3D vegetation structure extraction from lidar remote sensing

Abstract

Large footprint (15m-25m diameter) lidar records the full lidar energy return (lidar waveform), when laser energy penetrates into vegetation canopy. The full lidar waveforms are directly linked with the three-dimensional characterization of vegetation structure. Recently studies on vegetation structure parameter retrievals from large foot-print lidar found direct relationships between vegetation structure parameters such as tree height, stem diameter, above ground biomass and full lidar waveforms. But these studies are mainly limited to empirical studies and these relationships vary for different sites. To better understand the link between large foot print lidar waveforms and vegetation structure parameters, we applied a vegetation Geometric-Optical and Radiative Transfer (GORT) model to simulate vegetation lidar waveforms with 3-D vegetation structure parameters as inputs. We evaluated the performance of the GORT model in conifer forests using the data collected by Scanning Lidar Imager of Canopies by Echo Recovery (SLICER) and found that GORT simulates large-footprint vegetation lidar waveforms well. To better retrieve 3-D vegetation structure parameters, we investigate the sensitivity of waveforms to GORT input parameters. Our analysis shows that lidar waveforms is most sensitive to the tree density, then to the foliage density and the least to the tree size. A stochastic inversion method will be implemented for inversion.