Snow surface temperature, radiative forcing and snow depth as determinants of snow density

Wednesday, 17 December 2014: 11:20 AM
Peter B. Kirchner, University of California Los Angeles, Joint Institute for Regional Earth Systems Science and Engineering, Los Angeles, CA, United States, Thomas H Painter, NASA Jet Propulsion Laboratory, Pasadena, CA, United States, McKenzie Skiles, University of California Los Angeles, Los Angeles, CA, United States and Jeffrey S Deems, University of Colorado, Boulder, CO, United States
Watershed scale observations of snow water equivalence (SWE) are becoming increasingly important globally as the quantity and timing of snowmelt has become less predictable. In the Colorado River watershed, where dust deposition can hasten snowmelt by several weeks, the need for these observations is critical. While advances in measuring snow depth and albedo from the NASA Airborne Snow Observatory have greatly improved our ability to constrain snow depth and radiative forcing, we have yet to develop a method for remotely observing snow density, which is required for calculating SWE.

We evaluate measured and modeled variables of snow- infrared surface temperature, radiative forcing and snow depth as predictors of snow density. We use 10 seasons of in situ measured snow surface temperature, cumulative modeled dust in snow radiative forcing, snow depth and manually measured snow density from locations in the Rocky Mountains of southwestern Colorado. We also use measured snow depth and SWE from the 2013 and 2014 water years, from 23-35 locations stratified by modeled downwelling short wave radiation, and evaluate them as predictors of snow density.

Our analysis shows that daily mean snow surface temperature (R2 0.61, p = <0.001) and cumulative radiative forcing (R2 0.54, p = <0.001) individually have significant coefficients of determination whereas snow depth alone was not significant. Multiple regression with all three variables (R2 0.84, p = <0.001) was the best predictor of density. Furthermore, when snowpack conditions were isothermal at 0° C, the diurnal coefficient of variation, of measured hourly surface temperature, exhibited consistently high variance. In 2013 we found significant correlations between spatially distributed measurements of snow density (R2 0.33, p = <0.001) and modeled downwelling short wave radiation. However, in 2014 the correlation was very low, supporting our hypothesis that seasonal differences in dust driven radiative forcing are also important for determining the density. These results suggest the properties of snow surface temperature and radiative forcing, when combined with snow depth, are highly predictive of snow density, indicating the potential for inference of SWE from remote sensing measurements of snow depth, snow infrared surface temperature and spectral albedo.