Monitoring Seasonal Growth of Eucalyptus Plantation under Different Forest Age and Slopes Based on Multi-Temporal UAV Stereo Images

Abstract

Eucalyptus grows rapidly and can grow up to 10 m per year, making them an important source of wood supply. Guangxi is a typical karst terrain, with hills and plains being the main growth areas for Eucalyptus. The differences in terrain can affect the seasonal growth of Eucalyptus plantations at different forest ages, which in turn affects the stocking of Eucalyptus. Currently, indiscriminate logging and management not only cause resource waste but also bring economic losses to operators. In this study, an unmanned aerial vehicle (UAV) was used to obtain multi-temporal stereo images to investigate the seasonal growth differences in structural parameters of individual Eucalyptus trees under different forest ages and slopes, providing data support for the precise management of Eucalyptus resources. The results showed that the tree height, crown width, diameter at breast height (DBH), and aboveground biomass (AGB) of individual trees, extracted based on UAV stereo images, are highly correlated with the field measured data, with an R² of 0.99, 0.78, 0.75, and 0.92, and RMSE of 0.21 m, 0.16 m, 1.17 cm, and 3.79 kg/tree, respectively. The growth of Eucalyptus structural parameters varies in different seasons, with faster growth in spring and autumn, accounting for 76.39%, 73.75%, 73.65%, and 73.68% of the total annual growth, respectively. The growth of different structural parameters of individual trees is closely related to forest age, with tree height, crown width, and DBH gradually slowing down with the increase in forest age, while AGB shows a trend of first increasing and then decreasing. The differences in terrain also have a certain impact on the growth of individual Eucalyptus tree structural parameters. For individual 1-year-old and 3-year-old Eucalyptus trees, those located on gentle slopes grow faster in spring and autumn than those located on flat land. However, those located on flat land grow faster in summer and winter than those located on gentle slopes. For tree height, crown width, DBH, and AGB, the maximum annual growth differences between Eucalyptus trees on gentle slopes and flat are 3.17 m, 0.26 m, 1.9 cm, and 9.27 kg/tree, respectively. The results indicate that based on UAV stereo images, the individual tree structural parameters in Eucalyptus plantations under different forest ages and slopes can be extracted, as well as seasonal growth can be monitored, providing data support for logging, planting, and management of Eucalyptus plantations.

1. Introduction

Eucalyptus has the advantages of fast growth, high yield, and good wood properties. It is not only a vital source of short fiber in pulp and paper production [1], but also plays an important role in ecological security [2], regulating climate change, and ensuring national timber security [3]. Currently, eucalypti are extensively cultivated in China [4]. Guangxi, China, is mainly characterized by mountainous and hilly basins, with a mild climate and abundant rainfall, making it ideal for the planting and growth of Eucalyptus. Since the introduction of Eucalyptus in the 19th century, it has a history of over 130 years and has become the largest province in China in terms of Eucalyptus planting area. Due to the rapid growth and suitable climate conditions, eucalypti can grow 2–3 m in one quarter and up to 10 m in one year. They can be harvested within 4–5 years, making significant contributions to maintaining national wood security and alleviating the pressure on natural forest timber production [5]. However, due to its rapid growth, traditional forest resource survey data cannot meet the data requirements of Eucalyptus production management. Therefore, how to accurately and quickly obtain the growth of Eucalyptus of various forest ages in different terrains and seasons is the premise of achieving effective planting and logging of Eucalyptus trees, which is of great significance for the precise management of Eucalyptus in Guangxi.

Light detection and ranging (LiDAR) can penetrate the forest canopy and accurately obtain information on both the horizontal and vertical structures of forests, making it an important data source for estimating the structural parameters of individual trees in forests [6,7]. It has been widely used in the field of forest resource inventory and has achieved good results [8,9]. For example, Nelson et al. [10] utilized airborne laser scanning (ALS) to estimate the biomass and volume of a pine forest in southern Georgia, and the results indicated that the estimated biomass and volume were within 2.0% and 2.6% of the true values, respectively. Nilsson et al. [11] estimated the vertical structural parameters of a forest on the Scottish coast using LiDAR data with different densities, and the results showed that the estimated tree height was 2.1–3.7 m lower than the measured tree height. Leite et al. [12] utilized ALS data for assessing Eucalyptus volume through both the area method and individual tree method. The results showed that the estimated volume based on the area method had an R² = 0.95 and RMSE = 14.4%, while the estimated volume based on the individual tree method had an R² = 0.94 and RMSE = 16.4%. Although LiDAR can achieve an accurate estimation of individual tree structural parameters, it has not been widely applied in the practice of Eucalyptus plantation management due to its relatively high data acquisition cost and the rapid growth of Eucalyptus.

In addition to LiDAR data, high-resolution images are also the main data source for obtaining forest horizontal structural parameters. For example, Hu et al. [13] estimated the crown width of trees based on high-resolution images, and the correlation coefficient R² between the estimated crown width and the reference value was between 0.75 and 0.78. Dong et al. [14] improved the YOLOv7 algorithm using a simple parameter-free attention model and SIoU module and then, extracted crown width using the improved YOLOv7 algorithm based on high-resolution images. The result showed that R² = 0.83. You et al. [15] compared the differences in estimating individual tree structure parameters using LiDAR and unmanned aerial vehicle (UAV) stereo images and found that the R² of the crown width estimated based on LiDAR and high-resolution images were 0.93 and 0.88, with an RMSE of 0.79 and 0.89 m, respectively. Although low-cost high-resolution images can accurately extract forest structural parameters and monitor dynamic changes, light saturation and lack of vertical structural information remain the main factors limiting the widespread application of high-resolution images.

