Method and device for implementing forest canopy density estimation model

Abstract

The embodiment of the present specification provides a method and device for implementing a bamboo forest canopy closure estimation model, wherein the method includes: determining a sample plot and obtaining canopy closure data of the sample plot, obtaining soil samples and measuring nutrient content, and obtaining forest stand data and terrain factor data of the sample plot; based on the soil nutrient content data, forest stand data and terrain factor data, determining the better-fitting prediction variables, i.e., input parameters, of the bamboo forest canopy closure estimation model through variance analysis; based on the canopy closure data of the sample plot, selecting the better-fitting prediction variables and introducing random effect parameters at the regional level, using the optimal basic model to construct a nonlinear mixed-effect canopy closure model NLME-CD for bamboo forests; using four sampling strategies to select the best number of sample plots in each region, slope, aspect or altitude, evaluating the best number of sample plots in each region, slope, aspect or altitude predicted by the four sampling strategies CD for calibration, and calibrating the NLME-CD model using the empirical best linear unbiased prediction theory; using the leave-one-out cross-validation LOOCV method to evaluate the NLME-CD model to obtain the optimal NLME-CD model.

Description

Technical Field

[0001] This document relates to the field of computer technology, and in particular to a method and device for implementing a forest canopy density estimation model. Background Art

[0002] Bamboo forests play an important role in climate regulation and ecological function. Moso bamboo grows rapidly and accumulates carbon at high rates, making it considered a significant carbon sink. Its growth and distribution are influenced by site conditions and environmental factors. Moso bamboo canopy density (CD), defined as the vertical projection of the canopy covering the ground, is a crucial medium for the exchange of light, water, and gases between bamboo and the environment, and plays a crucial role in maintaining the function of bamboo forest ecosystems. CD is also a key indicator of bamboo vitality, reflecting individual adaptation strategies to competition, environmental conditions, and climate change.

[0003] In the existing technology, the research focuses on: (1) how to quickly and accurately obtain CD data; (2) the impact of CD on forest growth; (3) the impact of tree species and stand density on CD, etc.

[0004] While forest surveys are time-consuming, labor-intensive, and costly, it is possible to develop canopy density models suitable for predicting stand canopy density using measurements from a limited number of samples representing the growing conditions of the species of interest. However, because the data are collected from growing conditions at different locations, spatial autocorrelation and data heterogeneity are problematic. In such cases, mixed-effects modeling approaches are necessary to mitigate biases caused by these factors. Therefore, developing nonlinear mixed-effects CD models is of great significance.

[0005] Variations in site conditions and environmental factors, such as altitude, slope, and aspect, can affect soil temperature, precipitation, and chemical and physical properties. This significantly influences the growth and distribution of trees and bamboo, thus impacting stand structure, such as canopy size, site productivity, and carbon conservation. Because moso bamboo grows rapidly and accumulates more carbon, research into the development of carbon conservation models for moso bamboo is crucial.

[0006] Differences in CD are caused by tree species, climate, soil factors, and competition. Previous studies have primarily focused on estimating CD using small-scale areas, stand factors, and least-squares parameter estimation. However, overfitting can lead to biased estimates, unrepresentative CD estimates due to small areas, and inadequate consideration of the impact of soil nutrient content on CD. These are some of the current challenges in CD estimation. Therefore, considering appropriate factors and parameter methods is a key technical bottleneck that urgently needs to be addressed in estimating canopy density in bamboo forests. Summary of the Invention

[0007] The purpose of the present invention is to provide a method and device for realizing a forest canopy density estimation model, aiming to solve the above-mentioned problems in the prior art.

[0008] The present invention provides a method for implementing a forest canopy density estimation model, comprising:

[0009] Determine a sample plot and obtain canopy density data of the sample plot, obtain soil samples and measure nutrient content, and obtain stand data and terrain factor data of the sample plot;

[0010] Based on the soil sample data, the forest stand data and the terrain factor data, determining the prediction variables, i.e., the input parameters, that have a better fit with the bamboo forest canopy density estimation model through variance analysis;

[0011] Based on the canopy density data of the sample plot, the selected well-fitting predictor variables and the introduction of random effect parameters at the regional level, a bamboo forest canopy density estimation model is constructed using the optimal basic model, wherein the bamboo forest canopy density estimation model is a nonlinear mixed effect canopy density estimation model NLME-CD for moso bamboo forests;

[0012] Four sampling strategies were used to select the optimal number of plots per region, slope, aspect, or altitude. The optimal number of plots per region, slope, aspect, or altitude for calibrating CD predictions for the four sampling strategies was evaluated. The NLME-CD model was calibrated using the empirical best linear unbiased prediction theory.

