An Evaluation of Terrain-Based Downscaling of MODIS-Based Fractional Snow- Covered Area Datasets Over the Tuolumne River, CA, Based on LiDAR-Derived Snow Data (Extended Abstract)

Remotely-sensed snow covered area (SCA) datasets with both high spatial and temporal resolutions are needed for research, planning, and management of hydrologic and ecologic resources. MODIS-based products provide good temporal resolution (daily), but on coarse scale grids (~463 m). This coarse spatial scale can be refined through applying downscaling procedures, which consist of using the fractional snow cover area product (fSCA, the percentage snow cover within a MODIS pixel area) to assign binary (presence/absence) SCA data on higher resolution grids. Current methods rely on representing ablation effects on snow spatial variability by using topographic radiation-derived slope factors and relative elevation as primary indicators of snow presence/absence (Walters et al., 2015), or a degree-day approach (Li et al., 2015). In both studies satellite-derived data were utilized for model input and validation. Uncertainty associated with the input and validation data in assessing downscaling performance could be better understood if reliable, platform-independent fine scale SCA data were available. Here, we propose such a framework for testing and developing a new downscaling procedure based on LiDAR-derived snow depth data collected over the Tuolumne River watershed, CA. Our new downscaling procedure is based on terrain-derived indices that are representative of both ablation and accumulation, drivers of snow spatial variability in complex terrain. The use of the LiDAR-derived dataset has several advantages over using the satellite data. First, the validation data is more accurate, as LiDAR-derived data are high resolution (1-3m). Second, accurate fSCA datasets can be derived from the high resolution LiDAR-derived snow data at the MODIS scale to be used as input data. Third, the downscaling performance can also be tested over vegetated areas, where LiDAR-derived data is presumably more accurate than the satellite data.