Assessing the Utility of Green LiDAR for Characterizing Forest Canopy Structure and Stream Bathymetry in Riparian Zones.

Abstract:

Forested riparian zones serve many ecosystem functions from species habitat through stream shading and large woody debris recruitment, to improvements in water quality. Moreover, stream depth and bathymetry in forested environments is difficult and costly to measure in the field, but critically important for stream-dwelling organisms. Green (bathymetric) LiDAR (G-L) can be used to characterize stream bathymetry, but little is known of its ability to accurately characterize stream bathymetry in narrow (width less than 5 m), heavily forested streams. Canopy characterization with green LiDAR is also poorly understood. We compared canopy and digital elevation models (DEMs) derived from green and near-infrared LiDAR (NIR-L) to field measurements in a narrow, forested stream in Oregon, USA, as well as comparing the two canopy models and DEMs to each other along the length of the stream and to estimates of leaf area index. We observed that the canopy models from the G-L are lower in accuracy compared to NIR-L canopy models. Canopy height models from the G-L were up to 26% less accurate in dense stands, compared to the NIR-L accuracy of 94%. We attribute these errors in part to the lower quality of DEMs generated from the G-L as compared to the NIR-L DEMs. As for bathymetry, the G-L DEM was 0.05 cm higher in elevation than the field measured stream elevation, while the NIR-L ground model was 0.17mm higher. The elevation difference tended to increase with stream depth for both types of LiDAR-derived DEMs, but stream depth only explained a small portion of the variability (coefficient of determination equals 0.09 for NIR-L DEM and 0.05 for G-L DEM). Our results suggest that G-L may be limited in accurately characterizing the bathymetry of narrow streams in heavily forested environments due to difficulty penetrating canopy and interactions with complex topography.