Next Article in Journal
The Need to Establish a Social and Economic Database of Private Forest Owners: The Case of Lithuania
Previous Article in Journal
Somatic Embryogenesis Induction and Genetic Stability Assessment of Plants Regenerated from Immature Seeds of Akebia trifoliate (Thunb.) Koidz
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Forests
Volume 14
Issue 3
10.3390/f14030475
forests-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Estimation of Aboveground Biomass of Individual Trees by Backpack LiDAR Based on Parameter-Optimized Quantitative Structural Models (AdQSM)

by
A Ruhan
1,2,
Wala Du
3,4,
Hong Ying
1,2,
Baocheng Wei
1,2,
Yu Shan
1,2,* and
Haiyan Dai
5
1
College of Geographic Science, Inner Mongolia Normal University, Hohhot 010022, China
2
Inner Mongolia Key Laboratory of Remote Sensing and Geographic Information Systems, Inner Mongolia Normal University, Hohhot 010022, China
3
Chinese Academy of Agricultural Sciences Grassland Research Institute, Hohhot 010022, China
4
Arshan Forest and Grassland Disaster Prevention and Mitigation Field Scientific Observation and Research Station of Inner Mongolia Autonomous Region, Arshan 137400, China
5
Inner Mongolia Ecology and Agriculture Meteorological Centre, Hohhot 010051, China
*
Author to whom correspondence should be addressed.
Forests 2023, 14(3), 475; https://doi.org/10.3390/f14030475
Submission received: 10 January 2023 / Revised: 18 February 2023 / Accepted: 23 February 2023 / Published: 27 February 2023
(This article belongs to the Special Issue Forest Regeneration and Landscape Resilience after Wildfire)
Browse Figures
Versions Notes

Abstract

:
Forest aboveground biomass (AGB) plays a key role in assessing forest productivity. In this study, we extracted individual tree structural parameters using backpack LiDAR, assessed their accuracy using terrestrial laser scanning (TLS) data and field measurements as reference values, and reconstructed 3D models of trees based on parameter-optimized quantitative structural models (AdQSM). The individual tree AGB was estimated based on individual tree volumes obtained from the tree model reconstruction, combined with the basic wood density values of specific tree species. In addition, the AGB calculated using the allometric biomass models was validated to explore the feasibility of nondestructive estimation of individual tree AGB by backpack LiDAR. We found that (1) the backpack LiDAR point cloud extracted individual tree diameter at breast height (DBH) with high accuracy. In contrast, the accuracy of the tree height extraction was low; (2) the optimal parameter values of the AdQSM reconstruction models for Larix gmelinii and Betula platyphylla were HS = 0.4 m and HS = 0.6 m, respectively; (3) the individual tree AGB estimated based on the backpack LiDAR and AdQSM fit well with the reference values. Our study confirms that backpack LiDAR can nondestructively estimate individual tree AGB, which can provide a reliable basis for further forest resource management and carbon stock estimation.

1. Introduction

As an important component of terrestrial ecosystems, the estimation of forest biomass is important for the study of net primary productivity and carbon cycling in forests. Traditional methods of measuring forest aboveground biomass (AGB) include destructive sampling methods, which are destructive to the ecosystem and time-consuming. Light detection and ranging (LiDAR), an emerging active remote sensing technology, is widely used for AGB estimation because it can directly extract information on the vertical structure of forests. Depending on the platform, LiDAR mainly includes space-borne LiDAR, airborne LiDAR and ground-based LiDAR (terrestrial laser scanning (TLS), vehicle-borne LiDAR, and backpack and handheld LiDAR), enabling the estimation of AGB at different spatial scales, such as area [1,2], stand [3], and individual tree [4]. Ground-based LiDAR is useful for obtaining forest understory structural parameters and is, therefore, mostly used for estimating AGB at the individual tree scale.
Terrestrial laser scanning (TLS) is capable of acquiring high-density 3D-point cloud data of target trees and has been widely used to estimate AGB at the individual tree scale [5,6,7]. Ghimire et al. [8] used high-precision TLS data to obtain individual tree structural parameters, which were then substituted into species-specific allometric biomass models to estimate the AGB of a Mediterranean coniferous forest, demonstrating the accuracy and validity of a nondestructive approach to forest AGB estimation. Beyene [9] used TLS to collect point cloud data from tropical rainforests in Malaysia and used tree parameters derived from TLS data to estimate AGB, confirming that TLS can accurately estimate forest parameters and AGB in dense tropical rainforests. However, the fixed measurement method of TLS limits its spatial flexibility, and the data acquisition time is too long. The backpack LiDAR system combines one or more laser scanners, an inertial measurement unit (IMU), and a global navigation satellite system (GNSS) tracker in a single unit, with simultaneous localization and mapping (SLAM) technology compensating for the limitations of TLS, such as low mobility and multisite cloud mapping. However, backpack LiDAR is currently focused on potential applications in forest inventories and the accuracy of acquiring tree structure parameters [10,11,12]. Hyyppä et al. [13] found that backpack LiDAR was more accurate in acquiring the structural parameters of individual trees than a handheld LiDAR scanner. Oveland et al. [14] compared three different (TLS, backpack, and handheld) LiDAR scanning methods in terms of estimating diameter at breast height (DBH), acquiring individual tree locations and data acquisition efficiency. It was found that the backpack LiDAR was the most efficient, and the estimated DBH was the closest to the true value.
Among the biomass estimation models, the allometric biomass model [15] is currently the most widely used empirical model and is widely used as reliable ground verification data instead of measured biomass. Some scholars [16] have used a combination of backpack LiDAR and airborne LiDAR to extract individual tree structural parameters and then fed them into the allometric biomass model to obtain sample-scale AGB. However, these methods still rely on a limited number of stand parameter factors to build an empirical model to estimate biomass. No studies have yet explored the feasibility of the direct use of backpack LiDAR point cloud data for AGB estimation. Quantitative structure models (QSMs) can reconstruct 3D models of trees based on high-density laser point clouds [17], in which the volume of the tree is obtained directly from the point cloud data and then combined with the basic wood density of a particular tree species to further estimate AGB. Unlike allometric biomass models, this method is based on the true biomorphic structure of a particular tree species to estimate AGB [18]. The QSM algorithm has been extensively investigated, particularly in tropical forests, and has been shown to be an effective method for the nondestructive estimation of AGB for individual trees. Gonzalez de Tanago et al. [19] used TLS point clouds to construct a 3D tree model to estimate the AGB of large trees in tropical forests and compared it with destructively measured AGB to determine the applicability of the QSM algorithm for estimating AGB in complex forest stands in tropical regions. Krishna Moorthy et al. [20] estimated the tree volume and AGB of vine stems in tropical forests using the TLS–QSM method and found that the stem volume obtained from a 3D tree model was highly correlated with the reference volume. Lau et al. [21] showed that the AGB estimated based on the TLS–QSM method was more consistent with the measured validation data than the estimates obtained from the subtropical allometric biomass models. Du et al. [22] and Fan et al. [23] further extended the QSM by proposing the AdTree and AdQSM methods. The methods provide a geometric basis for the automatic, detailed, and accurate 3D reconstruction of real trees and are able to quantify tree parameters such as DBH, tree height, and volume. Using LiDAR point cloud and photogrammetry point cloud data and destructive sampling data, the accuracy of the AdQSM method in estimating individual tree DBH, tree height, volume, and AGB was verified [24,25]. Zhang et al. [26] validated the potential of backpack LiDAR point cloud data to extract branch information from fruit trees by comparing the measured values with QSM estimates of fruit tree samples and found that multiple orders of branch information were underestimated to varying degrees. However, this method has not been applied to the study of individual tree AGB estimation from backpack LiDAR point cloud data, and the reconstruction results of the AdQSM method vary for different tree species with different morphological structures, which directly affects the accuracy of the final AGB estimation [27].
Located in the northernmost part of China, Greater Khingan is a major component of the Northeast Forest Region, and its forest ecosystems are critical to the regional and national carbon balance. Recently, there has been a growing concern about carbon emissions and environmental sustainability [28]. Therefore, accurate estimation of biomass in the region is important for studying carbon stocks in Greater Khingan and evaluating the role of China’s temperate forest ecosystems in the global carbon cycle. Therefore, we collected backpack LiDAR point cloud data from natural forest sample plots in Duraer National Forest at the southern foot of Greater Khingan, extracted individual tree structural parameters, and estimated individual tree volume and AGB based on the AdQSM, using TLS data and field measurements as reference values to verify their accuracy. The objectives of this study are (1) to compare the differences between the two LiDAR scanning techniques and the accuracy of extracting individual tree structural parameters; (2) to optimize the AdQSM parameters for Larix gmelinii and Betula platyphylla individual tree AGB estimation; and (3) to use the backpack LiDAR–AdQSM method to estimate individual tree AGB and compare this with the AGB calculated by the allometric biomass models to explore the feasibility of estimating individual tree AGB from backpack LiDAR point cloud data. This study proposes an efficient and nondestructive method for estimating individual tree AGB, which provides a scientific basis for further rational forest management and carbon stock estimation.

