Frontiers Estimating carbon sequestration potential and optimizing management strategies for Moso

Estimating carbon sequestration potential and optimizing management strategies for Moso bamboo ( Phyllostachys pubescens ) forests using machine learning

Phyllostachys Pubescens's evaluation and management strategy of carbon absorption potential in beech forests play an important role in enhancing its characteristics and promoting sustainable development. However, it is enough to simulate the change in carbon absorption potential of Moso Buna forest, and to select and optimize the appropriate management measures based on the lon g-term dataline obtained in a delicate fixe d-choice survey. do not have. As a result, this study applied endless surveys and climate data in the fixe d-optional section of Zhejiang Province from 2004 to 2019. In order to build a dope model of the mosa beech forest, we compare only four types of methods, random forests, supporting vectors, XGBOOST, and BP neural networks. The final goal was to determine an excellent algorithm model. Separately, we predicted the important features of the Moso Buna Forest's future carbon accumulation and the required amount of terrestrial carbon. Next, the optimal management strategy was formulated based on the monitoring data. In fact, the carbon accumulation model built using the XGBoost method obtained the following results. Phyllostachys Pubescens's evaluation and management strategy of carbon absorption potential in beech forests play an important role in enhancing its characteristics and promoting sustainable development. However, it is enough to simulate the change in carbon absorption potential of Moso Buna forest, and to select and optimize the appropriate management measures based on the lon g-term dataline obtained in a delicate fixe d-choice survey. do not have. As a result, this study applied endless surveys and climate data in the fixe d-optional section of Zhejiang Province from 2004 to 2019. In order to build a dope model of the mosa beech forest, we compare only four types of methods, random forests, supporting vectors, XGBOOST, and BP neural networks. The final goal was to determine an excellent algorithm model. Separately, we predicted the important features of the Moso Buna Forest's future carbon accumulation and the required amount of terrestrial carbon. Subsequently, the optimal management strategy was formulated based on monitoring data. In fact, the carbon accumulation model built using the XGBoost method obtained the following results. Phyllostachys Pubescens's evaluation and management strategy of carbon absorption potential in beech forests play an important role in enhancing its characteristics and promoting sustainable development. However, it is enough to simulate the change in carbon absorption potential of Moso Buna forest, and to select and optimize the appropriate management measures based on the lon g-term dataline obtained in a delicate fixe d-choice survey. do not have. As a result, this study applied endless surveys and climate data in the fixe d-optional section of Zhejiang Province from 2004 to 2019. In order to build a dope model of the mosa beech forest, we compare only four types of methods, random forests, supporting vectors, XGBOOST, and BP neural networks. The final goal was to determine an excellent algorithm model. Separately, we predicted the important features of the Moso Buna Forest's future carbon accumulation and the required amount of terrestrial carbon. Next, the optimal management strategy was formulated based on the monitoring data. In fact, the carbon accumulation model built using the XGBoost method obtained the following results.

1 Introduction

Reeds are an important component of forest resources related to the beech subfamily of the Gramineae family. In the world, they are found in 87 families and up to 1, 500 orders (Ramakrishnan et al.). In tropical and subtropical regions, this forest suggests important financial, environmental and social significance (Terefe et al.). In many states, the area of beech forests is increasing with the decline of forest area (Yang et al.). China has the opportunity to taste the richest and largest bamboo resources in the world. According to the 9th National Forest Resource Inventory, the total area of beech forests in the country is 6. 411. 600 hectares, of which Moso bamboo is 4. 677. 800 hectares. 677. 800 hectares, accounting for 72. 96% of the total beech forests (Xu Y. et al. 2023). Moso bamboo (Phyllostachys pubescens (Phyllostachysns)) is a type of bamboo widely known in southern China, which has become an important forest resource in the region thanks to its unique rejection and exceptional opportunity to eat carbon (Li et al. It is considered one of the promising plants for absorbing carbon. Apart from this, bamboo products have become widely known among ordinary consumers due to the rapid increase of the progressive beech industry. It plays a remarkable role in fighting poverty and revitalizing rural areas that cultivate sugarcane (Pan et al.

In previous research (Mao et al., 2017b; Lv et al., 2023; zhou et al., 2023; Kaam et al., 2024) Sensing and modeling methods have been used. Most of these studies are studying the spatial and distribution of carbon reserved carbon reserves in Mouzo bamboo forests using remote sensing and spatial analysis methods using remote sensing data, images, and ground measurements of Landsat and Landsat TM. (XU et al. Landsat data has many advantages in evaluating the carbon reserves of the Moso bamboo grove, such as mult i-spectra information, ful l-scale cover rate, time row analysis, open use possibility. It is an effective tool for studying changes in carbon reserves in Moso bamboo forests (Li et al., 2019; Qi et al., 2022). Integrated ecosystems, including carbon cycle (Liu et al., 2021; Camargo García et al., YU et al., 2023). Integrated and parameters of the selected models were ultimately enabled

In order to clearly predict the amount of carbon requirements of the Moso Buna Forest and determine the appropriate management strategy, a lon g-term survey was conducted using a fixed selection section and climate data. At the same time, the purpose was to identify a clearer evaluation model using various machine survey methods to determine the amount of carbon of Moso Buna forest. In order to predict the amount of carbon accumulation, a machine learning method is used to determine appropriate data such as terrain, climate conditions, and vegetation indicators. After that, the creation of a feature model is optimized and adjusted, and a method of achieving higher accuracy for prediction is selected (Silatsa et al., 2020; reiersen et al., 2022; Uniyal et al.) The type of machine learning method distinguishes different accuracy values when handling a set of data. In fact, earlier research on Mosobuna forest biomass and carbon accumulation proved that the model built with the XGBoost method support was considered more accurate (Li X. et al., 2018, 2023). ; Gomes et al., 2019). At the same time, partial logging of the mozozobuna forest functions as an important strategy for maintaining productivity and production, increasing the amount of carbon absorption, clearly predicts the amount of carbon requirements of the Moso Buna forest, and appropriate. In order to determine the management strategy, a lon g-term survey was conducted using fixed selection sections and climate data. At the same time, the purpose was to identify a clearer evaluation model using various machine survey methods to determine the amount of carbon of Moso Buna forest. In order to predict the amount of carbon accumulation, a machine learning method is used to determine appropriate data such as terrain, climate conditions, and vegetation indicators. After that, the creation of a feature model is optimized and adjusted, and a method of achieving higher accuracy for prediction is selected (Silatsa et al., 2020; reiersen et al., 2022; Uniyal et al.) The type of machine learning method distinguishes different accuracy values when handling a set of data. In fact, earlier research on Mosobuna forest biomass and carbon accumulation proved that the model built with the XGBoost method support was considered more accurate (Li X. et al., 2018, 2023). ; Gomes et al., 2019). At the same time, partial logging of Mozobuna forests functions as an important strategy for maintaining productivity and production, increasing the amount of carbon absorption, clearly predicting the amount of carbon requirements of the Monsho Buna forest, and appropriate management strategy. In order to determine, a lon g-term survey was conducted using fixed selection sections and climate data. At the same time, the purpose was to identify a clearer evaluation model using various machine survey methods to determine the amount of carbon of Moso Buna forest. In order to predict the amount of carbon accumulation, a machine learning method is used to determine appropriate data such as terrain, climate conditions, and vegetation indicators. After that, the creation of a feature model is optimized and adjusted, and a method of achieving higher accuracy for prediction is selected (Silatsa et al., 2020; reiersen et al., 2022; Uniyal et al.) The type of machine learning method distinguishes different accuracy values when handling a set of data. In fact, earlier research on Mosobuna forest biomass and carbon accumulation proved that the model built with the XGBoost method support was considered more accurate (Li X. et al., 2018, 2023). ; Gomes et al., 2019). At the same time, partial logging of Mozobuna forests functions as an important strategy for maintaining productivity and production, and increases the amount of carbon absorption.

2 Materials and methods

2.1 Study site

The survey was conducted in Zhejiang Province, China, south of the Yangtze River Delta, and on the southeast coast. The range is from 27 ° 02 'to 31 ° 11' and 123 ° 10 'to east sutra 118 ° 01' (Fig. 1). However, the periphery of Zhejiang Province is one of the smallest ministry in China and can be sampled with a wonderful forest coating rate of 61 %. The periphery of Zhejiang Province has an uneven terrain that is characterized by many mountains and a limited number of waters. There are many mountains in the western part, and the east is composed of hills and plains. The surrounding area of Zhejiang Province is located in the subtropical zone, belongs to the monsoon humid climate, the temperature is low, the four seasons are clear, and the sunshine and precipitation are abundant. The average annual temperature in this area is 15 to 18 ° C, the annual sunshine time is 1100 to 2200 hours, and the average annual rainfall is 1100 to 2000 mm (yin et al.) The annual temperature is low in January, the highest in July. I will. Apart from this, there is a large amount of precipitation in May and June. There are abundant vegetable resources around Zhejiang Province, and the resources are lon g-term.

