A linearly approximated iterative Gaussian decomposition method for waveform LiDAR processing

摘要: Abstract Full-waveform LiDAR (FWL) decomposition results often act as the basis for key LiDAR-derived products, example canopy height, biomass and carbon pool estimation, leaf area index calculation under detection. To date, prevailing method FWL product creation is Gaussian Decomposition (GD) based on a non-linear Levenberg-Marquardt (LM) optimization node parameter estimation. GD follows “greedy” approach that may leave weak nodes undetected, merge multiple into one or separate noisy single ones. In this manuscript, we propose an alternative called Linearly Approximated Iterative (LAIGD method). The novelty of LAIGD it multi-step “slow-and-steady” iterative structure, where new are quickly discovered adjusted using linear fitting technique before they forwarded optimization. Two experiments were conducted, real full-waveform data from NASA’s land, vegetation, ice sensor (LVIS) another synthetic containing different number degrees overlap to assess performance in variable signal complexity. LVIS revealed considerable improvements RMSE (44.8% lower), RSE (56.3% lower) rRMSE (74.3% values compared benchmark method. These further confirmed with data. Furthermore, proposed reduces execution times half, important consideration there plans global coverage upcoming Global Ecosystem Dynamics Investigation International Space Station.