2. Materials and Methods

2.1. Study Area

The study area is located in Duraer National Forest, Arxan City, Xing’an League, Inner Mongolia Autonomous Region (119°28′–121°01′ E, 47°15′–47°35′ N). The forestry area is located in northwestern Tianchi town, Arxan city, with a total area of approximately 44,400 hm2 and a forest cover of 61%. Duraer National Forest is located at the western foot of the middle section of Greater Khingan, bordering Hulunbeier city in Inner Mongolia to the north and east and the Oriental Province in Mongolia to the west, making it an important natural ecological reserve in China. The average altitude is 1100 m (Figure 1), and it has low-to-medium mountainous terrain. The climate type is a temperate continental monsoon climate with an annual precipitation of approximately 437 mm and an average annual temperature of 1.48 °C. The main tree species in the study area are Larix gmelinii (Rupr.) Kuzen, Betula platyphylla Suk, and Populus davidiana Dode.

2.2. Data

2.2.1. Field Data

We selected four natural forest sample plots in Duraer Forest based on different tree species and different stand densities, with sample Plots 1 and 2 in Larix gmelinii forest and sample Plots 3 and 4 in Betula platyphylla forest (Figure 2). Each sample plot was set up as a rectangle of 10 m × 40 m, and the basic information of the sample plots is shown in Table 1. Individual trees with DBHs greater than 5 cm were selected and tagged, and the DBH and tree height were measured several times using a DBH ruler and laser altimeter and then averaged to reduce errors. Information such as species category, stand naturalness, and denseness was recorded. At the same time, the coordinates of the center point and the four corners of the sample were obtained by using the differential satellite station technique of the GNSS receiver, which was used to crop and register the point cloud data at a later stage.

2.2.2. Backpack LiDAR Data

The backpack LiDAR device we used was the Green Valley International’s model LiBackpack DGC50 (Green Valley, Beijing, China) with specifications shown in Table 2 and we followed a “S”-shaped-designed hiking route to collect point clouds of trees within the sample plots (Figure 1B). The surveyor connected a mobile phone to the backpack LiDAR to view the number of search satellites on the phone and to view the point cloud data collection in real time while collecting data. It was important to keep the backpack as steady as possible during the acquisition process, taking care to walk slowly around corners and in an arc to ensure good quality point cloud data. To obtain the point cloud data with absolute coordinates, a GNSS receiver (Haida Zhong, Wuhan, China) was used to obtain the absolute coordinates of a point outside the sample area with a stable GPS signal, and a reference station was set up at this point to obtain static data. Finally, the original point cloud data and the trajectory file as well as the GNSS static data were imported into LiFUser BP software (Green Valley, Beijing, China) for trajectory solving to obtain the point cloud data containing absolute geographic location information.

2.2.3. TLS Data

We used a FARO FOCUS 3D 120 terrestrial laser scanner (Faro Technologies Company, Lake Mary, FL, USA) to scan the four sample plots using a multistation layout with the following scanner profile parameters: 20 m outdoor scanning, color scanning mode, and a scanning time of approximately 7 min per station. Six stations were evenly distributed within each sample plot for multidirectional scanning to ensure that each tree in the sample plot was adequately scanned, as shown in Figure 1B. The point cloud data from the multistation scans were registered in Faro Scene software (Faro Technologies Company, Lake Mary, FL, USA) and aligned within 1 cm.

2.2.4. Point Cloud Data

The solved and registered backpack LiDAR and TLS point cloud data were imported into LiDAR360 V5.4 software (Green Valley, Beijing, China). First, the point cloud data were cropped by sample extent and filtered after removing redundancy and noise, and ground points were classified using the improved progressive TIN densification filtering algorithm proposed by Zhao et al. [29]. Then, a digital elevation model (DEM) with a resolution of 0.5 m was generated by irregular triangular mesh interpolation. The elevation value Z of each point cloud was subtracted from the corresponding DEM elevation value found, and the point cloud was normalized based on the DEM to eliminate the influence of topography on tree height estimation (Figure 3).

2.3. Methods

2.3.1. Individual Tree Segmentation and Parameter Extraction

To obtain the structural parameters of individual trees and to achieve accurate 3D modeling of individual trees, individual tree segmentation of the point cloud for all trees in the sample plot was needed. First, the normalized point cloud data were identified by the comparative shortest-path algorithm (CSP) [30] to identify the spatial clustering of standing trees. Then, point cloud slices at 1.2–1.3 m of the trunk were extracted, and the fitted cylinder method was used to batch fit the DBH. Manual screening of the DBH fit results was required to manually refit the poorly fitted individual trees to ensure individual tree DBH accuracy (Figure 4). Finally, the fitted DBH values were used as seed points for individual tree segmentation. The CSP algorithm searched for points within the radius of the DBH or the nearest point as the initial cluster of seed points for subsequent segmentation based on the 3D coordinates of the seed points (Figure 5). The accuracy of the individual tree segmentation directly affects the accuracy of the individual tree modeling and, hence, the accuracy of the individual tree AGB estimation. Therefore, the segmented individual tree point clouds needed to be checked one by one, and the misclassified individual trees were manually re-segmented. Tree height was estimated from the difference in elevation between the top and bottom point clouds of the segmented individual tree. All these steps were performed in LiDAR360 V5.4 software (Green Valley, Beijing, China).

