SPATIOTEMPORAL VARIABILITY OF SNOW DEPLETION CURVES DERIVED FROM SNODAS FOR THE CONTERMINOUS UNITED STATES, 2004-2013

Assessment of water resources at a national scale is critical for understanding their vulnerability to future change in policy and climate. Broad-scale hydrologic models used for such assessments include the U.S. Geological Survey National Hydrologic Model (NHM). The NHM represents snow processes through snow depletion curves (SDCs) which relate snow water equivalent to snow covered area over a snowmelt season for a given hydrologic response unit (HRU). Currently, there are two SDCs used in the NHM: one for HRUs above treeline and one for HRUs below treeline. This study revisits the assumption that these two SDCs adequately describe the spatiotemporal variability of snow process in the conterminous United States (CONUS). CONUS-scale evaluation of snow process variability requires consistently-derived data at a broad-scale over many snow seasons. The operational snow model Snow Data Assimilation System (SNODAS) provided data for this analysis. SNODAS uses remotely-sensed and station data to produce modeled snow water equivalent and snow covered area at a 1-kilometer (km) grid for CONUS. Daily gridded SNODAS data were aggregated to the HRU scale (n=109,951) for each snowmelt season from October 1, 2004 to October 1, 2013 (nine seasons). Annual SDCs were created using a consistent methodology for each year and each HRU. The annual SDC closest to the median value of SDCs for each HRU was chosen to represent snowmelt processes in that HRU. Representative SDCs were grouped, aggregated, and compared to assess variability. Grouping methods included a k-mean clustering algorithm of SNODAS and DEM-derived characteristics, and a hierarchical clustering analysis. Second-order polynomial regression was used to generalize all SDCs within a group into a single SDC to compare across groups. Statistically-derived clusters were also mapped to assess geographic variability and grouping. Results show that SDC variability is not adequately described in with the two existing groups at the CONUS-scale.