Foundation laser radar-based binary wood volume equation forest measurement method

Abstract

The invention provides a binary wood volume equation forest measurement method based on a ground-based laser radar. The method comprises the following steps: acquiring complete point cloud information scanned by the ground-based laser radar; processing the point cloud data to obtain tree structure parameters; fitting a plurality of cutting degree equation models through tree structure parameters to obtain a single-tree optimal cutting degree equation model; and a single-wood cutting degree-binary wood volume equation is constructed based on the single-wood optimal cutting degree equation, and the single-wood volume is calculated. According to the method, the traditional under-forest data acquisition work can be simplified, the tree species cutting degree equation can be updated for the sample plot forest more timely and accurately, the permanent damage to the tree caused by ground survey can be reduced, effective technical support is provided for the survey of the stand volume of a man-made forest and the estimation of the forest stand outturn percentage, and the method is suitable for popularization and application. Compared with a traditional method for estimating the volume and the stock of the single wood through a wood stock table, the method is higher in precision and better in flexibility, can be better used in the production practice process, and greatly saves manpower, material resources and financial resources.

Description

Technical Field

[0001] The invention belongs to the field of forest timber volume measurement, and in particular relates to a forest timber measurement method based on a binary timber volume equation using a ground-based laser radar. Background Art

[0002] The traditional method of estimating tree volume is to obtain tree structural parameters by manually or mechanically cutting down a large number of trees, and then calculate the volume through the compiled one-dimensional and two-dimensional volume tables, and then deduce the volume of the entire forest stand. Since the compilation of the volume table is quite old, the traditional method not only cannot update the volume table of the forest in time, but also leads to low accuracy of the estimated volume. In the current volume table, the volume deviation of many tree species has exceeded the error limit of ±3%, which is not suitable for the current requirements for accurate monitoring of forest resources and the status of forest resources.

[0003] At present, taking Mongolian oak as an example, there are few studies on the taper equation and timber volume of some tree species combined with lidar at home and abroad, and there is a lack of fast and effective methods to estimate timber volume. Summary of the invention

[0004] The purpose of the present invention is to provide a forest measurement method based on binary volume equation of ground-based laser radar.

[0005] The technical scheme of the present invention is as follows:

[0006] A method for measuring trees using a binary volume equation based on ground-based laser radar, the method comprising:

[0007] Obtain the complete point cloud information of the sample site scanned by the ground-based laser radar;

[0008] By processing the point cloud data, the trunk structure parameters are obtained;

[0009] The optimal taper equation model for a single tree is obtained by fitting multiple taper equation models through trunk structural parameters.

[0010] Based on the single tree optimal taper equation, the single tree taper-binary volume equation is constructed to calculate the single tree volume.

[0011] Furthermore, the point cloud information scanned by the ground-based laser radar is obtained as follows:

[0012] Set up multiple laser radars at the experimental site;

[0013] Target balls are placed between the laser radar stations so that each laser radar station can see the three target balls without obstruction. The point cloud data obtained by each laser radar is spliced ​​at multiple stations according to the position of the target balls to obtain the complete point cloud information of the experimental site.

[0014] Furthermore, the optimal taper equation model for a single tree is obtained by fitting multiple taper equation models through trunk structural parameters, specifically including:

[0015] Based on the trunk structure parameters, multiple different types of taper equation models are selected as basic models, nonlinear regression fitting model parameters are performed through NLS nonlinear function, and the optimal taper model is selected through multiple indicator factors.

[0016] Furthermore, the trunk structure parameters are obtained as follows:

[0017] First, the complete point cloud information is downsampled and denoised;

[0018] The denoised complete point cloud information is filtered through the CSF algorithm elevation normalization to remove the influence of terrain undulation on the elevation value of the point cloud information;

[0019] The filtered point cloud information is segmented into individual trees and branches and leaves are separated through the distance-based Euclidean clustering algorithm, and the point cloud trunk of the individual tree is extracted. The highest value of the segmented single tree point cloud trunk is taken as the tree height. The trunks of the sample tree point cloud at different heights are sliced ​​and projected onto a two-dimensional plane. The Ransac algorithm is used for circle fitting to extract the breast diameter, center coordinates of the circle and trunk diameter values ​​at different heights.

