Integrating Unmanned aerial vehicle Laser Scanning (ULS) and Terrestrial Laser Scanning (TLS) data in complex forest environments remains a significant challenge. Despite the availability of numerous registration algorithms, robust comparative studies are limited by the lack of reliable multi-platform benchmark datasets. In this study, we introduce the first multiplatform benchmark dataset for ULS-TLS point cloud registration in forests, encompassing 17 plots from seven diverse regions with about 1.56 billion points. The dataset is categorized into three difficulty levels based on overlap ratio and rigid overlap.We evaluated the performance of five registration algorithms against this benchmark. Chen2022 achieved the highest accuracy with a 100% success rate across all difficulty levels. While Wu2024 demonstrated robust performance in lower difficulties but faced challenges in more complex scenarios. We also found that terrain variations and rigid overlap significantly impacted registration accuracy, particularly for algorithms reliant on individual tree positions such as Hyypp¨a2021 and Feng2024. These findings underscore the need for improved data collection strategy, ground filtering techniques, and feature matching algorithms to enhance performance in challenging environments. We present the first openly accessible multi-platform benchmark dataset for forested regions and anticipate that future research will expand this work to additional areas. The dataset can be downloaded from: DatasetDownloadLink.

Accurate acquisition of tree species and forest age information is crucial for forest ecosystem protection and sustainable development. Although such information can be obtained through remote sensing technology, accurate extraction of tree species and forest age still faces significant challenges due to the remote sensing data characteristics and the complex and variable mountainous natural geographical environment. This article uses the Landsat-NDVI long-term series data from 1991 to 2021 to address this problem. A growth characteristic change model is established for the main dominant tree species in Shangri-La city, including Pinus yunnanensis, Pinus densata Mast, Picea & Abies, and Quercus acutissima, throughout the growth cycle in vertical zones at different altitudes. Besides, the study compared the normalized difference vegetation index (NDVI) time-series data of the area to be identified with the growth characteristic model to match the appropriate growth range automatically, and their similarity is determined to complete tree species classification. The forest age is then obtained based on the optimal matching point of the NDVI time-series data in the growth model. This method can simultaneously obtain tree species and forest age information and acquire forest age when the image duration exceeds the remote sensing image record. Ultimately, the overall accuracy of tree species classification reached 83.42%, and the forest age fitting results also showed a high degree of correlation, with a coefficient of determination ($R^{2}$) of 0.92. The root mean square error (RMSE) of forest age in Shangri-La, dominantly covered by mature forests/overmature forests, has dropped to 10.59. This study significantly improves remote sensing technology’s accuracy and application efficiency in multiparameter inversion of forest resources while providing a new technical means for forest resource management and ecological protection.

As the most complex community structure in terrestrial ecosystems, tropical rainforests still employ traditional manual surveys to obtain spatial information on trees, which are time-consuming and laborious with significant errors. Light detection and ranging (LiDAR) can provide high-quality 3-D point cloud data, but using it to extract structural information of trees and shrubs in complex forests remains a technical challenge for point cloud mapping. Therefore, this article proposes a solution for realizing the accurate separation of LiDAR point cloud data for trees and shrubs within complex forests such as tropical rainforests. The method first preprocesses the data to obtain understory point cloud data. The local curvature (L-curvature) features are then utilized to perform preliminary separation. Then, the growth segmentation is performed based on the normal vector and the curvature magnitude to obtain the clustered objects. Accordingly, trees and shrubs are identified according to the features of the segmented clustered objects. Finally, eight tropical rainforest plots were selected to evaluate and analyze the performance of the method. The research results indicate that the accuracy of tree extraction in tropical rainforests using this method can reach over 91%. The accuracy and efficiency of this method for tree-shrub separation are superior to other methods. This study provides essential support for investigating forest understory vegetation characteristics and the spatial growth distribution of tropical rainforest trees and shrubs. It lays a foundation for the subsequent application and promotion of terrestrial LiDAR when investigating vegetation information in complex forest scenarios.

Ice, cloud, and land elevation satellite (ICESat-2)/Advanced Topographic Laser Altimeter System (ATLAS) multibeam micropulse photoncounting light detection and ranging (LiDAR) can be effectively applied to extract forest canopy height. However, the ICESat-2/ATLAS photon point cloud interfered with the signal-to-noise ratio (SNR), fraction vegetation coverage (FVC), and terrain slope. The main challenge of this research is to extract high-precision canopy heights. Therefore, this article improves the canopy height extraction method based on the ICESat-2/ATL08 theoretical algorithm. First, an adaptive filter, Threshold Segmentation based on Spatial Clustering and Bimodal Reconstruction (TS-SCABR), is proposed, which can adapt to different SNR scenarios. Then, combined with the gradient method, the discontinuous data are detrended in sections to eliminate the edge mutation problem of the detrended data. Based on the detrended data, the iterative filtering algorithm of the local terrain is employed to fit the ground curve, and the mutation detection and empirical mode decomposition (EMD)-digital smoothing polynomial (DISPO) filtering remove the pseudoground photons to identify the data of ground and nonground photons accurately. Finally, the percentile statistics method is utilized to extract the canopy-top photons from the nonground photons according to their elevation difference. The results indicate that, under different natural conditions, the improved algorithm has better adaptability than the previous algorithm. Compared with the original ATL08 ATBD algorithm, the canopy height accuracy is significantly improved, especially in low FVC and high slope scenarios. When the FVC is lower than 25%,$R_{2}$increases by 50.3%, and the root mean square error (RMSE) is reduced by 2.175 m, and when the slope is higher than 45°, it increases by 41.7%, and the RMSE is reduced by 2.159 m. Therefore, the algorithm has apparent advantages in inverting the canopy height in a mountainous environment with lush forests.

Land cover classification is important for effectively protecting and developing land resources. This study investigates the joint use of the Ice, Cloud, and land Elevation Satellite-2 (ICESat-2) data and Landsat 8 Operational Land Imager (OLI) data in land cover classification with random forest (RF) in Yunnan province, China, to explore the application potential of photon counting Lidar data in land cover classification. The contributions of this paper are: (1) The joint use of ICESat-2 and Landsat 8 image datasets can provide better land cover classification accuracy, achieving 10% and 3% accuracy gains for five types(forest/low-vegetation/water/construction land/barren) and four types (vegetation/water/construction-land/barren)of land cover, respectively. (2) The proposed feature selection improves the overall accuracy by 1.5% and 1% for five and four land cover types, respectively. (3) The accuracy of the land cover classification reached 82% and 98% for five and four types of land cover. (4) The terrain factors, the number of canopy photons, and solar conditions significantly impact land cover classification for a complex terrain area.