[0013] The leave-one-out cross-validation (LOOCV) method was used to evaluate the NLME-CD model and obtain the optimal NLME-CD model.

[0014] The present invention provides a device for implementing a forest canopy density estimation model, comprising:

[0015] An acquisition module is used to determine a sample plot and obtain canopy density data of the sample plot, obtain soil samples and measure nutrient content, and obtain forest stand data and terrain factor data of the sample plot;

[0016] A variable determination module is used to determine, based on the soil nutrient content data, the forest stand data and the terrain factor data, a prediction variable that has a better fit with the bamboo forest canopy density estimation model, i.e., an input parameter, through variance analysis;

[0017] A model construction module is used to construct a bamboo forest canopy density estimation model using an optimal basic model based on the canopy density data of the sample plot, the selected well-fitting predictor variables, and the introduction of random effect parameters at the regional level, wherein the bamboo forest canopy density estimation model is a nonlinear mixed effect canopy density estimation model NLME-CD for moso bamboo forests;

[0018] a parameter estimation module for selecting the optimal number of plots per region, slope, aspect, or elevation using the four sampling strategies, evaluating the optimal number of plots per region, slope, aspect, or elevation for calibrating CD predictions using the four sampling strategies, and calibrating the NLME-CD model using empirical best linear unbiased prediction theory;

[0019] The model evaluation module is used to evaluate the NLME-CD model using the leave-one-out cross validation (LOOCV) method to obtain the optimal NLME-CD model.

[0020] The present invention proposes a systematic method for estimating bamboo forest canopy density for the first time, effectively addressing the time-consuming and costly nature of plot surveys, model fitting bias, and the failure to consider soil nutrient content. The method recommends adding nitrogen fertilizer to promote bamboo forest growth and development. The technical solutions of the present invention contribute to a deeper understanding of bamboo forest adaptation strategies to competition and environmental pressures, and can help improve forest management strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate one or more embodiments of this specification or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0022] Figure 1 Flowchart of a method for implementing a bamboo forest canopy density estimation model according to an embodiment of the present invention;

[0023] Figure 2 is a schematic diagram of each bamboo plot used for CD shooting and a schematic diagram of a CD image according to an embodiment of the present invention;

[0024] Figure 3 is a schematic diagram of a software interface of a computing CD according to an embodiment of the present invention;

[0025] Figure 4 Schematic diagram of the effects of aspect, altitude, and slope on the density of a bamboo forest according to an embodiment of the present invention;

[0026] Figure 5 is a schematic diagram of the relationship between the CD of an embodiment of the present invention and the stand variables, soil variables, and environmental variables used in the analysis;

[0027] Figure 6is a diagram illustrating the root mean square error (RMSE) and total relative error (TRE) of an ordinary least squares (OLS) model, a model with a mean response (M-response), and models with four sampling strategies and sample sizes in each region according to an embodiment of the present invention;

[0028] Figure 7 is a diagram illustrating the root mean square error (RMSE) and total relative error (TRE) of an ordinary least squares (OLS) model, a model with a mean response (M-response), and models with four sampling strategies and sample sizes at each altitude according to an embodiment of the present invention;

[0029] Figure 8 is a schematic diagram of the root mean square error (RMSE) and total relative error (TRE) of an ordinary least squares (OLS) model, a model with a mean response (M-response), and a model with four sampling strategies and sample sizes at each slope aspect according to an embodiment of the present invention;

[0030] Figure 9 is a diagram illustrating the root mean square error (RMSE) and total relative error (TRE) of an ordinary least squares (OLS) model, a model with a mean response (M-response), and models with four sampling strategies and sample sizes at each slope according to an embodiment of the present invention;

[0031] Figure 10 2 is a schematic diagram of a device for implementing a bamboo forest canopy density estimation model according to an embodiment of the present invention. DETAILED DESCRIPTION

[0032] In order to enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the technical solutions in one or more embodiments of this specification will be clearly and completely described below in conjunction with the drawings in one or more embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of this document.

