Method for measuring wood density of coniferous tree species with different diameter grades

Abstract

The invention discloses a method for measuring the wood density of coniferous tree species with different diameter levels, and relates to the technical field of forest ecology. Compared with the previous method for measuring the wood density of the tree, the method disclosed by the invention solves the problems that the current destructive sampling method is time-consuming and labor-consuming and has huge damage to the tree; the normal growth of the tree is not damaged by collecting the tree core through the growth cone, so that the seedlings and young trees are protected, and the application range is expanded; and relevant data is acquired through a series of measurement operations, and an empirical model is built by combining with improved bat-optimized Bayes, so that the wood density of tree species with different diameter levels can be rapidly measured, and reliable data is provided for forest resource research and management. Meanwhile, consumption of manpower and material resources is reduced, working efficiency is improved, and the method has important significance in promoting sustainable development of forestry.

Description

Technical Field

[0001] The invention relates to the technical field of forest ecology, in particular to a method for measuring the wood density of coniferous tree species with different diameter grades. Background Art

[0002] The wood density of trees is closely related to their growth environment and can reflect their adaptability to climate change. Therefore, wood density is an important indicator for quantitatively assessing tree growth, wood quality, and ecosystem function, and is crucial for constructing and predicting ecological models. Therefore, quickly and accurately measuring the wood density of trees is of great significance for optimizing timber harvesting and utilization strategies and achieving sustainable forest management. It can also provide a scientific basis for predicting and responding to climate change. However, destructive sampling methods are currently commonly used to measure the wood density of trees. Although this method is accurate, it is time-consuming and labor-intensive and causes great damage to trees, especially causing the death of seedlings and even saplings. These limitations limit the applicability of destructive sampling methods. Therefore, a method suitable for quickly, efficiently, and accurately measuring the wood density of trees of different plant sizes under non-destructive conditions is urgently needed.

[0003] In order to solve the above problems, the present invention proposes a method for measuring the wood density of coniferous tree species with different diameter grades. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for measuring the wood density of coniferous tree species of different diameter grades to solve the problems raised in the background technology:

[0005] Currently, destructive sampling is commonly used to determine the wood density of trees. Although this method is accurate, it is time-consuming and labor-intensive and causes great damage to trees, especially causing the death of seedlings and even young trees. These limitations limit the applicability of destructive sampling methods.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] A method for measuring the wood density of coniferous tree species of different diameter grades comprises the following steps:

[0008] S1: Select a study area containing several sampling points and take coniferous tree species in the study area as the research objects;

[0009] S2: Divide the diameter at breast height (DBH) of the selected coniferous tree species into different classes, and collect tree samples within each DBH class range at each sampling point;

[0010] S3: For each sample tree, a tree core was collected using a growth cone;

[0011] S4: For each tree core sampled, the volume of the core was measured using the water displacement method. The core sample was then placed in a 65°C oven and dried. After drying, its dry weight was measured. The wood density of each tree sample was calculated based on the ratio of dry weight to volume.

[0012] S5: 75% of the data were randomly selected to construct an empirical model for predicting the wood density of red pine of different diameter grades, and the remaining 25% of the data were used to verify the prediction accuracy of the empirical model.

[0013] Preferably, the diameter at breast height of the coniferous tree species ranges from 0.2 to 100 cm.

[0014] Preferably, the empirical model of wood density of Korean pine with different diameter grades is specifically constructed as follows:

[0015] Determine the dependent variable y: wood density; independent variable x: tree diameter at breast height; establish a linear relationship y = ω0 + ω1x; obtain a set of data points containing x values ​​and y values ​​(x1, y1), (x2, y2), Λ, (x n ,y n ); For the parameter model ω={ω0,ω1}, assume that the prior distribution is normal distribution, that is, p(ω)~N(μ,δ), where μ is the prior mean vector and δ is the prior covariance matrix;

[0016] Given the data points and model parameters, for the linear regression model, assume that the noise obeys the normal distribution, that is:

[0017]

[0018] in,

[0019] Then the likelihood function is p(y|X,ω)~N(Xω,σ 2 I), where X is a matrix consisting of the diameters at breast height of trees; I is the identity matrix;

