Analysis of Waveform Lidar Data Using Shape-Based Metrics

Models that use large-footprint waveform light detection and ranging (lidar) to estimate forest height, structure, and biomass have typically used either point data extracted from the waveforms or cumulative distributions of the waveform energy, disregarding potential information latent within the waveform shape. Shape-based metrics such as the centroid $C$ and the radius of gyration $RG$ can capture features missed by height-based metrics that are likely related to forest structure and biomass. Noise analyses demonstrated the relative insensitivity of $C$ and $RG$ , supporting the hypothesis that these metrics could be used to identify similar shapes within noisy waveforms [such as the Laser Vegetation Imaging Sensor (LVIS) and Geoscience Laser Altimeter Sensor (GLAS)] or to discriminate among waveforms with different underlying shapes. These findings suggest that $C$ and $RG$ can be successfully used in future lidar studies of forest structure and that further research should be conducted to develop additional shape-based metrics, as well as to investigate the relationship between forest structure and lidar waveform shape.