Exploring the Potential of GEDI in Characterizing Tree Height Composition Based on Advanced Radiative Transfer Model Simulations

Tree height composition describes the relative abundance of trees in different height levels and performs as a critical characteristic for community ecology. The recent launched full-waveform spaceborne LiDAR (Light Detection and Ranging), i.e., Global Ecosystem Dynamics Investigation (GEDI), can map canopy height, but whether this observation reflects tree height composition remains untested. In this study, we firstly conduct numerical simulations to explore to what extent tree height composition can be obtained from GEDI waveform signals. We simulate waveforms for diverse forest scenarios using GEDI simulator coupled with LESS (LargE-Scale remote sensing data and image Simulation), a state-of-the-art radiative transfer model. We devise a minimalistic model, Tree generation based on Asymmetric generalized Gaussian (TAG), for customizing tree objects to accelerate forest scene creation. The results demonstrate that tree objects generated by TAG perform similarly in LiDAR simulation with objects from commercial 3-dimensional software. Results of simulated GEDI waveforms reasonably respond to the variation of crown architectures in even-aged forests. GEDI waveforms have an acceptable ability to identify different height layers within multi-layer forests, except for fir forests with a cone-shaped crown. The shape metric of waveforms reflects the height of each layer, while retrieval accuracy decreases with the increases in height variations within each layer. A 5-m interval between layers is the minimum requirement so that the different height layers can be separated. A mixture of different tree species reduces the retrieval accuracy of tree height layers. We also utilize real GEDI observations to retrieve tree heights in multi-height-layer forests. The findings indicate that GEDI waveforms are also efficient in identifying tree height composition in practical forest scenarios. Overall, results from this study demonstrate that GEDI waveforms can reflect the height composition within typical forest stands.