[0033] Method Example

[0034] According to an embodiment of the present invention, a method for implementing a bamboo forest canopy density estimation model is provided. Figure 1 FIG. 1 is a flow chart of a method for implementing a bamboo forest canopy density estimation model according to an embodiment of the present invention. Figure 1 As shown, the method for implementing the bamboo forest canopy density estimation model according to an embodiment of the present invention specifically includes:

[0035] Step S101, determining a sample plot and obtaining canopy density data of the sample plot, obtaining soil samples and measuring nutrient content; and obtaining forest stand data and terrain factor data of the sample plot; specifically including:

[0036] Identify the sample plots as areas of bamboo forests that can cover areas with significant changes in bamboo forest density, site productivity, topography, and environment;

[0037] The bamboo forest canopy density was determined by image analysis of vertical snapshots of the forest canopy taken at predetermined times using a digital camera with a fisheye lens;

[0038] Obtaining the total carbon content (TC), organic carbon content (SOC), total nitrogen content (TN), and total phosphorus content (TP) of a soil composite sample, wherein the soil composite sample is obtained by mixing soil samples collected from five randomly selected points in the sample plot, drying them in a cool place in the laboratory, removing roots and debris, and sieving them using a sieve;

[0039] Acquire forest stand data and terrain factor data of the sample plot.

[0040] Step S102, based on the soil sample nutrient content data, forest stand data and terrain factor data, the variance analysis method is used to determine the prediction variables that fit the bamboo forest canopy density estimation model better, that is, the input parameters. The input parameters are specifically: stand density (N), dominant height (DH), basal area (BA), soil total nitrogen content (TN) and organic carbon content (SOC).

[0041] Step S103, based on the canopy density data of the sample plots, selecting the best-fitting predictor variables and introducing the random effect parameters at the regional level, constructing a nonlinear mixed-effect canopy density model (NLME-CD) for moso bamboo forests using the optimal basic model; specifically, the following steps are performed:

[0042] Based on the canopy density data of the plots, the well-fitting predictor variables, and the introduction of random effect parameters at the regional level, a bamboo forest canopy density estimation model was constructed according to the optimal basic model shown in Formula 1:

[0043]

[0044] Where CD represents the density of bamboo forest, β and a are the parameter vectors to be estimated, x is the parameter vector of explanatory variables, including forest stand variables and soil nutrient content variables, i represents the i-th region, and j represents the j-th plot.

[0045] Step S104, using the four sampling strategies to select the optimal number of plots in each region, slope, aspect, or altitude, evaluating the optimal number of plots in each region, slope, aspect, or altitude for calibration and subject-specific CD (four sampling strategies) prediction, and calibrating the NLME-CD model using the empirical best linear unbiased prediction theory; specifically, including:

[0046] Four sampling strategies were used to select the optimal number of plots per region, slope, aspect, or altitude. Based on the selected plots, the Lindstrom and Bates algorithm implemented in the NLME function in R software was used to estimate the optimal number of plots per region, slope, aspect, or altitude for calibrating CD predictions using maximum likelihood estimation. The four sampling strategies were as follows: randomly selecting 1–5 base area (BA) quadrats per region, slope, aspect, or altitude; selecting 1–5 quadrats with the maximum BA per region, slope, aspect, or altitude; selecting 1–5 quadrats with the average BA per region, slope, aspect, or altitude; and selecting 1–5 quadrats with the minimum BA per region, slope, aspect, or altitude.

[0047] Step S105 , using a leave-one-out cross validation (LOOCV) method to evaluate the NLME-CD model and obtain the optimal NLME-CD model.

[0048] In the embodiment of the present invention, after executing step S105, the method further includes:

[0049] Obtain the better-fitting prediction variables of the area to be estimated and introduce the random effect parameters at the regional level, input the better-fitting prediction variables and the regional level random effects into the optimal canopy density basic model, and construct a nonlinear mixed effect canopy density estimation model NLME-CD for bamboo forests with better fitting effect.

[0050] The technical solution described in this embodiment aims to reduce survey costs and accurately estimate bamboo stand canopy density (CD). By evaluating the effects of different altitudes, slopes, slope positions, stand variables, and soil nutrients on bamboo stand canopy density (CD), a nonlinear mixed-effect model incorporating stand factors, topographic factors, and soil nutrients is developed. This model also accurately determines the number of sample plots at different altitudes, slopes, and aspects required for CD calibration and prediction. Quantitatively assessing the response of CD to altitude, slope, and aspect within the Chinese bamboo distribution region, as well as the impact of stand characteristics and soil nutrient content on CD, will provide a deeper understanding of bamboo forest adaptation strategies under competitive and environmental pressures and improve forest management.