2.3.2. AdQSM Reconstruction Method for Individual Trees

The AdQSM method is a method for reconstructing individual tree structure models based on point cloud data, which was proposed by Fan et al. after improving and extending the AdTree algorithm [23]. This method is based on the Adtree algorithm and uses the minimum spanning tree algorithm (MST) to extract the initial skeleton of the tree from the input point cloud and to crop it. A series of cylinders are then fitted to approximate the trunk and branch geometry. Thus, a closed convex envelope polyhedron was obtained, enabling the accurate calculation of tree (branch and trunk) volumes from the reconstructed structure. The individual tree model reconstruction was completed in the C++ AdQSM program (Figure 6), and tree (trunk, branch) volumes (Equation (1)) were obtained and used to further estimate the individual tree AGB.
Volume tree = ∑ i = 1 m ∑ j = 1 n ( | ( a × b ) · c | 6 )
where Volumetree denotes the total volume including the trunk and branches, m denotes the number of all convex polyhedra that make up the tree model, and n denotes the number of all triangular pyramids that make up the convex polyhedra. a, b, and c denote the three vectors that pass through the vertices of the triangular pyramids.

2.3.3. AdQSM Parameter Optimization

There are two important parameters in the AdQSM modeling process: Height_Segmentation (HS) and Cloud_Parameter (CP). HS represents the height of each segment of the trunk point cloud being segmented in meters, and CP represents the thinning rate of the point cloud processing. Fan et al. [23] defined the default values for the parameters HS and CP as 0.5 m and 0.003, respectively. Since the AdQSM modeling process is not sensitive to the input point cloud density, CP was set as the default value. The different trunk point cloud segmentation heights affect the radius of the initial cylinder of the trunk and, thus, the accuracy of the tree 3D reconstruction [23]. Therefore, 20 trees from each tree species were selected separately to test different parameter values of HS to determine the optimal parameter values applicable to each tree species.
Our aim was to select the optimal parameters applicable to the different tree species from the different test values of HS (0.4, 0.5, 0.6, 0.7, 0.8, 0.9, and 1.0). First, the individual tree was modeled using the different test values, and the tree volume was extracted. Second, the volume was multiplied by the basic wood density ρ of the respective species to obtain the individual tree AGB and compared with the AGB calculated by the allometric biomass models. Finally, the parameter value corresponding to the minimum value of root mean square error (RMSE) was defined as the optimal parameter applicable to the species.

2.3.4. Estimation of AGB

We estimated the individual tree AGB by multiplying the individual tree volume by the basic wood density ρ (Equation (2)). The biomass of the trunk and branch parts of the tree accounts for approximately 80%–90% of the total biomass, while the leaf biomass accounts for only 10% of the total biomass [18]. We, therefore, estimated the AGB, including the trunk and branch fractions, without considering the leaf biomass.
AGB = ρ V b
where ρ is the basic wood density of the specific tree species (g/cm3) and V b is the individual tree volume estimated by backpack–AdQSM (m3).
We used the AGB of individual trees calculated using the allometric biomass models for the corresponding tree species in a given region as a reference value and compared it with the AGB of individual trees estimated based on LiDAR data and the AdQSM 3D reconstructed tree model. The corresponding basic wood density [31] and allometric biomass models [32] for Larix gmelinii and Betula platyphylla are shown in Table 3. The DBH data were obtained from manual measurement data.
To verify the accuracy of extracting DBH and tree height from backpack LiDAR point cloud data and the feasibility of estimating AGB based on the AdQSM method, measured data and parameters estimated from TLS point clouds were used as reference values. The coefficient of determination (R2) and root mean square error (RMSE) were used to evaluate the extraction accuracy. The equations are as follows:
R 2 = 1 − ∑ i = 1 n ( x i − x i ^ ) 2 ∑ i = 1 n ( x i − x i ¯ ) 2
RMSE = 1 n ∑ i = 1 n ( x i − x i ^ ) 2
where x i is the measured value of the individual tree parameter, x i ^ is the estimated value of the individual tree parameter, x i ¯ is the sample mean, and n is the number of samples. The larger the value of R2 is, the better the fit. The smaller the value of RMSE is, the higher the accuracy of the individual tree parameter estimation.

3. Results

3.1. Accuracy Assessment of LiDAR-Extracted DBH and Tree Height

The accuracy of the backpack LiDAR-extracted individual tree structure parameters (DBH, tree height) was evaluated using field measurements as a reference value, with extracted DBH values ranging from 5.2 to 29.9 cm and tree height ranging from 4.3 to 14.9 m. Figure 7A shows that the backpack LiDAR-extracted DBH had a high correlation with the measured values. This indicates that the individual tree DBH extracted by the backpack LiDAR could be obtained with good correlation and high accuracy when considering the measured data. The R2 value was greater than 0.97 in all four sample plots, and the RMSE was between 0.43 and 0.84 cm, indicating that the individual tree DBHs extracted by the backpack LiDAR have a certain reliability. From Figure 7B, we can see that the DBH extracted by TLS had a good linear relationship with the field measurements. The R2 value was greater than 0.97 for all four sample plots, and the RMSE was between 0.42 and 0.53 cm, which means that the extraction accuracy of this study is comparable to existing studies [33,34,35].
Figure 7C shows that the correlation between tree height extracted by the backpack LiDAR and field measurements was low. The R2 values of the four sample plots were all less than 0.4, and the RMSE ranged from 1.38 m to 3.12 m. Figure 7D shows that the correlation between tree height extracted from TLS data and field measurements was also low. R2 values were less than 0.6 in all four plots, and the RMSE ranged from 0.71 m to 1.75 m. This indicates that in a natural forest environment with high stand density and mutual shading of branches and leaves, LiDAR had difficulty accurately detecting the top of the tree canopy, resulting in large differences between the tree height values extracted from the point cloud data and the measured data.

3.2. AdQSM Parameter Optimization Results

We used different parameter values of HS in AdQSM to estimate individual tree AGB and defined the parameter corresponding to the minimum value of RMSE as the optimal parameter applicable to Larix gmelinii and Betula platyphylla species. Table 4 shows the results of individual tree AGB estimation for Larix gmelinii and Betula platyphylla with different parameter values of HS in AdQSM. The highest individual tree AGB accuracy was obtained for the Larix gmelinii reconstruction model with HS set to 0.4 m, with an RMSE value of 29.51 kg. The highest individual tree AGB accuracy was obtained for the Betula platyphylla reconstruction model with HS set to 0.6 m, with an RMSE value of 28.86 kg. Therefore, the optimal parameter values for the Larix gmelinii and Betula platyphylla reconstruction models were HS = 0.4 m and HS = 0.6 m, respectively.

3.3. Estimation of Aboveground Biomass

We randomly selected 20 Larix gmelinii and 20 Betula platyphylla trees from four sample plots to assess the accuracy of AGB estimation based on backpack LiDAR point cloud data and the AdQSM method (backpack–AdQSM). The reference AGB values for an individual tree of Larix gmelinii ranged from 10.9 kg to 293.1 kg, and the reference AGB values for an individual tree of Betula platyphylla ranged from 8.7 kg to 301.7 kg. The AGB of an individual tree of Larix gmelinii estimated by backpack–AdQSM ranged from 8.8 kg to 327.5 kg, and the AGB of an individual tree of Betula platyphylla estimated by backpack–AdQSM ranged from 3.9 kg to 432 kg. Figure 8 shows the regression of the individual tree AGB estimated by backpack–AdQSM on the AGB calculated by the allometric biomass models. Figure 8A shows that the Larix gmelinii individual tree AGB estimated by backpack–AdQSM correlates well with the reference value, with an R2 value of 0.84 and an RMSE of 29.51 kg. The distribution of residuals for the estimated Larix gmelinii individual tree AGB is shown in Figure 8B. It can be seen that most residuals for individual Larix gmelinii trees AGB are more consistently and evenly distributed on either side of the line y = 0 and are not over- or underestimated. The R2 value between the estimated and reference values of Betula platyphylla individual tree AGB expressed by the linear regression model in Figure 8C was 0.86, and the RMSE was 28.86 kg. From Figure 8D, it can be seen that the residual values of Betula platyphylla individual tree AGB were mostly in a range of −30 to 30 kg. The Betula platyphylla individual tree AGB estimated by backpack–AdQSM was generally large compared to the reference value, and as the AGB reference value increased, the Betula platyphylla AGB was again underestimated.

