Spatiotemporal Variability in Topographic and Vegetative Controls on Basin-Wide Snow Distribution in the Tuolumne River Basin

An accurate empirical characterization of topographic and vegetative controls on snow distribution can lead to a greater understanding of the underlying physical processes and an increased ability to downscale lower-resolution observations. As improved water resource forecast methods are sought to address climate-driven nonstationarity in snow distributions, constraining our uncertainty in topographic and vegetative controls on these distributions becomes imperative. The Airborne Snow Observatory (ASO) LiDAR-based observation campaign provides a novel dataset with the necessary spatiotemporal extent and resolution for rigorous assessment of spatiotemporal variance in topographic and vegetative controls. In this study, we examine ASO measurements from 2013-2016 in the Tuolumne River Basin, exploring relationships to topographic and vegetation features derived from analogous snow-free LiDAR flights. To address nonlinearities in these relationships, we use single and ensemble regression tree approaches and assess metrics of feature importance, while for greater interpretability, we assess parameter values from multiple linear regression. These complementary analyses are performed for each flight date in 2013-2016 at resolutions between 3 and 500m. They are performed globally and for each of the 13 HUC12-level watersheds within the study area. Feature importance and parameter values are compared across features and across intra-seasonal, inter-seasonal, spatial, and model scale dimensions. Initial results demonstrate a consistent pattern to the changing influence of topographic and vegetative features over intra-annual timescales. They support previous findings that elevational gradients dominate local topographic and vegetative features in controlling both depth and SWE yet suggest a declining importance of elevation in the ablation period. Together, topographic and vegetative features explain more of the spatial distribution of depth and SWE observed during accumulation than during ablation, suggesting stronger controls in this period. Additional findings addressing spatial and scale-based variability will be presented, and all results will be discussed in the context of the generalizability of controls for the downscaling of snowpack observations and model estimates.