Figure 1. 293 points in the unpleasant test area of Moso Beech Forest in Zhejiang Province in China, China. The survey was conducted in Zhejiang Province, located on the southeast coast south of the Yangtze River Delta. The range is from 27 ° 02 'to 31 ° 11' and 123 ° 10 'to east sutra 118 ° 01' (Fig. 1). However, the periphery of Zhejiang Province is one of the smallest ministry in China and can be sampled with a wonderful forest coating rate of 61 %. The periphery of Zhejiang Province has an uneven terrain that is characterized by many mountains and a limited number of waters. There are many mountains in the western part, and the east is composed of hills and plains. The surrounding area of Zhejiang Province is located in the subtropical zone, belongs to the monsoon humid climate, the temperature is low, the four seasons are clear, and the sunshine and precipitation are abundant. The average annual temperature in this area is 15 to 18 ° C, the annual sunshine time is 1100 to 2200 hours, and the average annual rainfall is 1100 to 2000 mm (yin et al.) The annual temperature is low in January, the highest in July. I will. Apart from this, the amount of precipitation is large in May and June. There are abundant vegetable resources around Zhejiang Province, and the resources are lon g-term.

2.2 Data collection

2.2.1 Fixed sample data

Figure 1. 293 points in the unpleasant test area of Moso Beech Forest in Zhejiang Province in China, China. The survey was conducted in Zhejiang Province, located on the southeast coast south of the Yangtze River Delta. The range is from 27 ° 02 'to 31 ° 11' and 123 ° 10 'to east sutra 118 ° 01' (Fig. 1). However, the periphery of Zhejiang Province is one of the smallest ministry in China and can be sampled with a wonderful forest coating rate of 61 %. The periphery of Zhejiang Province has an uneven terrain that is characterized by many mountains and a limited number of waters. There are many mountains in the western part, and the east is composed of hills and plains. The surrounding area of Zhejiang Province is located in the subtropical zone, belongs to the monsoon humid climate, the temperature is low, the four seasons are clear, and the sunshine and precipitation are abundant. The average annual temperature in this area is 15-18 ° C, the annual sunshine hours are 1100 to 2200 hours, and the average annual precipitation is 1100 to 2000 mm (yin et al.) The annual temperature is low in January, the highest in July. I will. Apart from this, there is a large amount of precipitation in May and June. There are abundant vegetable resources around Zhejiang Province, and the resources are lon g-term.

2.2.2 Climate data collection

Figure 1. 293 points in the unprophic test area of Moso Bae Beech Forest in Zhejiang Province, China, China.

2.3 Methods steps

Zhejiang Province established a continuous forest inventory system in 1979, and repeated surveys were conducted at five-year intervals. A total of 4, 250 fixed sampling sites were set up, evenly spaced at intervals of 4 km north-south and 6 km east-west. The fields were square, with each side measuring 28 and 28 m (Xu et al.). This study used continuous survey data collected at these fixed sample sites from 2004 to 2019. The sample sites included 293 sites, which consisted mainly of pure Moso bamboo forests. For each site, several elements were recorded, including height, slope, diameter at breast height (DBH), bamboo age, and number of bamboo stalks. Height, slope, and slope direction were measured using a portable GPS, total station, and compass meter. Current-year and 1-year culms were classified as 1 du, 2-3-year culms as 2 du, and 3-4-year culms as 3 du. The climate data used in this study were obtained from ClimateAP software, a standalone application for MS Windows that reduces unscaled data as a basis for historical and future climate variables from 1901 to 2100 (Wang et al., 2017). This software can generate climate data for multiple locations and periods. In this study, historical climate data from 2004 to 2019 and future climate data from 2024 to 2060 were obtained. The dataset included important parameters such as annual mean temperature, annual mean precipitation, and annual heat and humidity index for all 293 sample sites of Moso bamboo forests. Zhejiang Province established a continuous forest inventory system in 1979 and conducted repeated surveys at 5-year intervals. A total of 4, 250 fixed observation sites were set up, evenly spaced at intervals of 4 km north-south and 6 km east-west. The fields were square, with each side measuring 28 by 28 m (Xu et al.). This study used continuous survey data collected at these fixed sample sites from 2004 to 2019. The sample sites included 293 sites consisting mainly of pure Moso bamboo forests. For each site, several elements were recorded, including height, slope, diameter at breast height (DBH), bamboo age, and number of bamboo stalks. Height, slope, and slope direction were measured using a portable GPS, total station, and compass meter. Current year and 1st year culms were classified as 1 du, 2-3 year culms as 2 du, and 3-4 year culms as 3 du.

The climate data used in this study were obtained from ClimateAP software, a standalone application for MS Windows that reduces unscaled data as a basis for historical and future climate variables from 1901 to 2100 (Wang et al., 2017). The software can generate climate data for multiple locations and time periods. In this study, we obtained historical climate data from 2004 to 2019 and future climate data from 2024 to 2060. The dataset included important parameters such as annual mean temperature, annual mean precipitation, and annual heat and humidity index for all 293 sample sites of Moso bamboo forests. Zhejiang Province established a continuous forest inventory system in 1979 and conducted repeated surveys at 5-year intervals. A total of 4, 250 fixed observation sites were set up, evenly spaced at intervals of 4 km north-south and 6 km east-west. The fields were square, with each side measuring 28 and 28 m (Xu et al.). This study used continuous survey data collected at these fixed sample sites from 2004 to 2019. The sample sites included 293 sites that consisted mainly of pure Moso bamboo forests. For each site, several elements were recorded, including height, slope, diameter at breast height (DBH), bamboo age, and number of bamboo stalks. Height, slope, and slope direction were measured using a portable GPS, total station, and compass meter. Current year and 1st year culms were classified as 1 du, 2-3 year culms as 2 du, and 3-4 year culms as 3 du.

2.4 Model selection and construction

2.4.1 Random forest

Climate data used in this study were obtained from ClimateAP software, a standalone application for MS Windows that reduces unscaled data as a basis for historical and future climate variables from 1901 to 2100 (Wang et al., 2017). This software can generate climate data for multiple locations and time periods. In this study, historical climate data from 2004 to 2019 and future climate data from 2024 to 2060 were obtained. The dataset included important parameters such as annual mean temperature, annual mean precipitation, and annual heat and humidity index for all 293 sample sites of Moso bamboo forests.

First, the data obtained from the fixed survey of the field and the climate data were organized and divided into training and testing data in a ratio of 7:3. Only four methods, namely Random Forest (RF), Support Machine Gun (SVM), XGBOOST, and BPNN neural network, were used to construct the carbon of aboveground suction in the beech forest of Misao. These models were trained and tested to select good models. Then, based on the models, the main driving characteristics of aboveground carbon in the beech forest of Misao were selected. The data of the sections from the fixed selection were linked to establish the relationship between the moving parameters of the beech forest of Moso with normal management methods over time and to hold the future importance. By the method of inputting all the characteristics into the model, the carbon accumulation under various management methods in the future was predicted, which was implemented and compared with the results (Figure 2).

2.4.2 Support vector machine

Sketch 2. The process of evaluation of aboveground carbon accumulation in the proletarian MOSO beech forest and rational selection of control measures.

Random forest (RF) is a technique proposed by Breiman (2001) to control oar-and-compass studies based on inference trees. It may be used for both systematics, such as regression (Zamani Joharestani et al.). RF uses two important techniques: bag construction and random selection of symptoms. In the first line, a given number of inferences was constructed by the introduction of a large number of bootspectacle selections. This was followed by random selection and collection of symptoms, in which some symptoms were taken from the initial training data set to study the model. In each inference tree, nodes were shared to make fresh nodes, but this process was secondary to these conditions, such as the height of the tree, the small number of plate nodes, etc. After constructing the number of inference trees, irrevocable monitoring was performed by the averaging method. This allowed us to reduce the overfitting of the highest correlations and symptoms, and increase the adaptability and representativeness of the model.

In this study, mainly using RF regression analysis, the effects of various factors on the ground carbon accumulation of Moso bamboo grove were evaluated. Fixed sample parcels and climate data were used as learning data, and different decisive trees were built using a different learning subset. Finally, the average value of the predicted value of the internal carbon accumulation obtained from the tree was set as the final estimation of the ground carbon accumulation on the sample site._ϵThe support vector machine was introduced by Vapnik in 1995 (Joachims, 1998). The support vector is used for classification, but is called a support vector regression (SVR) when solving a regression problem. SVR assumes that the loss function is calculated, assuming that there is an absolute value between the model output and the true value. The SVR is like creating a section zone around the sample, and is considered accurately predicted when the training sample enters the section. However, the outer points of the section band are lost, and the size is determined by the length of the red line. The degree of flexibility can change on both sides of the distance band (Awad et al.) SVR problems can be formulated as follows (equation 1):

min w, b 1 2 || w || 2 + c ∑ IK = 1 m l? (f (X IK) -Y IK) (1)

2.4.3 XGBoost

Here, C is a regularization constant, L

It is a no n-sensitive loss.

In this study, a nonlinear kernel SVM was used because the number of samples was large compared to the characteristics dimension. The fixed plot sample was used as an input, and the regression model was used to predict the carbon storage value. The purpose was to minimize the absolute value of the difference between the predictive value and the actual carbon accumulation. In this approach, we aimed to bring the carbon accumulation on the ground at the sample point to the actual value as much as possible. In this study, mainly using RF regression analysis, the effects of various factors on the ground carbon accumulation of Moso bamboo grove were evaluated. Fixed sample parcels and climate data were used as learning data, and different decisive trees were built using a different learning subset. Finally, the average value of the predicted value of the internal carbon accumulation obtained from the tree was set as the final estimation of the ground carbon accumulation on the sample site.