[0051] The above technical solutions of the embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0052] To address the technical bottleneck in estimating the canopy density of moso bamboo stands, the embodiment of the present invention sets up sample plots in areas with abundant moso bamboo resources based on the data of the ninth forest inventory, where the bamboo forests are not disturbed by human activities, and conducts a random sampling survey of the bamboo forests. Figure 2 Schematic diagram of each bamboo plot and CD image for CD shooting according to an embodiment of the present invention, wherein the left figure indicates the position where the CD image is shot in each sample figure; Figure 3 Schematic diagram of the software interface of the computing CD according to an embodiment of the present invention. Figure 2 、 3 As shown, each plot was 20 × 20 m in size. Surveys were conducted within the plots to measure bamboo length per bamboo, biomass, dominant height, canopy density, and topographic factors. Slope, aspect ratio (AS), location, latitude and longitude, and elevation (EL) were measured using GPS. Factor analysis was used to construct a correlation model for correlation analysis. The results showed that canopy density (CD) showed a trend of increasing first and then decreasing with increasing altitude, with a turning point at 600–700 m. CD was positively correlated with stand density (N) and basal area (BA), and negatively correlated with dominant height (DH), altitude, slope, total nitrogen (TN), and soil organic carbon (SOC) (p < 0.05). A nonlinear mixed effects (NLME) CD model included stand density (N), dominant height (DH), basal area (BA), soil total nitrogen (TN), and soil organic carbon (SOC) as key predictors. Modeling random effects at the regional level significantly improved the model's fitting accuracy. Empirical best linear unbiased prediction theory was applied to estimate random effects for response calibration. Among several strategies used to estimate random effects, calibration results showed improved accuracy with increasing the number of selected plots. To save measurement costs and time, using two plots with medium basal area (BA) per region, slope, and elevation, or using two plots with the largest basal area (BA) at different slope aspects, provided higher accuracy. The NLME-CD model can reduce field measurement requirements and support forest managers in developing more effective bamboo forest management strategies.

[0053] (1) Determine the terrain factors with greater influence through correlation analysis.

[0054] Three-way analysis of variance (ANOVA) showed that altitude (p=0.0002) and aspect (p=0.0360) had significant effects on canopy density (CD), but their interaction was not significant, and the effect of altitude was stronger than that of aspect.

[0055] The box plots of CD (n=259) at different altitudes, slopes and downslopes are shown in Figure 2. Figure 4As shown in the figure, significant differences exist between sunny, shady, and semi-shady slopes. Across slope gradients and aspects, the overall trend is: shady > semi-shady > semi-sunny > sunny. At different altitudes, CD shows a trend of first decreasing and then increasing (EL6), then decreasing with increasing altitude. CD decreases with increasing slope (although the difference is not significant).

[0056] CD was affected by a combination of stand, soil, and topographic factors. CD was significantly positively correlated with stand density (N) and basal area (BA) (p < 0.05), and negatively correlated with dominant height (DH), altitude, slope, soil total nitrogen (TN), and organic carbon content (SOC) (p < 0.05). However, CD was not significantly correlated with temperature difference (TD) and total carbon content (TC) (p > 0.05).

[0057] (2) Determine the variables retained in the model through variance analysis.

[0058] like Figure 5 As shown, the present embodiment considers the effects of stand variables describing stand size and vitality, competition and site quality, and soil nutrient variables (TC, SOC, TN and TP) on CD. The relationship between variables (matrix plot) and variance influence factors (VIF) was graphically tested, and all significantly correlated variables were excluded. In the CD model, only variables with VIF < 10 were used as predictors. Only five variables: stand density (N), dominant height (DH), basal area (BA), total nitrogen content (TN) and organic carbon content (SOC) of soil were found to have a significant impact on CD changes and were used as predictors to develop a comprehensive nonlinear mixed-effect CD model.

[0059] (3) Model construction.

[0060] A nonlinear mixed-effects (NLME) CD model was developed using the basic model shown in Equation 1, introducing random effects at the regional level (assuming provinces were used as regions). Each NLME model consisted of a fixed effect parameter and regional random effects, and the optimal basic model was used to fit the complete data. The model with the minimum Akaike Information Criterion (AIC) and maximum log-likelihood (LL) was selected for further analysis in this embodiment of the present invention.