[0020] Furthermore, the multiple different types of taper equation models include a simple taper equation, a piecewise taper equation and a variable parameter taper equation.

[0021] Furthermore, the multiple index factors are the determination coefficient R 2 , root mean square error RMSE, mean absolute error MAE and Akaike information criterion AIC.

[0022] Furthermore, the single wood taper-binary volume equation is:

[0023]

[0024] Where V, D, and H are the volume, DBH, tree height, and the height from the trunk base to any diameter, respectively. a1, a2, a3, and a4 are the parameters to be fitted in the model.

[0025] Compared with the prior art, the present invention has the following advantages:

[0026] The present invention can not only simplify the traditional work of collecting understory data, but also update the taper equations for tree species in sample plots more timely and accurately, and can also reduce the permanent damage to trees caused by ground surveys, providing effective technical support for the stock volume survey of artificial forests and the estimation of stand yield. Compared with the traditional method of estimating the volume and stock volume of individual trees through volume tables, the present invention has higher accuracy and better flexibility, can be better used in production practice, and greatly saves manpower, material and financial resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The accompanying drawings generally illustrate various embodiments by way of example and not limitation, and together with the description and claims, serve to illustrate the embodiments of the invention. Where appropriate, the same reference numerals are used throughout the drawings to refer to the same or similar parts. Such embodiments are illustrative and are not intended to be exhaustive or exclusive embodiments of the present apparatus or method.

[0028] Figure 1 A schematic diagram of the process of the present invention is shown;

[0029] Figure 2 A schematic diagram of extracting a tree trunk by separating branches and leaves of the present invention is shown;

[0030] Figure 3 A schematic diagram of extracting the single tree diameter at breast height using the Ransac algorithm of the present invention is shown;

[0031] Figure 4 A schematic diagram of matching a single wood part of the present invention is shown;

[0032] Figure 5 A schematic diagram showing a comparative analysis of the actual measured value and the estimated value of the height of Mongolian oak of the present invention is shown;

[0033] Figure 6 A schematic diagram showing the correlation and error comparison analysis between the Mongolian Oak taper-binary volume equation model of the present invention and other volume calculation methods. DETAILED DESCRIPTION

[0034] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0035] 1.1 Overview of the study area

[0036] The study area is located in the Urban Forestry Demonstration Base of Northeast Forestry University (126°634′~126°637′E, 45°72′~45°73′N). It is located in Harbin City. The climate type is a temperate continental monsoon climate with long winters and short summers. The average annual precipitation is 569.1mm. Precipitation is mainly concentrated in June-September, accounting for 60% of the annual precipitation in summer. The concentrated snowfall period is from November to January of the following year. The annual average temperature is 5.5℃. The terrain of the experimental forest farm is undulating. The main tree species are Mongolian oak (Quercusmongolica), Korean pine (Pinus koraiensis), Dahurian larch (Larix gmelinii) and Manchurian ash (Fraxinusmandschurica Rupr). The Mongolian oak sample plot used is high in the northeast and low in the southwest, with fewer shrubs and weeds under the forest.

[0037] 1.2 Data Acquisition

[0038] The sample plot was selected from the Inner Mongolian Oak sample plot of the Experimental Forest Farm of Northeast Forestry University. Taking the Mongolian Oak sample plot as an example, the Mongolian Oak sample plot was planted with direct seeding in the spring of 1959. The overall area of ​​the Mongolian Oak sample plot is about 40m×40m. According to the experimental needs, it was divided into four 20m×20m sample plots for data acquisition.