[0020] The relationship between the posterior distribution p(ω|y,X) and the prior distribution and likelihood function is: p(ω|y,X)∝p(y|X,ω)p(ω);

[0021] The mean μ of the posterior distribution n and covariance δ n Calculated by the following formula:

[0022]

[0023] μ n =δ n (X T σ -2 y+δ -1 μ)

[0024] Where T is the transpose of the vector;

[0025] The negative logarithm of the posterior distribution is used as the optimization objective function J(ω):

[0026]

[0027] Among them, γ is the offset parameter;

[0028] The parameter model is optimized based on the improved bat optimization algorithm. The position of the individual bat corresponds to the parameter model ω={ω0,ω1} of the empirical model, the speed corresponds to the adjustment amount of the model parameters; the pulse frequency corresponds to the search step size; the pulse loudness corresponds to the acceptance level of the current optimal solution; and the pulse rate corresponds to the balance of the algorithm.

[0029] The improved bat optimization algorithm is as follows:

[0030] In the d-dimensional space, the position and velocity of the individual bat are updated as follows:

[0031] h j =h min +(h max -h min )α

[0032] v j (t) = v j (t-1)+(s j (t)-s best )h j *w

[0033] s j (t) = s j (t-1)+v j (t)

[0034]

[0035] Where h is the pulse frequency; j is the individual; h max and h min are the maximum pulse frequency and the minimum pulse frequency respectively; α is the random vector in the interval; v is the speed of the individual bat; t is the current iteration number; s is the position of the individual bat; s best is the optimal position of the individual bat; w is the inertia weight; w max and w min are the maximum and minimum values ​​of the inertia weight respectively; τ is the maximum number of iterations; ψ is a random number in the interval [0,1];

[0036] During the local search process, a random walk method is used to generate a new optimal solution around the selected optimal solution:

[0037]

[0038] Among them, s odd is the optimal position of the selected bat individual; s new To generate a new optimal solution around the selected optimal solution; β is a random number in the interval [-1,1]; is the average pulse loudness of all bat individuals in the current iteration;

[0039] ε(t+1)=λε(t)

[0040]

[0041] Where ε is the pulse loudness emitted by an individual bat; λ and η are both variation factors; is the pulse rate emitted by individual bats; is the initial pulse rate emitted by the individual bat;

[0042] An empirical model of wood density of red pine with different diameter grades was constructed based on the parameter model ω={ω0,ω1} obtained by improving the individual optimization of bats.

[0043] Preferably, the empirical model for constructing the prediction of wood density of Korean pine of different diameter grades is applicable to coniferous species at 126.8°-130.4° east longitude and 42.6°-49.5° north latitude.

[0044] Preferably, the prediction accuracy is specifically verified as follows:

[0045]

[0046] Among them, FC is the prediction accuracy of wood density; y i (i∈[1,n]) is the measured wood density; is the wood density value predicted based on the empirical model.

[0047] Compared with the prior art, the present invention provides a method for measuring the wood density of coniferous tree species of different diameter grades, which has the following beneficial effects:

[0048] This method uses growth cones to collect cores, preserving the normal growth of trees and protecting seedlings and saplings, thus expanding its scope of application. Furthermore, through a series of measurement operations to collect relevant data, combined with an improved bat-optimized Bayesian empirical model, it can rapidly determine the wood density of trees of different diameters, providing reliable data for forest resource research and management. Furthermore, it reduces labor and material resources, improves work efficiency, and is of great significance for promoting the sustainable development of forestry. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is a flow chart of the method mentioned in Example 1 of the present invention;

[0050] Figure 2 This is a schematic diagram of the diameter at breast height distribution of Korean pine-like trees mentioned in Example 1 of the present invention;

[0051] Figure 3 This is a schematic diagram of the empirical model for predicting the density of red pine wood of different diameter grades mentioned in Example 1 of the present invention. DETAILED DESCRIPTION

