Prediction method for urban forest overground biomass different-speed growth simulation
By combining field measurement and drone remote sensing data, a power-law model is constructed to estimate urban forest over-ground biomass, solving the problem of traditional model over-fitting, and achieving high-accurate over-ground biomass monitoring and urban forest carbon sequestration capacity assessment.
Patent Information
- Application Number
- CN202510044916.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-13
- Publication Date
- 2025-05-13
AI Technical Summary
Traditional linear models and machine learning models are prone to overfitting problems when estimating biomass on urban forest land, resulting in large monitoring errors and making it difficult to achieve spatially continuous biomass predictions on urban forest land.
By measuring the above-ground biomass in urban forests on-site, obtaining the diversity parameters of urban forest structures, building a power-law model to estimate the above-ground biomass, and optimizing the model parameters through nonlinear least squares method and forward stepwise selection method.
It effectively reduces the error in on-ground biomass estimation, avoids data overfitting, realizes spatially continuous on-ground biomass monitoring in urban forests, and improves the accuracy of assessment of carbon sequestration capabilities in urban forests.
Smart Images
Figure CN119992378A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the fields of landscape ecology remote sensing, forestry carbon sink remote sensing and landscape gardening digital technology, and particularly relates to a prediction method for allometric growth simulation of above-ground biomass in urban forests. Background Art
[0002] Urban forests can store large amounts of carbon through canopy CO2 and oxygen exchange, play a key role in the global carbon cycle and climate change mitigation, and are the main source of residents' health and well-being. In recent decades, the rapid development of cities has caused a rapid expansion of impervious surfaces around the world, destroying nearly 51,018 km 2 Forests have exacerbated the extreme trend of global warming. Accurately quantifying the aboveground biomass of urban forests has long been an important task for assessing forest carbon storage capacity and maintaining urban carbon balance. Due to the complex environment within the city, limited site equipment and insufficient forest resources to support large-scale monitoring, traditional aboveground biomass measurement methods based on in-situ observations (such as flux tower monitoring) and destructive logging are limited. The allometric growth equation method based on breast diameter or tree height relies on time-consuming and labor-intensive data collection. Traditional empirical models and digital models (such as Citygreen, iTree) are poorly transferable in urban forests. How to achieve spatially continuous monitoring of aboveground biomass in urban forests remains a major challenge.
[0003] In recent years, the rapid development of remote sensing technology has provided new opportunities for monitoring aboveground biomass in urban forests. UAV-borne LiDAR and multispectral have unprecedented advantages in quantifying different structures of urban forests, such as physical arrangement, three-dimensional volume, and physiological characteristics. Through regional-based methods, different structural characteristics of urban forests, such as canopy height, coverage and openness, heterogeneity, and stand density, can be parameterized. However, the aboveground biomass monitoring of urban forests combined with remote sensing technology currently mostly uses linear models (such as stepwise regression and full subset regression) or machine learning models (such as random forests and support vector machines), which focus too much on the input of structural parameters, which can easily cause overfitting of the model and lead to large errors in the monitoring of aboveground biomass in urban forests.
[0004] A large number of studies have shown that there is a relative growth relationship between the aboveground biomass and tree structure in urban forests. In the entire urban forest, the size, structure and spatial layout of trees are the specific manifestations of functionally invariant xylem elements bundled together to conduct water and nutrients from the trunks and branches upward to the leaves of each tree. The scale invariance of this relative growth relationship allows the aboveground biomass of individual trees, populations and communities in urban forests to be expressed in the form of a power function based on structural characteristics at different scales. In particular, when allometric growth modeling is performed using different aspects of urban forest structural parameters, it can not only greatly reduce the estimation error of aboveground biomass, but also avoid data overfitting. However, there is no technology to accurately monitor aboveground biomass through the allometric growth relationship of urban forest structural diversity. Summary of the invention
[0005] The present invention provides a prediction method for allometric growth simulation of urban forest aboveground biomass, which is used to solve the overfitting problem when traditional linear models and machine learning models are used to estimate the aboveground biomass of urban forests, and to solve the difficult problem of predicting the aboveground biomass of spatially continuous urban forests.
[0006] The present invention is implemented by the following technical methods:
[0007] Step 1: Field measurement of aboveground biomass of urban forests: sample plots were set up in the dominant communities in the study area, and RTK was used to locate the corner points of the sample plots. The height and DBH structure of the trees in the sample plots were then measured. Based on the measured structural parameters, the aboveground biomass of urban forests in the sample plots was calculated using the allometric growth model.
