Method for estimating single-tree biomass of different genotypes of catalpa bignonial based on unmanned aerial vehicle remote sensing

Abstract

The application discloses a method for estimating the biomass of different genotypes of Catalpa speciosa individual trees based on unmanned aerial vehicle remote sensing, which comprises the following steps: acquiring unmanned aerial vehicle LiDAR data and ground measurement data of the biomass of Catalpa speciosa in a preset area; constructing and screening a biomass inversion basic model according to the unmanned aerial vehicle LiDAR data and the ground measurement data of the biomass of Catalpa speciosa; introducing a dummy variable model to optimize the basic model, taking the genotypes of Catalpa speciosa as random effects, and constructing a nonlinear mixed effect individual tree biomass estimation model based on unmanned aerial vehicle LiDAR; and inputting the unmanned aerial vehicle LiDAR data of different genotypes of Catalpa speciosa in a region to be processed into the nonlinear mixed effect individual tree biomass estimation model based on unmanned aerial vehicle LiDAR data obtained above, so as to obtain the biomass of Catalpa speciosa individual trees. The technical scheme of the application can effectively estimate the biomass of different genotypes of Catalpa speciosa individual trees in a large range, and can assist in Catalpa speciosa breeding.

Description

Technical Field

[0001] This invention belongs to the field of remote sensing image processing technology, and in particular relates to a method for estimating the biomass of individual catalpa trees of different genotypes based on UAV remote sensing. Background Technology

[0002] Forests, as the main body of terrestrial ecosystems, play an irreplaceable role in maintaining regional ecological environments and global carbon balance, and mitigating global warming. Biomass is a key biophysical parameter for assessing and simulating terrestrial carbon storage and dynamic changes, and is of great significance for ecological research, developing forestry management strategies, and evaluating forest growth dynamics and carbon cycling. Tree height and biomass, as two basic variables for predicting tree biomass, can usually be obtained through ground measurements, but this process can be both time-consuming and costly. Meanwhile, accurate biomass assessment is crucial for monitoring forest resources and implementing sustainable development strategies. This paper addresses the difficulty in quantifying the interaction between genetic heterogeneity and environmental factors in estimating the biomass of Catalpa bungei. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a method for predicting the biomass of individual catalpa trees of different genotypes based on UAV remote sensing.

[0004] To achieve the above objectives, the present invention adopts the following technical solution:

[0005] A method for estimating the biomass of individual catalpa trees of different genotypes based on UAV remote sensing includes:

[0006] Step S1: Acquire UAV LiDAR data and ground measurement data of catalpa tree biomass in the preset area;

[0007] Step S2: Construct and screen basic models of single-tree biomass based on UAV LiDAR. A dummy variable model is introduced to optimize the above basic model. Using different genotypes of Catalpa trees as random effects, a nonlinear mixed-effect single-tree biomass prediction model based on UAV LiDAR data is constructed.

[0008] Step S3: Input the UAV LiDAR data of different genotypes of Catalpa trees in the area to be processed into the nonlinear mixed-effect single-tree biomass prediction model based on the UAV LiDAR data obtained above, and obtain the single-tree biomass of Catalpa trees.

[0009] Preferably, in step S1, the UAV LiDAR data is preprocessed, which includes: noise reduction, ground point classification, point cloud normalization and point cloud segmentation, to obtain UAV LiDAR single tree height data and crown width data.

[0010] Preferably, step S2 includes:

[0011] Based on UAV LiDAR data and ground measurement data of Catalpa tree biomass, a basic model of Catalpa tree biomass and LiDAR tree height and crown width was constructed. The basic model forms include: Logistic, Exponential, Power, and Richards model.

[0012] Based on the model's coefficient of determination (R²) 2 The basic model for the biomass inversion of individual catalpa trees was screened using indicators such as root mean square error (RMSE) and total relative error (TRE). The above basic model was optimized by introducing a planting density dummy variable model, resulting in a power function dummy variable biomass model.

[0013] Based on ground-based biomass measurement data of Catalpa trees and LiDAR data of individual tree height and crown width from UAVs, a nonlinear mixed-effects Catalpa tree biomass prediction model based on UAV LiDAR data was constructed, using different genotypes of Catalpa trees as random effects and employing a power function dummy variable biomass model.

[0014] As a preferred approach, the nonlinear mixed-effect model for estimating the biomass of individual catalpa trees based on UAV LiDAR data is as follows:

[0015]

[0016] Where AGB represents the biomass of Catalpa trees, y1 is a dummy variable representing the density of the sample plot, and y1 takes the value of 1 for high planting density and 0 for low planting density; L H For lidar tree height, L CD ε represents the lidar canopy amplitude, β1 to β4 are effect parameters, u1 and u2 are the random effects of the catalpa genotype on β2 and β3, respectively, and ε is the error term.