The support vector machine was introduced by Vapnik in 1995 (Joachims, 1998). The support vector is used for classification, but is called a support vector regression (SVR) when solving a regression problem. SVR assumes that the loss function is calculated, assuming that there is an absolute value between the model output and the true value. The SVR is like creating a section zone around the sample, and is considered accurately predicted when the training sample enters the section. However, the outer points of the section band are lost, and the size is determined by the length of the red line. The degree of flexibility can change on both sides of the distance band (Awad et al.) SVR problems can be formulated as follows (equation 1):

min w, b 1 2 || w || 2 + c ∑ IK = 1 m l? (f (X IK) -Y IK) (1)_{Here, C is a regularization constant, L}It is no n-sensitive loss._{Here, C is a regularization constant, L}The support vector machine was introduced by Vapnik in 1995 (Joachims, 1998). The support vector is used for classification, but is called a support vector regression (SVR) when solving a regression problem. SVR assumes that the loss function is calculated, assuming that there is an absolute value between the model output and the true value. The SVR is like creating a section zone around the sample, and is considered accurately predicted when the training sample enters the section. However, the outer points of the section band are lost, and the size is determined by the length of the red line. The degree of flexibility can change on both sides of the distance band (Awad et al.) SVR problems can be formulated as follows (equation 1):_{min w, b 1 2 || w || 2 + c ∑ IK = 1 m l? (f (X IK) -Y IK) (1)}Here, C is a regularization constant, L

It is no n-sensitive loss.

2.4.4 BP neural network

XGBoost is a non-closed-source machine learning blueprint created by Chen et al. It effectively sells the GBDT method, introduces algorithmic and technical improvements, and has been widely applied in many machine learning competitions and contests, achieving impressive results (Chen and Guestrin, 2016). It uses a gradient boosting method but differs from GBDT by defining a motivation function. Its ultimate goal is to make the predicted meaning of a tree population representing carbon stocks more literally match the actual values. Each time a new tree is added, it goes through the process of learning a new function to adjust each day's residual between the last monitoring and the true significance. In the final result, the estimates corresponding to the tree leaf nodes are summed to obtain a standard predicted value. The motivation function consists of two components: a loss function that considers the gap between the predicted value and the true significance, and a regularization term that determines the complexity (Eqs. 2, 3) (Chen et al.).

O b j ( t ) = ∑ I = 1 n l ( y I , y ^ I ( t - 1 ) + f t ( X I ) ) + Ω ( f t ) + c o n s t a n t ( 2 )

2.5 Model validation and carbon stock calculation

L ( ∅ ) = ∑ ik l ( y ik ′ - y ik ) + ∑ k Ω ( f t ) ( 3 )

Figure 3. Block diagram of the XGBoost method. where j i is the true significance, y

i_{Here, C is a regularization constant, L}T

) is the difficulty of the weak learner, and L(φ) is the loss function.

In the study provided, the fixed plot data and climate data were first input into the model. Each time a new tree of conclusions was connected, the task was to fit the residuals between the predicted areal carbon stocks obtained in the previous iteration. This iterative process aimed to sequentially minimize the loss function, which actually led to carbon stock values that were close to the actual values.

In 1958, Rumelhart proposed a BP (Back Propagation) learning method using reverse slip propagation for the BP network. The principle is to introduce a mistake in the output layer to estimate the mistakes of the previous layer, introduce this mistake in order to estimate the following mistakes, and obtain estimated mistakes for other layers. The (Li et al.) Error that wanted to continue to the ups was distributed into each cell, adjusting the weight of each cell, and continued to be repeated until the optimal error or a certain number of repetitions reached. The BP neural network (BPNN) has three or more layers: input, output, and hidden layers. Each neuron of the thre e-layer network is completely connected to each neuron in the front layer. There is no direct connection between neurons within the boundary of the first layer (Fig. 4)._{Sketch 4. Major processes of VR neural networks.}In this study, the input layer consists of eight characteristics: site data, weather data, and topographic data. These characteristics include average trunk density, average hig h-area, average age, average annual temperature, average annual rainfall, annual water index, slope, and altitude. These eight characteristics entered the input layer, passed through the recycling process, and ultimately contributed to the calculation of carbon accumulation on the ground.

2.6 Future parameter forecasts and management measures

In this study, a model of carbon accumulation of carbon of Biomass on the ground was built using four types of machine learning methods: random forest, support vector machine, XGBOOST, and BP neural network. The performance of the model was evaluated on the test set by comparing the four regression performance indicators: R two square, average tw o-tae (MSE), average uniform uniform (RMSE), and an average unconditional error (MAE). be. This evaluation was verified using the training simulation of the data set (Saltelli, 2002; Liemohn et al.) The formula of the four regression performance measurements (equation 4-7).

R 2 = 1 -∑ i = 0 m (y i-y ^) 2 ∑ i = 0 m (y i-y ¯) 2 (4)

M s e = 1 m ∑ i = 1 m (y i-y ^) 2 (5)

2.7 Data analysis

R m S e = ∑ i = 1 m (y i-y ^ i) 2 m (6)

3 Results

3.1 Construction and prediction effects of different machine learning carbon stock models

M s e = 1 m ∑ i = 0 m-1 | Y i-y ^ i | (7)

Here, M is the number of samples of the inspection kit, y

Is the current significance of the test kit sample, Y ^ is the prediction sign of the test kit sample, and the Y ¯ is the average significance of the test kit sample.

The biomass of each bambo o-mun sect bamboo in the pilot plot was designed to match the first plant biomass model. Next, the pilot plot in the pilot plot, the aerial biomass of the bamboo Meng sect bamboo was combined to seek a pilot site aerial biomass share. Finally, the biomass was multiplied by the recalculating coefficient of 0, 5042 to seek the carbon accumulation on the ground of the MOSO beech forest (equal 8) (Zhou, 2006).

3.2 Main driving factors and future parameter models of aboveground carbon stocks in Moso bamboo forests

C m O S O = (747. 787 D 2. 771 (0. 148 a 0. 028 + A) 5. 555 + 3. 772) × 0. × 0. 5042 (8)

Here c

Meso

D is the bamboo sect's chest height (cm), and A is the age of the bamboo sect (DU).< 0.01). With increased abandonment management time of, the average breast diameter of the sample plots decreased: y = −0.098 x + 10.414 ( R 2 = 0.378), while the average age class of the sample plots continued to rise: y = 0.069 x + 2.261 ( R 2 = 0.598).

The relationship between average breast height in the Moso Buna Forest and average bite-clip density was installed in normal aging based on data collected from 293 surveys on the 2004-2019 stage. In order to evaluate these characteristics for each year using various control scenarios, a model related to perforation in the middle and middle boots density was used. In order to conform to the general model of the Moso bamboo rising, these traits were predicted as 13, 0cm of the dialect of raising the chest, and the stump density value of 4. 500 kul/ha was 4. 500 kul/ha. In fact, the benefits of these values did not affect the subsequent expansion. In fact, it presented the height and inclination of the relatively measured sampling section, with the support of climate programs, the average annual temperature and annual average sediment. The models of each parameter at the survey point as part of the scales of the rejection were derived from the model developed by Yin et al. (2019). All predictable features have been integrated into the ground carbon model built with the XGBoost method support.

3.3 Dynamic prediction of aboveground carbon stock in Moso bamboo forests in Zhejiang Province under different management measures

In real time, it is believed that the generally accepted wood Lasmoso practice is abstinent from bamboo swings 1-3 du and more than bamboo 4 du. However, it was not understood about the specific logging strategy of bamboo 4DU. Therefore, in order to learn and improve the logging strategy of Moso's beech forest, the felling strategy of bamboo 4DU or higher was revealed. At the same time, the story of the accumulation of carbon accumulation in the upper basement of the beech forest of Moso's beech, which is progressing, reflects the importance of measures to adjust the carbon accumulation on the ground beech forest.

A management scenario was devised with a three-stage felling survey and layout for abandoned farmland management. Measure I: Prohibit felling of bamboo 1-3 du, and all bamboo 4 du or more are subject to felling. Measure II: Prohibit planoing of bamboo 1-3 du, and 50% planoing of bamboo 4 du or more. Measure III: Prohibit felling of bamboo 1-3 du, 30% felling of bamboo 4 du, and 80% felling of bamboo 5 du or more. Measure IV: Brownfield management.

Pycharm software (Pycharm 3. 10) and Excel 2021 were used for statistical analysis. The average carbon stock in the vegetation at the sampling site was multiplied by the area of Mosuo-beech forest to calculate the artificial carbon stock of Mosuo-beech forest in Zhejiang Province. The XGBoost method was then applied to predict the carbon stock. We used Python's XGBoost convolution to build the aboveground carbon stock model, and Bayes's bayes_opt was used for feature options and five-fold cross-validation. Finally, the monitoring results were obtained by averaging the results of all individual regressions. We used Pycharm's Matplotlib convolution to visualize predicted vs. observed values, boxplots, moment correlation tests, and graphs of thermal correlation systems. Each variable was normalized, and the significance of symptoms was paid for with the feature quality support of Python's scikit-learn library. The results were visualized using matplotlib support. The predicted characteristics from all site surveys, such as average height profile, average shrub density, tree age, and climate, were applied as input values for the carbon stock model, and the output values were used to analyze global changes.

