Publication Date


Type of Culminating Activity


Degree Title

Master of Science in Hydrologic Sciences



Major Advisor

Alejandro N. Flores, Ph.D.


Snow and ice are substantial components of the global energy balance and hydrologic cycle. Seasonal snow covers an area of 47 million km2 at its average maximum extent, 98% of which occurs across the Northern Hemisphere. The earth’s radiation budget is largely controlled by the fraction of absorbed solar energy, a parameter that is dependent upon snow surface albedo. Mountain snowpacks act as natural reservoirs, storing large quantities of water throughout the winter until eventual release during the melting phase. Accurate characterization of snow-covered area (SCA) and snow water equivalent (SWE) in such terrain could substantially improve the estimation of timing and volume of melt water runoff. However, knowledge of these hydrologic states is limited in part by scarcely populated in situ observation networks and logistical constraints in field survey sampling. Thus, satellite remote sensing observations are often employed in conjunction with simulation models to improve the estimation of snowpack states and resultant fluxes. This study attempts to merge complementary datasets in order to predict spatially variable snow processes at high resolution in basins exhibiting complex terrain. Specifically, the goal is to provide a means to downscale existing remote sensing and snow modeling datasets using computationally efficient methods that utilize physiographic information regarding terrain and land cover.

A linear combination model is proposed for downscaling fractional SCA from the Moderate Resolution Imaging Spectroradiometer (MODIS) instrument from its native resolution (500 m) to a hillslope-scale resolution (e.g., 10-30 m), preserving the predicted snow cover fraction at the basin scale. The model is calibrated to 30 m Landsat observations using elevation and incoming solar radiation indices for a study area in southwestern Idaho. Validation is performed with data not used during calibration. Results depict favorable model performance when comparing downscaled MODIS snow cover to Landsat binary observations. An “ideal” validation test is performed in which Landsat aggregate 500 m snow fraction informs the model with similarly positive results. The use of such an algorithm might benefit applications from flood forecasting to SWE reconstruction.

In a snowmelt modeling application, the satellite-derived snow cover downscaling algorithm is applied as a binary mask to constrain spatial melt runoff data from the SNOw Data Assimilation System (SNODAS). Differential solar radiation, forest canopy, and snow albedo estimates are also used to further downscale the modeled melt. Comparison with available field lysimeter data show proper spatial disaggregation of modeled melt onto opposing hillslopes, though timing and magnitude issues exist. Implications for resolving snowmelt at hillslope scales are briefly discussed.

Included in

Hydrology Commons