Advancements in structure from motion (SFM) and multi-view stereo (MVS) techniques have made it feasible to acquire high-density point cloud data using UAV stereo images [16]. This method not only enables the acquisition of horizontal and vertical structural parameters of forests but also significantly lowers data acquisition costs in comparison to ALS data. It provides an economically feasible alternative for monitoring the growth and changes in forest structural parameters over short time spans [17,18]. For example, Krause et al. [19] employed UAV images for individual tree height extraction, resulting in an RMSE of 0.48 m when compared to the measured data. Tang et al. [20] used multi-temporal UAV-RGB stereo images to monitor the monthly growth of tree height in subtropical mixed forests. The R² of extracting individual tree heights based on UAV-RGB stereo images is 0.99, and the RMSE is 0.36 m. The total growth of tree height within a year is 46.5 cm, and the fastest growth changes occur in May and June. The above research indicates that it is possible to extract the growth and changes in forest individual tree structural parameters over short time spans based on multi-temporal UAV stereo images. Nevertheless, prior research has primarily concentrated on the examination of individual tree heights, without considering the differences in growth and changes in different forest structural parameters under different forest ages, slopes, and seasonal conditions.

Accordingly, this study aims to collect multi-temporal UAV images using a lightweight UAV equipped with consumer-grade RGB cameras, extract individual tree structural parameters and seasonal growth changes in Eucalyptus plantations from different forest ages and slopes, and analyze the impact of slope and forest age on the growth changes. The specific goals are outlined as follows: (1) to verify the feasibility of extracting structural parameters of individual Eucalyptus trees from UAV images; (2) to explore the differences in growth changes in the structural parameters of individual Eucalyptus trees in different seasons; (3) to explore the differences in growth changes in the structural parameters of Eucalyptus at different forest ages; and (4) to explore the differences in growth changes in the structural parameters of individual Eucalyptus trees on different slopes.

2. Materials and Methods

2.1. Study Area

The study site is located in Luzhai County, Liuzhou City, Guangxi Zhuang Autonomous Region, China (109°30′0″ E–110°15′0″ E, 24°15′0″ N–25°0′0″ N) (Figure 1). It is in the central part of Guangxi, with a typical karst landform and a mild climate. The annual mean high temperature is 26 °C and the annual mean low temperature is 17 °C. In 2022, there were 146 days of rainfall throughout the year, with a total rainfall of 1379.8 mm. It is an important Eucalyptus planting area in China, with 1.4 million acres of Eucalyptus currently planted, and an annual harvest of 1.21 million cubic meters.

2.2. Data Introduction

2.2.1. Field Data

On 25 July 2022, field data collection was conducted in two study areas, including information such as the spatial position, slope, forest age, tree height, diameter at breast height (DBH), and ground point elevation of individual trees. Among them, the spatial position of individual trees, ground point elevation, and slope are acquired through a combined measurement approach involving a total station and real-time kinematics (RTKs). Forest age information was obtained by asking the owner of the Eucalyptus plantation (Figure 2). Tree height was obtained by ultrasonic altimeter Vertex IV, and DBH was measured by a diameter ruler. The crown width was measured by a tape measure, and the slope was measured by a geological compass. A total of 402 individual Eucalyptus trees of four age groups, including less than one-year-old, one-year-old, two-years-old, and three-years-old, were collected in this field data collection. The specific statistical results are shown in Table 1, and the statistical results of forest age and slope in each forest in this study area are shown in Table 2.

2.2.2. UAV Data

The stereo images of the Eucalyptus plantation were collected using the DJI Phantom 4 RTK UAV in July and October of 2022, and February, May, and July of 2023. The weather of flights was windless or breezy, using a five-direction flight mode, with a flight height of 70 m and a flight speed of 5.7 m/s. The forward overlap was set to 80%, while the side overlap was set to 70%. The WGS84 was used as the coordinate system. To ensure the positioning accuracy of the UAV image and the quality of the subsequent production of canopy height models (CHM), the RTK service type of UAV was configured as Network RTK.

2.3. Method

2.3.1. UAV Data Processing

Based on the UAV images obtained from five time periods, the UAV images were firstly processed using ContextCapture 4.4.12 software to generate a digital surface model (DSM) with a resolution of 3 cm. Subsequently, the UAV images were processed using Pix4D 4.7.5 software to obtain point cloud data for the study area. We classified the point cloud data and utilized the classified point cloud for interpolation to generate a digital elevation model (DEM) with a spatial resolution of 3 cm. Finally, we subtracted the DSM from DEM to obtain CHM. This data serves as the foundation for the subsequent acquisition of growth changes in the structural parameters of individual Eucalyptus trees of different forest ages and slopes.

2.3.2. Individual Tree Crown Segmentation

The watershed segmentation algorithm is a mathematical morphology segmentation method based on topological theory. Its basic idea is to transform a grayscale image into a gradient image, where the gradient value is regarded as a high and low mountain range, and the local minimum value and its neighborhood are regarded as a catchment basin. As these basins fill with ‘water’ and the water level rises, the lower gradient areas become submerged. Once the rising water stabilizes, segmentation boundaries or dividing lines can be identified, allowing the image to be segmented into distinct regions [21]. Because of the watershed algorithm’s strong performance in detecting faint edges, it has found successful application in research related to individual tree segmentation, yielding satisfactory results [22,23,24]. To obtain the tree height of Eucalyptus at different forest ages in various seasons, individual tree segmentation was performed on the CHMs at five time periods. When deploying the watershed algorithm, parameters such as the maximum tree height, minimum tree height, Gaussian smoothing factor, and smoothing window radius were required to be set. The maximum tree height and minimum tree height are set based on the corresponding CHM of the study area. Through a series of experiments, it was found that the Gaussian smoothing factor is set to 9 and the smoothing window radius is set to 111 pixels, the watershed algorithm has relatively good individual tree segmentation results.

To assess the precision of segmentation outcomes using the watershed, visual interpretation is employed. The main evaluation metrics used are recall (r), precision (p), and the F-score (F). The specific formulas are as follows [25,26]:

(1)$r=\frac{n_{TP}}{n_{TP}+n_{FN}}\times 100\%$

(2)$p=\frac{n_{TP}}{n_{TP}+n_{FP}}\times 100\%$

(3)$F=\frac{2rp}{r+p}\times 100\%$

where n_TP stands for the count of accurately segmented; n_FN for the unsegmented count; and n_FP for the over-segmented count.

