Determining the Flow State of Channels Under Vegetation With Airborne Lidar

Abstract

AbstractHeadwater channels are vital to ecological health, water quality, and watershed connectivity. However, the geographic extent and the temporal wetting and drying dynamics of headwater catchments are not thoroughly understood, primarily due to occlusive vegetation. Light Detection and Ranging (lidar) data can be utilized to address this knowledge gap. Airborne lidar is capable of penetrating vegetation and has been used as a spatial and spectral tool for water body delineation. In the work presented here we: (a) develop a novel means of normalizing lidar ground return intensity to allow for comparison across data sets; (b) demonstrate a statistically significant reduction in median return intensity of wet stream reaches under dense vegetation relative to dry channel reaches; and (c) leverage this reduction to create classified maps of wet and dry channel networks in densely vegetated drainages. Across four study basins spanning over 100 km2 we observed an average reduction in median intensity of 41.7% and 72.2% for areas with and without dense vegetation, respectively. Optimal thresholds for delineating wet from dry stream reaches were determined using probability density functions. Resulting classified maps yielded overall accuracies ranging from 87.3% to 95.3% when compared to the National Hydrography Dataset via stratified random sampling. This study demonstrates that remote delineation of channel flow states in densely vegetated areas is possible, thus allowing for better consideration, designation, and conservation of headwater channels.