[0061]

[0062] β and a are the parameter vectors to be estimated, x is the parameter vector of the explanatory variables, including the stand variables and soil nutrient content variables, i represents the i-th region, and j represents the j-th plot.

[0063] (4) Parameter estimation.

[0064] The parameters in the NLME-CD model were estimated by maximum likelihood using the Lindstrom and Bates (LB) algorithm implemented in the NLME function of R software.

[0065] (5) Response calibration of the NLME CD model.

[0066] The present invention considers two different scenarios for predicting random effects: (1) only the fixed effect part or no random effects are called M responses, and (2) when the NLME model includes random effects estimated from local measurements, it will be a subject-specific or marginal model. The random effects are estimated using the empirical best linear unbiased prediction (EBLUP) theory. To account for the potential variability of CD, the present invention uses the following strategy to select sample plots for each region, slope, aspect, or elevation:

[0067] One to five base area (BA) quadrats were randomly selected for each region, slope, aspect, or elevation;

[0068] 1–5 quadrats with maximum BA per region, slope, aspect, or elevation;

[0069] Average BA of 1–5 quadrats per region, slope, aspect, or elevation;

[0070] Minimum BA 1-5 quadrats per area, slope, aspect or elevation.

[0071] The results showed that regardless of the region, altitude, aspect and slope, as the number of plots increased, RMSE and TRE showed a downward trend (e.g. Figure 6-9 Compared with the other three strategies, sampling in different regions resulted in 1-5 sample points with medium basal area (BA) with the smallest RMSE ( Figure 6 Selecting two plots with medium basal area (BA) yielded the largest reductions in RMSE and TRE, by 10.37% and 30.78%, respectively. Using two plots with medium BA across provinces, slopes, and altitudes, and two plots with the largest BA across different slope aspects, more accurately estimated random effects than other strategies.

[0072] (6) Evaluation of the NLME CD model.

[0073] The effectiveness of the NLME-CD model could be evaluated using an independent dataset. However, since this dataset is unavailable, the present invention employed leave-one-out cross-validation (LOOCV) to validate the fixed portion of the NLME-CD model, as this provides an unbiased error estimate. The results show that the NLME-CD model has higher prediction accuracy than the M response, and the statistics of the former are smaller than those of the latter. This suggests that this region has a greater random influence on CD.

[0074] Finally, the embodiment of the present invention determines that what the present invention actually wants to protect is a method for estimating the density of bamboo forests, which is achieved by the following technical means:

[0075] (1) Sample plot setting: The sample plot layout ensures that it covers the bamboo forest area where the bamboo forest density, site productivity, topography and environment have changed significantly;

[0076] (2) Acquisition of canopy density data: Canopy density (CD) was determined by taking vertical snapshots of the forest canopy between 8:00 and 10:00 a.m. using a digital camera with a fisheye lens. A total of 10 observations were made at 1.5 m above the ground ( Figure 2 ). Use image analysis to obtain CD, and take 10 pictures for each sample ( Figure 2 ,Left).

[0077] (3) Collection of soil samples: From July to September 2023, soil samples were collected from the bamboo forest after the end of high growth using a stainless steel soil drill at a depth of 0 to 30 cm on all sample plots. Each soil sample was collected from five randomly selected points in the sample area, and the samples were mixed together to form a composite sample that could represent the entire sample area for laboratory analysis. The soil samples were dried in a cool place in the laboratory, and roots and debris were removed. The soil samples were then sieved through a 0.15 mm sieve to determine total carbon (TC), soil organic carbon (SOC), total nitrogen (TN), and total phosphorus (TP).

[0078] (4) Variable screening: VIF test.

[0079] (5) Model construction: nonlinear mixed effects model.

[0080] (6) Model calibration: Empirical Best Linear Unbiased Prediction (EBLUP); four sampling strategies.

[0081] (7) Model evaluation: leave-one-out cross validation (LOOCV).

