patch range.
[0083] 3. Forest cover parameter S: The NDVI index of the study area was calculated using Sentinel-2 data, and then the index kNDVI was calculated. The mean kNDVI value of the sampling map patch was statistically analyzed, and the forest cover parameter S was further calculated.
[0084] 4. Carbon storage parameter C: Using the forest spatial zoning results, the forestry inventory sampling sites were divided into 15 single-layer forests and 15 multi-layer forest sampling sites. The carbon storage within the sampling sites was calculated using the "allometric growth equation" based on the diameter at breast height, tree species, and number of trees.
[0085] S5. Construct carbon storage accounting models by zoning: Randomly select 10 single-layer forests and 10 multi-layer forests from the 15 single-layer forest sampling sites and 15 multi-layer forest sampling sites for modeling, and the remaining sites for validation. Construct carbon storage accounting physical models for single-layer forests and multi-layer forests respectively: Single-layer forest: ; Multi-layered forest: . And the model parameters and determination coefficient R are obtained by least squares. 2 , the calculation formula is shown in formula (7), as shown in Table 2, the R 2 All are above 0.9.
[0086] Table 2. Parameter solution results S6. Regional forest carbon stock accounting: Using a standard grid, Sentinel-2 remote sensing image data, DEM / DSM data, and point cloud data of the study area, and the three parameters E, H, and S required by the model, carbon stocks were calculated separately based on the parameters solved by the carbon storage model constructed based on different forest structures (single-layer forest and multi-layer forest). The carbon stocks of single-layer forest and multi-layer forest were then combined to obtain the spatial distribution map of forest carbon stocks in the study area, as shown in the figure. Figure 5 As shown, carbon storage gradually increases from the red area to the yellow area to the green area.
[0087] S7. Accuracy verification: Based on the spatial simulation results of carbon storage in the entire region, the remaining 5 single-layer forest and 5 multi-layer forest plots were used to verify the accuracy of the model. The accuracy indicators RMSE and RMSE% were calculated using the calculation formulas shown in formulas (8) and (9).
[0088] The calculation results show that the RMSE of single-layer forest and multi-layer forest are 374.18 kg and 270.01 kg respectively, and the RMSE% are 17.47% and 27.33% respectively.
[0089] The embodiments of this application use a physical model to estimate carbon reserves, providing clear model interpretation and understanding of the relationship between carbon reserves and various forest structural parameters. By integrating multi-source data (including remote sensing data, ground monitoring data, and LiDAR point cloud data), a highly accurate estimate of regional forest carbon reserves is achieved, providing a scientific basis for forest management and carbon reserve monitoring. This highly accurate carbon reserve estimation method can provide strong technical support for regional forest planning, management, carbon trading, and climate change response efforts.
[0090] The present application embodiment provides a device for estimating forest carbon reserves, such as Figure 6 As shown, the forest carbon stock estimation device 60 may include: an acquisition module 601, a calculation module 602, a partitioning module 603, an estimation module 604 and a generation module 605; The acquisition module 601 is used to acquire multi-source data of the target area; wherein the multi-source data includes regional-scale point cloud data and forest, grassland and wetland survey data; A calculation module 602 is configured to calculate the regional grid canopy entropy of the target area based on the point cloud data; wherein the regional grid canopy entropy is used to characterize the complexity of the forest structure in the target area; Partitioning module 603 is used to obtain forest extent data of the target area based on the forest, grassland and wetland survey data, and spatially partition the forest extent data based on the regional grid canopy entropy and the pre-generated canopy entropy classification threshold to obtain the spatial distribution range of the single-layer forest and the multi-layer forest in the target area; An estimation module 604 is configured to estimate the carbon stock of a single-layer forest within the spatial distribution range of the single-layer forest based on the physical model for calculating the carbon stock of the single-layer forest, and to estimate the carbon stock of a multi-layer forest within the spatial distribution range of the multi-layer forest based on the physical model for calculating the carbon stock of the multi-layer forest; The generating module 605 is used to generate forest carbon storage spatial distribution data of the target area based on the merging of the single-layer forest carbon storage and the multi-layer forest carbon storage.