4. Discussion

In this study, we used backpack LiDAR point cloud data to extract individual tree structural parameters and reconstructed a 3D model of the trees based on the parameter-optimized AdQSM method to estimate the individual tree AGB based on the reconstructed individual tree volume combined with the basic wood density values of specific tree species to explore the accuracy and potential of backpack LiDAR for the nondestructive estimation of individual tree AGB in temperate natural forest inventories.
Compared to traditional manual DBH measurements and TLS DBH measurements, backpack LiDAR allows for accurate scanning and real-time data integration during movement, providing a more flexible and efficient way to collect data for forest inventories [13,14] and is suitable for surveying large and complex forest areas [12]. To start with, manual measurement requires standing on the high-angle side of the tree base and measuring the DBH at 1.3 m using a DBH ruler, which is recorded in the sample plot record sheet, which can be subject to human error and can be made more difficult by practical conditions such as leaning trunk growth or interlocking branches. Secondly, TLS data collection requires multiple stations to be set up within the measurement site, which can make the operation more difficult, time-consuming, and costly in the case of dense natural forests or sloping stands. Backpack LiDAR as a portable device can compensate for these shortcomings; data collection only requires one surveyor to carry the equipment across the measurement site, greatly reducing the time and cost and improving efficiency. As shown in Table 5, when collecting point cloud data from a 10 m × 40 m sample, the data collection process is as follows: a manual survey takes, on average, about 36 min, TLS takes about 42 min, while backpack LiDAR only takes about 6 min to complete the operation. The traditional manual survey method requires the contents of the record sheet to be entered into excel, which takes approximately 5 min. The point cloud data collected by TLS requires the registration of each site’s point cloud into a single map, which takes about 50 min longer. Backpack LiDAR point cloud data pre-processing took approximately 25 min. Comparing the average area measured per person in one minute, the manual survey method has a time efficiency of 3.3 m2/min and the backpack LiDAR has a time efficiency 13.5 m2/min. The time efficiency of the backpack LiDAR method is approximately 4.2 times higher than the manual survey method, demonstrating the time efficiency of the backpack LiDAR method. The Larix gmelinii sample site was densely branched, and to avoid scratching the LiDAR module, the branches had to be cut down in advance along the designed route to ensure safe operation of the backpack LiDAR. Therefore, the acquisition times of backpack LiDAR data collection vary depending on the forest stand and need to be further validated in more operational environments. Internal data processing time varies depending on the size of the tree dataset and computer configuration.
The forest resource survey work also has to take into account practical and cost-effectiveness issues while meeting time efficiency. The cost of inputs in this study consists of two types, the first being labor costs (including travel, labor, etc.). Traditional manual survey work requires a minimum of three people to perform the work; at a total of RMB 2000 per person per day, the total labor cost is approximately RMB 6000/day, while the backpack LiDAR requires only one person to operate; the labor cost is approximately RMB 2000/day. The second is the cost of equipment. The cost of equipment required for manual surveying is approximately RMB 2000 (laser altimeter, DBH ruler, etc.), while the cost of backpack LiDAR equipment is approximately RMB 100,000. Both types of equipment have no wear and tear costs at a later stage. We assume that, based on the time taken to measure 1 ha of data by manual survey methods and backpack LiDAR, the average person working 8 h a day would spend approximately 6.3 days on manual surveys and 1.5 days on the backpack LiDAR method. The labor cost for the manual survey method is approximately 12,600 RMB and the labor cost for the backpack LiDAR method is approximately 3000 RMB. The total cost includes labor costs and fixed equipment costs. As can be seen from Figure 9, the backpack LiDAR method has a larger initial investment compared to the traditional manual survey method, mainly due to the high price of the equipment. With the development of LiDAR technology, the price of the equipment will become more and more affordable, and its application in various industries will reach greater popularity. Therefore, the future cost of backpack LiDAR equipment has the trend of reducing year by year. Although the traditional manual survey method has a low initial input cost, as the area of forest resource surveys increases, the labor cost increases. Therefore, this study concludes that backpack LiDAR is sufficiently advantageous, both in terms of time efficiency and cost-effectiveness.
Our results demonstrate that the DBH extraction accuracy of the backpack LiDAR method (RMSE = 0.43 cm–0.84 cm) is comparable to that of TLS (RMSE = 0.42 cm–0.53 cm). Samples 1 and 2 (Larix gmelinii) both had an R2 value greater than 0.97 and a mean RMSE of 0.68 cm, and the higher accuracy of DBH extraction (R2 = 0.99, RMSE = 0.44 cm) for Samples 3 and 4 (Betula platyphylla) may be because Betula platyphylla has fewer branches at 1.3 m than the Larix gmelinii trunk and, therefore, has less impact on the accuracy of the DBH fit. We compared the results of this study with previous studies (Table 6) and found that the extraction accuracies of individual tree DBHs based on backpack LiDAR point cloud data were all between 1.0 and 3.9 cm, and the extraction accuracies of individual tree DBHs based on TLS point cloud data were between 1.2 and 6.2 cm. The extraction accuracies in this study were all higher than the results of previous studies. Compared to the DBH extraction accuracy, the backpack LiDAR-extracted tree height values were less correlated with field measurements (mean R2 = 0.38, RMSE = 2.08 m) and lower than the TLS-extracted tree height values (mean R2 = 0.47, RMSE = 1.38 m). The reason for the less favorable backpack LiDAR tree height extraction results may be because in natural forests with dense canopies, the laser signal does not detect the true tree tops, thus reducing the accuracy of the backpack LiDAR-extracted tree heights, which also confirms the results of Bauwens et al. [36] and Cabo et al. [37]. From the statistics in Table 6, previous studies based on backpack LiDAR-extracted tree heights reported RMSEs between 1.4 and 2.8 m and R2 values between 0.41 and 0.68. Other scholars have also observed low accuracy of LiDAR tree height extraction and have proposed various explanations, such as those by Hartley et al. [38], who suggested that this may be due to an overestimation of the height of dead trees in the sample plots, which leads to a reduction in the accuracy of tree height estimation for the whole sample plot. It has also been suggested that field-measured tree height values are not reliable [39]. Some studies have shown that the error in field measurements can be as high as 5% [40].
It has been demonstrated that the AdQSM method can provide reliable and accurate AGB estimates for large tropical trees [24]. Our results demonstrate that the AdQSM method is also applicable to the estimation of AGB of Larix gmelinii and Betula platyphylla in temperate natural forests, where the staggered growth of Larix gmelinii branches affects the accuracy of individual tree segmentation, which in turn directly affects the accuracy of individual tree modeling [24]. This ultimately led to a lower accuracy of the estimation of the AGB of Larix gmelinii trees based on the backpack LiDAR point cloud data (R2 = 0.84, RMSE = 29.51 kg). In contrast, Betula platyphylla biophysical morphology was relatively simple, and the point cloud data were less influenced by branching, resulting in a slightly higher individual tree AGB estimation accuracy than that of Larix gmelinii (R2 = 0.86, RMSE = 28.86 kg). Previous studies have also estimated individual tree AGB for multiple tree species in temperate forests in northern China based on TLS point cloud data and the QSM algorithm [18], with an estimation accuracy higher than that of our study (CCC value of 0.97 and RMSE of 17.4 kg). This may be because our study area was in a natural forest with a more complex forest structure and more severe vegetation shading compared to forest stands previously studied. This resulted in a lower accuracy of individual tree AGB estimation. In a previous study, based on TLS point cloud data and the QSM algorithm, estimates of individual tree AGB were generally in ranges of R2 = 0.91 to 0.98 and rRMSE = 16.1% to 28.37% [19,20,21,44]. Estimation of individual tree AGB based on backpack LiDAR point cloud data and the QSM algorithm was slightly less accurate than TLS compared to these previous studies. In contrast to existing methods in the literature that used backpack LiDAR to estimate AGB [16], the method in this study is based on a realistic biomorphological structure model of the tree to estimate individual tree AGB. The potential of inputting backpack LiDAR point cloud data into the QSM algorithm for estimating individual tree AGB was demonstrated for the first time and extends the generality of the AdQSM method. This method allows not only the estimation of individual tree volume and AGB but also the monitoring of volume growth [45] as well as the quantitative analysis of structural information of fruit tree branches and trunks [26].
There are still many uncertainties in our study. First, the poor GPS signal during data acquisition by the backpack LiDAR can directly affect the quality of the track file, which can lead to point cloud solution failure or solutions of point cloud data with large errors in absolute coordinates. Moreover, acquiring point cloud data with high-precision absolute coordinates is crucial to the positioning of individual trees in the sample area. Therefore, how backpack LiDAR can efficiently collect absolute georeferenced point cloud data with high accuracy in dense forests with no GPS signal is a key focus for future research. Second, due to the low accuracy of backpack LiDAR in extracting tree height, future research should test the potential of backpack LiDAR in more diverse forest environments to explore the upper limit of forest height that it can detect and whether fusing backpack LiDAR and UAV–LiDAR data can improve the accuracy of tree height estimation. For this study, there were two main sources of error affecting the accuracy of forest AGB estimation: (i) We did not check destructively sampled individual tree AGB to verify the accuracy of the estimation but used AGB calculated by an allometric biomass model with diameter at breast height (D) as the independent variable as the reference data. The model may be subject to some error [46]; and (ii) due to the different biophysical and morphological structures of different tree species, the parameters set in the AdQSM individual tree modeling also differed, and there was also poor sensitivity and robustness in the individual tree reconstruction process. In addition, errors in the basic wood density could have an impact on the results of individual tree AGB estimation. Future research will, therefore, focus on further reducing these sources of error and the need for more tree species to optimize the AdQSM parameters to improve the accuracy of large-scale nondestructive estimation of tree timber volume and AGB based on backpack LiDAR point cloud data.

