Evaluating GEDI Geolocation Uncertainty against Airborne LiDAR: Implications for Footprint-level Analyses

Abstract:

The NASA Global Ecosystem Dynamics Investigation (GEDI) instrument, launched to the International Space Station in December 2018, has since acquired >3 billion measurements of forest structure. The GEDI mission collated a large global database of airborne LiDAR (ALS) acquisitions for simulation of GEDI waveforms coincident with in-situ observations and for pre-launch calibration of biomass algorithms. Here we implement an approach for colocation of these ALS data with on-orbit GEDI waveforms for post-launch calibration and validation of GEDI Level 1 and Level 2 products in the presence of geolocation error. Characterizing GEDI’s geolocation uncertainty is also crucial for key science applications, such as comparing GEDI with past ALS and field datasets to detect forest change or, more generally, integrating GEDI with other spatial datasets to address ecological questions. We independently assess GEDI geolocation error against the global ALS database, which encompasses 212 sites across six continents. We verify these results against equivalent GEDI geolocation error estimates derived from the full waveform NASA Land, Vegetation, Ice Sensor (LVIS). Using the colocated GEDI and ALS observations, we also evaluate the accuracy of GEDI height metrics and relate errors to variation in canopy cover, terrain slope, and plant functional type, highlighting implications for validation of GEDI footprint products and integrating footprint-level GEDI observations with other datasets.