The Relation of Millimeter-Wavelength Backscatter to Surface Snow Properties

Although model results indicate that microwave backscatter and emission from snowcover should depend on dry-snow grain size and wet-snow surface roughness, there has been little or no experimental verification. In this paper, helicopter-mounted radar measure-ade ments of 94-GHz backscatter from snowcover, and ground-truth measurements of snow surface roughness, wetness, grain size, and porosity, are analyzed. For each of six polarization combinations, and separately for dry snow and wet snow, spatial mean values <σ°> of the backscatter coefficient σ° are fit to linear combinations of the cosine of incidence angle and the snow variables, the latter in the order in which the explained variance in <σ°> is most increased, provided the increase is significant. However, the significance of an included snow variable is considered questionable if the predicted response of <σ°> to that variable is small compared with the spatial standard deviations of or (typically 4-5 dB). This is the case for dry-snow grain size, porosity, and for some polarization combinations, wetness. Only the response to wet-snow surface roughness is consistently comparable in magnitude to the standard deviations of σ°. Dry-snow surface roughness, wet-snow grain size, and for most polarization combinations, porosity, made no significant improvement to the explained variances. An examination of residuals shows that the linear model is adequate for the <σ°> response to cosine of incidence angle and snow surface roughness.