[0082] In summary, with the help of the technical solutions of the embodiments of the present invention, a systematic method for estimating bamboo forest canopy density has been proposed for the first time through systematic research. This effectively solves the problems of time-consuming and labor-intensive plot surveys, high costs, model fitting bias, and failure to consider soil nutrient content. The study recommends adding nitrogen fertilizer to promote the growth and development of bamboo forests. The technical solutions of the embodiments of the present invention help to gain a deeper understanding of the adaptive strategies of bamboo forests in responding to competition and environmental pressures, and help improve forest management strategies.

[0083] Device Example 1

[0084] According to an embodiment of the present invention, a device for implementing a bamboo forest canopy density estimation model is provided. Figure 10 Schematic diagram of a device for realizing a bamboo forest canopy density estimation model according to an embodiment of the present invention. Figure 10 As shown, the device for implementing the bamboo forest canopy density estimation model according to an embodiment of the present invention specifically includes:

[0085] The acquisition module 110 is used to determine the sample plot and obtain the density data of the sample plot, obtain soil samples and measure the nutrient content; and obtain the forest stand data and terrain factor data of the sample plot. The acquisition module 110 is specifically used to:

[0086] Identify the sample plots as areas of bamboo forests that can cover areas with significant changes in bamboo forest density, site productivity, topography, and environment;

[0087] The bamboo forest canopy density was determined by image analysis of vertical snapshots of the forest canopy taken at predetermined times using a digital camera with a fisheye lens;

[0088] Obtaining the total carbon content (TC), organic carbon content (SOC), total nitrogen content (TN), and total phosphorus content (TP) of a soil composite sample, wherein the soil composite sample is obtained by mixing soil samples collected from five randomly selected points in the sample plot, drying them in a cool place in the laboratory, removing roots and debris, and sieving them using a sieve;

[0089] obtaining forest stand data and topographic data of the sample plot;

[0090] The variable determination module 112 is used to determine the prediction variables, i.e., input parameters, that have a good fit with the bamboo forest canopy density estimation model based on the soil nutrient content data, the forest stand data, and the terrain factor data through variance analysis. The variable determination module 112 is specifically used to:

[0091] Based on the soil sample nutrient content data, stand data and terrain factor data, the variance analysis method is used to determine the better fitting prediction variables of the bamboo forest canopy density estimation model, i.e., the input parameters, wherein the input parameters are specifically: stand density (N), dominant height (DH), basal area (BA), soil total nitrogen content (TN) and organic carbon content (SOC);

[0092] The model construction module 114 is used to construct a bamboo forest canopy density estimation model using the optimal basic model based on the canopy density data of the sample plot, the selected well-fitting predictor variables, and the introduction of regional-level random effect parameters, wherein the bamboo forest canopy density estimation model is a nonlinear mixed-effect canopy density estimation model NLME-CD for moso bamboo forests; the model construction module 114 is specifically used to:

[0093] Based on the canopy density data of the plots, the well-fitted predictor variables, and the introduction of regional-level random effect parameters, a bamboo forest canopy density estimation model was constructed according to the optimal basic model shown in Formula 1:

[0094]

[0095] Where CD represents the density of bamboo forest, β and a are the parameter vectors to be estimated, x is the parameter vector of explanatory variables, including forest stand variables and soil nutrient content variables, i represents the i-th region, and j represents the j-th plot;

[0096] The parameter estimation module 116 is configured to select the optimal number of plots in each region, slope, aspect, or altitude using the four sampling strategies, evaluate the optimal number of plots in each region, slope, aspect, or altitude for calibrating CD predictions using the four sampling strategies, and calibrate the NLME-CD model using the empirical best linear unbiased prediction theory. The parameter estimation module 116 is specifically configured to:

[0097] Four sampling strategies were used to select the optimal number of plots for each region, slope, aspect, or altitude. Based on the selected plots, the Lindstrom and Bates algorithm implemented in the NLME function of R software was used to evaluate the optimal number of plots for each region, slope, aspect, or altitude for calibration and prediction of subject-specific CD (four sampling strategies) by maximum likelihood estimation. The four sampling strategies were as follows: randomly selecting 1–5 base area (BA) plots for each region, slope, aspect, or altitude; taking 1–5 plots with the maximum BA for each region, slope, aspect, or altitude; taking 1–5 plots with the average BA for each region, slope, aspect, or altitude; and taking 1–5 plots with the minimum BA for each region, slope, aspect, or altitude.

[0098] The model evaluation module 118 is used to evaluate the NLME-CD model using the leave-one-out cross validation (LOOCV) method to obtain the optimal NLME-CD model.

