The Global Ecosystem Dynamics Investigation (GEDI) is the first spaceborne LiDAR mission designed to map Earth’s tropical and temperate forests. It samples LiDAR waveforms of approximately 25m diameter footprints over the Earth’s surface. When registered forests, GEDI waveforms display the height profile to an area equivalent to a forest inventory plot - i.e. the allocation of biomass per height stratum between ground and top of canopy. The high vertical resolution of GEDI allows it to measure features never before possible from space at a global scale, with several data products related to forest structure being released. Recent studies used high resolution airborne and terrestrial LiDAR to measure forest structural complexity (or structural heterogeneity) through some custom complexity index, which is essentially a single metric that measures the occupation and variability of structural components in 3D space. A structural complexity product using GEDI data is also under development by the mission’s team. In this work we set the foundations for the first versions of this new forest structural complexity product. We used airborne laser scanning (ALS) point clouds distributed across the world with known GEDI overlaps and calculated a 3D canopy entropy index (complexity) from point cloud cuts with known GEDI shot intersections with high data quality. We then modeled the measured complexity using only GEDI metrics from the overlapping waveforms. We measured the performance of GEDI for predicting both horizontal and vertical components of the 3D canopy entropy and trained domain specific models based on the continental region and plant functional types of the training data. GEDI was able to predict the 3D canopy entropy precisely and spatially consistently across geographical domains, with high correlation to other structural metrics derived from the waveform, such as foliage height diversity and above ground biomass.

Airborne laser scanning has emerged as an effective technology for describing forest structural complexity, including the measurement of tree height inequality, leaf area and biomass. These measures have improved description and conceptual understanding of forest habitats, and their implications on all taxa resident in forest ecosystems. However, measurement of forest structural complexity evaluated in the second dimension, that is, in-homogeneity in any measure horizontally across space, has received little attention despite its ecological importance.

By deriving a new measure, δLAI, we develop a method for measuring horizontal forest structural complexity, and specifically the spatial in-homogeneity of basal area. To achieve this, we model lidar-measured leaf area index (LAI) with the generalised beta distribution, accounting for censorship due to maximum point densities and describing a wide array of distribution shapes, including multi-modal (U) distributions. The derivation of δ, a measure of dispersion for the standard beta distribution, is extended to variables constrained between any two real numbers. We describe the maximum likelihood estimate of the generalised δ for empirical data, and apply the method to lidar-measured LAI. We then validate the utility of δLAI with manually collected mapped forest data and spatial point pattern statistics.

Using empirical data, we find the LAI closely matches the theoretical generalised beta distribution, and find typically used summary statistics are biased by the mean matching theoretical functions. We demonstrate δLAI as an unbiased alternative, and find close correlations with in-homogeneity of first-order marked point pattern statistics calculated from manually mapped forests, as well as differing mean δLAI for clustered, random and regular spaced forests, as measured by second-order point pattern statistics.

These findings demonstrate δLAI as an effective measure of horizontal forest structural complexity, with the potential to explain habitat-organism interactions of forest dwelling taxa, disturbance mechanics from wind-throw and fire, and forest micro-climatic mediation.

Across ecology and meteorology, measurement of stand-level Plant Structural Traits (PSTs) is key to understanding forests at global scale. Here, we investigate what is required for PSTs such as canopy cover and foliage profiles to be derived from ICESat-2 data. To do so, we test the ability of an ICESat-2 simulator, based on pre-launch ICESat-2 simulations (Neuenschwander and Magruder, 2016) and the GEDI simulator (Hancock et al, 2019), to replicate PST measurements retrieved from observed ICESat-2 data. Airborne Laser Scanning data are used to identify the vertical profile of vegetation within ICESat-2 footprints, from which photons are Poisson-sampled. Canopy height, canopy cover, Relative Height metrics and Plant Area Volume Density profiles are calculated from both simulated and observed ICESat-2 data. PSTs are derived from simulated and observed ICESat-2 tracks at a range of sites and forest types. We find that simulating similar stand-level PFTs to those calculated from observed ICESat-2 data requires accurate classification of canopy and ground photons, a well-constrained canopy:ground reflectance ratio, and removal of noise prior to PFT calculation. This research suggests that, with global mapping of leaf angle distribution and canopy and ground reflectances, alongside correct ground classification, ICESat-2 data could be used to provide global canopy cover and foliage profile measurements.

Forest edges represent the transition zone between the forest interior and the open countryside and provide several ecological functions. They contribute to biodiversity as habitats for plant and animal species, regulate fluxes of nutrients and pollutants between surrounding agricultural areas and the forest, or regulate the microclimate. To ensure maintenance of these edge functions, a reliable and frequent assessment and monitoring of their conditions is required, particularly for degraded edges which should be restored.
Regular, nationwide, publicly available airborne laser scanning (ALS) acquisitions (15-20 points/m²) as scheduled for Switzerland in the future are ideal to monitor forest edge conservation and renaturation efforts. To this means, we provided a map of forest edge structure characteristics for entire Switzerland (total forest edge length ~187,000 km). We derived canopy height variability, presence or absence of the shrub layer and forest edge slope from the latest nationwide ALS acquisition, thus information about the horizontal and vertical structure in the forest edge zone. These ALS structure metrics were chosen such that they are as close as possible to the definition of the forest edge metrics which are collected as part of the Swiss National Forest Inventory (NFI). Additionally, we derived light regime information and the horizontal visibility into the forest interior which are two metrics that go beyond standard parameters.
However, acquisition characteristics of different ALS flight campaigns differ regarding flight plans, time of acquisition, and sensor specifications. Variations in these acquisition characteristics result in differences in the point clouds (point densities and height distribution). Therefore, we put a particular emphasis on the evaluation of the robustness of the derived ALS metrics. The goal of our study is to verify that the detected differences in the ALS structure metrics reflect real structure changes in the forest edge instead of different acquisition characteristics.

Understory vegetation plays a vital role in forest ecosystem structure and functioning and provision of ecological services. Image-based retrieval of understory vegetation structural variables (e.g., fCover and leaf area index (LAI)) is extremely difficult due to the occlusion effect of woody plants on the understory elements. Airborne small-footprint LiDAR technology offers a 3D insight into forest structure due to pulse penetration. However, the discrete-return point cloud has quite a limited ability to characterize the understory vegetation given its very limited number of pulse returns. Full-waveform LiDAR enables the continuous characterization of forest vertical profile and provides radiometric intensity information controlled by vegetation and soil backscattering coefficients, facilitating the radiometric-based retrieval of structural variables. In our study, we proposed an energy dimidiate model (SDM) that assumes the lower-layer mixture energy is composed of pure understory vegetation (PUV) energy and pure soil (PS) energy. Understory fCover can be retrieved based on the SDM and radiative transfer equation, and the understory LAI can be retrieved based on the gap probability model and the complementary of fCover In the implementation, we obtained the PS energy by a half-Gaussian fitting of the half of PS waveform. Then we calculated the product (J₀*ρ_u) of emitted pulse energy (J₀) and understory vegetation backscattering coefficient (ρ_u) by fitting a linear relationship between PUV energy and PS energy within a local area. Finally, with a hypothesis of Gaussian random variable of J₀*ρ_u, we retrieved the fCover by a Maximum Posterior Probability Estimation method under the Bayesian framework. Our algorithm was tested based on a simulation dataset and a field aerial LiDAR dataset acquired in a boreal forest ecosystem. Results showed that both fCover and LAI estimates are more consistent with the reference values compared to the previously-reported point-cloud -based method.