[0039] 1.2.1 Sample plot measurement Obtain the actual measurement data of the standing trees in the sample plot. During the measurement of the standing tree diameter at breast height, use a breast diameter ruler to measure three times at the 1.3m position of the standing tree trunk, record and take the average value as the measured breast diameter of the standing tree. If there are knots, bumps or other abnormal conditions at breast height, use the average value of the place where the upper and lower distances are equal and the trunk shape is relatively normal to calculate the breast diameter. During the actual measurement of the standing tree height and standing tree position, use the forest locator to obtain the highest point of the standing tree trunk and the relative position of the standing tree. The value of the highest point of the standing tree trunk is measured three times and the average value is taken as the tree height value. The distance and angle between the standing tree and the forest locator are used as the basis for calculating the relative position of the standing tree in the measurement plot.

[0040] 1.2.2 Laser scanning data acquisition Laser scanning data was collected in April 2022. The laser scanning data acquisition equipment was FARO (Focus 3D×330), with a scanning speed of 976,000 points / s, a scanning range of 0.6-330m, a wavelength of 550nm, and an infrared laser as the scanning light source. FARO laser radar was used to obtain point cloud data of the sample plots in 4 experimental sample plots. Five laser radar stations were set up in each sample plot, and target balls were placed between the laser radar stations. The point cloud data of the experimental sample plots obtained were spliced ​​at multiple stations according to the position of the target balls to obtain complete point cloud information of the sample plots.

[0041] 1.3 Data processing

[0042] There are 251 standing trees in the sample plot. According to the principle of 1:2 experimental sample selection, 80 standing trees were selected from 4 experimental plots, and the other 171 standing trees were selected as alternative test samples to verify the accuracy of the established taper equation model. The point cloud data storage format obtained by FARO was converted into las format, and then the multi-site cloud was aligned and spliced ​​by the position of the target ball in SCENE software. The obtained complete point cloud data was imported into the compiled Cloudcompare2.12.4 software for sample point cloud clipping.

[0043] 2 Research Methods

[0044] 2.1TLS single tree parameter extraction

[0045] 2.1.1 Parameter extraction

[0046] After preprocessing the point cloud data obtained by TLS, the Cloth Simulation Filter (CSF) algorithm was used to extract the DEM data, and the elevation normalization was performed to remove the influence of terrain undulation on the elevation value of the point cloud data, and the point cloud data was filtered. The single tree segmentation was performed using the distance-based Euclidean clustering algorithm, and the branches and leaves of the standing trees were separated based on density and curvature. The point cloud trunk of the single tree was extracted, and the highest z value of the segmented single tree point cloud trunk was used as the tree height. The trunks of the sample tree point cloud at different heights were sliced ​​and projected onto a two-dimensional plane. The slice thickness was 10 cm. The Ransac algorithm was used to perform circle fitting to extract the breast diameter and the center coordinates of the circle, and the relative position of the standing trees was obtained through the forest locator.

[0047] 2.2 Evaluation of single tree parameter extraction accuracy

[0048] The data of DBH and tree height measured under the forest were compared with the DBH and tree height extracted from the point cloud data obtained by LiDAR, and the diameter at relative height extracted from TLS point cloud was compared with the diameter calculated by the Mongolian oak taper equation model. The determination coefficient (R 2 ) and root mean square error (RMSE) were used to evaluate the accuracy of tree height, breast diameter and relative height diameter extracted from TLS data.

[0049] 2.3 Model construction

[0050] 2.3.1 Construction of the taper equation model

[0051] The taper equations that describe the changes in tree trunk shape are divided into three categories: simple taper equations, segmented taper equations, and variable parameter taper equations. The overall trunk shape of Quercus mongolica is relatively sharp, and the diameter decreases from bottom to top. Due to the particularity of Quercus mongolica, the trunk shape changes greatly above the diameter at breast height. According to previous studies on the trunk shape of Quercus mongolica, the changes in the number of breast-height shapes are affected by both the diameter at breast height and the tree height. The characteristics of the changes in the number of breast-height shapes can be described by the function When selecting and constructing the taper equation, considering the characteristics of stem shape change, 7 representative taper equations at home and abroad were selected, as shown in Table 1.