[0017] This invention is based on tree height (L) extracted from UAV LiDAR data. H ) and crown width (L CD It can effectively predict the biomass of individual catalpa trees of different genotypes; incorporating genotype differences as a random effect into the model can significantly improve prediction accuracy. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating a method for predicting the biomass of individual catalpa trees of different genotypes based on UAV remote sensing, as described in an embodiment of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0022] Example 1:

[0023] like Figure 1 As shown, this embodiment of the invention provides a method for estimating the biomass of individual catalpa trees of different genotypes based on UAV remote sensing, including:

[0024] Step S1: Acquire UAV LiDAR data and ground measurement data of catalpa tree biomass in the preset area;

[0025] Step S2: Construct and screen basic models of single-tree biomass based on UAV LiDAR. A dummy variable model is introduced to optimize the above basic model. Using different genotypes of Catalpa trees as random effects, a nonlinear mixed-effect single-tree biomass prediction model based on UAV LiDAR data is constructed.

[0026] Step S3: Input the UAV LiDAR data of different genotypes of Catalpa trees in the area to be processed into the nonlinear mixed-effect single-tree biomass prediction model based on the UAV LiDAR data obtained above, and obtain the single-tree biomass of Catalpa trees.

[0027] As one embodiment of the present invention, in step S1, a drone is used to collect catalpa tree data in a preset area. The drone model is Bumblebee BB-4, and the lidar model is Rig Laser Radar AS-1300HL.

[0028] Furthermore, the UAV LiDAR data is preprocessed, including noise reduction, ground point classification, point cloud normalization, and point cloud segmentation, to obtain UAV LiDAR data on individual tree height and crown width.

[0029] The noise reduction process is as follows:

[0030] The raw lidar point cloud is denoised to remove outliers caused by sensor errors, atmospheric interference, or non-vegetation objects (such as birds and machinery). The denoising operation is performed by combining a height threshold with point density filtering to effectively improve the overall quality of the point cloud data and the accuracy of subsequent classification.

[0031] Ground points are classified as follows:

[0032] The denoised point cloud data was processed using a ground point classification algorithm, primarily based on gradient changes, echo characteristics, and neighborhood relationships to distinguish between ground and non-ground points. This classification method comprehensively considers local terrain undulation features, enabling accurate extraction of ground points even under complex terrain conditions. This step lays the foundation for subsequent terrain correction and vegetation structure analysis.

[0033] Point cloud normalization is as follows:

[0034] After ground point extraction, all point cloud data are normalized based on the Digital Terrain Model (DTM), that is, the height values ​​of each point are converted to heights relative to the ground surface. The normalization operation eliminates the influence of terrain undulation on the extraction of vegetation parameters such as tree height, thereby ensuring the consistency and comparability of tree height measurements.

[0035] Point cloud segmentation is as follows:

[0036] Normalized point cloud data was used for individual tree segmentation using a Canopy Height Model (CHM) and local maxima detection method. The specific process included generating CHM images, identifying local maxima as potential canopy centers, and combining watershed algorithms or region growing algorithms to delineate individual tree contours. During segmentation, segmentation parameters were adaptively adjusted based on information such as point cloud density, canopy width, and height variations to improve the accuracy of individual tree identification in complex forest stand environments.

[0037] Extracting the LiDAR tree height (L) of the drone H ) and crown width (L CD The data is as follows:

[0038] Tree height is obtained by extracting ground and non-ground points of trees from LiDAR data and calculating the maximum vertical distance from the ground to the treetop; crown width is obtained by identifying non-ground points of the crown, obtaining the area of ​​the projected area of ​​a single tree crown, and calculating the average diameter of a single tree crown.

[0039] As one embodiment of the present invention, step S2 includes:

[0040] Step S21: Extract the UAV LiDAR tree height (L) based on the UAV LiDAR data. H) and crown width (L CD Tree height is obtained by extracting ground and non-ground points of trees from LiDAR data and calculating the maximum vertical distance from the ground to the treetop; crown width is obtained by identifying non-ground points of the crown, obtaining the area of ​​the projected area of ​​a single tree crown, and calculating the average diameter of a single tree crown.

