Bringing All Lidar Data Together: Investigations of Spatially Coincident Terrestrial, Airborne, and Satellite Lidar Data for Deriving Vegetation Structure Metrics

Abstract

The overall goal of this study was to compare forest structure metrics obtained by processing ICESat waveform data and spatially coincident discrete-return airborne lidar and ground-based laser scanner data over varied terrain conditions covered by forests. For mostly flat-terrain conditions, we investigated lidar data over forests in east Texas, which are characteristics for most of the south-eastern United States, thus allowing terrain to be largely excluded as a source of error. For sloped-terrain, we used lidar data over forests in Oregon. In addition to forest height measurements, we investigated biomass estimates derived from waveform height metrics and also compared terrain elevation measurements and canopy parameters. Specific objectives were to: (1) compare terrain elevations and canopy height parameters derived from ICESat and airborne lidar; (2) compare vertical forest structure metrics and above ground biomass estimates derived with data from sensors on all three platforms, ground, airborne, and satellite, with reference data collected by field measurements; and (3) discuss advantages and limitations of lidar data on all three platforms for characterizing forest vegetation. Over flat terrain, results indicated a very strong correlation for terrain elevations between ICESat and airborne lidar, with R-square values of 0.98 and sub-meter RMSE. ICESat height variables were able to explain 80% of the variance associated with the reference biomass derived from airborne lidar, with an RMSE of 37.7 Mg/ha. Most of the models for height metrics had R-square values above 0.9. Results from ongoing investigations for sloped-terrain are expected to establish practical procedures for improving analysis of waveform data in vegetation studies.