2.3.3. Extraction of Individual Eucalyptus Tree Structural Parameters

Based on the results of individual tree segmentation, the maximum value of the CHM in the region corresponding to the individual tree segmentation vector result is extracted using local maximum values as the tree height, thus achieving the extraction of individual tree height. Concurrently, by evaluating the diameters of the segmentation results in both east–west and north–south directions and computing their average, the crown width of each tree was ascertained. To obtain the individual tree above-ground biomass (AGB) of Eucalyptus, a tree height–DBH statistical model, established based on field-measured data, is inverted to obtain the individual tree DBH, and then the individual tree AGB is obtained based on the biomass allometric growth equation. The specific computational formula is as follow [27]:

(4)$$\begin{gathered} \mathrm{A}\mathrm{G}\mathrm{B}=0.0241{\left(D^{2}H\right)}^{0.9768}+0.0049{\left(D^{2}H\right)}^{0.8449} \\ +0.0007{\left(D^{2}H\right)}^{0.9355}+0.0064{\left(D^{2}H\right)}^{0.8738} \end{gathered}$$

where AGB represents the above-ground biomass of the individual Eucalyptus tree (kg/tree), D denotes the diameter at breast height of the individual tree (cm), and H signifies the height of the individual tree (m).

To measure the accuracy of individual tree structural parameter extraction, R² and RMSE were used for evaluation, as shown in the following formula [19]:

(5)$R^2=1-\frac{\sum _{i=1}^n{\left(Y_i-y_i\right)}^2}{\sum _{i=1}^n{\left(Y_i-\overline{Y}\right)}^2}$

where $y_i$ represents the estimated value, $Y_i$ represents the measured value, and $\overline{Y}$ represents the average value.

(6)$RMSE=\sqrt{\frac{1}{n}\sum _{i=1}^n{\left(Y_i-y_i\right)}^2}$

where RMSE represents root mean square error, $y_i$ represents the estimated value, and $Y_i$ represents the measured value.

2.3.4. Extraction of Seasonal Growth in Structural Parameters of Individual Eucalyptus Trees under Different Forest Ages and Slopes

Based on the extraction results of individual tree height, crown width, DBH, and AGB in different time phases, seasonal differential calculations were performed to obtain the growth changes in different seasons, to explore the growth changes in structural parameters, such as tree height, crown width, DBH, and AGB of individual Eucalyptus trees under different forest ages and slopes.

3. Results

3.1. UAV Data Processing Results

The DEM, DSM, and CHM generated based on the UAV stereo images are shown in Figure 3.

As depicted in Figure 3, the elevation changes of the DEM in Study Area A is relatively large, at 27 m, which is greater than the elevation changes in Study Area B. Therefore, the slope of Study Area A is greater than that of Study Area B. However, the difference in tree height between Study Areas A and B is relatively small, with CHM ranging from 0 to 20 m.

To further quantify the accuracy of the DEM generation by interpolating point clouds sourced from UAV imagery, a correlation analysis was conducted between the elevation values of interpolated DEM and field-measured elevation. The results are shown in Figure 4.

The results shown in Figure 4 suggest that the elevation values from the DEM generated by interpolating point clouds derived from UAV stereo images have a high correlation with the field-measured elevations, with an R² of 0.99 and an RMSE of 0.20 m. The results indicate that UAV stereo images can generate accurate DEMs under Eucalyptus plantations, enabling accurate tree height extraction in subsequent analyses.

3.2. Results of Individual Tree Segmentation

Based on the CHM, the watershed segmentation algorithm was used for individual tree segmentation, and the results are shown in Figure 5.

As shown in Figure 5, the watershed algorithm can effectively segment the majority of the Eucalyptus plantation, but there may still be varying degrees of over-segmentation and under-segmentation in some areas, as highlighted in the designated regions of Figure 4.

To quantitatively analyze the results of individual tree segmentation, 85 sample points were selected for validation in each sample plot, and the correct segmentation, over-segmentation, and under-segmentation of each forest were statistically analyzed. The recall (r), accuracy rate (p), and F-score (F) measures were used for evaluation. The statistical results are shown in Table 3.

From the results shown in Table 3, the F-scores of various individual tree segmentation results are above 90%, with a mean F-score of 96.63%. This indicates that the watershed algorithm can achieve high-precision individual tree segmentation in Eucalyptus plantations, rendering overall satisfactory tree segmentation results.

To summarize, the watershed algorithm can effectively achieve individual tree segmentation in Eucalyptus plantations. However, as depicted in Figure 4, the segmented individual tree crown vector results often divide the gaps between tree crowns into individual tree crowns, resulting in the calculated crown area being larger than the actual crown area. Therefore, directly utilizing the watershed individual tree segmentation results presents challenges in accurately measuring the changes in individual tree crown width. In order to accurately quantify the seasonal growth changes in individual tree crown diameter in the future, it is necessary to manually correct the segmentation results of individual trees.

3.3. Results of Individual Tree Structural Parameter Extraction

• (1). Results of Individual Tree Structural Parameter Extraction

To validate the precision of extracting individual tree height, crown width, DBH, and AGB using UAV data, a correlation analysis was carried out to compare the extraction results with field measurements. The results are shown in Figure 6.

The results depicted in Figure 6 indicate the potential for precise Eucalyptus structural parameter extraction using UAV stereo imagery, with the highest extraction accuracy for tree height, R² = 0.99, RMSE = 0.21m, followed by AGB results, R² = 0.92, RMSE = 3.79 kg/tree. While the results for individual crown width and DBH are relatively poor, with R² values of 0.78 and 0.75, and RMSE values of 0.16 m and 1.17 cm, respectively.