[0052] Table 1 Taper equation model

[0053]

[0054]

[0055] Note: a0, a1, a2, ..., a8 are the parameters to be estimated; d is the trunk diameter corresponding to any relative height of the trunk; D is the breast diameter; H is the tree height; h is the height from the base of the trunk to any diameter; q=h / H, Z=(H-h) / H, X=(1-q(1 / 3)) / (1-(1.3 / H)(1 / 3)), L=H-h, t=1.3 / h.

[0056] According to the principle of 1:2 experimental sample selection, 80 analytical trees were randomly selected from the four experimental plots, and the other 171 trees were used as alternative test samples. The randomly selected standing trees already included the measured data under the forest (DBH and tree height). The model data characteristics are shown in Table 2. The taper equation model was fitted with the nls function nonlinearly using R language to fit the model parameters.

[0057] Table 2 Model data feature statistics

[0058]

[0059] 2.3.2 Accuracy test and evaluation of the model The accuracy test and evaluation of the taper equation model is carried out using the coefficient of determination (R 2 ), root mean square error (RMSE), mean absolute error (MAE) and Akaike information criterion (AIC) are used as indicator factors to evaluate whether the model is optimal.

[0060] 2.3.3 Estimation of volume of a single timber

[0061] Using the taper equation model proposed by Meng Xianyu (1982) to integrate the two-variable volume quadrature formula, integrating the taper equation from 0 to H yields

[0062]

[0063] The total volume equation of the trunk. That is, formula (1). Using the measured tree height and breast diameter data, the two-variable volume equation of Quercus mongolica obtained by the taper equation is compared with the three current methods for calculating the volume of Quercus mongolica. The two-variable volume equation of the taper equation is compared with the one-variable and two-variable volume equations of northern Quercus mongolica standing trees, such as formulas (2) and (3), and the volume calculated by the central cross-sectional area, with a segment length of 1m, such as formula (4). Compare the differences with the three and conduct a significance test. Use the correlation coefficient R 2 Represents the correlation with the other three volume estimation methods.

[0064]

[0065] V1=0.0006*D 2.003 (2)

[0066] V2=0.000062*D 1.65438 *H 1.14264 (3)

[0067]

[0068] Where: g i is the central cross-sectional area of ​​the ith section; l is the length of the section; g' is the cross-sectional area at the bottom of the tip; l' is the length of the tip; V, D, H are the timber volume, DBH and tree height, respectively.

[0069] 3 Results Analysis

[0070] 3.1 Comparative analysis of TLS extracted parameters and measured values

[0071] 3.1.1 Single Tree Extraction Accuracy After the preprocessing of the point cloud data obtained by the LiDAR, the single tree matching was performed by the forest locator and the tree location information map obtained by TLS. The forest locator obtained the relative position of the standing trees and found the standing trees based on the DBH position through distance and angle. The number of trees measured in the sample plot was 251, and 239 single trees were successfully matched, with a matching accuracy of 95.21%.

[0072] 3.1.2 Accuracy of DBH and Tree Height Extraction The equation model was constructed based on the structural parameters of the 80 selected modeling trees. The average DBH of the Mongolian oak in the entire experimental plot was 16.44 cm, and the average tree height was 15.05 m. The measured average DBH of the Mongolian oak selected for modeling was 16.79 cm, and the measured average tree height was 15.60 cm. The average DBH of the Mongolian oak selected for modeling obtained through TLS point cloud data was 16.83 cm, and the average tree height was 15.86 m. The measured DBH and tree height of the four plots were compared with the TLS estimated DBH and tree height. Figure 5 shown.

[0073] During the actual measurement, it was found that the points with large deviations were due to the presence of knots and depressions at the breast diameter of some tree trunks, which caused some deviations between the measured values ​​and the values ​​extracted from the TLS point cloud. The R language was used to detect the discrete points of the data, and the breast diameter with deviations was adjusted by visual inspection under the forest and the standard deviation discrimination method. The actual standard under the forest was used for points exceeding two times the standard deviation and they were eliminated. There is a certain deviation between the tree height extracted from the point cloud data and the actual tree height. The reason for the deviation is that when the tree height is measured manually under the forest, sometimes the leaves block it, and it is impossible to find the highest point of the trunk very accurately. The deviation is large and recorded as an abnormal point. However, the problem can be eliminated by eliminating discrete points using the double standard deviation method.