[0099] The device further comprises:

[0100] The estimation module is used to obtain the well-fitting prediction variables of the estimated area and introduce the random effect parameters at the regional level, input the well-fitting prediction variables and the regional level random effects into the optimal canopy density basic model, and construct the nonlinear mixed effect canopy density estimation model NLME-CD with good fitting effect.

[0101] The embodiment of the present invention is an apparatus embodiment corresponding to the above-mentioned method embodiment. The specific operations of each module can be understood by referring to the description of the method embodiment, which will not be repeated here.

[0102] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for implementing a bamboo forest canopy density estimation model, characterized in that: include: Determine a sample plot and obtain canopy density data of the sample plot, obtain soil samples and measure nutrient content, and obtain stand data and terrain factor data of the sample plot; Based on the soil sample data, the forest stand data and the terrain factor data, determining the prediction variables, i.e., input parameters, that best fit the bamboo forest canopy density estimation model through variance analysis; Based on the canopy density data of the sample plot, the selected best-fitting predictor variables, and the introduction of regional-level random effect parameters, a bamboo forest canopy density estimation model was constructed using the optimal basic model, wherein the bamboo forest canopy density estimation model was the nonlinear mixed-effect canopy density estimation model NLME-CD for moso bamboo forests; Four sampling strategies were used to select the optimal number of plots per region, slope, aspect, or altitude. The optimal number of plots per region, slope, aspect, or altitude for calibrating CD predictions for the four sampling strategies was evaluated. The NLME-CD model was calibrated using the empirical best linear unbiased prediction theory. The leave-one-out cross-validation (LOOCV) method was used to evaluate the NLME-CD model and obtain the optimal NLME-CD model.

2. The method according to claim 1, characterized in that The method further comprises: The best fitting prediction variables of the area to be estimated and the random effect parameters at the regional level are obtained, and the best fitting prediction variables and the regional level random effects are input into the optimal canopy density basic model to construct the nonlinear mixed effect canopy density estimation model NLME-CD with the best fitting effect for bamboo forests.

3. The method according to claim 1, characterized in that Determining a sample plot and obtaining canopy density data for the sample plot, obtaining soil samples from the sample plot and measuring nutrient content, and obtaining stand data and terrain factor data for the sample plot specifically include: Identify the sample plots as areas of bamboo forests that can cover areas with significant changes in bamboo forest density, site productivity, topography, and environment; The bamboo forest canopy density was determined by image analysis of vertical snapshots of the forest canopy taken at predetermined times using a digital camera with a fisheye lens; Obtaining the total carbon content TC, organic carbon content SOC, total nitrogen content TN, and total phosphorus content TP of a soil composite sample, wherein the soil composite sample is obtained by mixing soil samples collected from five randomly selected points in the sample plot, drying in a cool place in the laboratory, removing roots and debris, and sieving using a sieve; Acquire forest stand data and terrain factor data of the sample plot.

4. The method according to claim 1, wherein The best fitting prediction variables are specifically: stand density N, dominant height DH, basal area BA, soil total nitrogen content TN and soil organic carbon content SOC.

5. The method according to claim 1, wherein Based on the canopy density data of the sample plot, the best fitting predictor variables and the introduction of regional level random effect parameters, the optimal basic model is used to construct the nonlinear mixed effect canopy density estimation model NLME-CD of bamboo forest, which specifically includes: Based on the canopy density data of the sample plot, the best-fitting predictor variables, and the introduction of regional-level random effect parameters, a bamboo forest canopy density estimation model was constructed according to the optimal basic model shown in Formula 1: Where CD represents the density of bamboo forest, β and a are the parameter vectors to be estimated, x is the parameter vector of explanatory variables, including forest stand variables and soil nutrient content variables, i represents the i-th region, and j represents the j-th plot.

6. The method according to claim 1, wherein Four sampling strategies were used to select the optimal number of plots per region, slope, aspect, or altitude. The optimal number of plots per region, slope, aspect, or altitude for calibrating CD predictions for the four sampling strategies was evaluated. The NLME-CD model was calibrated using the empirical best linear unbiased prediction theory. Specifically, the following steps were performed: The optimal number of plots in each region, slope, aspect, or elevation was selected using the four sampling strategies. Based on the selected plots, the Lindstrom and Bates algorithm implemented in the NLME function of R software was used to evaluate the optimal number of plots in each region, slope, aspect, or elevation for calibrating the CD predictions of the four sampling strategies with maximum likelihood estimation.

