Characterizing Snow Water Equivalent uncertainty in vegetated regions for the Airborne Snow Observatory

Snow depth uncertainty in the Airborne Snow Observatory's (ASO - http://aso.jpl.nasa.gov) 50 m Snow Depth products have been shown to increase with increasing terrain slope and vegetation density in the Senator Beck basin in Colorado. These snow depth uncertainties are directly propagated into the ASO 50 m SWE estimates. With improved characterization of uncertainty, such SWE estimates may be potentially corrected to improve the retrievals. However, methods for applying this type of correction have not yet been developed. In this study, we used the ASO SnowEx 2017 collections for Senator Beck with high-density lidar retrievals (26 points/m^2) as the best representation of spatially distributed Snow Water Equivalent (SWE), or "truth". We then used point decimation to "thin" the lidar return density to (1.5 points/m^2), which is equivalent to ASO's typical operations. When comparing the two basin-integrated SWE estimates at 50 m resolution, ASO's operational condition was estimated to have 5% SWE volumetric error. Most of the areas with greater uncertainty were situated in forests, where lidar point density is diminished. To adjust for these SWE uncertainties, we used several combinations of well-known physical and survey characteristics, such as vegetation, ground point density (< 1 points/m^2), and slope, to identify locations for potential error correction. In this study, we explore techniques for improving the accuracy in ASO 50 m SWE estimates that are provided to water managers for future ASO operations while maintaining the delicate trade-off and balance between snow depth accuracy, flight time, and cost.