[0074] Taking the measured DBH and tree height as an example, the estimated data refer to the structural parameters obtained by the point cloud data obtained by TLS. The measured and estimated data of DBH and tree height of 80 trees in four plots were compared and analyzed, and it was found that the two showed a strong linear relationship with a correlation coefficient R 2 After adjustment, they both reached 0.98 and 0.97, and the root mean square error of the diameter at breast height was adjusted from 0.853cm to 0.564cm. The root mean square error of the tree height was also adjusted from 0.189m to 0.166m, which also shows that the data obtained this time is of high accuracy and the error caused by the data processing method is small, and the point cloud data can be used for trunk diameter extraction research.

[0075] 3.2 Model fitting accuracy analysis

[0076] According to the characteristics of the sample plots, the average tree height of the four sample plots was between 12-21m. The study chose to extract the trunk diameters of 12 different heights of the trunks of the four sample plots to fit the taper equation. According to the characteristics of the trunk shape of Mongolian oak, the taper equation was constructed using the diameters of 1.3m and relative heights of 0.1H, 0.2H, 0.3H, 0.4H, 0.5H, 0.6H, 0.7H, 0.8H and 0.9H, respectively.

[0077] By fitting the taper equation model with the trunk diameter at 11 different heights of the tree trunk, the fitting effect of the basic model is shown in Table 3. The indicators of the fitting model accuracy are R-squared (coefficient of determination), root mean square error (RMSE), mean absolute error (MAE) and Akaike information criterion (AIC). The AIC indicator not only requires the explanatory power of the data, but also comprehensively judges the complexity and goodness of fit of the model. Therefore, the four indicators are used to measure the overall accuracy of the model fitting.

[0078] Table 3 Basic model fitting indicators

[0079] Serial number Equation Type R-squared RMSE MAE AIC 1 Simple 0.9667 0.4488 0.4043 27.20 2 Simple 0.9509 0.6511 0.5792 25.71 3 Simple 0.8943 0.9535 0.8564 30.25 4 Variable Index 0.9925 0.3813 0.2721 27.86 5 Variable Index 0.9783 0.4351 0.3926 29.35 6 Variable Index 0.9750 0.4646 0.3556 27.06 7 Segmentation 0.9619 0.5735 0.4830 26.22

[0080] Among them, the variable exponential taper equation of Meng Xianyu (1982) with sequence number 4 in the variable exponential equation has a better comprehensive fitting effect on the change of the stem curve of Quercus mongolica in the experimental plot, with R-squared reaching 0.99, RMSE of 0.38, MAE of 0.27, and AIC of 27.86. The model parameters are shown in Table 4 below.

[0081] Table 4 Parameters of taper equation

[0082]

[0083]

[0084] 3.3 Verification of stem diameter extraction at different heights

[0085] The accuracy of the constructed Mongolian oak taper equation model and the trunk diameter structural parameters extracted from the point cloud data of 20 random test samples was verified. The volume analysis of 80 random trees in the alternative test sample data in the experimental plot was compared with the one-dimensional and two-dimensional volume equations of northern Mongolian oak standing trees, the central cross-sectional area method and the two-dimensional quadrature of the Mongolian oak taper equation model. After comparing the accuracy of the four quadrature equations, the correlation coefficients of the tapered equation model of Meng Xianyu (1982) with the one-dimensional, two-dimensional and central cross-sectional area quadrature of Mongolian oak were 0.982, 0.986 and 0.991 respectively. There was no significant difference between the two-dimensional quadrature of the taper equation and the other three methods for calculating the volume (P>0.1), which proved that the two-dimensional quadrature derived from the taper equation of Mongolian oak had a high accuracy in calculating the volume of a single tree, and could adapt to different regional environments by changing parameters, and could be directly used in actual production practice activities.

