Testing Land-Vegetation retrieval algorithms for the ICESat-2 mission

The upcoming spaceborne satellite, the Ice, Cloud and land Elevation Satellite 2 (ICESat-2), will provide topography and canopy profiles at the global scale using photon counting LiDAR. To prepare for the mission launch, the aim of this research is to develop a framework for retrieving ground and canopy height in different forest types and noise levels using two ICESat-2 testbed sensor data: MABEL (Multiple Altimeter Beam Experimental Lidar) data and simulated ICESat-2 data. The first step of the framework is to reduce as many noise photons as possible through grid statistical methods and cluster analysis. Subsequently, we employed the overlapping moving windows and estimated quantile heights in each window to characterize the possible ground and canopy top using the filtered photons. Both MABEL and simulated ICESat-2 data generated satisfactory results with reasonable accuracy, while the results of simulated ICESat-2 data were better than that of MABEL data with smaller root mean square errors (RMSEs). For example, the RMSEs of canopy top identification in various vegetation using simulated ICESat-2 data were less than 3.78 m comparing to 6.48 m for the MABE data. It is anticipated that the methodology will advance data processing of the ICESat-2 mission and expand potential applications of ICESat-2 data once available such as mapping vegetation canopy heights.