[0052] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0053] This method uses growth cones to collect tree cores, preserving the normal growth of trees and protecting seedlings and saplings, thus expanding its scope of application. Furthermore, through a series of measurement operations to collect relevant data, combined with an improved bat-optimized Bayesian empirical model, it can rapidly determine the wood density of tree species of different diameters, providing reliable data for forest resource research and management. Furthermore, it reduces labor and material resources, improves work efficiency, and is of great significance for promoting the sustainable development of forestry. Specifically, it includes the following:

[0054] Example 1:

[0055] See also Figure 1-3 The present invention provides a method for measuring the wood density of coniferous tree species of different diameter grades, comprising the following steps:

[0056] S1: Select a study area with several sampling points and focus on coniferous tree species within the study area; the details are as follows:

[0057] In this example, Korean pine was used as the research object. Five sampling points were located throughout the distribution area of ​​broad-leaved Korean pine forests in Northeast my country. From the south to the north, they are the Changbai Mountain National Nature Reserve in Jilin (42.38°N, 128.08°E), the Northeast Yew National Nature Reserve in Muling, Heilongjiang (43.48°N, 130.24°E), the Liangshui National Nature Reserve in Heilongjiang (47.18°N, 128.89°E), the Fenglin National Nature Reserve in Heilongjiang (48.06°N, 129.12°E), and the Shengshan National Nature Reserve in Heilongjiang (49.30°N, 126.48°E).

[0058] S2: Classify the diameter at breast height (DBH) of the selected coniferous species and collect samples within each DBH class range at each sampling point; the details are as follows:

[0059] At the end of August 2019, at each sampling point, sample trees were selected according to four diameter at breast height (DBH) levels, including 0-10 cm, 10-30 cm, 30-60 cm, and 60-100 cm. 10-20 sample trees were selected for each level. Figure 2 , and finally 260 sample trees (DBH ranged from 0.2 to 100 cm) were selected from the five sampling points.

[0060] S3: For each sample tree, collect the core using the growth cone; the details are as follows:

[0061] For each sample tree, a tree core with a length of 1 cm was collected using a growth cone with an inner diameter of 5.15 mm.

[0062] S4: For each tree core sampled, the volume of the core was measured using the water displacement method. The core sample was then placed in a 65°C oven and dried. After drying, its dry weight was measured. The wood density of each tree sample was calculated based on the ratio of dry weight to volume. The details are as follows:

[0063] For each tree core sampled, the volume of the core was first measured using the drainage method after it was brought back to the laboratory. 3 ), then put the tree core samples into a 65℃ oven, dry them and measure their dry weight (g). The wood density of each sample tree was calculated based on the ratio of dry weight to volume, with a total of 260 data.

[0064] S5: 75% of the data was randomly selected to construct an empirical model for predicting the wood density of red pine of different diameters, and the remaining 25% of the data was used to verify the prediction accuracy of the empirical model. The details are as follows:

[0065] 75% of the data were randomly selected to construct an empirical model for predicting the wood density of red pine of different diameters. The empirical model for the wood density of red pine of different diameters was constructed as follows:

[0066] Determine the dependent variable y: wood density; independent variable x: tree diameter at breast height; establish a linear relationship y = ω0 + ω1x; obtain a set of data points containing x values ​​and y values ​​(x1, y1), (x2, y2), Λ, (x n ,y n ); For the parameter model ω={ω0,ω1}, assume that the prior distribution is normal distribution, that is, p(ω)~N(μ,δ), where μ is the prior mean vector and δ is the prior covariance matrix;

[0067] Given the data points and model parameters, for the linear regression model, assume that the noise obeys the normal distribution, that is:

[0068]

[0069] in,

[0070] Then the likelihood function is p(yX,ω)~N(Xω,σ 2 I), where X is a matrix consisting of the diameters at breast height of trees; I is the identity matrix;

[0071] The relationship between the posterior distribution p(ω|y,X) and the prior distribution and likelihood function is: p(ω|y,X)∝p(y|X,ω)p(ω);

[0072] The mean μ of the posterior distribution n and covariance δ n Calculated by the following formula:

[0073]

[0074] μ n =δ n (X T σ -2 y+δ -1 μ)

[0075] Where T is the transpose of the vector;

[0076] The negative logarithm of the posterior distribution is used as the optimization objective function J(ω):

[0077]

[0078] Among them, γ is the offset parameter;

[0079] The parameter model is optimized based on the improved bat optimization algorithm. The position of the individual bat corresponds to the parameter model ω={ω0,ω1} of the empirical model, the speed corresponds to the adjustment amount of the model parameters; the pulse frequency corresponds to the search step size; the pulse loudness corresponds to the acceptance level of the current optimal solution; and the pulse rate corresponds to the balance of the algorithm.