[0091] In an embodiment of the present application, a possible implementation is provided. When calculating the regional grid canopy entropy of the target area based on the point cloud data, the calculation module 602 is used to: Divide the point cloud data into blocks based on the standard grid to generate grid point cloud data; The regional grid canopy entropy of the target area is calculated based on the grid point cloud data.
[0092] The present application provides a possible implementation method, in which the canopy entropy classification threshold is generated based on the following method: Acquire sample plot image data of multiple forest sample plots within the target area; Based on the standard grid, multiple canopy plots were selected from the sample area image data, and the canopy type and sample canopy entropy of each canopy plot were determined; the canopy types included single-layer forest and multi-layer forest; The canopy entropy classification threshold is determined based on the canopy type and sample canopy entropy of each canopy plot; among them, the canopy entropy classification threshold is used to divide single-layer forests and multi-layer forests.
[0093] A possible implementation method is provided in an embodiment of the present application, wherein the multi-source data further includes regional-scale multispectral image data and forest resource inventory data; The above-mentioned physical models for carbon stock accounting in single-layer forests and multi-layer forests are constructed based on the following methods: Using multispectral image data, multiple forest sampling sites were selected from multiple sampling sites in the forest resource inventory data; Construct modeling samples and validation samples based on each forest sampling site and the canopy type corresponding to the forest sampling site; The model parameters were calculated based on multi-source data, and the physical models for carbon stock accounting of single-layer forest and multi-layer forest were constructed according to the model parameters, modeling samples and verification samples.
[0094] In an embodiment of the present application, a possible implementation is provided. When the estimation module 604 calculates model parameters based on multi-source data, it is configured to: Based on the sampling range of the forest sampling site, the canopy entropy of the sampling site is calculated in combination with the point cloud data; Interpolation processing is performed based on point cloud data to obtain the digital surface model and digital elevation model of the target area; The average tree height of the forest sampling site was calculated based on the difference between the digital surface model and the digital elevation model; Calculate forest cover parameters based on multispectral image data; Calculate the carbon storage of forest sampling sites based on the spatial distribution range of single-layer forests and multi-layer forests; The canopy entropy of the sampling site, the average tree height of the forest sampling site, the forest cover parameter and the carbon storage of the forest sampling site were used as model parameters.
[0095] In one embodiment of the present application, a possible implementation is provided. When calculating the average tree height of a forest sampling site based on the difference between a digital surface model and a digital elevation model, the estimation module 604 is configured to: The target area is determined based on the height parameters in the standard grid based on the difference between the digital surface model and the digital elevation model; According to the sampling range of the forest sampling site, the average height of trees in the forest sampling site was obtained based on the height parameter statistics of the standard grid.
[0096] In one embodiment of the present application, a possible implementation is provided. When calculating forest cover parameters based on multispectral image data, the estimation module 604 is configured to: Calculate the normalized vegetation index of the target area using multispectral image data; The kernel normalized difference vegetation index was calculated based on the normalized difference vegetation index; The average kernel normalized difference vegetation index within the sampling range of the forest sampling site was calculated; Forest cover parameters were calculated based on the mean kernel normalized difference vegetation index.
[0097] In an embodiment of the present application, a possible implementation is provided. When calculating the carbon storage of a forest sampling site based on the spatial distribution range of a single-layer forest and the spatial distribution range of a multi-layer forest, the estimation module 604 is configured to: Based on the spatial distribution range of single-layer forests and multi-layer forests, the canopy type of each forest sampling site was determined; Based on each forest sampling site and the canopy type corresponding to the forest sampling site, the carbon storage of the forest sampling site was calculated using the allometric growth equation.
[0098] The device of the embodiment of the present application can execute the method provided by the embodiment of the present application, and its implementation principle is similar. The actions performed by each module in the device of each embodiment of the present application correspond to the steps in the method of each embodiment of the present application. For the detailed functional description of each module of the device, please refer to the description in the corresponding method shown in the previous text, and will not be repeated here.