[0008] Step 2: Obtaining the structural diversity parameters of urban forests. Obtain the original point cloud and multispectral data of urban forests within the study area through drone-borne laser radar and drone-borne multispectral data. Preprocess the two types of data and extract six types of structural parameters of urban forest canopy height, coverage and openness, heterogeneity, stand density, three-dimensional volume and physiological characteristics in the sample plot from the preprocessed point cloud data and multispectral data to form a set of structural diversity parameters for predicting the aboveground biomass of urban forests.
[0009] Step 3: Construction of an urban forest aboveground biomass estimation model. Based on the aboveground biomass of urban forests in the sample plot obtained in step 1 and the urban forest structural diversity index obtained in step 2, correlation analysis was first performed on the relationship between each structural diversity index and the aboveground biomass obtained in step 1. The structural parameters with the best correlation in each structural category were determined as the optimal parameters for estimating the aboveground biomass of urban forests. A power law model for estimating aboveground biomass was constructed. All combinations of optimal parameters were tested using the nonlinear least squares method and forward stepwise selection method to determine the optimal model for estimating the aboveground biomass of urban forests.
[0010] Step 4: Monitoring of aboveground biomass of urban forests. Based on the optimal model obtained in step 3, the aboveground biomass of urban forests in the study area was calculated.
[0011] Moreover, the sample plot size in step 1 is 10m×10m, and the sample plots are evenly distributed in the study area.
[0012] Moreover, the preprocessing of the two types of data obtained in the step 2 includes applying the lidR package in the R software to perform point cloud filtering, digital ground model interpolation, and normalization on the original point cloud data to extract the canopy height model CHM; and performing geometric correction, image registration, and image fusion on the multispectral data in the Pix4Dmapper photogrammetry software to generate an orthophoto map.
[0013] Moreover, extracting structural parameters from the preprocessed point cloud data and multispectral data in the step 2 means: according to the sample corner points in step 1, drawing the sample shape file along the sample corner points using ArcGIS software, then importing the sample shape file into R software, and applying a region-based method to extract urban forest structural parameters from the preprocessed point cloud data and multispectral data.
[0014] Furthermore, the six categories of structural parameters are extracted in step 2 to form a structural diversity parameter set for predicting the aboveground biomass of urban forests, including the following indicators:
[0015] Moreover, the power law model formula in step 3 is as follows:
[0016]
[0017] In the formula, AGB is the aboveground biomass of urban forests, a0 is the estimated coefficient of AGB, and P h is the canopy height structural parameter, P c is the coverage or openness structural parameter, P d is the stand density structure parameter, P sh is the heterogeneous structural parameter, P v is the three-dimensional volume structure parameter, and Pt is the physiological characteristic structure parameter.
[0018] Moreover, in the step 3, all combinations of optimal parameters are tested by nonlinear least squares method and forward stepwise selection method, including: applying nonlinear least squares method in R software to derive coefficients of power law model, adding optimal parameters as model variables in sequence starting from zero feature model by forward stepwise selection method; performing bootstrapping resampling for each model 1000 times, and the goodness of fit of the model after resampling R 2 , root mean square error RMSE and mean absolute error MAE are used to evaluate the model performance.
[0019] Moreover, as mentioned above, the model fit goodness R 2 The calculation formula is as follows:
[0020]
[0021] In the formula, n is the number of samples, y is i and represent the aboveground biomass of urban forest obtained in step 1 and the aboveground biomass of urban forest calculated by the model in sample plot i, respectively. Represents the average value of aboveground biomass of urban forest obtained in step 1.
[0022] Moreover, as mentioned above, the calculation formula of the model root mean square error RMSE is as follows:
[0023]
[0024] In the formula, n is the number of samples, y is i and They represent the aboveground biomass of urban forest obtained in step 1 in sample plot i and the aboveground biomass of urban forest calculated by the model.
[0025] Moreover, as mentioned above, the calculation formula of the model mean absolute error MAE is as follows:
[0026]
[0027] In the formula, n is the number of samples, y is i and They represent the aboveground biomass of urban forest obtained in step 1 in sample plot i and the aboveground biomass of urban forest calculated by the model.
[0028] Moreover, in step 4, the fishing net tool in ArcGIS software is used to create a 10m×10m grid shape file covering the study area, and the regional-based method is applied by R software to extract the urban forest structural parameters of the study area from the preprocessed point cloud data and multispectral data, and the aboveground biomass of the urban forest in the study area is calculated according to the optimal model of step 3.
[0029] The present invention uses ground surveys, UAV remote sensing data and geographic information system technology in a comprehensive manner, based on the physical significance of biomass production and metabolism in urban forest growth. It can conveniently, quickly and accurately calculate the aboveground biomass of urban forests, and realize spatially continuous aboveground biomass monitoring of urban forests, which has important guiding significance for improving the carbon fixation capacity of urban forests.