5. Conclusions

In this study, we used natural forest backpack LiDAR point cloud data from the Duraer National Forest at the southern foot of Greater Khingan to explore the accuracy of its extraction of individual tree structural parameters and the feasibility of estimating individual tree volume and AGB based on the optimized AdQSM method. The following conclusions were drawn:
(1)
Compared with manual survey methods and TLS technology, backpack LiDAR can provide a more efficient way of acquiring data for forest inventory. This enables large-scale inventories of forest resources and further development of forest management plans;
(2)
Backpack LiDAR can accurately measure the canopy understory structural parameters, where DBH measurement accuracy is the highest and comparable to TLS accuracy, while backpack LiDAR-extracted tree height values are less correlated with the actual measured data;
(3)
Using individual tree point cloud data of Larix gmelinii and Betula platyphylla species, the HS parameters in the AdQSM reconstruction model were optimized, and after testing different parameter values, HS = 0.4 m and HS = 0.6 m were defined as the optimal parameter values applicable to Larix gmelinii and Betula platyphylla species;
(4)
The AGB of individual trees estimated based on the backpack–AdQSM method correlated well with the AGB of individual trees calculated by the allometric biomass models. This indicates that the backpack–AdQSM method effectively accounts for the biophysical structure of individual trees and is suitable for the accurate estimation of individual tree AGB. This nondestructive estimation method can be further used to test and calibrate new allochthonous growth models and provide a valuable reference for the accurate estimation of carbon stocks in temperate natural forest ecosystems.

Author Contributions

A.R.; methodology, software, writing—original draft preparation, W.D.; validation, W.D., H.Y. and B.W.; formal analysis, H.D.; writing—review and editing, Y.S.; data curation, resources. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the Inner Mongolia “Rejuvenate Inner Mongolia Through Science and Technology” Action Key Special Project (Grant No. 2020ZD0028), Inner Mongolia Autonomous Region “14th Five-Year Plan” Key Research and Development and Achievement Transformation Program in Social Public Welfare (Grant No. 2022YFSH0027), “Forest and grassland fire monitoring and early warning and emergency management system”, an autonomous region science and technology innovation guidance award fund project, The Central Leading Local Science and Technology Development Funds “Integrated Demonstration of Ecological Protection and Comprehensive Utilization of Resources Technology in Alshan”, “Key Technology Research on Forest and Grassland Fire Risk Assessment” a project for the introduction of high-level talents in Inner Mongolia Autonomous Region in 2021, Project of Introducing High-level Talents to Inner Mongolia Normal University (Grant No. 2020YJRC050), “Fundamental Research Funds for the Inner Mongolia Normal University” (Grant No. 2022JBXC017), “Graduate Students’ Re-search & Innovation Fund of Inner Mongolia Normal University” (Grant No. CXJJS22133).

Data Availability Statement

Not applicable.

Acknowledgments

We are grateful to the fieldwork support from the Inner Mongolia Key Laboratory of Remote Sensing and Geographic Information Systems. And thanks to the support of the Arshan Forest and Grassland Disaster Prevention and Mitigation Field Scientific Observation and Research Station of Inner Mongolia Autonomous Region and the National Natural Science Foundation of China (No. 42201374).