• (2). Extraction Results of Individual Structural Parameters for Eucalyptus Across Different Forest Ages

A total of 85 Eucalyptus plantation samples were selected from different forest age plots and based on the extraction results of individual tree structural parameters, a distribution map of the individual structural parameters for Eucalyptus plantations of different forest ages was plotted. The results are shown in Figure 7.

As depicted in Figure 7, the differences in tree height, DBH, and AGB among different forest ages of Eucalyptus trees are relatively large. For example, the mean tree heights of one-month-old, one-year-old, two-year-old, and three-year-old Eucalyptus forests are 1.96 m, 10.7 m, 13.8 m, and 16.7 m, respectively. However, the differences in individual crown width among the different forest ages of Eucalyptus trees are relatively small, especially for one-year-old, two-year-old, and three-year-old forests, with a mean value of 2.17 m, 2.33 m, and 2.33 m, respectively. This is mainly because the planting spacing of Eucalyptus plantations is usually 2.0 m × 3.0 m, which limits the growth of the crown width.

3.4. Seasonal Growth Results of Individual Eucalyptus Structural Parameters under Different Forest Ages and Slope Conditions

• (1). The Seasonal Growth Results of Individual Tree Structural Parameters in Eucalyptus Plantations with Different Forest Ages

Based on the extraction results of individual structural parameters for five periods, the seasonal growth changes in individual structural parameters in Eucalyptus plantations of different forest ages were calculated, as shown in Figure 8.

As shown in Figure 8, for different individual tree structural parameters of Eucalyptus plantations, the growth changes are generally significant in spring and autumn, while the growth changes in summer are relatively small, and the growth in winter is the slowest. For individual tree height, crown width, and DBH, the difference in growth changes among the four seasons of a one-month-old Eucalyptus is the most significant. The growth changes in the tree height in spring, summer, autumn, and winter are 2.20 m, 0.86 m, 2.35 m, and 0.61 m, respectively. The growth changes in the crown diameter are 0.47 m, 0.20 m, 0.45 m, and 0.16 m, respectively. The growth changes in the DHB are 1.32 cm, 0.52 cm, 1.19 cm, and 0.36 cm, respectively. For AGB, there is the most significant difference in growth changes among the four seasons of one-year-old Eucalyptus. The growth changes in AGB in spring, summer, autumn, and winter are 8.65 kg/tree, 3.20 kg/tree, 8.00 kg/tree, and 1.74 kg/tree, respectively. For different forest ages of Eucalyptus plantations, the average annual growth of individual tree height, crown width, and DBH gradually decreases with the increase in forest age, while the average annual growth of Eucalyptus AGB first increases, and then decreases with the increase in forest age.

• (2). The Seasonal Growth Results of Individual Tree Structural Parameters in Eucalyptus Plantations under Different Slopes

Seasonal changes in the growth from structural parameters of Eucalyptus under various slope conditions (flat and gentle slope) were determined using data extracted from five different periods. The results are illustrated in Figure 9.

According to the results shown in Figure 9, for Eucalyptus trees aged one-year-old and three-years-old, Eucalyptus trees on gentle slopes grow faster than those on flat land in spring and autumn. The maximum difference in height growth is 0.61 m/season, the maximum difference in crown growth is 0.02 m/season, the maximum difference in DBH growth is 0.36 cm/season, and the maximum difference in AGB growth is 3.74 kg/tree. However, in summer and winter, Eucalyptus trees on flat land grow faster than those on gentle slopes. For one-month-old Eucalyptus, Eucalyptus trees on gentle slopes grow faster than those on flat land in any season. In summary, Eucalyptus trees on gentle slopes grow better than those on flat land, and gentle slopes are more suitable for Eucalyptus planting.

4. Discussion

4.1. Analysis of Individual Tree Structure Parameter Extraction Results from UAV-RGB Images

UAVs equipped with consumer-grade RGB cameras have been widely employed for extracting forest structural parameters, primarily due to their cost-effective data acquisition, and they have yielded favorable results [28,29]. For example, Hao et al. [30] used a UAV equipped with inexpensive RGB cameras to estimate the height of young trees, and the estimated and measured heights were R² = 0.95 and RMSE = 0.12 m. Panagiotidis et al. [24] used the stereo images obtained with UAV to estimate the tree height and crown width and compared the estimated results with field-measured values. The findings indicated that the RMSE% of the crown was between 14.29 and 18.56, while the RMSE% of tree height was between 11.42 and 12.62. The above research reveals that UAV stereo imagery allows for high-precision extraction of structural parameters. The results of this study are basically consistent with previous research results. In this study, based on UAV stereo images, the extraction of tree height, crown width, DBH, and AGB of Eucalyptus plantations was conducted. The R² between the extracted individual tree structural parameters and the measured values were 0.99, 0.78, 0.75, and 0.92, respectively, and the RMSE were 0.21 m, 0.16 m, 1.17 cm, and 3.79 kg/tree, respectively. The results show that precise extraction of structural parameters is attainable using UAV-RGB imagery, establishing a dependable data foundation for tracking seasonal variations in Eucalyptus of tree height, crown width, DBH, and AGBs in future research.

4.2. Analysis of Growth Differences in Eucalyptus Plantations in Different Seasons

Precipitation and temperature are key factors affecting tree growth. Studies have shown that precipitation is positively correlated with tree growth [31], and in some cases, their impact on tree growth even exceeds that of temperature. The precipitation and temperature vary across different seasons, leading to distinct growth differences in Eucalyptus throughout different seasons. Based on multi-temporal UAV stereo images, this study extracts seasonal growth changes in the structural parameters of the Eucalyptus. The findings indicate that Eucalyptus plantations grow faster in spring and autumn, while they grow slower in summer and winter. This is mainly due to the combined influence of local precipitation and temperature. From July 2022 to July 2023, the total precipitation in the study area was 652.5 mm, while the precipitation in spring and autumn was 328.2 mm and 230.4 mm, respectively, accounting for 50% and 35% of the total precipitation. Therefore, the abundant precipitation in spring and autumn enables Eucalyptus to grow relatively faster in spring and autumn. Although Eucalyptus exhibits strong drought resistance, it is a species with high water demand for its growth, particularly during its active growth season. Consequently, its growth change during the dry season is notably lower compared to that in the rainy season [32,33]. Previous studies have shown the growth of Eucalyptus is positively correlated with temperature [34], and lower temperatures affect the growth of Eucalyptus, so it grows slower in winter. Therefore, considering the seasonal growth changes in Eucalyptus and the conditions of temperature and precipitation, it is recommended to cut down Eucalyptus in autumn.

