Assessing impact of geolocation error on canopy height measurement performance of GEDI

NASAs Global Ecosystem Dynamics Investigation (GEDI) is designed to provide high-resolution measurements of forest structure and topography between 52 N and S latitudes. However, insufficient horizontal geolocation accuracy may limit science applications of footprint-level products, as early adopters have found misalignments with in-situ field data and high-resolution imagery used for calibration and validation (Cal/Val) purposes. Here we developed a new means to rapidly evaluate and mitigate the impact of geolocation error on the performance of GEDIs forest height estimates over northeastern US. By integrating high-resolution airborne lidar data collected through the nationwide 3D Elevation Program, we provided optimal geolocation adjustments of GEDI at per beam level and tracked their performances over time. Our results suggest that the first release of GEDI L1-2 product can have large geolocation errors (e.g. > 25m) and its impact on canopy height measurement could drastically vary in space and time. For every 10m of GEDIs geolocation error, there is in average an increase of 0.29m in canopy height measurement error and a reduction of 0.043 in r2 in comparison to statewide 1-m leaf-off lidar. It is possible to substantially improve GEDI performance by applying per-beam level geolocation adjustments, with the biggest improvement documented over sparsely vegetated areas (r2 = 0.11 over urban area and 0.10 over other sparsely vegetated land). The second release of GEDI data has achieved a much-improved geolocation accuracy (~10m) that should be able to meet requirements from many science applications tolerant to moderate geolocation errors. Our approach has provided a short-term solution for an enhanced Cal/Val strategy for GEDI and can potentially create an alternative pathway to generate and validate biomass products by better linking GEDI footprint samples directly with in-situ data collections.