[0080] The improved bat optimization algorithm is as follows:

[0081] In the d-dimensional space, the position and velocity of the individual bat are updated as follows:

[0082] h j =h min +(h max -h min )α

[0083] v j (t) = v j (t-1)+(s j (t)-s best )h j *w

[0084] sj (t) = s j (t-1)+v j (t)

[0085]

[0086] Where h is the pulse frequency; j is the individual; h max and h min are the maximum pulse frequency and the minimum pulse frequency respectively; α is the random vector in the interval; v is the speed of the individual bat; t is the current iteration number; s is the position of the individual bat; s best is the optimal position of the individual bat; w is the inertia weight; w max and w min are the maximum and minimum values ​​of the inertia weight respectively; τ is the maximum number of iterations; ψ is a random number in the interval [0,1];

[0087] During the local search process, a random walk method is used to generate a new optimal solution around the selected optimal solution:

[0088]

[0089] Among them, s odd is the optimal position of the selected bat individual; s new To generate a new optimal solution around the selected optimal solution; β is a random number in the interval [-1,1]; is the average pulse loudness of all bat individuals in the current iteration;

[0090] ε(t+1)=λε(t)

[0091]

[0092] Where ε is the pulse loudness emitted by an individual bat; λ and η are both variation factors; is the pulse rate emitted by individual bats; is the initial pulse rate emitted by the individual bat;

[0093] An empirical model of wood density of red pine with different diameter grades was constructed based on the parameter model ω={ω0,ω1} obtained by improving the individual optimization of bats.

[0094] Based on the above construction steps and collected data, the empirical model of wood density of different diameter grades of red pine is finally constructed as follows:

[0095] y=0.009x+0.3452

[0096] Where y is the wood density and x is the tree's diameter at breast height. Figure 2The remaining 25% of the data is used to verify the prediction accuracy of the empirical model. The specific verification of the prediction accuracy is as follows:

[0097]

[0098] Among them, FC is the prediction accuracy of wood density; y i (i∈[1,n]) is the measured wood density; is the wood density value predicted based on the empirical model.

[0099] Among the five sampling points, the highest prediction accuracy was in Fenglin, reaching 99.9%, the lowest prediction accuracy was 84.3%, and the average prediction accuracy was 91.1%-95.6%. For details, please refer to Table 1:

[0100] Table 1 Statistical analysis of the prediction accuracy (FC%) of Korean pine wood density at different sampling points

[0101] Sampling point Maximum Minimum average value Standard deviation Coefficient of variation Changbai Mountain 99.3 84.3 94.2 4.9 5.2 Muling 99.2 89.1 94.5 3.3 3.5 cold water 99 85.1 91.1 4.1 4.5 Fenglin 99.9 87.7 95.6 4.2 4.4 Katsuyama 99 85.6 93.1 4.0 4.3

[0102] In summary, the empirical model proposed in the present invention can accurately predict the wood density of Korean pine of different diameters (DBH = 0.2-100 cm) in different regions, which provides good technical support for the rapid, efficient and accurate determination of the wood density of coniferous species of different diameters under non-destructive conditions.

[0103] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A method for determining the wood density of coniferous tree species of different diameter grades, characterized in that: The steps include: S1: Select a research area containing several sampling points and take coniferous tree species in the research area as the research object; S2: Divide the DBH grades of the selected coniferous tree species and collect sample trees in each DBH grade range at each sampling point; S3: For each sample tree, a tree core was collected using a growth cone; S4: For each sample tree core, the volume of the core was measured using the water displacement method. The core samples were placed in a 65°C oven and dried. The dry weight was measured and the wood density of each sample tree was calculated based on the ratio of dry weight to volume. S5: 75% of the data were randomly selected to construct an empirical model for predicting the wood density of red pine with different diameters, and the remaining 25% of the data were used to verify the prediction accuracy of the empirical model.