4.3. Analysis of Growth Differences in Eucalyptus Plantations at Different Forest Ages

Previous studies have shown that forest age has a significant impact on tree growth. This study conducts growth monitoring of Eucalyptus plantations of different forest ages using multi-temporal UAV stereo images. The results show that as the forest age increases, the growth changes in height, crown width, and DBH of individual Eucalyptus trees gradually slow down, while the growth changes in AGB first increase, and then decrease. The growth changes in the structure parameters of individual Eucalyptus trees are mainly related to resource competition [35,36]. During the seedling stage, the Eucalyptus grows faster due to the large spatial distance between trees and sufficient light. As the trees grow, the crown expands, the light decreases, and the spatial competition intensifies. The demand for resources may exceed the amount that the environment can provide, resulting in a slowdown in the growth of tree height, crown width, and DBH. As the crown expands, the resources available per unit of leaf area will also decrease, which will limit the growth of the tree. In addition, as the trees grow, their physiological functions also undergo changes, such as a slowdown in cell division, leading to a decrease in growth. The growth changes in AGB in individual Eucalyptus trees were consistent with previous similar research conclusions. For example, Askne et al. [37] monitored the growth of AGB in northern forests based on TanDEM-X and found that the growth change in AGB in both young and elderly forests is slower, while the growth change in AGB in intermediate forest ages is faster. Neeff et al. [38] established a forest growth model for secondary forests in the central Amazon region and found that the AGB accumulated rapidly in the early stages, and the growth change in AGB gradually decreased and eventually stagnated as the forest ages increased. This could be attributed to the smaller size of Eucalyptus seedlings during the seedling stages, and AGB is closely related to the volume and density of the trees, resulting in a relatively low AGB. Due to the rapid growth change in Eucalyptus, it can reach approximately 10 m within a year, thereby accelerating the growth of AGB. However, as the forest ages increase, the competition among Eucalyptus intensifies, leading to a deceleration in tree height and DBH growth, consequently affecting the growth of AGB [39,40].

4.4. Analysis of Growth Differences in Eucalyptus Plantations in Different Slope

The slope has a certain influence on the distribution of sunlight and precipitation, thus the change in slope is also one of the important factors affecting the growth of Eucalyptus plantations. This study investigated the differences in the growth of Eucalyptus on flat land and gentle slopes. It was found that Eucalyptus on gentle slopes grows faster than on flat land. This is mainly caused by the different natural conditions brought about by different slopes. The gentle slope terrain helps to form a relatively loose soil structure, which facilitates water infiltration and discharge, providing a better drainage environment, allowing Eucalyptus to grow in a moist but not too wet environment, reducing the risk of water decay at the roots of the trees. Meanwhile, well-drained soil can ensure sufficient oxygen supply to the roots, which is crucial for the growth of trees. Due to the relatively small slope of the gentle slope, the soil will not suffer from the loss of water and nutrients. While flat land has almost no slope, and in contrast, the soil has poor drainage, which affects the growth of Eucalyptus due to the accumulation of water. In addition, the slope may affect the angle and duration of sunlight exposure. Eucalyptus on gentle slopes can receive relatively longer amounts of sunlight throughout the day, while Eucalyptus needs sufficient sunlight for photosynthesis. Therefore, the difference in sunlight exposure may affect their growth. In summary, Eucalyptus on gentle slopes may benefit from better soil conditions and superior sunlight exposure, making it grow faster on gentle slopes than on flat slopes. Therefore, in the future, Eucalyptus can be planted in hilly and gentle slope areas to achieve better economic and ecological benefits.

5. Conclusions

In this study, the multi-temporal UAV stereo images were collected using a lightweight UAV equipped with inexpensive consumer-grade RGB cameras. The structural parameters and seasonal growth changes in Eucalyptus of different forest ages and slopes were extracted and the effects of slope and forest age on the growth changes also were analyzed. The main conclusions are as follows:

• (1). Based on UAV images, it is possible to achieve high-precision extraction of structural parameters of individual trees in Eucalyptus plantations, with an extraction accuracy of R² = 0.99, RMSE = 0.21 m for individual tree height, an extraction accuracy of R² = 0.78, RMSE = 0.16 m for crown width, an extraction accuracy of R² = 0.75, RMSE = 1.17 cm for DBH, and an extraction accuracy of R² = 0.92, RMSE = 3.79 kg/tree for AGB.

• (2). The growth changes in the structural parameters of individual Eucalyptus trees vary in different seasons, with faster growth in spring and autumn and slower growth in summer and winter. The growth of tree height, crown width, DBH, and AGB in spring and autumn account for 76.39%, 73.75%, 73.65%, and 73.68% of the total annual growth, respectively.

• (3). The growth of different structural parameters of individual Eucalyptus trees is closely related to forest age. The growth of tree height, crown width, and DBH gradually slows down with increasing forest age, while AGB shows a trend of first increasing, and then decreasing. When the forest age is one-year-old, the growth change in AGB is the fastest.

• (4). The terrain has a certain impact on the growth of structural parameters of individual Eucalyptus trees. For Eucalyptus trees of one and three-years-old, Eucalyptus trees located on gentle slopes in spring and autumn grow faster than those on flat land. However, in summer and winter, Eucalyptus trees on flat land grow faster than those on gentle slopes. For Eucalyptus trees of one-month-old, Eucalyptus trees on gentle slopes grow faster than those on flat land in any season. For individual tree height, crown width, DBH, and AGB, the maximum annual growth differences between Eucalyptus trees on gentle slopes and flat land are 3.17 m, 0.26 m, 1.9 cm, and 9.27 kg/plant, respectively.

