Estimating snow volume in mountain catchments using Aerial LiDAR

Numerous hydrologic problems would benefit from knowing the true volume of snow in a mountainous catchment. Unfortunately, field based techniques commonly fall short due to their inability to adequately capture the spatial variability of snow in complex terrain. LiDAR has been used to map snow cover by differencing digital elevation models of snow free and snow covered ground. Although this technique has shown promise, the accuracy of the method depends on the reliability and accuracy of the snow-free LiDAR-derived ground elevations used to generate the DEM. Uncertainties have been detected in shrub-dominated areas where LiDAR-processing algorithms seem unable to separate shrub canopy from ground. Misclassification of ground elevation can lead to errors in LiDAR-derived snow depth, including negative depth estimates in locations where shrubs are matted down and completely buried during snow deposition. LiDAR-derived snow depth estimates are also subject to other errors including overestimates in areas with dense shrubs, missing data around tree bowls, and misclassification of terrain features by processing algorithms. Inaccuracies in the x, y, and z directions can translate into significant errors in snow volume estimates, especially on steep slopes. To assess the reliability of LiDAR derived snow estimates, associated errors need to be quantified. This study attempts to identify and quantify the errors associated with LiDAR derived snow volume estimates for the mountainous terrain of the Dry Creek Experimental Watershed, ID. Preliminary results indicate that the errors can be on the order of magnitude equal to the snow depth.