Canopy height extraction over mountainous areas from GEDI lidar deconvoluted waveforms

The extraction of canopy heights from spaceborne lidar received waveforms over mountainous areas is a challenging task, as the waveform signal broadening and overlapping effects make the vegetation and ground returns difficult to identify. This study aims to obtain the deconvoluted waveform from the received waveform using a Richardson–Lucy deconvolution algorithm for better separation of the vegetation and ground returns from overlapped waveform components. Specifically, the canopy top elevation is extracted from the start bin of the deconvoluted waveform and the ground elevation is calculated as the elevation centroid of the ground return, which is determined by shuttle radar topography mission (SRTM) data and deconvoluted waveform. The canopy height is the difference between the canopy top and the ground elevations. Eight track data of the GEDI lidar over the rugged Larkspur Mountain of Colorado are employed to validate the proposed method. The results demonstrate that the mean error, mean absolute error, and root mean square error of derived canopy heights over rugged terrain with surface slopes more than 20° are greatly reduced by 86.7%, 23.2%, and 19.5% relative to GEDI canopy height products. The results prove that the proposed method is more applicable to derive canopy heights over rugged mountainous areas.