To enhance the accuracy of the model, the Bayes Optimization Law was used for the strict fitting of each model. Using the Gaussian process, it was clarified to keep the previous parameter information into into account and always determine the optimal parameters. This approach is most preferred because it is effective in minimizing repetitive and hig h-speed updates. When using the mechanism of the support vector for modeling, the Bezof tuning was performed and the model was verified five times. In the 20th repetition, the model achieved the minimum average circuit error corresponding to the parameter value of "C" 0, 1029 and the parameter value of "γ" 0, 2253. When building a model using the RF determination base algorithm, the effectiveness of the model is exercised by the hyper parameter "n_esestimators", "max_depth" and "min_samples_leaf". Therefore, when setting a parameter, focus on these two hyperparameters. When setting the parameter "n_estimators", change the range of 100-500 in step 10, "Max_depth" is changed to 2-10 in step 1, and "min_samples_leaf" is 5-10 in step 1. I changed it. The XGboostal algorithm is known for its effective implementation of the gradient, but is very sensitive to "n_estimators" and "max_depth". To enhance the accuracy of the model, the Bayes Optimization Law was used for the strict fitting of each model. Using the Gaussian process, it was clarified to keep the previous parameter information into into account and always determine the optimal parameters. This approach is most preferred because it is effective in minimizing repetitive and hig h-speed updates. When using the mechanism of the support vector for modeling, the Bezof tuning was performed and the model was verified five times. In the 20th repetition, the model achieved the minimum average circuit error corresponding to the parameter value of "C" 0, 1029 and the parameter value of "γ" 0, 2253. When building a model using the RF determination base algorithm, the effectiveness of the model is exercised by the hyper parameter "n_esestimators", "max_depth" and "min_samples_leaf". Therefore, when setting a parameter, focus on these two hyperparameters. When setting the parameter "n_estimators", change the range of 100-500 in step 10, "Max_depth" is changed to 2-10 in step 1, and "min_samples_leaf" is 5-10 in step 1. I changed it. The XGboostal algorithm is known for its effective implementation of the gradient, but is very sensitive to "n_estimators" and "max_depth". To enhance the accuracy of the model, the Bayes Optimization Law was used for the strict fitting of each model. Using the Gaussian process, it was clarified to keep the previous parameter information into into account and always determine the optimal parameters. This approach is most preferred because it is effective in minimizing and hig h-speed renewal. When using the mechanism of the support vector for modeling, the Bezof tuning was performed and the model was verified five times. In the 20th repetition, the model achieved the minimum average circuit error corresponding to the parameter value of "C" 0, 1029 and the parameter value of "γ" 0, 2253. When building a model using the RF determination base algorithm, the effectiveness of the model is exercised by the hyper parameter "n_esestimators", "max_depth" and "min_samples_leaf". Therefore, when setting a parameter, focus on these two hyperparameters. When setting the parameter "n_estimators", change the range of 100-500 in step 10, "Max_depth" is changed to 2-10 in step 1, and "min_samples_leaf" is 5-10 in step 1. I changed it. The XGboostal algorithm is known for its effective implementation of the gradient, but is very sensitive to "n_estimators" and "max_depth".

4 Discussion

4.1 Evaluation and validation of machine learning for carbon stock prediction in Moso bamboo forests

The evaluation data of each regression performance index of the four models in the test set showed that the above-ground model for predicting carbon stocks in Moso bamboo forests constructed using the XGBoost algorithm was obviously superior to other models. The model using the XGBoost algorithm achieved R2 0, 9895, MSE 0, 0104, RMSE 0, 1059, and MAE 0, 0665 (Table 1). The loss function curves of the training set and test set of the XGBoost algorithm show that as the number of iterations increases, the RMSE decreases and eventually becomes flat (Fig. 5). To visualize the predicted and actual values of the samples, a line graph was created comparing the actual and predicted values (Fig. 6). The results showed that the model constructed using the XGBoost algorithm could predict the above-ground carbon stocks of Moso bamboo forests more accurately (Fig. 6C).

4.2 Impact of stand structure factors on aboveground carbon stocks in Moso bamboo forests

Table 1. Indicators and results of regression efficiency evaluation of different carbon stock models in the test set.

4.3 Estimating the aboveground carbon storage potential of Moso bamboo forests

Figure 5. Loss function curves of the aboveground carbon stock model of Moso bamboo forests using the XGBoost algorithm.

Figure 6. Comparison of models constructed with four different machine learning methods, namely, support vector machine (A), random forest (B), XGBoost (C), and BP neural network (D). The red dotted line indicates the predicted value, and the blue solid line indicates the actual value.

4.4 Uncertainty analysis of carbon stock predictions

Eight variables, including topography, survey site, and climate, were combined to estimate the effect of each variable on aboveground carbon stock in the Moso beech forest. In addition, the correlation between these variables and aboveground carbon stock was examined using a model constructed using the XGBoost method. As a result, the correlation between stem density and carbon stock was 0. 85, the correlation between diameter at breast height and carbon stock was 0. 47, and the correlation between tree age and carbon stock was 0. 22 (Fig. 7A). The moments acting on aboveground carbon stock in the Moso beech forest were ranked from most significant to least significant as follows: average stem density, average tree height, average tree age, average annual precipitation, average annual temperature, average annual heat and humidity index, average annual heat and humidity index, ray, height (Fig. 7B). Canopy density, DBH, and tree age had a large effect on aboveground carbon stock in the Moso beech forest, while stand conditions and climate had little effect. It was also observed that the carbon stocks at these sites gradually increased with the increase in the average DBH and average density of the Moso bamboo varieties in the study area (Figure 8).

4.5 Management implications for Moso bamboo forests

Figure 7. Correlation between aboveground carbon stocks and their factors in the Moso beech forest, and the inherent importance of each factor. (A) Different colors suggest the importance of the correlation: scarlet is negative correlation, blue is positive correlation. (B) Every length has every meaning, and the larger the length, the higher the importance of the data.

5 Conclusion

Sketch 8. Relationship between carbon reserves and average diameter at breast height in the sample plots. Different colors indicate the significance of the average diameter at the study area, and dark points indicate outliers.

Data availability statement

For the Moso beech forest test plot, increment curves were constructed for the last 4 years. The average DBH of the test plot corresponded to the equation y = 170, 78 ln x - 1290, 4 (r 2 = 0. 9964) (Fig. 9a), and the average shrub density of the test plot corresponded to the equation y = 188, 781 ln x - 1 × 10 6 (r 2 = 0. 9985) (Fig. 9b). The compliance of age classes in the beech forests of the Moscow region remained mainly measured during normal domestic felling, and the middle age classes of the pole showed conditional strength. When the abandonment of the Moso beech forests began, a strong linear longevity was observed between the dispatch density and the years of abandonment: y = 340, 66 x + 2272. 763 (r 2 = 0. 700, p

Author contributions

Sketch 9. Relationship between the cause and duration of the tree of the whole Moso beech forest under normal management. (a) The change over time of the average apex of the breast height peak of the whole Moso beech forest sample, (b) The change over time of the average density of the stem of the whole Moso beech forest sample. Snowboards are the average average, and the dark areas appear in th e-normal difference.

Funding

According to the announcement of the importance of forest resources and their environmental functions in Zhejiang Province, the Moso original forest in Zhejiang Province was 653. 500 hectares in 2004, 727. 000 hectares in 2009, 775. 800 hectares in 2014, and 814. 300 hectares. Using these data, the aboveground carbon supply was evaluated and was 8, 51 ± 4, 68 tenge C, 9, 36 ± 5, 37 tenge C, 12, 80 ± 6, 89 tenge C, and 17, 00 ± 8, 35 tg c. As the area of Mozong beech forests has been increasing year by year, the increase rate of Mozong beech forest area is predicted to be at the 0, 88% level based on the database of Mozong beech forest area in recent years. By combining the significance of each parameter of the aboveground carbon model constructed with the aid of the XGBOOST method, the aboveground carbon stock of beech forests in Mozong, Zhejiang Province from 2024 to 2060 was predicted using all kinds of management measures.

Acknowledgments

As shown in Table 2, the accumulation of Mosuo beech forests in Zhejiang Province in the management scenarios of Meriro I, II, and II is increasing every day, peaking 40 years later, and subsequent fluctuations are repeated. Stabilize. However, the time and size of the peak and its size vary depending on these different management measures. In the case of countermeasures I, the amount of carbon accumulation of the mosuo beech forest in Zhejiang Province increases every day, reaching the peak of 31, 54 ± 9, 60 tg C in 2053 (Fig. 10a). In the countermeasures II, the peak of Mosuo beech forests in Zhejiang Province increases daily and reaches the peak of 33, 56 ± 7, 53 TG C in 2050 (Fig. 10B). In the measurement II, the accumulation of ground carbon forests in Zhejiang Province increases daily, reaching the peak of 36, 25 ± 8, 47 TG C in 2046 (Fig. 10C). The important time to reach the peak depends on the measurement method, and the following order is: The size of the peak of carbon accumulation corresponds to that order: Cutting bamboo grove 1-3DU prohibited, logging. There is no doubt that management measures to secure 30 % of bamboo forest 4DU and 80 % of fell bamboo forest 5DU or more will increase the amount of carbon accumulation of fell bamboo grove.

Conflict of interest

Table 2 Predicted ground carbon accumulation (TG C) for 2024-2060 in Mosuo Bauca Forest in Zohe Province Province.