7. The method according to claim 1 or 6, characterized in that The four sampling strategies are as follows: randomly selecting 1-5 base area BA plots in each region, slope, aspect or altitude; taking 1-5 plots with the maximum BA in each region, slope, aspect or altitude; taking 1-5 plots with the average BA in each region, slope, aspect or altitude; and taking 1-5 plots with the minimum BA in each region, slope, aspect or altitude.

8. A device for implementing a bamboo forest canopy density estimation model, characterized in that: Specifically include: An acquisition module is used to determine a sample plot and obtain canopy density data of the sample plot, obtain soil samples and measure nutrient content, and obtain forest stand data and terrain factor data of the sample plot; A variable determination module is used to determine the best fitting prediction variable, i.e., input parameter, of the bamboo forest canopy density estimation model based on the soil nutrient content data, the forest stand data, and the terrain factor data through variance analysis; A model construction module is used to construct a bamboo forest canopy density estimation model using the optimal basic model based on the canopy density data of the sample plot, the selected best-fitting predictor variables, and the introduction of regional-level random effect parameters, wherein the bamboo forest canopy density estimation model is a nonlinear mixed-effect canopy density estimation model NLME-CD for moso bamboo forests; a parameter estimation module for selecting the optimal number of plots per region, slope, aspect, or elevation using the four sampling strategies, evaluating the optimal number of plots per region, slope, aspect, or elevation for calibrating CD predictions using the four sampling strategies, and calibrating the NLME-CD model using empirical best linear unbiased prediction theory; The model evaluation module is used to evaluate the NLME-CD model using the leave-one-out cross validation (LOOCV) method to obtain the optimal NLME-CD model.

9. The device according to claim 8, characterized in that The device further comprises: The estimation module is used to obtain the best-fitting prediction variables of the area to be estimated and introduce random effect parameters at the regional level, input the best-fitting prediction variables and the regional level random effects into the optimal canopy density basic model, and construct the best-fitting nonlinear mixed effect canopy density estimation model NLME-CD for bamboo forests.

10. The device according to claim 8, characterized in that The acquisition module is specifically used for: Identify the sample plots as areas of bamboo forests that can cover areas with significant changes in bamboo forest density, site productivity, topography, and environment; The bamboo forest canopy density was determined by image analysis of vertical snapshots of the forest canopy taken at predetermined times using a digital camera with a fisheye lens; Obtaining the total carbon content TC, organic carbon content SOC, total nitrogen content TN, and total phosphorus content TP of a soil composite sample, wherein the soil composite sample is obtained by mixing soil samples collected from five randomly selected points in the sample plot, drying in a cool place in the laboratory, removing roots and debris, and sieving using a sieve; obtaining forest stand data and topographic data of the sample plot; The variable determination module is specifically used for: Based on the soil nutrient content data, stand data, and topographic factor data, the best fitting prediction variables, i.e., input parameters, of the bamboo forest canopy density estimation model are determined by variance analysis, wherein the fixed input parameters specifically include: stand density N, dominant height DH, basal area BA, soil total nitrogen content TN, and soil organic carbon content SOC; The model building module is specifically used to: Based on the canopy density data of the sample plot, the best-fitting predictor variables, and the introduction of regional-level random effect parameters, a bamboo forest canopy density estimation model was constructed according to the optimal basic model shown in Formula 1: Where CD represents the density of bamboo forest, β and a are the parameter vectors to be estimated, x is the parameter vector of explanatory variables, including forest stand variables and soil nutrient content variables, i represents the i-th region, and j represents the j-th plot; The parameter estimation module is specifically used for: Four sampling strategies were used to select the optimal number of plots in each region, slope, aspect, or altitude. Based on the selected plots, the Lindstrom and Bates algorithm implemented in the NLME function of R software was used to evaluate the optimal number of plots in each region, slope, aspect, and altitude for calibrating the CD predictions of the four sampling strategies by maximum likelihood estimation. Among them, 1-5 base area (BA) plots were randomly selected in each region, slope, aspect, or altitude; 1-5 plots with the maximum BA in each region, slope, aspect, or altitude were taken; 1-5 plots with the average BA in each region, slope, aspect, or altitude were taken; and 1-5 plots with the minimum BA in each region, slope, aspect, or altitude were taken.

Images