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Tian, X.; Su, Z.; Chen, E.; Li, Z.; van der Tol, C.; Guo, J.; He, Q. Estimation of forest above-ground biomass using multi-parameter remote sensing data over a cold and arid area. Int. J. Appl. Earth Obs. Geoinf. 2012, 14, 160–168. [Google Scholar] [CrossRef]
  2. Tsui, O.W.; Coops, N.C.; Wulder, M.A.; Marshall, P.L. Integrating airborne LiDAR and space-borne radar via multivariate kriging to estimate above-ground biomass. Remote Sens. Environ. 2013, 139, 340–352. [Google Scholar] [CrossRef]
  3. He, Q.-S.; Cao, C.-X.; Chen, E.-X.; Sun, G.-Q.; Ling, F.-L.; Pang, Y.; Zhang, H.; Ni, W.-J.; Xu, M.; Li, Z.-Y.; et al. Forest stand biomass estimation using ALOS PALSAR data based on LiDAR-derived prior knowledge in the Qilian Mountain, western China. Int. J. Remote Sens. 2011, 33, 710–729. [Google Scholar] [CrossRef]
  4. Popescu, S.C. Estimating biomass of individual pine trees using airborne lidar. Biomass Bioenergy 2007, 31, 646–655. [Google Scholar] [CrossRef]
  5. Feliciano, E.A.; Wdowinski, S.; Potts, M.D. Assessing Mangrove Above-Ground Biomass and Structure using Terrestrial Laser Scanning: A Case Study in the Everglades National Park. Wetlands 2014, 34, 955–968. [Google Scholar] [CrossRef]
  6. Olschofsky, K.; Mues, V.; Köhl, M. Operational assessment of aboveground tree volume and biomass by terrestrial laser scanning. Comput. Electron. Agric. 2016, 127, 699–707. [Google Scholar] [CrossRef] [Green Version]
  7. Zimbres, B.; Shimbo, J.; Bustamante, M.; Levick, S.; Miranda, S.; Roitman, I.; Silvério, D.; Gomes, L.; Fagg, C.; Alencar, A. Savanna vegetation structure in the Brazilian Cerrado allows for the accurate estimation of aboveground biomass using terrestrial laser scanning. For. Ecol. Manag. 2020, 458, 117798. [Google Scholar] [CrossRef]
  8. Ghimire, S.; Xystrakis, F.; Koutsias, N. Using Terrestrial Laser Scanning to Measure Forest Inventory Parameters in a Mediterranean Coniferous Stand of Western Greece. PFG–J. Photogramm. Remote Sens. Geoinf. Sci. 2017, 85, 213–225. [Google Scholar] [CrossRef]
  9. Beyene, S.M. Estimation of Forest Variable and Aboveground Biomass using Terrestrial Laser Scanning in the Tropical Rainforest. J. Indian Soc. Remote Sens. 2020, 48, 853–863. [Google Scholar] [CrossRef]
  10. Xie, Y.; Zhang, J.; Chen, X.; Pang, S.; Zeng, H.; Shen, Z. Accuracy assessment and error analysis for diameter at breast height measurement of trees obtained using a novel backpack LiDAR system. For. Ecosyst. 2020, 7, 33. [Google Scholar] [CrossRef]
  11. Liu, L.; Zhang, A.; Xiao, S.; Hu, S.; He, N.; Pang, H.; Zhang, X.; Yang, S. Single Tree Segmentation and Diameter at Breast Height Estimation With Mobile LiDAR. IEEE Access 2021, 9, 24314–24325. [Google Scholar] [CrossRef]
  12. Xie, Y.; Yang, T.; Wang, X.; Chen, X.; Pang, S.; Hu, J.; Wang, A.; Chen, L.; Shen, Z. Applying a Portable Backpack Lidar to Measure and Locate Trees in a Nature Forest Plot: Accuracy and Error Analyses. Remote Sens. 2022, 14, 1806. [Google Scholar] [CrossRef]
  13. Hyyppä, E.; Yu, X.; Kaartinen, H.; Hakala, T.; Kukko, A.; Vastaranta, M.; Hyyppä, J. Comparison of Backpack, Handheld, Under-Canopy UAV, and Above-Canopy UAV Laser Scanning for Field Reference Data Collection in Boreal Forests. Remote Sens. 2020, 12, 3327. [Google Scholar] [CrossRef]
  14. Oveland, I.; Hauglin, M.; Giannetti, F.; Schipper Kjørsvik, N.; Gobakken, T. Comparing Three Different Ground Based Laser Scanning Methods for Tree Stem Detection. Remote Sens. 2018, 10, 538. [Google Scholar] [CrossRef] [Green Version]
  15. Wang, C. Biomass allometric equations for 10 co-occurring tree species in Chinese temperate forests. For. Ecol. Manag. 2006, 222, 9–16. [Google Scholar] [CrossRef]
  16. Lu, J.; Wang, H.; Qin, S.; Cao, L.; Pu, R.; Li, G.; Sun, J. Estimation of aboveground biomass of Robinia pseudoacacia forest in the Yellow River Delta based on UAV and Backpack LiDAR point clouds. Int. J. Appl. Earth Obs. Geoinf. 2020, 86, 102014. [Google Scholar] [CrossRef]
  17. Raumonen, P.; Kaasalainen, M.; Åkerblom, M.; Kaasalainen, S.; Kaartinen, H.; Vastaranta, M.; Holopainen, M.; Disney, M.; Lewis, P. Fast Automatic Precision Tree Models from Terrestrial Laser Scanner Data. Remote Sens. 2013, 5, 491–520. [Google Scholar] [CrossRef] [Green Version]
  18. Chen, S.; Feng, Z.; Chen, P.; Ullah Khan, T.; Lian, Y. Nondestructive Estimation of the Above-Ground Biomass of Multiple Tree Species in Boreal Forests of China Using Terrestrial Laser Scanning. Forests 2019, 10, 936. [Google Scholar] [CrossRef] [Green Version]
  19. Gonzalez de Tanago, J.; Lau, A.; Bartholomeus, H.; Herold, M.; Avitabile, V.; Raumonen, P.; Martius, C.; Goodman, R.C.; Disney, M.; Manuri, S.; et al. Estimation of above-ground biomass of large tropical trees with terrestrial LiDAR. Methods Ecol. Evol. 2017, 9, 223–234. [Google Scholar] [CrossRef] [Green Version]
  20. Krishna Moorthy, S.M.; Raumonen, P.; Van den Bulcke, J.; Calders, K.; Verbeeck, H. Terrestrial laser scanning for non-destructive estimates of liana stem biomass. For. Ecol. Manag. 2020, 456, 117751. [Google Scholar] [CrossRef]
  21. Lau, A.; Calders, K.; Bartholomeus, H.; Martius, C.; Raumonen, P.; Herold, M.; Vicari, M.; Sukhdeo, H.; Singh, J.; Goodman, R. Tree Biomass Equations from Terrestrial LiDAR: A Case Study in Guyana. Forests 2019, 10, 527. [Google Scholar] [CrossRef] [Green Version]
  22. Du, S.; Lindenbergh, R.; Ledoux, H.; Stoter, J.; Nan, L. AdTree: Accurate, Detailed, and Automatic Modelling of Laser-Scanned Trees. Remote Sens. 2019, 11, 2074. [Google Scholar] [CrossRef] [Green Version]
  23. Fan, G.; Nan, L.; Chen, F.; Dong, Y.; Wang, Z.; Li, H.; Chen, D. A New Quantitative Approach to Tree Attributes Estimation Based on LiDAR Point Clouds. Remote Sens. 2020, 12, 1779. [Google Scholar] [CrossRef]
  24. Fan, G.; Nan, L.; Dong, Y.; Su, X.; Chen, F. AdQSM: A New Method for Estimating Above-Ground Biomass from TLS Point Clouds. Remote Sens. 2020, 12, 3089. [Google Scholar] [CrossRef]
  25. Dong, Y.; Fan, G.; Zhou, Z.; Liu, J.; Wang, Y.; Chen, F. Low Cost Automatic Reconstruction of Tree Structure by AdQSM with Terrestrial Close-Range Photogrammetry. Forests 2021, 12, 1020. [Google Scholar] [CrossRef]