[0030] The method for monitoring aboveground biomass of urban forests of the present invention is simple, and data acquisition and calculation are convenient, and can be widely applied to the systematic management of urban forest structures and ecological functions. BRIEF DESCRIPTION OF THE DRAWINGS
[0031] Figure 1 Schematic diagram of the process for implementing the present invention.
[0032] Figure 2 It is a correlation analysis diagram between various structural diversity indicators in the embodiment and the aboveground biomass measured in the field.
[0033] Figure 3 is the prediction accuracy of the optimal model in the embodiment.
[0034] Figure 4 It is the biomass prediction result of Harbin Urban Forestry Demonstration Base in the embodiment. DETAILED DESCRIPTION
[0035] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0036] A prediction method for allometric growth simulation of aboveground biomass of urban forests is implemented in the following steps:
[0037] Step 1: Field measurement of aboveground biomass of urban forests: sample plots were set up in the dominant communities in the study area, and RTK was used to locate the corner points of the sample plots. The height and DBH structure of the trees in the sample plots were then measured. Based on the measured structural parameters, the aboveground biomass of urban forests in the sample plots was calculated using the allometric growth model.
[0038] Step 2: Obtaining the structural diversity parameters of urban forests. Obtain the original point cloud and multispectral data of urban forests within the study area through drone-borne laser radar and drone-borne multispectral data. Preprocess the two types of data and extract six types of structural parameters of urban forest canopy height, coverage and openness, heterogeneity, stand density, three-dimensional volume and physiological characteristics in the sample plot from the preprocessed point cloud data and multispectral data to form a set of structural diversity parameters for predicting the aboveground biomass of urban forests.
[0039] Step 3: Construction of an urban forest aboveground biomass estimation model. Based on the aboveground biomass of urban forests in the sample plot obtained in step 1 and the urban forest structural diversity index obtained in step 2, correlation analysis was first performed on the relationship between each structural diversity index and the aboveground biomass obtained in step 1. The structural parameters with the best correlation in each structural category were determined as the optimal parameters for estimating the aboveground biomass of urban forests. A power law model for estimating aboveground biomass was constructed. All combinations of optimal parameters were tested using the nonlinear least squares method and forward stepwise selection method to determine the optimal model for estimating the aboveground biomass of urban forests.
[0040] Step 4: Monitoring of aboveground biomass of urban forests. Based on the optimal model obtained in step 3, the aboveground biomass of urban forests in the study area was calculated.
[0041] The following takes the experimental data of Harbin City, Heilongjiang Province as an example to further illustrate the technical solution of the present invention.
[0042] Step 1: Field measurement of aboveground biomass of urban forests. In this embodiment, Harbin Urban Forestry Demonstration Base is selected as the research area, sample plots are arranged in the dominant communities in the research area, and RTK is used to locate the corner points of the sample plots, and then the tree height and breast diameter structure information of the trees in the sample plots are measured. Based on the measured structural parameters, the aboveground biomass of urban forests in the sample plots is calculated using the allometric growth model;
[0043] Step 2: Obtaining the structural diversity parameters of urban forests. Obtain the original point cloud and multispectral data of urban forests within the study area through drone-borne laser radar and drone-borne multispectral data. Preprocess the two types of data and extract six types of structural parameters of urban forest canopy height, coverage and openness, heterogeneity, stand density, three-dimensional volume and physiological characteristics in the sample plot from the preprocessed point cloud data and multispectral data to form a set of structural diversity parameters for predicting the aboveground biomass of urban forests.
[0044] Step 3: Construction of an urban forest aboveground biomass estimation model. Based on the aboveground biomass of urban forests in the sample plot obtained in step 1 and the urban forest structural diversity index obtained in step 2, correlation analysis was first performed on the relationship between each structural diversity index and the aboveground biomass obtained in step 1. The structural parameters with the best correlation in each structural category were determined as the optimal parameters for estimating the aboveground biomass of urban forests. A power law model for estimating aboveground biomass was constructed. All combinations of optimal parameters were tested using the nonlinear least squares method and forward stepwise selection method to determine the optimal model for estimating the aboveground biomass of urban forests.
[0045] Step 4: Monitoring of aboveground biomass of urban forests. Based on the optimal model obtained in step 3, the aboveground biomass of urban forests in the study area was calculated.
[0046] In step 1 of this embodiment, 30 10m×10m urban forest plots are set, and the plots are evenly distributed within the study area.