[0099] The embodiment of the present application spatially partitions the forest range data through regional grid canopy entropy and pre-generated canopy entropy classification thresholds to obtain the spatial distribution range of single-layer forests and the spatial distribution range of multi-layer forests in the target area; then, based on different physical models for forest carbon stock accounting, the carbon stocks of single-layer forests and multi-layer forests are estimated separately within the spatial distribution ranges of forests with different canopy types; at the same time, by fusing multi-source data, effective quantification of the complexity of forest structure is achieved, and a high-precision estimation of regional forest carbon stocks is achieved.
[0100] In an embodiment of the present application, an electronic device is provided, including a memory, a processor, and a computer program stored on the memory. The processor executes the above-mentioned computer program to implement the steps of a forest carbon stock estimation method. Compared with the relevant technology, the following can be achieved: the embodiment of the present application spatially partitions the forest range data through regional grid canopy entropy and pre-generated canopy entropy classification thresholds to obtain the spatial distribution range of single-layer forests and the spatial distribution range of multi-layer forests in the target area; then, based on different forest carbon stock accounting physical models, the single-layer forest carbon stock and the multi-layer forest carbon stock are estimated respectively within the spatial distribution range of forests of different canopy types; at the same time, by fusing multi-source data, effective quantification of the complexity of forest structure is achieved, and high-precision estimation of regional forest carbon stocks is achieved.
[0101] In an embodiment of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method shown in the first aspect of the embodiment of the present application are implemented.
[0102] While the above description does not provide detailed technical details regarding the patterning of each layer, those skilled in the art will appreciate that various technical means can be employed to form layers, regions, and the like in desired shapes. Furthermore, those skilled in the art may devise methods that differ from those described above to achieve the same structure. Furthermore, while each embodiment has been described separately, this does not mean that the measures in each embodiment cannot be advantageously combined.
[0103] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present application.
[0104] Obviously, those skilled in the art may make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalents, this application is intended to include these modifications and variations.
Claims
1. A method for estimating forest carbon stocks, characterized in that: The method comprises: Acquire multi-source data of the target area; wherein the multi-source data includes regional-scale point cloud data and forest, grassland and wetland survey data; Calculating the regional grid canopy entropy of the target area based on the point cloud data; wherein the regional grid canopy entropy is used to characterize the complexity of the forest structure in the target area; Acquire forest extent data of the target area based on the forest, grassland and wetland survey data, and spatially partition the forest extent data based on the regional grid canopy entropy and a pre-generated canopy entropy classification threshold to obtain the spatial distribution range of single-layer forests and multi-layer forests in the target area; Within the spatial distribution range of the single-layer forest, the carbon stock of the single-layer forest is estimated based on the physical model for carbon stock accounting of the single-layer forest; within the spatial distribution range of the multi-layer forest, the carbon stock of the multi-layer forest is estimated based on the physical model for carbon stock accounting of the multi-layer forest; The spatial distribution data of forest carbon reserves in the target area are generated based on the merging of the single-layer forest carbon reserves and the multi-layer forest carbon reserves.
2. The method according to claim 1, wherein calculating the regional grid canopy entropy of the target area based on the point cloud data comprises: Performing block processing on the point cloud data based on a standard grid to generate grid point cloud data; The regional grid canopy entropy of the target area is calculated based on the grid point cloud data.
3. The method according to claim 1, wherein the canopy entropy classification threshold is generated based on the following method: Acquire sample plot image data of multiple forest sample plots within the target area; A plurality of canopy plots are screened from the sample area image data based on a standard grid, and the canopy type and sample canopy entropy of each canopy plot are determined; wherein, The canopy types include single-story forests and multi-story forests; A canopy entropy classification threshold is determined based on the canopy type and the sample canopy entropy of each canopy sample plot; wherein the canopy entropy classification threshold is used to divide single-layer forests and multi-layer forests.