Publisher’s note

Sketch 10. Predicted ground carbon accumulation from 2024 to 2060 in the Mosuo Beech Forest in Zhejiang Province. The feature (A-D) is the ground carbon accumulation predicted under four possible measures according to this. Blue dashes are 95%trust separation, and the points that correspond to black dashes are all possible, shor t-term carbon accumulation and peak carbon accumulation. As shown in Table 2, the accumulation of Mosuo beech forests in Zhejiang Province in the management scenarios of Meriro I, I, and II is increasing every day at first, peaking in 40 years, and slightly later. Stable while repeating fluctuations. However, the time and size of the peak and its size vary depending on these different management measures. In the case of countermeasures I, the amount of carbon accumulation of the mosuo beech forest in Zhejiang Province increases every day, reaching the peak of 31, 54 ± 9, 60 tg C in 2053 (Fig. 10a). In the countermeasures II, the peak of Mosuo beech forests in Zhejiang Province increases daily and reaches the peak of 33, 56 ± 7, 53 TG C in 2050 (Fig. 10B). In the measurement II, the accumulation of ground carbon forests in Zhejiang Province increases daily, reaching the peak of 36, 25 ± 8, 47 TG C in 2046 (Fig. 10C). The important time to reach the peak depends on the measurement method, and the following order is: The size of the peak of carbon accumulation corresponds to that order: Cutting bamboo grove 1-3DU prohibited, logging. There is no doubt that management measures to secure 30 % of bamboo forest 4DU and 80 % of fell bamboo forest 5DU or more will increase the amount of carbon accumulation of fell bamboo grove.

References

Table 2 Predicted ground carbon accumulation (TG C) for 2024-2060 in Mosuo Bauca Forest in Zohe Province Province.

Sketch 10. Predicted ground carbon accumulation from 2024 to 2060 in the Mosuo Beech Forest in Zhejiang Province. The feature (A-D) is the ground carbon accumulation predicted under four possible measures according to this. Blue dashes are 95%trust separation, and the points that correspond to black dashes are all possible, shor t-term carbon accumulation and peak carbon accumulation. As shown in Table 2, the accumulation of Mosuo beech forests in Zhejiang Province in the management scenarios of Meriro I, II, and II is increasing every day, peaking 40 years later, and subsequent fluctuations are repeated. Stabilize. However, the time and size of the peak and its size vary depending on these different management measures. In the case of countermeasures I, the amount of carbon accumulation of the mosuo beech forest in Zhejiang Province increases every day, reaching the peak of 31, 54 ± 9, 60 tg C in 2053 (Fig. 10a). In the countermeasures II, the peak of Mosuo beech forests in Zhejiang Province increases daily and reaches the peak of 33, 56 ± 7, 53 TG C in 2050 (Fig. 10B). In the measurement II, the accumulation of ground carbon forests in Zhejiang Province increases daily, reaching the peak of 36, 25 ± 8, 47 TG C in 2046 (Fig. 10C). The important time to reach the peak depends on the measurement method, and the following order is: The size of the peak of carbon accumulation corresponds to that order: Cutting bamboo grove 1-3DU prohibited, logging. There is no doubt that management measures to secure 30 % of bamboo forest 4DU and 80 % of fell bamboo forest 5DU or more will increase the amount of carbon accumulation of fell bamboo grove.

Table 2 Predicted ground carbon accumulation (TG C) for 2024-2060 in Mosuo Bauca Forest in Zohe Province Province.

Although methods for assessing current forest carbon stocks at regional scales are well developed, predictions of future supply and carbon stock deviations in forests, especially aboveground carbon stocks in mozobeam forests, are still little studied (Ni, 2013; Hu et al. The model is based on data from in situ fixed soundings during forest inventory implementation, and also provides these points as interactions with area expansion, management measures, tree characteristics, topography and climate. Benefiting from a network of in-depth studies, Hu et al. (2015) predicted hookstocks of organic carbon supply under future scenarios of climate and land use configuration up to 2050, and Bulut (2023) made a prediction of aboveground biomass in the Mediterranean with SVM support. However, Suy M. et al. (2023) predicted changes in mangrove basal area and carbon supply to basals using RF, SVM and XGBOOST methods. Their results were supported by RF. and several others have proven to work in fitted regression, but in fact they contradict our results. The underlying cause of this rise is

The results showed that the average bark density, average diameter at breast height, and average age of Moso bamboo were important factors affecting carbon accumulation, while slope, altitude, and climate had no significant effect on carbon accumulation. Li et al. (2013) pointed out that the main factors affecting carbon accumulation in Moso bamboo forests were the number of stalks and average diameter at breast height, which is consistent with our results. On the other hand, Fan et al. (2012) argued that slope and aspect have a significant effect on carbon supply in Moso bamboo forests. This discrepancy may be related to the dominance of tree structure factors on carbon accumulation over topographic factors, and the smaller effect of slope and aspect on carbon accumulation. Stump density is considered to be one of the most important structural factors in Moso bamboo forests (Yu, 2011; Alam et al., 2012), and DBH of Moso bamboo is one of the important parameters for evaluating environmental biomass and carbon reserves (Bond-Lamberty et al. (2017, 2018) showed that the change in carbon reserves in the aboveground part of Moso bamboo forests was not correlated with the change in carbon reserves in the aboveground part of Moso bamboo forests.

Our evaluation shows the increase in terrestrial carbon supply in Zhejiang Province in the Moso Buna forest, applying the data of the infinite forest inventory and the area covered with the Moso Buna forest within the range. There is. From 1984 to 2014, Bamboo-Moso in Zhejiang Province in Zhejiang Province in Zhejiang Province's 9, 94 to 17, and 19 Tenji C in Zhejiang Province (Satoshi Province) in 2009. In 2012), it increased to 14, 11 TG (Xu, 2017) in 2014, which is basically consistent with our estimation. In previous studies, monitoring of the Moso Buna forest terrestrial carbon was based on data obtained from fixing surveys in the field and was not abundant. In this study, we utilized four types of indicators and predicted the ground carbon supply of the Moso Buna forest according to each layout. As the forecast results match, the amount of carbon accumulation in I, II, and III measures increases and eventually stabilizes. In IV measures, carbon accumulation peaked in 2033, and in 2034, the amount of carbon accumulation in the Moso Buna forest in Zhejiang Province was reduced, reducing the thre e-befor e-three measures. In this way, our evaluation shows the increase in terrestrial carbon supply in Zhejiang Province's Moso Buna Forest, the area covered by the four edges, the data of the infinite forest inventory and the Moso Buna forest. Is applied. From 1984 to 2014, Bamboo-Moso in Zhejiang Province in Zhejiang Province in Zhejiang Province's 9, 94 to 17, and 19 Tenji C in Zhejiang Province (Satoshi Province) in 2009. In 2012), it increased to 14, 11 TG (Xu, 2017) in 2014, which is basically consistent with our estimation. In previous studies, monitoring of the Moso Buna forest terrestrial carbon was based on data obtained from fixing surveys in the field and was not abundant. In this study, we utilized four types of indicators and predicted the ground carbon supply of the Moso Buna forest according to each layout. As the forecast results match, the amount of carbon accumulation in I, II, and III measures increases and eventually stabilizes. In IV measures, carbon accumulation peaked in 2033, and in 2034, the amount of carbon accumulation in the Moso Buna forest in Zhejiang Province was reduced, reducing the thre e-befor e-three measures. In this way, in our evaluation, the increase in terrestrial carbon supply in Zhejiang Province is applied to the four edges and the area covered by the infinite forest inventory data and the area covered by the Moso Buna forest. Is shown. From 1984 to 2014, Bamboo-Moso in Zhejiang Province in Zhejiang Province in Zhejiang Province's 9, 94 to 17, and 19 Tenji C in Zhejiang Province (Satoshi Province) in 2009. In 2012), it increased to 14, 11 TG (Xu, 2017) in 2014, which is basically consistent with our estimation. In previous studies, monitoring of the Moso Buna forest terrestrial carbon was based on data obtained from fixing surveys in the field and was not abundant. In this study, we utilized four types of indicators and predicted the ground carbon supply of the Moso Buna forest according to each layout. As the forecast results match, the amount of carbon accumulation in I, II, and III measures increases and eventually stabilizes. In IV measures, carbon accumulation peaked in 2033, and in 2034, the amount of carbon accumulation in the Moso Buna forest in Zhejiang Province was reduced, reducing the thre e-befor e-three measures. Like

The percentage of the age of Moso bamboo 3DU has no established aspect. According to our research, Meriro III, which provides 30 % of bamboo swing 1-3 DU, bamboo 4 du bamboo, bamboo 5 du, and 80 % of the highest access to 80 % of the other three measures and carbon absorption. He had high legal abilities. I believe that one of the reasons is that the age of beech forests, which triggered logging, to improve the illuminance of beech forests in the Moscow area, contribute to the photosynthesis, increase, and development of Moso bamboo bamboo. It is. Apart from this, logging reduces the outflow of Moso bamboo, improves air circulation, and promotes Moso bamboo breathing and metabolism. Apart from this, healthy logging promotes blood circulation and hig h-calorie drug updates, enhances fertility, forms a calorie residue suitable for lifting Moso bamboo (ZENG and 2019; Zheng etc., 2022). The percentage of the age of Moso bamboo 3DU has no established aspect. According to our research, Meriro III, which provides 30 % of bamboo swing 1-3 DU, bamboo 4 du bamboo, bamboo 5 du, and 80 % of the highest access to 80 % of the other three measures and carbon absorption. He had high legal abilities. I believe that one of the reasons is that the age of beech forests, which triggered logging, to improve the illuminance of beech forests in the Moscow area, contribute to the photosynthesis, increase, and development of Moso bamboo bamboo. It is. Apart from this, logging reduces the outflow of Moso bamboo, improves air circulation, and promotes Moso bamboo breathing and metabolism. Apart from this, healthy logging promotes blood circulation and hig h-calorie drug updates, enhances fertility, forms a calorie residue suitable for lifting Moso bamboo (ZENG and 2019; Zheng etc., 2022). The percentage of the age of Moso bamboo 3DU has no established aspect. According to our research, Meriro III, which provides 30 % of bamboo swing 1-3 DU, bamboo 4 du bamboo, bamboo 5 du, and 80 % of the highest access to 80 % of the other three measures and carbon absorption. He had high legal abilities. I believe that one of the reasons is that the age of beech forests, which triggered logging, to improve the illuminance of beech forests in the Moscow area, contribute to the photosynthesis, increase, and development of Moso bamboo bamboo. It is. Apart from this, logging reduces the outflow of Moso bamboo, improves air circulation, and promotes Moso bamboo breathing and metabolism. Apart from this, healthy logging promotes blood circulation and hig h-calorie drug updates, enhances fertility, forms a calorie residue suitable for lifting Moso bamboo (ZENG and 2019; Zheng etc., 2022).