[0041] Step S22: First, organize and clean the data to ensure the drone's LiDAR tree height (L) is correct. H ) and crown width (L CD The consistency and completeness of the data with ground-measured biomass data were ensured. Secondly, a basic regression model was constructed, using the measured biomass of a single tree as the response variable, and the tree height (L) in the LiDAR data as the response variable. H ) and crown width (L CD Using these three variables as independent variables, a nonlinear regression analysis is used to establish the relationship between them. Furthermore, by fitting the model and performing parameter estimation and validation, a basic model capable of accurately predicting individual tree biomass, such as the power function model, is selected.

[0042] Then, a dummy variable model was introduced to consider the impact of environmental factors such as different planting densities or fertilization practices on biomass inversion. Dummy variables were used to adjust the parameters in the basic model, enabling it to more accurately reflect biomass changes under different scenarios. Finally, by combining the optimization of the power function and dummy variables, a more accurate power function dummy variable model was obtained to describe the biomass inversion relationship.

[0043] Step S23: Based on UAV LiDAR data and ground-based biomass measurement data of Catalpa trees, using different genotypes of this tree species as random effects, a nonlinear mixed-effects single-tree biomass prediction model based on UAV LiDAR data is constructed using R software through a power function dummy variable model. The prediction of random effects adopts a method based on the empirical best linear unbiased prediction (EBLUP) theory. First, the tree height (L) from the UAV LiDAR data is obtained. H ) and crown width (L CD Data on variables such as tree species and tree genotype information were used, and the model was introduced as a random effect. Then, a nonlinear power function was used to describe the UAV LiDAR tree height (L). H ) and crown width (L CD The relationship between genotype biomass and measured biomass was investigated, and a mixed-effects model was fitted using the maximum likelihood estimation method. The random effect component of genotype was used to capture the differences in biomass caused by different genotypes. Through model fitting, diagnosis, and validation, a nonlinear mixed-effects model that can accurately predict the biomass of individual trees was finally obtained.

[0044] The nonlinear mixed-effect single-tree biomass prediction model based on UAV Lidar data is as follows:

[0045]

[0046] Where AGB represents the biomass of Catalpa trees; y1 is a dummy variable representing the density of the sample plot, with values ​​of 1 for high planting density and 0 for low planting density; L H For lidar tree height, L CD ε represents the lidar canopy amplitude; β1 to β4 are effect parameters; u1 and u2 are the random effects of the catalpa genotype on β2 and β3, respectively; ε is the error term.

[0047]

[0048] Among them, Γ i =In i In i For n i ×n i The identity matrix.

[0049] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for predicting the biomass of individual catalpa trees of different genotypes based on UAV remote sensing, characterized in that, include: Step S1: Acquire UAV LiDAR data and ground measurement data of catalpa tree biomass in the preset area; Step S2: Construct and screen basic models of single-tree biomass based on UAV LiDAR, introduce dummy variable models to optimize the above basic models, and construct a nonlinear mixed-effect single-tree biomass prediction model based on UAV LiDAR data with different genotypes of catalpa trees as random effects. Step S3: Input the UAV LiDAR data of different genotypes of Catalpa trees in the area to be processed into the nonlinear mixed-effect single-tree biomass prediction model based on UAV LiDAR data obtained above to obtain the single-tree biomass of Catalpa trees. In step S1, the UAV LiDAR data is preprocessed, including: noise reduction, ground point classification, point cloud normalization and point cloud segmentation, to obtain UAV LiDAR data on single tree height and crown width. Step S2 includes: Based on UAV LiDAR data and ground measurement data of Catalpa tree biomass, a basic model of Catalpa tree biomass and LiDAR tree height and crown width was constructed. The basic model forms include: Logistic, Exponential, Power, and Richards model. Based on the model determination coefficient (R²), root mean square error (RMSE), and total relative error (TRE), a basic model for the biomass inversion of individual catalpa trees was selected. The above basic model was optimized by introducing a planting density dummy variable model, resulting in a power function dummy variable biomass model. Based on ground-based biomass measurement data of Catalpa trees and LiDAR data of individual tree height and crown width from UAVs, a nonlinear mixed-effects Catalpa tree biomass prediction model based on UAV LiDAR data was constructed using different Catalpa tree genotypes as random effects and a power function dummy variable biomass model. The nonlinear mixed-effect model for estimating the biomass of individual catalpa trees based on UAV LiDAR data is as follows: ； in, AGB y1 represents the biomass of Catalpa trees, and y1 is a dummy variable for plot density. The value of y1 is 1 for high planting density and 0 for low planting density. L H For lidar single tree height, L CD For single-tree crown width using lidar, β 1 ~ β 4 For effect parameters, and The genotype pairs of Catalpa trees are respectively β 2 and β 3 random effects ε This is the error term.

Images