Comprehensive LiDAR simulation with efficient physically-based DART-Lux model (II): Validation with GEDI and ICESat-2 measurements at natural and urban landscapes

Abstract

LiDAR is a developed technology that has been widely used to measure the Earth's surface by acquiring accurate three-dimensional (3D) information. DART (Discrete Anisotropic Radiative Transfer) model developed a new LiDAR modeling method based on the Monte Carlo bidirectional path tracing mode named DART-Lux. Using the DART-RC (Ray Carlo) mode as a reference, DART-Lux shows consistency and efficiency for LiDAR signal modeling. This paper presents a further validation of DART-Lux LiDAR model for simulating actual LiDAR waveform and photon-counting measurements by considering two in-orbit spaceborne LiDAR systems: GEDI (Global Ecosystem Dynamics Investigation) and ICESat-2 (Ice, Cloud, and land Elevation Satellite-2). The validation experiments are conducted on accurate 3D descriptions of an urban landscape in Toulouse, France, and a natural forest landscape in Saihanba, China. The pulse-by-pulse comparisons of GEDI and simulated waveforms yield mean R² = 0.893, mean RMSE = 0.077. The simulated ICESat-2 photon counting shows accuracies of signal photon frequency (R² = 0.950, RMSE = 0.465 pts./pulse) and noise photon frequency (R² = 0.820, RMSE = 0.247 pts./pulse). Results in GEDI and ICESat-2 overlapping footprints illustrate the usefulness of DART-Lux for studying their height retrieval inconsistency. Furthermore, sensitivity studies conducted with DART-Lux reveal performance and limitation of GEDI and ICESat-2 in height measurement. This study confirms the accuracy of DART-Lux for simulating actual LiDAR signals and provides valuable insights for exploitation of GEDI and ICESat-2.