Forest structural characteristics and their changes have a strong relation with stand complexity which can be altered by various silvicultural activities and natural disturbances. Capturing forest stand complexity is limited into simple forest characteristics that can be measured with conventional means. However, stand complexity could be estimated using airborne laser scanning (ALS) data with increased scalability and novel characteristics.

Therefore, this study investigates how low density ALS point clouds can be utilized to estimate forest structural diversity and how stand complexity develops over time in different forest types. Study was conducted in Evo (southern Finland) where the ALS data was collected in 2012 and 2019 covering 34 sample plots with size of 32x32m with point density < 1 point/m² in both data acquisitions. Point clouds were analysed in 1 m height horizontal layers creating a canopy height model (CHM) with 5m resolution. CHM was then reclassified in four categories based on photosynthetically active radiation - open gap, euphotic zone, oligophotic zone and closed gap pixels. Light availability categories were used to create Canopy Volume Profiles and the effect of different forest types was evaluated by dividing the plots into various high diversity levels (Gini coefficient) and species diversity values (Shannon index) based on field inventory data.

Results show that plots with larger height diversity and species richness have more open gaps, indicating more mixed and variating stand structures with larger canopy surface. However, open gap proportion considerably reduces over time displaying the process when forest stands without human interventions are getting more structurally homogeneous. Euphotic light zone proportion increased over time in all forest diversity types proving that the upper canopy part is getting more vegetated with larger and denser branches which results in larger number of laser returns.

Riparian systems are the interface between aquatic and terrestrial environments, and riparian vegetation includes all vegetation directly adjacent to a stream network. In the Pacific Northwest of North America, riparian vegetation is critical for salmon habitat protection. Airborne laser scanning (ALS) is able to characterize the three-dimensional attributes of forest structure at the watershed level. We hypothesize that ALS can determine structural attributes unique to riparian and upland forests, which is an important aspect of riparian conservation.

In this study, ALS data were acquired using a Reigl Q1560 dual-channel LiDAR system for two watersheds on Vancouver Island, Canada, with a point density of 25 points per m². ALS data were used to derive 15 metrics describing vegetation height, complexity, and canopy structure (e.g. mean height, rumple index, canopy cover, etcetera) at a 20 m spatial resolution. The 15 selected ALS metrics were sampled at 45 paired plots located in riparian and upland stands, and were tested for significant differences (p<0.05) using a Wilcoxon test.

Significant differences between ALS metrics at riparian and upland plots were found in over half of the ALS metrics tested, however, these metrics varied between the two watersheds. Euphotic canopy volume, canopy cover, rumple index, and four height metrics (standard deviation of height and 20^th, 50^th, and 95^th percentile) were metrics that significantly described riparian areas in both watersheds. Euphotic canopy volume, which describes filled voxels in the upper 65% of the canopy, was determined to be the most important variable for delineating riparian areas.

ALS is a useful tool in characterizing the structure of riparian vegetation and estimating its location at the watershed level. As the transition between riparian and upland forest is difficult to detect using traditional field methods, these results suggest ALS data could be beneficial in delineating areas of riparian vegetation.

The recent explosion in availability of high resolution remote sensing technologies and, crucially, the tools to analyse the 3D data they produce is leading to substantial interest in using them for large scale ecological forest monitoring. The level of detail contained in the entire 3D shape of trees, fully captured within these data, can generate a wide range of metrics of interest to ecologists, but the potential of different metrics to bring new ecological insight has not been fully explored.

Working across a range of European forest ecosystems, we have constructed a unique 3D dataset of European forest structural properties from passive and active sensors. We have used AI approaches to segment individual trees for ecological analysis (here presented by Owen et al.). In this presentation, we show how the three dimensional structure of individual trees and whole plots varies across mixed European forests sampled across Mediterranean, temperate and boreal ecosystems. We discuss the value of this approach for improving our fundamental ecological understanding of forest functioning.

Forest is an integral component of natural ecosystems that humans and living organisms depend on. Thus, understanding the forest health status is very important in forest ecology, for forest maintenance and management, which aids in a healthy regulated environment. Therefore, quantifying and evaluating deadwood is critical for forest health assessment while equally serving as indicators for biodiversity. Deep learning (DL) has achieved unprecedented accuracy in automated forest measurement. However, LiDAR-based DL applications in forest health, especially in mapping deadwood, are still in early ages despite the established significance of LiDAR in multi-sensor fusion for training DL models. This study fuses Airborne Laser Scanning (ALS) point cloud and aerial CIR imagery to classify single coniferous trees into five decay levels while mutually evaluating two DL techniques: a 3D point cloud-based PointNet and 2D projected point clouds based on CNN. Initially, the ALS point cloud was fused with the CIR imagery to extract (NIR, R, G) colors for the point cloud. Then, a semi-automatic single-tree segmentation approach was used for individual tree segmentation. These 3D-based individual trees were used as the dataset source for training in the PointNet algorithm. The 3D points cloud-based individual trees were further projected in 2D four orthogonal planes displaying each tree in four side views, which served as datasets for training the CNN algorithm. Mutual evaluation of the trained models showed promising results for both algorithms, although the 2D-based CNN (OA = 90.9%) outperformed the 3D-based PointNet (OA = 80.6%). These findings further reveal the relevance of the spectral image texture in classifying tree decay levels. Therefore, our models can be considered for sustainable automatic determination of tree decay levels and deadwood assessment from LiDAR data coupled with DL, especially in large-scale forestry. The proposed method can serve as a rigorous tool for monitoring biodiversity and ecology.

Entropy, a measure of disorder or randomness, has been proposed as a valuable metric for evaluating tropical forests' complexity and diversity. Specifically, entropy can be used to quantify the spatial distribution and heterogeneity of vegetation, which is essential for understanding the functional roles of tropical forests in carbon sequestration, water regulation, and habitat provision. In recent years, Light Detection and Ranging (LiDAR) technology has emerged as a powerful tool for quantifying the three-dimensional structure of tropical forests, enabling the study of their ecological processes and linkages to entropy.

The main objectives of this study are 1) To present a new entropy indicator based on the Gini coefficient and Lorenz curves and 2) To evaluate this new entropy indicator, the Lorenz-entropy index, as a function of ecological successional. This research was conducted in a tropical dry forest under different levels of ecological succession in Costa Rica and two rainforest ecosystems: one in Costa Rica and another in Panama that are in ecological climax.

To evaluate these objectives, we used waveform-derived canopy height from LVIS to determine each study area's Gini coefficient from the Lorenz curve. Our results demonstrate that as structural complexity increases, there is continual growth of the Gini coefficient to its maximum entropy of 0.33. At the same time, the Lorenz-entropy index reaches its peak. This resulted in a positive correlation between structural complexity and Lorenz entropy, making it evident that entropy and complexity will increase as successional stages progress from early to climax.

Our findings highlight the potential of LiDAR to improve our understanding of the relationship between forest structure and entropy in tropical forests. Specifically, LiDAR can identify areas of high entropy or complexity, which may be necessary for maintaining biodiversity and supporting ecosystem services such as carbon storage, nutrient cycling, and productivity.