Comparison of Interpolation Techniques and Model Simulations of Snow Water Equivalent in the Sierra Nevada

Abstract:

Seasonal water storage in mountain snowpack is a critical water resource but its variability is highly uncertain in time and space. Here we compare methods for analyzing the spatial distribution of snow water equivalent (SWE) over the Sierra Nevada Mountains. Using progressively more complex techniques, we interpolate measurements from over 250 snow courses and 100 automatic snow sensors to obtain a map of SWE over the entire mountain range. Interpolation methods range from simple regressions utilizing only latitude, longitude, and elevation, to more complex approaches that additionally consider temperature, precipitation, aspect, and others. We explore the differences between maps obtained from interpolations of only snow courses, only snow pillows, and the combined datasets. We additionally consider a model simulation from the Weather Research and Forecast (WRF) model run at 3-km resolution. Biases from the interpolations and the WRF simulation are determined through evaluation against an existing 90-m resolution product based upon data assimilation and SWE reconstruction techniques. Comparisons of the SWE maps will allow us to determine if one method overestimates or underestimates with regards to the others and whether interpolation is useful for a spatially heterogeneous variable such as SWE. We select three different years to study in order to capture a range of snowpack conditions. Water year 2009 (October 2008 through September 2009) was an average snow accumulation year for the region. We also select water year 2005 for a high accumulation year and water year 2014 for a low accumulation year. It is likely that each method will result in different SWE estimates, revealing the uncertainty involved in snow hydrology.