4. The method according to claim 1, wherein the multi-source data further comprises regional-scale multispectral image data and forest resource inventory data; The single-layer forest carbon stock accounting physical model and the multi-layer forest carbon stock accounting physical model are constructed based on the following method: screening a plurality of forest sampling sites from a plurality of sampling sites in the forest resource inventory data using the multispectral image data; constructing modeling samples and verification samples according to each of the forest sampling sites and the canopy types corresponding to the forest sampling sites; Model parameters are calculated based on the multi-source data, and a single-layer forest carbon stock accounting physical model and a multi-layer forest carbon stock accounting physical model are respectively constructed according to the model parameters, the modeling samples and the verification samples.
5. The method according to claim 4, wherein the calculating model parameters based on the multi-source data comprises: Calculating the canopy entropy of the sampling site based on the sampling range of the forest sampling site and combining the point cloud data; Performing interpolation processing based on the point cloud data to obtain a digital surface model and a digital elevation model of the target area; Calculating the average height of trees in the forest sampling site based on the difference between the digital surface model and the digital elevation model; Calculating forest cover parameters based on the multispectral image data; Calculating the carbon storage of the forest sampling site based on the spatial distribution range of the single-layer forest and the spatial distribution range of the multi-layer forest; The canopy entropy of the sampling site, the average tree height of the forest sampling site, the forest cover parameter and the carbon storage of the forest sampling site are used as the model parameters.
6. The method according to claim 5, wherein the calculating the average height of trees in the forest sampling site based on the difference between the digital surface model and the digital elevation model comprises: Determining a height parameter of the target area based on a standard grid based on a difference between the digital surface model and the digital elevation model; For the sampling range of the forest sampling site, the average height of trees in the forest sampling site is obtained based on the height parameter statistics of the standard grid.
7. The method according to claim 5, wherein calculating forest cover parameters based on the multispectral image data comprises: Calculating the normalized vegetation index of the target area using the multispectral image data; Calculating a kernel normalized difference vegetation index based on the normalized difference vegetation index; Calculate the average kernel normalized difference vegetation index within the sampling range of the forest sampling site; Forest cover parameters are calculated based on the mean kernel normalized difference vegetation index.
8. The method according to claim 5, wherein the carbon storage of the forest sampling site is calculated based on the spatial distribution range of the single-layer forest and the spatial distribution range of the multi-layer forest, comprising: Determining the canopy type of each forest sampling site based on the spatial distribution range of the single-layer forest and the spatial distribution range of the multi-layer forest; Based on each of the forest sampling sites and the canopy type corresponding to the forest sampling site, the carbon storage of the forest sampling site is calculated using an allometric growth equation.
9. A device for estimating forest carbon reserves, characterized in that: The device comprises: An acquisition module is used to acquire multi-source data of the target area; wherein the multi-source data includes regional-scale point cloud data and forest, grassland and wetland survey data; A calculation module, configured to calculate the regional grid canopy entropy of the target area based on the point cloud data; wherein the regional grid canopy entropy is used to characterize the complexity of the forest structure in the target area; A partitioning module is used to obtain the forest range data of the target area based on the forest, grassland and wetland survey data, and spatially partition the forest range data based on the regional grid canopy entropy and a pre-generated canopy entropy classification threshold to obtain the spatial distribution range of the single-layer forest and the multi-layer forest in the target area; an estimation module, configured to estimate the carbon stock of a single-layer forest within the spatial distribution range of the single-layer forest based on a physical model for calculating carbon stock of a single-layer forest, and to estimate the carbon stock of a multi-layer forest within the spatial distribution range of the multi-layer forest based on a physical model for calculating carbon stock of a multi-layer forest; A generating module is used to generate forest carbon storage spatial distribution data of the target area based on the single-layer forest carbon storage and the multi-layer forest carbon storage.
10. An electronic device, characterized in that: The method is characterized in that it comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method according to any one of claims 1 to 8 when executing the computer program.
11. A storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.