[0086] The present invention utilizes the characteristics of ground-based laser radar to quickly, efficiently and accurately obtain target point cloud data, obtains point cloud information of Inner Mongolian oak species in the sample plot through on-site understory scanning, and non-destructively extracts forest structure parameters, wherein the correlation coefficients between the diameter at breast height and tree height extracted from the standing tree point cloud and the measured diameter at breast height and tree height reach 0.98 and 0.97, respectively, indicating that the extraction of forest structure parameters by ground-based laser radar can replace the actual measurement under the forest. Based on the ground-based laser radar point cloud data of the experimental forest of Northeast Forestry University, the structural parameters of a single tree are extracted, and the taper equation of the tree species is constructed using the nls nonlinear fitting function of the R language. The constructed taper equation and volume equation can be well used to estimate the change in trunk diameter and volume of Mongolian oak species, respectively. Therefore, the binary volume equation forest measurement method based on the ground-based laser radar of the present invention has a good application value in processing TLS data obtained under the forest, establishing the taper equation model, estimating the standing timber volume, and investigating forest resources.

[0087] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for measuring trees using a binary volume equation based on ground-based laser radar, characterized in that: The method comprises: Obtain complete point cloud data of the sample site scanned by ground-based laser radar; By processing the point cloud data, the trunk structure parameters are obtained; The optimal taper equation model for a single tree is obtained by fitting multiple taper equation models through trunk structural parameters. Based on the single tree optimal taper equation, the single tree taper-binary volume equation is constructed to calculate the single tree volume.

2. The method for measuring trees using binary volume equation based on ground-based laser radar according to claim 1 is characterized in that: The specific method for obtaining the point cloud data of the sample site scanned by the ground-based laser radar is as follows: Set up multiple laser radars at the experimental site; A target ball is placed between the laser radar stations, and the point cloud data obtained by each laser radar is spliced ​​at multiple stations according to the position of the target ball to obtain the complete point cloud information of the experimental site.

3. The method for measuring trees using binary volume equation based on ground-based laser radar according to claim 1, characterized in that: The optimal taper equation model for a single tree is obtained by fitting multiple taper equation models with trunk structural parameters, including: Based on the trunk structure parameters, multiple different types of taper equation models are selected as basic models, nonlinear regression fitting model parameters are performed through NLS nonlinear function, and the optimal taper model is selected through multiple indicator factors.

4. The method for measuring trees using binary volume equation based on ground-based laser radar according to claim 1, characterized in that: The trunk structure parameters are as follows: First, the complete point cloud information is downsampled and denoised; The denoised complete point cloud information is filtered through the CSF algorithm elevation normalization to remove the influence of terrain undulation on the elevation value of the point cloud information; The filtered point cloud information is segmented into individual trees and branches and leaves are separated through the distance-based Euclidean clustering algorithm, and the point cloud trunk of the individual tree is extracted. The highest value of the segmented single tree point cloud trunk is taken as the tree height. The trunks of the sample tree point cloud at different heights are sliced ​​and projected onto a two-dimensional plane. The Ransac algorithm is used for circle fitting to extract the breast diameter, center coordinates of the circle and trunk diameter values ​​at different heights.

5. The method for measuring trees using binary volume equation based on ground-based laser radar according to claim 3 is characterized in that: The multiple different types of taper equation models include a simple taper equation, a piecewise taper equation and a variable parameter taper equation.

6. The method for measuring trees using binary volume equation based on ground-based laser radar according to claim 3 is characterized in that: The multiple index factors are the determination coefficient R 2 , root mean square error RMSE, mean absolute error MAE and Akaike information criterion AIC.

7. The method for measuring trees using binary volume equation based on ground-based laser radar according to claim 1, characterized in that: The single-tree taper-two-variable volume equation constructed based on the single-tree optimal taper equation is as follows: Where V, D, H, and h are the timber volume, DBH, tree height, and the height from the trunk base to any diameter, respectively; a1, a2, a3, and a4 are the model parameters to be fitted.

Images