2. A method for determining the wood density of coniferous tree species of different diameter grades according to claim 1, characterized in that: The diameter at breast height of the coniferous tree species ranges from 0.2 to 100 cm.

3. A method for determining the wood density of coniferous tree species of different diameter grades according to claim 1, characterized in that: The empirical model of wood density of Korean pine with different diameters is specifically constructed as follows: Determine the dependent variable y: wood density; independent variable x: tree diameter at breast height; establish a linear relationship y = ω0 + ω1x; obtain a set of data points containing x values ​​and y values ​​(x1, y1), (x2, y2), Λ, (x n ,y n ); For the parameter model ω={ω0,ω1}, assume that the prior distribution is a normal distribution, that is, p(ω)~N(μ,δ), where μ is the prior mean vector and δ is the prior covariance matrix; Given the data points and model parameters, for the linear regression model, assume that the noise follows a normal distribution, that is: in, Then the likelihood function is p(y|X,ω)~N(Xω,σ 2 I), where X is a matrix consisting of the diameters at breast height of trees; I is the identity matrix; The relationship between the posterior distribution p(ω|y,X) and the prior distribution and likelihood function is: p(ω|y,X)∝p(y|X,ω)p(ω); The mean μ of the posterior distribution n and covariance δ n Calculated by the following formula: m n =d n (X T s -2 y+d -1 m) Where T is the transpose of the vector; The negative logarithm of the posterior distribution is used as the optimization objective function J(ω): Among them, γ is the offset parameter; Based on the improved bat optimization algorithm, the parameter model is optimized. The position of the individual bat corresponds to the parameter model ω={ω0,ω1} of the empirical model, the speed corresponds to the adjustment amount of the model parameters; the pulse frequency corresponds to the search step size; the pulse loudness corresponds to the acceptance degree of the current optimal solution; the pulse rate corresponds to the balance of the algorithm; The improved bat optimization algorithm is as follows: In the d-dimensional space, the position and velocity of the individual bat are updated as follows: h j =h min +(h max -h min )α v j (t)=v j (t-1)+(s j (t)-s best )h j *w s j (t)=s j (t-1)+v j (t) Where h is the pulse frequency; j is the individual; h max and h min are the maximum pulse frequency and the minimum pulse frequency respectively; α is the random vector in the interval; v is the speed of the individual bat; t is the current iteration number; s is the position of the individual bat; s best is the optimal position of the individual bat; w is the inertia weight; w max and w min are the maximum and minimum values ​​of the inertia weight respectively; τ is the maximum number of iterations; ψ is a random number in the interval [0,1]; During the local search process, a random walk is used to generate a new optimal solution around the selected optimal solution: Among them, s odd is the optimal position of the selected individual bat; s new To generate a new optimal solution around the selected optimal solution; β is a random number in the interval [-1,1]; is the average pulse loudness of all bat individuals in the current iteration; ε(t+1)=λε(t) Among them, ε is the pulse loudness emitted by individual bats; λ and η are both variation factors; is the pulse rate emitted by individual bats; is the initial pulse rate emitted by individual bats; Based on the parameter model ω={ω0,ω1} obtained by improving the individual optimization of bats, an empirical model of the wood density of red pine of different diameters was established.

4. A method for determining wood density of coniferous tree species of different diameter grades according to claim 3, characterized in that: The empirical model for constructing the prediction of wood density of Korean pine of different diameter grades is applicable to coniferous tree species at 126.8°130.4° east longitude and 42.6°49.5° north latitude.

5. A method for determining wood density of coniferous tree species of different diameter grades according to claim 4, characterized in that: The prediction accuracy is specifically verified as follows: Among them, FC is the prediction accuracy of wood density; y i (i∈[1,n]) is the measured wood density; is the wood density value predicted based on the empirical model.

Images