The results of this study imply the feasibility of monitoring seasonal variations in structural parameters within individual eucalypti through the use of multi-temporal UAV imagery and the effects of forest age and slope on the change in structural parameters of individual Eucalyptus trees. However, due to the limited number of forest ages and slope types selected in this study, it is impossible to further analyze the differences in the growth of Eucalyptus plantations caused by more slope types and more forest ages. Therefore, in the future, based on this study, more forest ages, slope types, and other influencing factors (such as soil type, altitude, and artificial fertilization) should be selected to analyze their impact on the growth changes in Eucalyptus plantations, in order to provide data reference and decision-making support for the sustainable management of Eucalyptus plantations.

Author Contributions

Conceptualization, X.T. and H.Y.; data curation, X.T., H.Y. and Q.Y.; formal analysis, X.T., P.L. and H.Y.; methodology, X.T., H.Y. and P.L.; supervision, S.J., J.D. and J.C.; validation, Y.L., J.C. and Q.Y.; writing—original draft preparation, X.T. and H.Y.; writing—review and editing, X.T. and H.Y. All authors have read and agreed to the published version of the manuscript.

Data Availability Statement

No new data were created or analyzed in this study. Data sharing is not applicable to this article.

Conflicts of Interest

The authors declare no conflict of interest.

Footnotes

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.

Figures and Tables

Figure 1. Study Area: (a) Study Area a; (b) Study Area b.

Figure 2. Eucalyptus plantations of different forest ages: (a) one month old; (b) one year old; (c) two years old; (d) three years old.

Figure 3. UAV stereo images processing results: (a) DEM; (b) DSM; (c) CHM. The first column is study area a, and the second column is study area b.

Figure 4. Results of elevation correlation analysis.

Figure 5. Results of individual tree segmentation: (a) Plot 1; (b) Plot 2; (c) Plot 3; (d) Plot 4; (e) Plot 5; (f) Plot 6.

Figure 6. Results of individual tree structural parameter extraction: (a) Tree height; (b) Crown width; (c) DBH; (d) AGB.

Figure 7. Box plot of the extraction results of individual Eucalyptus tree structural parameters of different forest ages: (a) Tree height; (b) Crown width; (c) DBH; (d) AGB.

Figure 8. Seasonal growth results of individual structural parameters in Eucalyptus plantations of different forest ages: (a) tree height; (b) crown width; (c) DBH; (d) AGB.

Figure 9. Seasonal growth changes in forest structural parameters in Eucalyptus plantations with different slopes: (a–c) Tree heights of one month old, one year old, and three years old; (d–f) Crown width of one month old, one year old, and three years old; (g–i) DBH of one month old, one year old, and three years old; (j–l) AGB of one month old, one year old, and three years old.

Table 1

Field data statistics results.

-	TH (m)	CW (m)	DBH (cm)
Maximum	17.98	2.77	13.8
Minimum	6.62	1.13	4.1
Average	13.71	2.17	9.4
Standard deviation	2.48	0.27	1.9

Note: TH stands for tree height, CW stands for crown width, and DBH stands for diameter at breast height.

Table 2

Statistical results of forest age and slope of each forest in the study area.

Plot Number	Forest Age	Slope (°)
1	Three years old	Gentle (5–10°)
2	One year old	Gentle (5–10°)
3	One month old	Gentle (5–10°)
4	One year old	Flat (0–5°)
5	One month old	Flat (0–5°)
6	Two years old	Flat (0–5°)
7	Three years old	Flat (0–5°)

Table 3

Statistics of individual tree segmentation results.

Plot Number	Forest Age	Quantity (Tree)	nTP	nFN	nFP	r (%)	p (%)	F (%)
1	Three years old	85	80	5	0	94.11	100	96.97
2	One year old	85	84	1	0	98.82	100	99.40
3	One month old	85	83	2	0	97.65	100	98.81
4	One year old	85	80	3	2	96.39	97.56	96.97
5	One month old	85	73	12	0	85.88	100	92.40
6	Two years old	85	74	7	4	91.36	94.87	93.08
7	Three years old	85	83	1	1	98.81	98.81	98.81

References

1. Silverio, F.O.; Barbosa, L.C.; Maltha, C.R.; Silvestre, A.J.; Pilo-Veloso, D.; Gomide, J.L. Characterization of lipophilic wood extractives from clones of Eucalyptus urograndis cultivate in Brazil. BioResources; 2007; 2, pp. 157-168. [DOI: https://dx.doi.org/10.15376/biores.2.2.157-168]

2. Dai, E.F.; Wang, X.L.; Zhu, J.J.; Xi, W.M. Quantifying ecosystem service trade-offs for plantation forest management to benefit provisioning and regulating services. Ecol. Evol.; 2017; 7, pp. 7807-7821. [DOI: https://dx.doi.org/10.1002/ece3.3286] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/29043036]

3. Palma, J.; Hakamada, R.; Moreira, G.G.; Nobre, S.; Rodriguez, L.C.E. Using 3PG to assess climate change impacts on management plan optimization of Eucalyptus plantations. A case study in Southern Brazil. Sci. Rep.; 2021; 11, 2708. [DOI: https://dx.doi.org/10.1038/s41598-021-81907-z]

4. Farias, A.A.; Soares, C.P.B.; Leite, H.G.; da Silva, G.F. Quantile regression: Prediction of growth and yield for a Eucalyptus plantation in northeast Brazil. Eur. J. For. Res.; 2021; 140, pp. 983-989. [DOI: https://dx.doi.org/10.1007/s10342-021-01380-1]

5. Zhang, Y.; Wang, X. Geographical spatial distribution and productivity dynamic change of Eucalyptus plantations in China. Sci. Rep.; 2021; 11, 19764. [DOI: https://dx.doi.org/10.1038/s41598-021-97089-7]

6. Maltamo, M.; Næsset, E.; Vauhkonen, J. Forestry Applications of Airborne Laser Scanning; Managing Forest Ecosystems Springer: Dordrecht, The Netherlands, 2014; Volume 27, 460.