In fact, in our research, in the first year, homework refusal, beech forest moso density remained at the applicable level, and thanks to the appropriate use of forest waste light and water without human intervention. (INOUE et al.) For a lon g-term no n-control period, excessive bamboo height has led to a decrease in light use and increased mortality in beech forests. (Bell, 1997) actually leads to a decrease in carbon accumulation on the surface. In fact, appropriate appropriate surveys have shown that the productivity and legal abilities of moso beech forests that absorb carbon are associated with control control measures. Partial logging of Moso bamboo is considered necessary to maintain the productivity of Moso bamboo (Kuehl et al., 2013; Mao et al., 2017b). By adjusting the aging structure of the mosa bamboo forest, it is possible to improve the carbon accumulation of carbon on the ground and achieve the sustainable formation of the mosa bamboo forest. MAO et al. (2017a) is a high-quality management model that enables bamboo fitting, and the perfect bamboo-swing-swing-swing capacity to grow carbon accumulation of beech forests by 74, 63 %. in this terrestrial carbon accumulation, in fact, in our research, the first year, the density of homework, and the density of beech forests remained at the applicable level, and in fact for forest waste without human intervention. Extr a-bamboo height is the possibility of light use for lon g-term control periods that have contributed to bamboo rejection thanks to the appropriate use of light and water. The decrease in, and the increased mortality rate in beech forests (bell, 1997) actually leads to a decrease in carbon accumulation on the surface. In fact, appropriate appropriate surveys have shown that the productivity and legal abilities of moso beech forests that absorb carbon are associated with control control measures. Partial logging of Moso bamboo is considered necessary to maintain the productivity of Moso bamboo (Kuehl et al., 2013; Mao et al., 2017b). By adjusting the aging structure of the mosa bamboo forest, it is possible to improve the carbon accumulation of carbon on the ground and achieve the sustainable formation of the mosa bamboo forest. MAO et al. (2017a) is a high-quality management model that enables bamboo fitting, and the perfect bamboo-swing-swing-swing capacity to grow carbon accumulation of beech forests by 74, 63 %. In fact, in this terrestrial carbon accumulation, in our research, the first year, then rejection of homework, and the density of beech forest posters remained at the application level, and the light of forest waste without human intervening. It indicates that it has contributed to bamboo rejection thanks to the appropriate use of water (INOUE et al.) For a lon g-term no n-control period, excess bamboo height decreases the possibility of light use. Connections, an increase in mortality in beech forests (bell, 1997) actually lead to a decrease in carbon accumulation on the surface. In fact, appropriate appropriate surveys have shown that the productivity and legal abilities of moso beech forests that absorb carbon are associated with control control measures. Partial logging of Moso bamboo is considered necessary to maintain the productivity of Moso bamboo (Kuehl et al., 2013; Mao et al., 2017b). By adjusting the aging structure of the mosa bamboo forest, it is possible to improve the carbon accumulation of carbon on the ground and achieve the sustainable formation of the mosa bamboo forest. MAO et al. (2017a) is a high-quality management model that enables bamboo fitting, and the perfect bamboo-swing-swing-swing capacity to grow carbon accumulation of beech forests by 74, 63 %. With this ground carbon accumulation

The main uncertainty in this study appears in three fields: data collection, model monitoring, and future transformer monitoring. When collecting data in areas where selection is fixed, artificial mistakes are inevitable. For example, the diameter of the Moso bamboo hill and the evaluation of the age of the tree, which causes an error. At the same time, climate data is obtained from climate software and includes the highest level of uncertainty about the origin of past and future weather data. The choice of model characteristics in all kinds of machine learning and optimization methods is caused by random mistakes that act on the accuracy of the model, leading to uncertainty in carbon accumulation monitoring (Liu et al.) Apart from this. Moso's beech forest suggests a special forest image that the number of years is huge and small. We predicted these characteristics of trees as the average dialect of chest altitude, the average density of the trunk, and the age of the trunk. This monitoring was simulated without considering artificial conditions, environmental impacts, or unconditional solid standards. These moments accurately influenced the carbon accumulation in the above section of the Mozo beech forest, and made important considerations for the confusion of beech forests. The main uncertainty in this study appears in three areas: data collection, model monitoring, and future transformation monitoring. When collecting data in areas where selection is fixed, artificial mistakes are inevitable. For example, the diameter of the Moso bamboo hill and the evaluation of the age of the tree, which causes an error. At the same time, climate data is obtained from climate software and includes the highest level of uncertainty about the origin of past and future weather data. The choice of model characteristics in all kinds of machine learning and optimization methods is caused by random mistakes that act on the accuracy of the model, leading to uncertainty in carbon accumulation monitoring (Liu et al.) Apart from this. Moso's beech forest suggests a special forest image that the number of years is huge and small. We predicted these characteristics of trees as the average dialect of chest altitude, the average density of the trunk, and the age of the trunk. This monitoring was simulated without considering artificial conditions, environmental impacts, or unconditional solid standards. These moments accurately influenced the carbon accumulation in the above section of the Mozo beech forest, and made important considerations for the confusion of beech forests. The main uncertainty in this study appears in three fields: data collection, model monitoring, and future transformer monitoring. When collecting data in areas where selection is fixed, artificial mistakes are inevitable. For example, the diameter of the Moso bamboo hills and the evaluation of the age of the tree, which causes an error. At the same time, climate data is obtained from climate software and includes the highest level of uncertainty about the origin of past and future weather data. The choice of model characteristics in all kinds of machine learning and optimization methods is caused by random mistakes that act on the accuracy of the model, leading to uncertainty in carbon accumulation monitoring (Liu et al.) Apart from this. Moso's beech forest suggests a special forest image that the number of years is huge and small. We predicted these characteristics of trees as the average dialect of chest high peaks, the average density of the trunk, and the age of the trunk. This monitoring was simulated without considering artificial conditions, environmental impacts, or unconditional solid standards. These moments were accurately influenced the carbon accumulation in the above section of Mozo beech forest, and made important considerations for the confusion of beech forests.

Moso bamboo forest is a unique forest that is different from other forests. The ability to bind carbon is extremely high, and the products that can be obtained are also essential for recycling and carbon recycling (LI et al.) According to the research so far, the productivity and carbon absorption capacity of Moso bamboo grove depends on the control measures. Selective logging (Li et al.) Sharp is an approach necessary to ensure the stability of Moso bamboo grove. By refusing management, the carbon reserves may increase rapidly in the test site, but the carbon reserves on the ground are still reduced. Nevertheless, the carbon reserved amount on the ground will continue to decrease due to the high density of the shrub and the small DBH due to lack of management (Yin et al., 2019). In our research, the bamboo size 3 du or less was saved during the Wheelhouse I, and the bamboo size 4 du or more was not completely cut. However, according to other studies, bamboo 4 du also has a certain level of photosynthesis. With the completely purified old bamboo, nutrients could not be supplied directly to the new bamboo, hindering growth and growth, and gradually increasing the carbon reserves in the above portion. Therefore, as part of the II-event, we decided to collect only 50 % of bamboo from bamboo with low photosynthetic ability. Moso bamboo grove is a unique forest different from other forests. The ability to bind carbon is extremely high, and the products that can be obtained are also essential for recycling and carbon recycling (LI et al.) According to the research so far, the productivity and carbon absorption capacity of Moso bamboo grove depends on the control measures. Selective logging (Li et al.) Sharp is an approach necessary to ensure the stability of Moso bamboo grove. By refusing management, the carbon reserves may increase rapidly in the test site, but the carbon reserves on the ground are still reduced. Nevertheless, the carbon reserved amount on the ground will continue to decrease due to the high density of the shrub and the small DBH due to lack of management (Yin et al., 2019). In our research, the bamboo size 3 du or less was saved during the Wheelhouse I, and the bamboo size 4 du or more was not completely cut. However, according to other studies, bamboo 4 du also has a certain level of photosynthesis. With the completely purified old bamboo, nutrients could not be supplied directly to the new bamboo, hinder growth and growth, and gradually increased the carbon reserves in the above portion. Therefore, as part of the II-event, we decided to collect only 50 % of bamboo from bamboo with low photosynthetic ability. Moso bamboo forest is a unique forest that is different from other forests. The ability to bind carbon is extremely high, and the products that can be obtained are also essential for recycling and carbon recycling (LI et al.) According to the research so far, the productivity and carbon absorption capacity of Moso bamboo grove depends on the control measures. Selective logging (Li et al.) Sharp is an approach necessary to ensure the stability of Moso bamboo grove. By refusing management, the carbon reserves may increase rapidly in the test site, but the carbon reserves on the ground are still reduced. Nevertheless, the carbon reserved amount on the ground will continue to decrease due to the high density of the shrub and the small DBH due to lack of management (Yin et al., 2019). In our research, the bamboo size 3 du or less was saved during the Wheelhouse I, and the bamboo size 4 du or more was not completely cut. However, according to other studies, bamboo 4 du also has a certain level of photosynthesis. With the completely purified old bamboo, nutrients could not be supplied directly to the new bamboo, hinder growth and growth, and gradually increased the carbon reserves in the above portion. Therefore, as part of the II-event, we decided to collect only 50 % of bamboo from bamboo with low photosynthetic ability.

