Modeling the Estimation Uncertainty of Terrain Matching Methods for Spaceborne Laser Altimetry Data

ICESat-2 carrying the first photon-counting laser altimeter shows unprecedented high-density observations on Earth surfaces and has been widely used in land mapping, forestry and biomass survey, sea-ice and polar monitoring, and marine environmental studies. In the geometry-based applications of ICESat-2, the terrain matching, which registers the laser altimetry data with other sources of topographic data, serves as a convenient and effective approach for accuracy verification or multiple data registration. However, how reliable the terrain matching results are and what factors affect their reliability have not been addressed by current studies. This study derives a theoretical estimation uncertainty model to link the terrain matching performance with the main influence factors, which include the reliefs, accuracies, and resolutions of the reference topography data, the random error magnitudes of laser altimetry data, and the amounts of laser footprints. The proposed model is quantitatively verified by bootstrapping experiments using 51 ICESat-2 tracks in six study sites. Across these tracks, the mean confidence levels whose matching errors fall within two and three times of the model-predicted uncertainties are 80.10%/83.20%/71.82% and 92.61%/93.95%/87.06% for x/y/z directions, indicating high reliability. This study can work as a guidance when implementing a terrain matching in somewhere with specific lidar data, providing quantitative performance labels for the terrain matching results.