[0047] In step 2 of this embodiment, the two types of data obtained are preprocessed, including applying the lidR package in the R software to perform point cloud filtering, digital ground model interpolation, and normalization on the original point cloud data to extract the canopy height model CHM; and performing geometric correction, image registration, and image fusion on the multispectral data in the Pix4Dmapper photogrammetry software to generate an orthophoto map.
[0048] Step 2 of this embodiment extracts structural parameters from the preprocessed point cloud data and multispectral data, which means: based on the sample corner points in step 1, the sample shape file is drawn along the sample corner points using ArcGIS software, and then the sample shape file is imported into R software, and the urban forest structural parameters are extracted from the preprocessed point cloud data and multispectral data using a region-based method.
[0049] In step 2 of this embodiment, six categories of structural parameters are extracted to form a structural diversity parameter set for predicting aboveground biomass of urban forests, including the following indicators:
[0050] In step 3 of this embodiment, correlation analysis is performed on the relationship between each structural diversity index and the aboveground biomass obtained in step 1. The correlation analysis results are as follows: Figure 2 As shown: Hmean, GFP, rumple, VAI, VVol and cv are the parameters with the best correlation with the aboveground biomass of urban forests among the six categories of structural parameters, namely urban forest canopy height, cover and openness, heterogeneity, stand density, three-dimensional volume and physiological characteristics, and are identified as the optimal parameters for estimating the aboveground biomass of urban forests.
[0051] The power law model formula constructed in step 3 of this embodiment is as follows:
[0052]
[0053] In the formula, AGB is the aboveground biomass of urban forests, a0 is the estimated coefficient of AGB, and P h is the canopy height structural parameter, P c is the coverage or openness structural parameter, P d is the stand density structure parameter, P sh is the heterogeneous structural parameter, P v is the three-dimensional volume structure parameter, P t It is a physiological characteristic structural parameter.
[0054] Step 3 of this embodiment tests all combinations of optimal parameters by nonlinear least squares method and forward stepwise selection method, including: applying nonlinear least squares method in R software to derive coefficients of power law model, adding optimal parameters as model variables in sequence starting from zero feature model by forward stepwise selection method; performing 1000 bootstrapping resampling for each model, and calculating the goodness of fit R of the resampled model. 2 , root mean square error RMSE and mean absolute error MAE are used to evaluate the model performance.
[0055] The model fit goodness of fit R in step 3 of this embodiment 2 The calculation formula is as follows:
[0056]
[0057] In the formula, n is the number of samples, y is i and They represent the aboveground biomass of urban forest obtained in step 1 and the aboveground biomass of urban forest calculated by the model in sample plot i respectively;
[0058] The calculation formula of the root mean square error RMSE of the model in step 3 of this embodiment is as follows:
[0059]
[0060] In the formula, n is the number of samples, y is i and They represent the aboveground biomass of urban forest obtained in step 1 in sample plot i and the aboveground biomass of urban forest calculated by the model.
[0061] The calculation formula of the mean absolute error MAE of the model in step 3 of this embodiment is as follows:
[0062]
[0063] In the formula, n is the number of samples, y is i and They represent the aboveground biomass of urban forest obtained in step 1 and the aboveground biomass of urban forest calculated by the model in sample plot i respectively;
[0064] The model performance evaluation results of step 3 of this embodiment show that among all the model variable combinations, Hmean, GFP, rumple, VAI, VVol and cv structural parameters are the best combination for estimating the aboveground biomass of urban forests. 2 =0.799, RMSE =2.790, MAE b =2.194, which can be applied to the estimation of aboveground biomass of urban forests in the study area.
[0065] In step 4 of this embodiment, the aboveground biomass of urban forests in the study area is calculated based on the optimal model obtained in step 3, and the aboveground biomass of urban forests in the study area is predicted to be spatially continuous, thereby clarifying the aboveground biomass pattern of urban forests in the study area. Figure 4 .
