Using unmanned aerial vehicles (UAV) as platforms for light detection and ranging (LiDAR) sensors offers the efficient operation and advantages of active remote sensing; hence, UAV-LiDAR plays an important role in forest resource investigations. However, high-precision individual tree segmentation, in which the most appropriate individual tree segmentation method and the optimal algorithm parameter settings must be determined, remains highly challenging when applied to multiple forest types. This article compared the applicability of methods based on a canopy height model (CHM) and a normalized point cloud (NPC) obtained from UAV-LiDAR point cloud data. The watershed algorithm, local maximum method, point cloud-based cluster segmentation, and layer stacking were used to segment individual trees and extract the tree height parameters from nine plots of three forest types. The individual tree segmentation results were evaluated based on experimental field data, and the sensitivity of the parameter settings in the segmentation methods was analyzed. Among all plots, the overall accuracy F of individual tree segmentation was between 0.621 and 1, the average RMSE of tree height extraction was 1.175 m, and the RMSE% was 12.54%. The results indicated that compared with the CHM-based methods, the NPC-based methods exhibited better performance in individual tree segmentation; additionally, the type and complexity of a forest influence the accuracy of individual tree segmentation, and point cloud-based cluster segmentation is the preferred scheme for individual tree segmentation, while layer stacking should be used as a supplement in multilayer forests and extremely complex heterogeneous forests. This research provides important guidance for the use of UAV-LiDAR to accurately obtain forest structure parameters and perform forest resource investigations. In addition, the methods compared in this paper can be employed to extract vegetation indices, such as the canopy height, leaf area index, and vegetation coverage.

The forest canopy height (FCH) plays a critical role in forest quality evaluation and resource management. The accurate and rapid estimation and mapping of the regional forest canopy height is crucial for understanding vegetation growth processes and the internal structure of the ecosystem. A stacking algorithm consisting of multiple linear regression (MLR), support vector machine (SVM), k-nearest neighbor (kNN), and random forest (RF) was used in this paper and demonstrated optimal performance in predicting the forest canopy height by synergizing Sentinel-2 images acquired from the cloud-based computation platform Google Earth Engine (GEE) with data from ICESat-2 (Ice, Cloud, and Land Elevation Satellite-2). This research was conducted to achieve continuous mapping of the canopy height of plantations in Saihanba Mechanical Forest Plantation, which is located in Chengde City, northern Hebei province, China. The results show that stacking achieved the best prediction accuracy for the forest canopy height, with an R² of 0.77 and a root mean square error (RMSE) of 1.96 m. Compared with MLR, SVM, kNN, and RF, the RMSE obtained by stacking was reduced by 25.2%, 24.9%, 22.8%, and 18.7%, respectively. Since Sentinel-2 images and ICESat-2 data are publicly available, this opens the door for the accurate mapping of the continuous distribution of the forest canopy height globally in the future.

Forest growing stem volume (GSV) reflects the richness of forest resources as well as the quality of forest ecosystems. Remote sensing technology enables robust and efficient GSV estimation as it greatly reduces the survey time and cost while facilitating periodic monitoring. Given its red edge bands and a short revisit time period, Sentinel-2 images were selected for the GSV estimation in Wangyedian forest farm, Inner Mongolia, China. The variable combination was shown to significantly affect the accuracy of the estimation model. After extracting spectral variables, texture features, and topographic factors, a stepwise random forest (SRF) method was proposed to select variable combinations and establish random forest regressions (RFR) for GSV estimation. The linear stepwise regression (LSR), Boruta, Variable Selection Using Random Forests (VSURF), and random forest (RF) methods were then used as references for comparison with the proposed SRF for selection of predictors and GSV estimation. Combined with the observed GSV data and the Sentinel-2 images, the distributions of GSV were generated by the RFR models with the variable combinations determined by the LSR, RF, Boruta, VSURF, and SRF. The results show that the texture features of Sentinel-2’s red edge bands can significantly improve the accuracy of GSV estimation. The SRF method can effectively select the optimal variable combination, and the SRF-based model results in the highest estimation accuracy with the decreases of relative root mean square error by 16.4%, 14.4%, 16.3%, and 10.6% compared with those from the LSR-, RF-, Boruta-, and VSURF-based models, respectively. The GSV distribution generated by the SRF-based model matched that of the field observations well. The results of this study are expected to provide a reference for GSV estimation of coniferous plantations.

Detecting and segmenting individual trees in forest ecosystems with high-density and overlapping crowns often results in bias due to the limitations of the commonly used canopy height model (CHM). To address such limitations, this paper proposes a new method to segment individual trees and extract tree structural parameters. The method involves the following key steps: (1) unmanned aerial vehicle (UAV)-scanned, high-density laser point clouds were classified, and a vegetation point cloud density model (VPCDM) was established by analyzing the spatial density distribution of the classified vegetation point cloud in the plane projection; and (2) a local maximum algorithm with an optimal window size was used to detect tree seed points and to extract tree heights, and an improved watershed algorithm was used to extract the tree crowns. The proposed method was tested at three sites with different canopy coverage rates in a pine-dominated forest in northern China. The results showed that (1) the kappa coefficient between the proposed VPCDM and the commonly used CHM was 0.79, indicating that performance of the VPCDM is comparable to that of the CHM; (2) the local maximum algorithm with the optimal window size could be used to segment individual trees and obtain optimal single-tree segmentation accuracy and detection rate results; and (3) compared with the original watershed algorithm, the improved watershed algorithm significantly increased the accuracy of canopy area extraction. In conclusion, the proposed VPCDM may provide an innovative data segmentation model for light detection and ranging (LiDAR)-based high-density point clouds and enhance the accuracy of parameter extraction.

Due to the cold climate and dramatically undulating altitude, the identification of dynamic vegetation trends and main drivers is essential to maintain the ecological balance in Tibet. The normalized difference vegetation index (NDVI), as the most commonly used greenness index, can effectively evaluate vegetation health and spatial patterns. MODIS-NDVI (Moderate-resolution Imaging Spectroradiometer-NDVI) data for Tibet from 2001 to 2020 were obtained and preprocessed on the Google Earth Engine (GEE) cloud platform. The Theil–Sen median method and Mann–Kendall test method were employed to investigate dynamic NDVI changes, and the Hurst exponent was used to predict future vegetation trends. In addition, the main drivers of NDVI changes were analyzed. The results indicated that (1) the vegetation NDVI in Tibet significantly increased from 2001 to 2020, and the annual average NDVI value fluctuated between 0.31 and 0.34 at an increase rate of 0.0007 year⁻¹; (2) the vegetation improvement area accounted for the largest share of the study area at 56.6%, followed by stable unchanged and degraded areas, with proportions of 27.5 and 15.9%, respectively. The overall variation coefficient of the NDVI in Tibet was low, with a mean value of 0.13; (3) The mean value of the Hurst exponent was 0.53, and the area of continuously improving regions accounted for 41.2% of the study area, indicating that the vegetation change trend was continuous in most areas; (4) The NDVI in Tibet indicated a high degree of spatial agglomeration. However, there existed obvious differences in the spatial distribution of NDVI aggregation areas, and the aggregation types mainly included the high-high and low-low types; and (5) Precipitation and population growth significantly contributed to vegetation cover improvement in western Tibet. In addition, the use of the GEE to obtain remote sensing data combined with time-series data analysis provides the potential to quickly obtain large-scale vegetation change trends.