Estimating the spatial distribution of snow in mountain basins using remote sensing and energy balance modeling

We present a modeling approach that couples information about snow cover duration from remote sensing with a distributed energy balance model to calculate the spatial distribution of snow water equivalence (SWE) in a 1.2 km 2 mountain basin at the peak of the accumulation season. In situ measurements of incident solar radiation, incident longwave radiation, air temperature, relative humidity, and wind speed were distributed around the basin on the basis of topography. Snow surface albedo was assumed to be spatially constant and to decrease with time. Distributed snow surface temperature was estimated as a function of modeled air temperature. We computed the energy balance for each pixel at hourly intervals using the estimated radiative fluxes and bulk‐aerodynamic turbulent‐energy flux algorithms from a snowpack energy and mass balance model. Fractional snow cover within each pixel was estimated from three multispectral images (Landsat thematic mapper), one at peak accumulation and two during snowmelt, using decision trees and a spectral mixture model; from these we computed snow cover duration at subpixel resolution. The total cumulative energy for snowmelt at each remote sensing date was weighted by the fraction of each pixel's area that lost its snow cover by that date to determine an initial SWE for each pixel. We tested the modeling approach in the well‐studied Emerald Lake basin in the southern Sierra Nevada. With no parameter fitting the modeled spatial pattern of SWE and the mean basin SWE agreed with intensive field survey data. As the modeling approach requires only a remote sensing time series and an ability to estimate the energy balance over the model domain, it should prove useful for computing SWE distributions at peak accumulation over larger areas, where extensive field measurements of SWE are not practical.