Claims
1. A prediction method for allometric growth simulation of aboveground biomass in urban forests, characterized in that The method proceeds as follows: Step 1: Field measurement of aboveground biomass of urban forests: sample plots were set up in the dominant communities in the study area, and RTK was used to locate the corner points of the sample plots. The height and DBH structure of the trees in the sample plots were then measured. Based on the measured structural parameters, the aboveground biomass of urban forests in the sample plots was calculated using the allometric growth model. Step 2: Obtaining the structural diversity parameters of urban forests. Obtain the original point cloud and multispectral data of urban forests within the study area through drone-borne laser radar and drone-borne multispectral data. Preprocess the two types of data and extract six types of structural parameters of urban forest canopy height, coverage and openness, heterogeneity, stand density, three-dimensional volume and physiological characteristics in the sample plot from the preprocessed point cloud data and multispectral data to form a set of structural diversity parameters for predicting the aboveground biomass of urban forests. Step 3: Construction of an urban forest aboveground biomass estimation model. Based on the aboveground biomass of urban forests in the sample plot obtained in step 1 and the urban forest structural diversity index obtained in step 2, correlation analysis was first performed on the relationship between each structural diversity index and the aboveground biomass obtained in step 1. The structural parameters with the best correlation in each structural category were determined as the optimal parameters for estimating the aboveground biomass of urban forests. A power law model for estimating aboveground biomass was constructed. All combinations of optimal parameters were tested using the nonlinear least squares method and forward stepwise selection method to determine the optimal model for estimating the aboveground biomass of urban forests. Step 4: Monitoring of aboveground biomass of urban forests. Based on the optimal model obtained in step 3, the aboveground biomass of urban forests in the study area was calculated.
2. The method for predicting the aboveground biomass of urban forests by allometric simulation according to claim 1, characterized in that The sample plot size described in step 1 is 10m×10m, and the samples are evenly distributed in the study area.
3. The method for predicting the aboveground biomass of urban forests by allometric simulation according to claim 1, characterized in that In step 2, the two types of data obtained are preprocessed, including: The lidR package was used in R software to perform point cloud filtering, digital ground model interpolation and normalization on the original point cloud data to extract the canopy height model (CHM). The multispectral data were geometrically corrected, image registered and fused in Pix4D mapper photogrammetry software to generate orthophotos.
4. The method for predicting the aboveground biomass of urban forests by allometric simulation according to claim 1, characterized in that Extracting structural parameters from the preprocessed point cloud data and multispectral data in step 2 means: based on the sample corner points in step 1, drawing the sample shape file along the sample corner points using ArcGIS software, then importing the sample shape file into R software, and applying a region-based method to extract urban forest structural parameters from the preprocessed point cloud data and multispectral data.
5. The method for predicting the aboveground biomass of urban forests by allometric simulation according to claim 1, characterized in that The six categories of structural parameters are extracted as described in step 2 to form a set of structural diversity parameters for predicting urban forest aboveground biomass, including the following indicators:
6. The method for predicting the aboveground biomass of urban forests by allometric simulation according to claim 1, characterized in that The power law model formula described in step 3 is as follows: In the formula, AGB is the aboveground biomass of urban forests, a0 is the estimated coefficient of AGB, and P h is the canopy height structural parameter, P c is the coverage or openness structural parameter, P d is the stand density structure parameter, P sh is the heterogeneous structure parameter, P v is the three-dimensional volume structure parameter, P t It is a physiological characteristic structural parameter.
7. The method for predicting the allometric growth of aboveground biomass of urban forests according to claim 1, characterized in that In step 3, all combinations of optimal parameters are tested by nonlinear least squares and forward stepwise selection, including: The nonlinear least squares method in R software was applied to derive the coefficients of the power law model, and the optimal parameters were added as model variables in sequence, starting from the zero-feature model, by the forward stepwise selection method; For each model, 1000 bootstrapping resamplings were performed, and the goodness of fit of the model after resampling was R 2 , root mean square error RMSE and mean absolute error MAE are used to evaluate the model performance.
8. The method for predicting the allometric growth of above-ground biomass in urban forests according to claim 1, characterized in that The model goodness of fit R 2 The calculation formula is as follows: In the formula, n is the number of samples, y is i and represent the aboveground biomass of urban forest obtained in step 1 and the aboveground biomass of urban forest calculated by the model in sample plot i, respectively. represents the average aboveground biomass of urban forests obtained in step 1; The calculation formula of the root mean square error RMSE is as follows: In the formula, n is the number of samples, y is i and They represent the aboveground biomass of urban forest obtained in step 1 and the aboveground biomass of urban forest calculated by the model in sample plot i respectively; The calculation formula of the mean absolute error MAE is as follows: In the formula, n is the number of samples, y is i and They represent the aboveground biomass of urban forest obtained in step 1 in sample plot i and the aboveground biomass of urban forest calculated by the model.
9. The method for predicting the allometric growth of aboveground biomass of urban forests according to claim 1, characterized in that In step 4, the fishing net tool in ArcGIS software was used to create a 10 m × 10 m grid shape file covering the study area. The regional-based method was applied through R software to extract the urban forest structural parameters of the study area from the preprocessed point cloud data and multispectral data, and the aboveground biomass of the urban forest in the study area was calculated based on the optimal model in step 3.