7. Naesset, E. Estimating timber volume of forest stands using airborne laser scanner data. Remote Sens. Environ.; 1997; 61, pp. 246-253. [DOI: https://dx.doi.org/10.1016/S0034-4257(97)00041-2]

8. Hyyppä, J.; Hyyppä, H.; Leckie, D.; Gougeon, F.; Yu, X.; Maltamo, M. Review of methods of small-footprint airborne laser scanning for extracting forest inventory data in boreal forests. Int. J. Remote Sens.; 2008; 29, pp. 1339-1366. [DOI: https://dx.doi.org/10.1080/01431160701736489]

9. Van Leeuwen, M.; Nieuwenhuis, M. Retrieval of forest structural parameters using LiDAR remote sensing. Eur. J. For. Res.; 2010; 129, pp. 749-770. [DOI: https://dx.doi.org/10.1007/s10342-010-0381-4]

10. Nelson, R.; Krabill, W.; Tonelli, J. Estimating forest biomass and volume using airborne laser data. Remote Sens. Environ.; 1988; 24, pp. 247-267. [DOI: https://dx.doi.org/10.1016/0034-4257(88)90028-4]

11. Nilsson, M. Estimation of tree heights and stand volume using an airborne lidar system. Remote Sens. Environ.; 1996; 56, pp. 1-7. [DOI: https://dx.doi.org/10.1016/0034-4257(95)00224-3]

12. Leite, R.V.; Amaral, C.H.D.; Pires, R.d.P.; Silva, C.A.; Soares, C.P.B.; Macedo, R.P.; Silva, A.A.L.d.; Broadbent, E.N.; Mohan, M.; Leite, H.G. Estimating Stem Volume in Eucalyptus Plantations Using Airborne LiDAR: A Comparison of Area- and Individual Tree-Based Approaches. Remote Sens.; 2020; 12, 1513. [DOI: https://dx.doi.org/10.3390/rs12091513]

13. Hu, L.; Xu, X.; Wang, J.; Xu, H. Individual tree crown width detection from unmanned aerial vehicle images using a revised local transect method. Ecol. Inform.; 2023; 75, 102086. [DOI: https://dx.doi.org/10.1016/j.ecoinf.2023.102086]

14. Dong, C.; Cai, C.; Chen, S.; Xu, H.; Yang, L.; Ji, J.; Huang, S.; Hung, I.-K.; Weng, Y.; Lou, X. Crown Width Extraction of Metasequoia glyptostroboides Using Improved YOLOv7 Based on UAV Images. Drones; 2023; 7, 336. [DOI: https://dx.doi.org/10.3390/drones7060336]

15. You, H.; Tang, X.; You, Q.; Liu, Y.; Chen, J.; Wang, F. Study on the Differences between the Extraction Results of the Structural Parameters of Individual Trees for Different Tree Species Based on UAV LiDAR and High-Resolution RGB Images. Drones; 2023; 7, 317. [DOI: https://dx.doi.org/10.3390/drones7050317]

16. Iglhaut, J.; Cabo, C.; Puliti, S.; Piermattei, L.; O’Connor, J.; Rosette, J. Structure from motion photogrammetry in forestry: A review. Curr. For. Rep.; 2019; 5, pp. 155-168. [DOI: https://dx.doi.org/10.1007/s40725-019-00094-3]

17. Pádua, L.; Hruška, J.; Bessa, J.; Adão, T.; Martins, L.M.; Gonçalves, J.A.; Peres, E.; Sousa, A.M.R.; Castro, J.P.; Sousa, J.J. Multi-Temporal Analysis of Forestry and Coastal Environments Using UASs. Remote Sens.; 2018; 10, 24. [DOI: https://dx.doi.org/10.3390/rs10010024]

18. Chen, J.; Chen, Z.; Huang, R.; You, H.; Han, X.; Yue, T.; Zhou, G. The Effects of Spatial Resolution and Resampling on the Classification Accuracy of Wetland Vegetation Species and Ground Objects: A Study Based on High Spatial Resolution UAV Images. Drones; 2023; 7, 61. [DOI: https://dx.doi.org/10.3390/drones7010061]

19. Krause, S.; Sanders, T.G.M.; Mund, J.-P.; Greve, K. UAV-Based Photogrammetric Tree Height Measurement for Intensive Forest Monitoring. Remote Sens.; 2019; 11, 758. [DOI: https://dx.doi.org/10.3390/rs11070758]

20. Tang, X.; You, H.; Liu, Y.; You, Q.; Chen, J. Monitoring of Monthly Height Growth of Individual Trees in a Subtropical Mixed Plantation Using UAV Data. Remote Sens.; 2023; 15, 326. [DOI: https://dx.doi.org/10.3390/rs15020326]

21. Chen, Q.; Baldocchi, D.; Gong, P.; Kelly, M. Isolating individual trees in a savanna woodland using small footprint lidar data. Photogramm. Eng. Remote Sens.; 2006; 72, pp. 923-932. [DOI: https://dx.doi.org/10.14358/PERS.72.8.923]

22. Miraki, M.; Sohrabi, H.; Fatehi, P.; Kneubuehler, M. Individual tree crown delineation from high-resolution UAV images in broadleaf forest. Ecol. Inform.; 2021; 61, 101207. [DOI: https://dx.doi.org/10.1016/j.ecoinf.2020.101207]

23. Nuijten, R.J.; Coops, N.C.; Goodbody, T.R.; Pelletier, G. Examining the multi-seasonal consistency of individual tree segmentation on deciduous stands using digital aerial photogrammetry (DAP) and unmanned aerial systems (UAS). Remote Sens.; 2019; 11, 739. [DOI: https://dx.doi.org/10.3390/rs11070739]