As a result, it was found that the carbon accumulation of Mosuo bamboo grove varies depending on the difference in management measures. Excellent planned management measures are useful for understanding the dynamics of carbon accumulation in the bamboo grove, leading to vast area management, and achieving China's 2 goals.

In this study, the optimal carbon accumulation model was built in the Moso bamboo forest in Zhejiang Province. They used continuous definition data, climate data, and machine learning algorithms at the fixed sample point of Zhejiang Province from 2004 to 2019. In this study, the amount of carbon accumulation on the ground of the Moso bamboo grove in Zhejiang Province was predicted under different management means and the optimal management strategy was identified. The optimal management measures were 1 to 3DU of bamboo grove, 30 % of bamboo grove 4DU, and 80 % of bamboo forest 5DU or more. With the optimal management strategy, the amount of carbon accumulation on the ground part of Zhejiang Province is expected to reach its peak in 2046, and carbon isolation capabilities can maintain a stable level. Furthermore, our results show that the deteriorated Moso bamboo grove may increase in the short term, but may not be maintained in the long term. This situation is disadvantageous for achieving carbon neutral goals. In addition, it was found that it is not only the climate to determine the change in carbon accumulation. Changes in carbon accumulation on the ground have a strong correlation between average trunk density, average breast height, and average tree height.

The original charges published in this study are included in this paper / supplementary material. Further questions are to the corresponding author. As a result, it was found that the carbon accumulation of Mosuo bamboo forests differs depending on the difference in management measures. Excellent planned management measures are useful for understanding the dynamics of carbon accumulation in the bamboo grove, leading to vast area management, and achieving China's 2 goals.

The original charges published in this study are included in this paper / supplementary material. Further questions are to the corresponding author. As a result, it was found that the carbon accumulation of Mosuo bamboo grove varies depending on the difference in management measures. Excellent planned management measures are useful for understanding the dynamics of carbon accumulation in the bamboo grove, leading to vast area management, and achieving China's 2 goals.

The original charges published in this study are included in this paper / supplementary material. Further questions are to the corresponding author.

SL: writing-original concept, writing-review and editing. NY: writing-review and editing. XS: writing-review and editing. XC: writing-review and editing. YS: writing-review and editing. GZ: writing-review and editing. LX: writing-review and editing.

The authors indicate financial support for the research, authorship and/or publication of the submitted notes. This work was supported by the National Natural Science Foundation of China (Grant No: 32001315).

We thank the editors and reviewers who contributed to the peer review of this work.

Indeed, according to the authors, the work was conducted without any paid or financial relationships that could be construed as a potential conflict of interest.

All statements in this notice are those of the authors and do not necessarily reflect the views of their affiliates, the publisher, the editors, or the reviewers. Any statements by any products reviewed in this book, or by their manufacturers, are not endorsed or approved by the publisher.

Alam, A., Kilpeläinen, A. and Kellomäki, S. (2012). Effects of initial stand density and thinning regime on CO2 emissions associated with production and management of energy tree species in boreal ecosystems. Eur. J. For. 131, 655-667.

Awad, M., Khanna, R., Awad, M., and Khanna, R. (2015). Vector regression using Efficient Learning. Mach. Theor. ConceptsAppl. Eng. System. 4, 67-80. doi: 1007/978-1-4302-5990-9_4.

Bell, DT (1997). 18-year changes in a deciduous forest along a stream in Illinois. J. Torrey Bot. 124, 174-188.

Bond-Lamberty, B., Wang, K., & amp; amp; Gower, ST (2002). Allometric comparison of above-ground and below-ground biomass and fusiform wood area of six boreal tree species in northern Manitoba. Can. J. For. doi: 10. 1139/x02-063.

Breiman, L. (2001). Random forests. Mach. Learning. 45, 5-32. doi: 10. 1023/A:1010933404324.

Bulut, S. (2023). Calabrian pine (Pinus brutia Ten.) in the Mediterranean region Turkey. Ecol. Ecol. 74, 101951. doi: 10. 101951

Camargo García, JC, Arango Arango, AM, & amp; amp; Trinh, L. (2023). Allometric equations to estimate the potential of beech forests as carbon sinks and their aboveground biomass. Environment. Developer Sustain. 26, 1-29.

Chen, T., and Gustrin, K. (2016). "Xgboost: a scalable tree boosting system," in Proceedings of the 22nd Acm Sigkdd International Conference on Knowledge Discovery and Data Mining, (New York, NY: ACM), 785-794. doi: 10. 1145/2939672. 2939785

Xgboost: Extreme gradient amplification. Bundle R version 0. 4 1, 1-4.

Chunxia, Z., and Xingcui, D. (1997). Study on productivity of Phyllostachys heterocycla var. pubesens. J. Bamboo Res. 16, 31-36.

Du, H., Mao, F., Zhou, G., Li, X., Xu, X., Ge, H., et al. Estimation and validation of spatiotemporal structure of aboveground carbon in beech forests by combining remote sensing data and an improved bio-OM-BGC model. IEEE J. Sel. Top. Appl. Earth Observ. 11, 2282-2295. doi: 10. 1109/JSTARS. 2018. 2817344

Fan, Y., Zhou, G., Shi, Y., Dong, D., and Zhou, Y. (2012). Hangzhou City, Zhejiang Province, Hangzhou City, Zhejiang Province, Hangzhou City, Zhejiang Province. J. Zhejiang A& amp; amp; F Univ. 29, 321-327.

Gomez, L. C., Faria, R. M., de Souza, E., Veloso, G. V., Schaefer, C. E. G. R., & amp; amp; Fernandes Filho, E. I. (2019). Modeling and mapping basal organic carbon stocks in Brazil. Geoderma 340, 337-350. doi: 10. 1016/j. geoderma. 2019. 01. 007

Grange, SC, Carslow, DC, Lewis, AC, Boleti, E., and Hueglin, C. (2018). A random forest-based meteorological regularization model for PM10 trend analysis in Switzerland. Atmos. Chem. Phys. 18, 6223-6239. doi: 10. 5194/acp-18-6223-2018

Han, N., Du, H., Zhou, G., Xu, X., Cui, R., and Gu, C. (2013). Spatiotemporal heterogeneity of aboveground carbon accumulation in Moso bamboo from Landsat Thematic Mapper images: A case study of Anji County, China. Int. J. Remote sensing. 34, 4917-4932. 2013. 782115

Hu, H., Wang, S., Guo, Z., Xu, B., and Fang, J. (2015). Matrix model with structured boundaries predicts a significant increase in carbon reserves in China's forest biomass between 2005 and 2050. Scientific Reports 5:11203.

Inoue, A., Sato, M., and Shima, H. (2018). Critical volume density response in beech forests: A case study of Phyllostachys pubescens forests in the Land of the Rising Sun. For. Ecol. Manage. 425, 138-144. doi: 10. 1016/j. foreco. 2018. 05. 044

Ji, B., Tao, J., Zhang, G., Yao, H., and Xu, J. (2012). Assurance of accuracy of forest vegetation biomass estimation in Zhejiang Province. J. Zhejiang A& amp; amp; F Univ. 29, 328-334.

Joachims, T. (1998). Practical application of large-scale SVM learning. Germany: TU Dortmund.

Kaam, R., Tchamb, M., Nforka, B. N. and Jomo, K. K. (2024). Remote sounding, allometry and carbon stocks in the Western Highlands of Cameroon. Adv. Bamboo Sci. 6: 100055. 1016/j. bamboo. 2023. 100055

Kang, F., Li, X., Du, H., Mao, F., Zhou, G., Xu, Y., et al. (2022). Spatiotemporal evolution of carbon fluxes from bamboo forests in China and their response to climate change based on the BPS model. External Sens. 14: 366. Doi: 10. 3390/rs14020366

Kuehl, Y., Li, Y. and Henley, G. (2013). The effect of selective cutting on carbon absorption potential in bamboo mos o-bamboo (Phylostachys pubescens). For. Trees Livelihoods 22, 1-18. Doi: 10. 1080/14728028. 2013. 773652

Li, C., Zhou, G., Shi, Y., Zhou, Y., Xu, L., Fan, Y., et al. Effects of different management practices on net carbon absorption potential in moso bamboo forest ecosystems. Sci. Silvae sin. 53, 1-9.

Li, J., Cheng, J., Shi, J. and Huang, F. (2012). A brief introduction to the neural network inverse distribution (BP) algorithm and its improvement. Adv. Comput. Sci. Inform. English. 2, 553-558. 1007/978-3-642-30223-7_87.

Lee, L., L., Lee, N., Lou, D. and Chen, J. (2019). Control of annual and additional leg distribution in bamboo forests in Moscow region and subtropical areas using Sentinel-2 and Landsat 8 pseudosequences. Remote Sens. Environ. 231: 111265. 1016/j. rse 2019. 111265 Li, L.-E., Lin, Y.-J. and Yen, T.-M. (2016). Allometric models to predict aerobatic bamboo (Bambusa stenostachya) biomass and evaluate carbon storage. Taiwan J. Sci. 31, 37-47. Li, P ., Zhou, G., Du, H., Lu, D., Mo, L., Xu, X., et al. Current and potential carbon reserves in Menso bamboo forests in China. J. Environ. Manage. 156, 89-96. 1016/j. jenvman 2015. 03. 030

Li, X., Du, H., Mao, F., Zhou, G., Chen, L., Xing, L., et al. Estimation of aerial biomass of bamboo forests using spatiotemporal data of Modis Lai, assimilated ENKF and machine learning algorithms. Agric. For. Meteorol. 256, 445-457. Doi: 10. 1016/j. agrformet. 2018. 04. 002

Li, Y., Han, N., Li, X., Du, H., Mao, F., Cui, L., et al. Spatiotemporal assessment of aboveground carbon stocks of bamboo forests based on Landsat data in Zhejiang Province, China. External Sens. 10, 898.