  26. Zhang, C.; Yang, G.; Jiang, Y.; Xu, B.; Li, X.; Zhu, Y.; Lei, L.; Chen, R.; Dong, Z.; Yang, H. Apple Tree Branch Information Extraction from Terrestrial Laser Scanning and Backpack-LiDAR. Remote Sens. 2020, 12, 3592. [Google Scholar] [CrossRef]
  27. Raumonen, P.; Casella, E.; Calders, K.; Murphy, S.; Åkerblom, M.; Kaasalainen, M. Massive-Scale Tree Modelling from Tls Data. ISPRS Ann. Photogramm. Remote Sens. Spat. Inf. Sci. 2015, II-3/W4, 189–196. [Google Scholar] [CrossRef] [Green Version]
  28. Brusca, S.; Famoso, F.; Lanzafame, R.; Messina, M.; Monforte, P. Placement optimization of biodiesel production plant by means of centroid mathematical method. Energy Procedia 2017, 126, 353–360. [Google Scholar] [CrossRef]
  29. Zhao, X.; Guo, Q.; Su, Y.; Xue, B. Improved progressive TIN densification filtering algorithm for airborne LiDAR data in forested areas. ISPRS J. Photogramm. Remote Sens. 2016, 117, 79–91. [Google Scholar] [CrossRef] [Green Version]
  30. Tao, S.; Wu, F.; Guo, Q.; Wang, Y.; Li, W.; Xue, B.; Hu, X.; Li, P.; Tian, D.; Li, C.; et al. Segmenting tree crowns from terrestrial and mobile LiDAR data by exploring ecological theories. ISPRS J. Photogramm. Remote Sens. 2015, 110, 66–76. [Google Scholar] [CrossRef] [Green Version]
  31. Meng, S. The Aboveground Biomass of the Main Tree Species in Daxing’anling Mountains. Ph.D. Thesis, Beijing Forestry University, Beijing, China, 2018. [Google Scholar]
  32. Chuanguo, C.; Junfeng, Z. Woody Biomass Manual of Typical Species in the Northeast of China; China Forestry Publishing House: Beijing, China, 1989. [Google Scholar]
  33. Liang, X.; Hyyppä, J.; Kaartinen, H.; Holopainen, M.; Melkas, T. Detecting Changes in Forest Structure over Time with Bi-Temporal Terrestrial Laser Scanning Data. ISPRS Int. J. Geo-Inf. 2012, 1, 242–255. [Google Scholar] [CrossRef]
  34. Liang, X.; Hyyppä, J. Automatic Stem Mapping by Merging Several Terrestrial Laser Scans at the Feature and Decision Levels. Sensors 2013, 13, 1614–1634. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  35. Kankare, V.; Liang, X.; Vastaranta, M.; Yu, X.; Holopainen, M.; Hyyppä, J. Diameter distribution estimation with laser scanning based multisource single tree inventory. ISPRS J. Photogramm. Remote Sens. 2015, 108, 161–171. [Google Scholar] [CrossRef]
  36. Bauwens, S.; Bartholomeus, H.; Calders, K.; Lejeune, P. Forest Inventory with Terrestrial LiDAR: A Comparison of Static and Hand-Held Mobile Laser Scanning. Forests 2016, 7, 127. [Google Scholar] [CrossRef] [Green Version]
  37. Cabo, C.; Del Pozo, S.; Rodríguez-Gonzálvez, P.; Ordóñez, C.; González-Aguilera, D. Comparing Terrestrial Laser Scanning (TLS) and Wearable Laser Scanning (WLS) for Individual Tree Modeling at Plot Level. Remote Sens. 2018, 10, 540. [Google Scholar] [CrossRef] [Green Version]
  38. Hartley, R.J.L.; Jayathunga, S.; Massam, P.D.; De Silva, D.; Estarija, H.J.; Davidson, S.J.; Wuraola, A.; Pearse, G.D. Assessing the Potential of Backpack-Mounted Mobile Laser Scanning Systems for Tree Phenotyping. Remote Sens. 2022, 14, 3344. [Google Scholar] [CrossRef]
  39. Su, Y.; Guo, Q.; Jin, S.; Guan, H.; Sun, X.; Ma, Q.; Hu, T.; Wang, R.; Li, Y. The Development and Evaluation of a Backpack LiDAR System for Accurate and Efficient Forest Inventory. IEEE Geosci. Remote Sens. Lett. 2021, 18, 1660–1664. [Google Scholar] [CrossRef]
  40. Daamen, W.J.R.-S.L. Results from a Check on Data Collection of the National Forest Survey in 1973–1977; Swedish University of Agricultural Sciences: Uppsala, Sweden, 1980. [Google Scholar]
  41. Hyyppä, E.; Kukko, A.; Kaijaluoto, R.; White, J.C.; Wulder, M.A.; Pyörälä, J.; Liang, X.; Yu, X.; Wang, Y.; Kaartinen, H.; et al. Accurate derivation of stem curve and volume using backpack mobile laser scanning. ISPRS J. Photogramm. Remote Sens. 2020, 161, 246–262. [Google Scholar] [CrossRef]
  42. Ko, C.; Lee, S.; Yim, J.; Kim, D.; Kang, J. Comparison of Forest Inventory Methods at Plot-Level between a Backpack Personal Laser Scanning (BPLS) and Conventional Equipment in Jeju Island, South Korea. Forests 2021, 12, 308. [Google Scholar] [CrossRef]
  43. Gao, S.; Zhang, Z.; Cao, L. Individual Tree Structural Parameter Extraction and Volume Table Creation Based on Near-Field LiDAR Data: A Case Study in a Subtropical Planted Forest. Sensors 2021, 21, 8162. [Google Scholar] [CrossRef]
  44. Calders, K.; Newnham, G.; Burt, A.; Murphy, S.; Raumonen, P.; Herold, M.; Culvenor, D.; Avitabile, V.; Disney, M.; Armston, J.; et al. Nondestructive estimates of above-ground biomass using terrestrial laser scanning. Methods Ecol. Evol. 2014, 6, 198–208. [Google Scholar] [CrossRef]
  45. Kaasalainen, S.; Krooks, A.; Liski, J.; Raumonen, P.; Kaartinen, H.; Kaasalainen, M.; Puttonen, E.; Anttila, K.; Mäkipää, R. Change Detection of Tree Biomass with Terrestrial Laser Scanning and Quantitative Structure Modelling. Remote Sens. 2014, 6, 3906–3922. [Google Scholar] [CrossRef] [Green Version]
  46. Dutcă, I.; McRoberts, R.E.; Næsset, E.; Blujdea, V.N. Accommodating heteroscedasticity in allometric biomass models. For. Ecol. Manag. 2021, 505, 119865. [Google Scholar] [CrossRef]
Forests 14 00475 g001 550
Figure 1. (A) Location of the study area, (B) the distribution of backpack LiDAR scanning tracks and TLS stations in the sample area.
Figure 1. (A) Location of the study area, (B) the distribution of backpack LiDAR scanning tracks and TLS stations in the sample area.
Forests 14 00475 g001
Forests 14 00475 g002 550
Figure 2. (A) Larix gmelinii plot. (B) Betula platyphylla plot.
Figure 2. (A) Larix gmelinii plot. (B) Betula platyphylla plot.
Forests 14 00475 g002
Forests 14 00475 g003 550
Figure 3. (A) Backpack LiDAR data acquisition and point cloud data. (B) TLS data acquisition and point cloud data.
Figure 3. (A) Backpack LiDAR data acquisition and point cloud data. (B) TLS data acquisition and point cloud data.
Forests 14 00475 g003
Forests 14 00475 g004 550
Figure 4. LiDAR point cloud data fitted to DBH. (A,B) Results of the backpack LiDAR point cloud data fitted to DBH. (C,D) Results of the TLS point cloud data fitted to DBH.
