Incorporation of Spectral and Directional Radiative Transfer
in a Snow Model
Graham Glendinning1 and Liz Morris2
Present radiative transfer methods in physically based energy budget models of snow do not include
adequate spectral or directional resolution to deal with the scattering of solar radiation. This paper
reports on results from an advanced physically based snow energy budget model (SNTHERM)
linked with a discrete ordinate radiative transfer model (DISORT) at nine wavelengths assuming
spherical snow grains. Scattering properties were averaged over a small range of grain sizes (as seen
in real snow) to eliminate interference-induced fluctuations.
A method was derived to split a single measurement of spectrally integrated solar radiation into its
direct and diffuse components at nine wavelengths. The split of radiation is required as input to the
radiative transfer model, and is produced as a weighted average from days of total cloud cover and
days of clear skies (from the 6S atmospheric radiative transfer model).
The fully linked model was tested on a data set from a field campaign on the Uranus Glacier, Antarc-
tica, December 1994February 1995. An automatic weather station provided meteorological inputs,
with additional measurements made of solar radiation. A vertical array of thermistors made continu-
ous measurements of snowpack temperature in the top meter of snow, though solar heating and
melting contaminated those thermistors near the surface. Snow pit data were used to initialize the
The combined model was tested against simpler radiative transfer parameterizations, such as the
method of Marks and fixed albedo and extinction techniques. Albedo predictions from the discrete
ordinate radiative transfer model show large variations in albedo with solar zenith angle and the
diffuse and direct spectral radiative split, though unfortunately no albedo measurements were avail-
able for direct comparisons. RMS deviations in measured and modeled temperatures at a depth of 70
cm were found to be 1.2 in. for the optimum fixed albedo (0.8) and extinction (50 m1), 2.0 in. for the
Marks method, and 0.6C for the DISORT method. The use of a spectral method was thus seen to
provide markedly superior results for snowpack temperatures over the test period.
The measurement period was of limited duration at a single location. Further testing is recommended
to demonstrate the benefits over longer time periods and at other test sites, where the snow grains
may be less rounded.
Institut fr Meteorologie und Geophysik, Universitt Innsbruck, Innrain 52, A-6020 Innsbruck, Austria
British Antarctic Survey, High Cross, Madingley Road, Cambridge, CB3 0ET, UK