24. Panagiotidis, D.; Abdollahnejad, A.; Surový, P.; Chiteculo, V. Determining tree height and crown diameter from high-resolution UAV imagery. Int. J. Remote Sens.; 2017; 38, pp. 2392-2410. [DOI: https://dx.doi.org/10.1080/01431161.2016.1264028]

25. Goldbergs, G.; Maier, S.W.; Levick, S.R.; Edwards, A. Efficiency of individual tree detection approaches based on light-weight and low-cost UAS imagery in Australian Savannas. Remote Sens.; 2018; 10, 161. [DOI: https://dx.doi.org/10.3390/rs10020161]

26. Yin, D.; Wang, L. How to assess the accuracy of the individual tree-based forest inventory derived from remotely sensed data: A review. Int. J. Remote Sens.; 2016; 37, pp. 4521-4553. [DOI: https://dx.doi.org/10.1080/01431161.2016.1214302]

27. The National Leading Group for Climate Change. Guidelines on Carbon Accounting and Monitoring for Forest Project: 2008. Available online: https://xueshu.baidu.com/usercenter/paper/show?paperid=f7e60bc546ba7c6be8e99b6c0aaa168d&site=xueshu_se (accessed on 1 November 2023).

28. Dalla Corte, A.P.; Rex, F.E.; Almeida, D.R.A.d.; Sanquetta, C.R.; Silva, C.A.; Moura, M.M.; Wilkinson, B.; Zambrano, A.M.A.; Cunha Neto, E.M.d.; Veras, H.F.P. et al. Measuring Individual Tree Diameter and Height Using GatorEye High-Density UAV-Lidar in an Integrated Crop-Livestock-Forest System. Remote Sens.; 2020; 12, 863. [DOI: https://dx.doi.org/10.3390/rs12050863]

29. da Cunha Neto, E.M.; Rex, F.E.; Veras, H.F.P.; Moura, M.M.; Sanquetta, C.R.; Käfer, P.S.; Sanquetta, M.N.I.; Zambrano, A.M.A.; Broadbent, E.N.; Dalla Corte, A.P. Using high-density UAV-Lidar for deriving tree height of Araucaria Angustifolia in an Urban Atlantic Rain Forest. Urban For. Urban Green.; 2021; 63, 127197. [DOI: https://dx.doi.org/10.1016/j.ufug.2021.127197]

30. Hao, Z.; Lin, L.; Post, C.J.; Jiang, Y.; Li, M.; Wei, N.; Yu, K.; Liu, J. Assessing tree height and density of a young forest using a consumer unmanned aerial vehicle (UAV). New Forest; 2021; 52, pp. 843-862. [DOI: https://dx.doi.org/10.1007/s11056-020-09827-w]

31. Wiles, G.C.; Mennett, C.R.; Jarvis, S.K.; D’Arrigo, R.D.; Wiesenberg, N.; Lawson, D.E. Tree-ring investigations into changing climatic responses of yellow-cedar, Glacier Bay, Alaska. Can. J. For. Res.; 2012; 42, pp. 814-819. [DOI: https://dx.doi.org/10.1139/x2012-028]

32. Wang, Z.; Du, A.; Xu, Y.; Zhu, W.; Zhang, J. Factors Limiting the Growth of Eucalyptus and the Characteristics of Growth and Water Use under Water and Fertilizer Management in the Dry Season of Leizhou Peninsula, China. Agronomy; 2019; 9, 590. [DOI: https://dx.doi.org/10.3390/agronomy9100590]

33. Stern, R.L.; Schaberg, P.G.; Rayback, S.A.; Hansen, C.F.; Murakami, P.F.; Hawley, G.J. Growth trends and environmental drivers of major tree species of the northern hardwood forest of eastern North America. J. For. Res.; 2023; 34, pp. 37-50. [DOI: https://dx.doi.org/10.1007/s11676-022-01553-7]

34. Mokochinski, J.B.; Mazzafera, P.; Sawaya, A.C.H.F.; Mumm, R.; de Vos, R.C.H.; Hall, R.D. Metabolic responses of Eucalyptus species to different temperature regimes. J. Integr. Plant Biol.; 2018; 60, pp. 397-411. [DOI: https://dx.doi.org/10.1111/jipb.12626] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/29247597]

35. Burkhart, H.E.; Tomé, M. Modeling Forest Trees and Stands; Springer Science & Business Media: Berlin/Heidelberg, Germany, 2012.

36. West, P.; Ratkowsky, D. Problems with models assessing influences of tree size and inter-tree competitive processes on individual tree growth: A cautionary tale. J. For. Res.; 2022; 33, pp. 565-577. [DOI: https://dx.doi.org/10.1007/s11676-021-01395-9]

37. Askne, J.I.H.; Persson, H.J.; Ulander, L.M.H. Biomass Growth from Multi-Temporal TanDEM-X Interferometric Synthetic Aperture Radar Observations of a Boreal Forest Site. Remote Sens.; 2018; 10, 603. [DOI: https://dx.doi.org/10.3390/rs10040603]

38. Neeff, T.; dos Santos, J.R. A growth model for secondary forest in Central Amazonia. For. Ecol. Manag.; 2005; 216, pp. 270-282. [DOI: https://dx.doi.org/10.1016/j.foreco.2005.05.039]

39. Matala, J.; Hynynen, J.; Miina, J.; Ojansuu, R.; Peltola, H.; Sievänen, R.; Väisänen, H.; Kellomäki, S. Comparison of a physiological model and a statistical model for prediction of growth and yield in boreal forests. Ecol. Model.; 2003; 161, pp. 95-116. [DOI: https://dx.doi.org/10.1016/S0304-3800(02)00297-1]

40. Kurz, W.A.; Stinson, G.; Rampley, G. Could increased boreal forest ecosystem productivity offset carbon losses from increased disturbances?. Philos. Trans. R. Soc. B Biol. Sci.; 2008; 363, pp. 2259-2268. [DOI: https://dx.doi.org/10.1098/rstb.2007.2198]