Li, Y., Xu, Z., Hao, Z., YAO, X., Zhang, Q., Huang, X., Others (2023). Hyper Spectral UAV A comparative study of RFE and machine learning algorithms in the lines of Moso bamboo (Phyllostachys Pubescens). Geocarto int. 38: 2207550. 2023. 2207550

Liemohn, M. W., McCollow, J. P., Jordanova, V. K., NGWIRA, C. M., Morley, S. K., C., C., et al. Sp. Weather 16, 2079-2102. DOI: 10. 1029/2018sw002067.

Lin, H., and Xiaoyong, H. (1993). Hazel Phyllostachys Pubescens's reasonable age structure and growth rule in the shooting period. J. Bamboo Res. 12, 52-56.

LIU, E. Ren, Z., YAO, H., and Zhou, G. (2021). Mult i-scale carbon accumulation estimation based on the mixture of Moso bamboo (PHYLLOSTACHYS edulis). Glob. Ecol. Retained. 31: E01856. Doi: 10. 1016/j. Gecco. 2021. E01856

LIU, E. SHI, Y., Li, Y., Zhou, G., and Yang, D. (2013). 1-7.

LIU, Y. Y., Mao, F. J. J. J. J. J., XING, L. Q., L. Q., L. F., Zheng, J. L., et Al. J. Appl. Ecol. 30, 1743-1753.

LIU, W., Chen, Z., and Hu, Y. (2022). Int. j. pres. Ship. Pip. 197: 104655. Doi: 10. 1016/j. IJPVP. 2022. 104655

Lv, Y., Han, N., and Du, H. (2023). SPOT-6 Image An estimation of bamboo forest ground carbon using an RGLM model based on the object-based multi-scale segmentation. Sensal port 15: 2566.

MAO, F., Du, H., Zhou, G., Li, Xu, XU, X., Li, P., et al. LAI assing and Beps model binding to analyze. Agriculture. For. Meteorol. 242, 96-108. DOI: 10. 1016/j.

Mao, F., Zhou, G., Li, P., P., H., Xu, Xu, X., Y., et. Bime-BGC's maximum carbon accumulation and yield of the Moso bamboo forest using a BGC model Optimization of optimization strategy for converting. J. ENVIRONMENT. Manage. 191, 126-135. Doi: 10. 1016/jenvman. 2017. 01. 016

Mao, F., Lee, P., Zhou, G., Du, H., Xu, X., Shi, J., etc. (2016). Development of a biome BGC model for modeling the control ecosystem of beech forests. J. Environ. Manage. 172, 29-39. 1016/j. jenvman 2015. 12. 013

Ni, J. (2013). Carbon accumulation in soil ecosystems in China: approaching a more definitive assessment. Clim. Change 119, 905-917. 1007/s10584-013-0767-7.

Ouimet, R., Camire, C., Brazeau, M. and Moore, J.-D. (2008). Evaluation of crude root biomass and content of Caloriver formulations in introduced sweet maple, pine and dark pine. Can. J. for. Res. 38, 92-100. Doi: 10. 1139/x07-134

Pan, K., Zhou, G., Schrestha, A. K., Chen, J., Kozak, R., Lee, N., et al. (2023). Biomass as a lead for nature-based conclusions (NBS) to reduce climate configuration consequences. Prod. Carbon credits climb. 11: 175.

Poorazimy, M., Shatae, S., McRoberts, R. E. and Mohammadi, J. (2020). Integrating data from gravity-free laser scanning, galactic radar data and digital aerial photography to assess carbon stocks in the above-mentioned intervals of a Hyrcanic forest, Iran. External sens. Environ. 240: 111669. 1016/j. rse. 2020. 111669

Qi, S., Song, B., Liu, K., Gun, P., Luo, J., Zhang, M., et al. (2022). Habitat segregation of beech forests in China by the introduction of Landsat 8 Pictures and cryptic archives from Google Earth. External Sens. 14: 762. Doi: 10. 3390/rs14030762

Ramakrishnan, M., Iryalya, K., Vinod, K. K., Sharm, A., Cho, J., Sathish, V. and others (2020). Genetics and genome of bamboo moso (Phyllostachys edulis): current position, future difficulties and biotechnological techniques for creating a stable beech industry. Food Energy Secur. 9: 229. 229

Rersen, G., Tao, D., Lutens, B., Klemmer, K., Amara, K., Steinogger, A. and others (2022). "Reforestree: a dataset for estimating tropical forest carbon stock with deep learning and aerial photography," Proceedings of the AAAI Conference on Artificial Intelligence, (Washington, DC: AAAI Press), 12119-12125.

Saltelli, A. (2002). A proper implementation of the sensitivity index calculation model. Comput. Phys. Commun. 145, 280-297. 1016/s0010-4655 (02) 00280-1

Shan, Z., Zhou, G., Du, H., Xu, H., Shi, J., Liu, J., et al. (2013). Estimation of distribution and aboveground carbon stock of Mozong beech forests based on remote sensing images from different sources. Int. J. Remote Sens. 34, 5351-5368. Doi: 10. 1080/01431161. 2013. 788260

SHI, Y., Xu, L., Ji, Ji, B., Zhou, G., H., et. Mosuo Bamboo Forests Using a Machine Learning Algorithm Couped with a STRUCTURAL Tion Model. for. Ecol. Manage. 429, 406-413. DOI: 10. 1016/j. Foreco. 2018. 07. 035

SILATSA, F. T., Yemefak, M., Tabi, HEUVELINK, G. B. M., & Amp; Leenaars, J. G. B. (2020). Organic carbon accumulation Estimate. Geoderma 367: 114260. DOI: 10. 1016/J. Geoderma. 2020. 114260

Tang, X., Zhao, X., Bai, Y., Tang, Z., Wang, W., Zhao, Y., et al. Carbon pool in the Chinese land ecosystem: New under intensive field survey Estimated value. Prop. Nat. Acad. Science USA 115, 4021-4026. DOI: 10. 1073/PNAS. 1700291115

Terefe, R., jian, l., and kunyong, y. (2019). J. Sci. RES. 24, 1-7. Doi: 10. 9734/JSRR/2019/V24I130145

YUNIYAL, S., Purohit, S., Charasia, K., Rao, SS, and Amminedu, E. (2022). Landsat 8 OLI in India and Jodple and Machinery Learning Algorithm 。 Urban Before. Urban Before. 67: 127445. DOI: 10. 1016/J. UFUG.

VLAHOGIANNIS, DM, Moura, S., and McFarlane, J. (2023). Intersense: XGBOOST model for identification of traffic control officers at intersections using GPS data collected from crowds. Transp. RES. Part C Orig. Technology 151: 104112. 2023. 104112

Wang, B., WEI, WJ, LIU, KJ, YU, WZ, WZ, NIU, H., and Man, Rz (2013). Biomass and carbon accumulation in the tropical tropical forest: its features and meaning. J. TROP. For. Science 25, 137-148.

Wang, T., Wang, G., Innes, JL, B., and Chen, B. (2017). Climateap: Applications for dynamic and future climate data in the Asi a-Pacific region 。 cover. Agriculture. Science English. 4, 448-458. DOI: 10. 15302/J-FAS E-201722.

WU, BS (1984). Preliminary analysis on the relationship between the underground structure and yield of bamboo grove. J. Bamboo Res. 3, 49-58.

XU, L., SHI, Y., Zhou, G., Xu, X., LIU, E., Zhou, et al. Promotion factor. Forest 8: 371.

XU, L., SHI, Y., Zhou, G., Xu, X., LIU, E., Zhou, et al. In Moso Buna Forest in Zhejiang Province, China Zhejiang Province and Dynamics. For. Ecol. Manage. 409, 479-488. DOI: 10. 1016/J. FORECO 2017. 057

XU, M., Sun, C., Du, Z. and ZHU, X. Sci. TOTAL ENVIRONMENT. 859: 160173. 10. 1016/J. SciTotenv. 160173

XU, XU, Du, H., Zhou, Zhou, G., Ge, H., Shi, Y., Zhou, Y., et al. In the case of beech forest mosa, a photo support with a landscopic landser mopper evaluation. Remote sensing. 32, 1431-1448. 10. 10. 1080/0143160351389.

XU, Y., Ge, X., Gao, G., Yang, Y., Hu, Y., Li, Z., Others (2023). All kinds of microorganisms and plant carbon for land organic carbon in untrained mozozobuna forests. Catena 233: 107481. Doi: 10. 1016/j. Catena. 2023. 107481.

Elim Poon - Journalist, Creative Writer

Last modified: 27.08.2024

This study underscores the significant influence of estimating carbon sequestration potential and optimizing management decisions on enhancing and. This study underscores the significant influence of estimating carbon sequestration potential and optimizing management decisions. Similar Papers · Estimating carbon sequestration potential and optimizing management strategies for Moso bamboo (Phyllostachys pubescens) forests using machine.

Play

accepted