Figure 4. LiDAR point cloud data fitted to DBH. (A,B) Results of the backpack LiDAR point cloud data fitted to DBH. (C,D) Results of the TLS point cloud data fitted to DBH.
Forests 14 00475 g004
Forests 14 00475 g005 550
Figure 5. Individual tree segmentation.
Figure 5. Individual tree segmentation.
Forests 14 00475 g005
Forests 14 00475 g006 550
Figure 6. (A) 3D structural model of Larix gmelinii individual tree point cloud data and AdQSM reconstruction, and (B) 3D structural model of Betula platyphylla individual tree point cloud data and AdQSM reconstruction.
Figure 6. (A) 3D structural model of Larix gmelinii individual tree point cloud data and AdQSM reconstruction, and (B) 3D structural model of Betula platyphylla individual tree point cloud data and AdQSM reconstruction.
Forests 14 00475 g006
Forests 14 00475 g007 550
Figure 7. Comparison of DBH (A) and tree height (C) extracted by backpack LiDAR with measured data, and comparison of DBH (B) and tree height (D) extracted from TLS data with measured data (four colors indicate each of the four sample plots).
Figure 7. Comparison of DBH (A) and tree height (C) extracted by backpack LiDAR with measured data, and comparison of DBH (B) and tree height (D) extracted from TLS data with measured data (four colors indicate each of the four sample plots).
Forests 14 00475 g007
Forests 14 00475 g008 550
Figure 8. Comparison of AGB results based on the backpack–AdQSM estimates with those calculated by the allometric biomass models. (A,B) Comparison of estimated individual tree AGB of Larix gmelinii with reference values and residual distributions. (C,D) Comparison of estimated individual tree AGB of Betula platyphylla with reference values and residual distributions, respectively.
Figure 8. Comparison of AGB results based on the backpack–AdQSM estimates with those calculated by the allometric biomass models. (A,B) Comparison of estimated individual tree AGB of Larix gmelinii with reference values and residual distributions. (C,D) Comparison of estimated individual tree AGB of Betula platyphylla with reference values and residual distributions, respectively.
Forests 14 00475 g008
Forests 14 00475 g009 550
Figure 9. Cost–benefit analysis diagram of manual measurement methods versus BLS survey methods.
Figure 9. Cost–benefit analysis diagram of manual measurement methods versus BLS survey methods.
Forests 14 00475 g009
Table
Table 1. Tree species and DBH parameters at the sample site.
Table 1. Tree species and DBH parameters at the sample site.
PlotSpeciesNumber of
Trees		DBH (cm)
MeanMinMax
1Larix gmelinii3019.58.535
2Larix gmelinii2916.98.627.5
3Betula platyphylla4912.75.829.8
4Betula platyphylla4413.95.526.7
Table
Table 2. Comparison of various parameters of the backpack LiDAR and TLS.
Table 2. Comparison of various parameters of the backpack LiDAR and TLS.
LiBackpack DGC50FARO FOCUS 3D 120
Laser SensorVelodyne VLP-16 × 2Faro
LiDAR Accuracy±3 cm±1 mm
Scan Range100 m120 m
Scan Rate600,000 (pts/s)97,600 (pts/s)
Scan modecolorcolor
Pixel1800 W7000 W
Viewing angel rangeHorizontal 360°
Vertical 180°		Horizontal 360°
Vertical 300°
Table
Table 3. Allometric biomass models and basic wood density values for each tree species.
Table 3. Allometric biomass models and basic wood density values for each tree species.
SpeciesAllometric Biomass ModelsWood Density (ρ) (g/cm3)
Larix gmeliniiAGB = 0.0277 × DBH2.7930.498
Betula platyphyllaAGB = 0.1905 × DBH2.2430.427
Table
Table 4. Estimating AGB results of different tree species with different HS parameters in AdQSM.
Table 4. Estimating AGB results of different tree species with different HS parameters in AdQSM.
SpeciesRMSE
(HS = 0.4)/kg		RMSE
(HS = 0.5)/kg		RMSE
(HS = 0.6)/kg		RMSE
(HS = 0.7)/kg		RMSE
(HS = 0.8)/kg		RMSE
(HS = 0.9)/kg		RMSE
(HS = 1.0)/kg
Larix gmelinii29.5155.95143.17114.42185.73104.3119.84
Betula platyphylla54.2165.6428.8645.8377.24126.81125.73
Table
Table 5. Comparison of the efficiency of manual measurement methods with BLS survey methods.
Table 5. Comparison of the efficiency of manual measurement methods with BLS survey methods.
Survey
Method		PersonnelPlotArea (m2)Time Consumption
Data CollectionData ProcessingTotalm2/min
Manual measurement3140030:165:1635:323.78
231:274:3035:573.75
342:146:0948:232.76
439:075:3644:433.00
BLS115:4221:0426:4615.12
25:2323:3628:5913.99
36:2426:1632:4012.35
46:0425:4231:4612.71
Table
Table 6. Comparison of existing research results between backpack LiDAR and TLS.
Table 6. Comparison of existing research results between backpack LiDAR and TLS.
ReferenceLiDAREquipmentDBHTree Height
R2RMSE (cm)R2 RMSE (m)
This studyBackpackVelodyne VLP-160.980.60.382.1
[14]BackpackVelodyne VLP 16-2.2--
[13]BackpackRiegl VUX-1HA-1.6-1.4
[10]BackpackVelodyne VLP 160.921.5--
[41]BackpackRiegl VUX-1HA---1.5
[16]BackpackVelodyne VLP-160.941.0--
[42]BackpackVelodyne VLP-16-1.3-1.7
[43]BackpackVelodyne VLP-160.961.5--
[39]BackpackVelodyne VLP-160.952.00.651.9
[38]BackpackVelodyne VLP-160.991.70.412.8
[12] BackpackVelodyne VLP 160.903.9--
This studyBackpackFaro Focus 3D x 1200.980.50.471.4
[14]TLSFaro Focus 3D x130-6.2--
[18]TLSFaro Focus S x 1500.971.20.91.3
[39]TLSRiegl VZ-4000.962.00.632.0
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Ruhan, A.; Du, W.; Ying, H.; Wei, B.; Shan, Y.; Dai, H. Estimation of Aboveground Biomass of Individual Trees by Backpack LiDAR Based on Parameter-Optimized Quantitative Structural Models (AdQSM). Forests 2023, 14, 475. https://doi.org/10.3390/f14030475

AMA Style

Ruhan A, Du W, Ying H, Wei B, Shan Y, Dai H. Estimation of Aboveground Biomass of Individual Trees by Backpack LiDAR Based on Parameter-Optimized Quantitative Structural Models (AdQSM). Forests. 2023; 14(3):475. https://doi.org/10.3390/f14030475

Chicago/Turabian Style

Ruhan, A, Wala Du, Hong Ying, Baocheng Wei, Yu Shan, and Haiyan Dai. 2023. "Estimation of Aboveground Biomass of Individual Trees by Backpack LiDAR Based on Parameter-Optimized Quantitative Structural Models (AdQSM)" Forests 14, no. 3: 475. https://doi.org/10.3390/f14030475

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop