2828 IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 49, NO. 8, AUGUST 2011
A Case Study of Using a Multilayered
Thermodynamical Snow Model for
Radiance Assimilation
Ally M. Toure, Kalifa Goïta, Alain Royer, Edward J. Kim, Michael Durand, Steven A. Margulis, and Huizhong Lu
Abstract—A microwave radiance assimilation (RA) scheme for
the retrieval of snow physical state variables requires a snow-
pack physical model (SM) coupled to a radiative transfer model.
In order to assimilate microwave brightness temperatures (Tbs)
at horizontal polarization (h-pol), an SM capable of resolving
melt–refreeze crusts is required. To date, it has not been shown
whether an RA scheme is tractable with the large number of
state variables present in such an SM or whether melt-refreeze
crust densities can be estimated. In this paper, an RA scheme is
presented using the CROCUS SM which is capable of resolving
melt-refreeze crusts. We assimilated both vertical (v) and hori-
zontal (h) Tbs at 18.7 and 36.5 GHz. We found that assimilating
Tb at both h-pol and vertical polarization (v-pol) into CROCUS
dramatically improved snow depth estimates, with a bias of 1.4 cm
compared to -7.3 cm reported by previous studies. Assimilation of
both h-pol and v-pol led to more accurate results than assimilation
of v-pol alone. The snow water equivalent (SWE) bias of the RA
scheme was 0.4 cm, while the bias of the SWE estimated by an
empirical retrieval algorithm was -2.9 cm. Characterization of
melt-refreeze crusts via an RA scheme is demonstrated here for
the first time; the RA scheme correctly identified the location of
melt-refreeze crusts observed in situ.
Index Terms—Assimilation, melt–refreeze crusts, radiance,
snow, snowpack model (SM).
I. I NTRODUCTION
S
EASONAL snow cover has a strong impact on cli-
mate, hydrological processes, and on human activities [1].
Snow is the frozen storage term in the water balance and is
also a valuable resource. Monitoring snow physical variables
Manuscript received June 14, 2010; revised November 3, 2010 and
January 22, 2011; accepted January 30, 2011. Date of publication April 29,
2011; date of current version July 22, 2011. This work was supported in
part by the National Science and Engineering Research Council of Canada,
by Environment Canada (Cryosphere System in Canada Program), and by
the National Aeronautics and Space Administration Terrestrial Hydrology
Program.
A. M. Toure and E. J. Kim are with the National Aeronautics and Space Ad-
ministration Goddard Space Flight Center, Global Modeling and Assimilation
Office Code 610.1, Greenbelt, MD 20771 USA (e-mail: ally.toure@nasa.gov;
edward.j.kim@nasa.gov).
K. Goïta, A. Royer, and H. Lu are with Université de Sherbrooke,
Sherbrooke, QC J1K 2R1, Canada (e-mail: kalifa.goita@usherbrooke.ca;
Alain.Royer@USherbrooke.ca; huizhong.lu@usherbrooke.ca).
M. Durand is with the School of Earth Sciences, The Ohio State University,
Columbus, OH 43210 USA (e-mail: durand.8@osu.edu).
S. A. Margulis is with the Department of Civil and Environmental En-
gineering, University of California, Los Angeles, CA 90095 USA (e-mail:
margulis@seas.ucla.edu).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TGRS.2011.2118761
(extent, water equivalent, and melting conditions) is essential
for weather and hydrological forecasts in many regions and pro-
vides information for meteorological forecasts, hydroelectric
power generation, fresh water supply, river traffic, irrigation,
runoff, and flood control [2].
Passive microwave-based estimates of terrestrial snow vari-
ables such as snow water equivalent (SWE) from satel-
lite observations have so far been generated primarily by
regression-based empirical methods. Snow variables retrieved
by such methods are sometimes used in product-based data
assimilation [3]–[5]. With this type of approach, the retrieval
and the forward data assimilation do not necessarily have con-
sistent physics or assumptions. Consistency is much easier to
achieve when radiances are directly assimilated. This radiance
assimilation (RA) approach has been used for years to retrieve
atmospheric parameters by the operational weather forecasting
community with great success [6]. This approach may also be
an effective way to retrieve snow physical parameters such as
snow depth and its water equivalent [7].
Snow RA requires accurate predictions of the brightness
temperature (Tb) emitted by the snowpack. This imposes re-
quirements on the key elements of the snow RA scheme: a
snowpack model (SM), a radiative transfer model (RTM), and a
data assimilation framework. In a previous study, Durand et al.
[8] demonstrated that the microwave emission model for lay-
ered snowpacks (MEMLS) [9] has the ability to accurately
predict the radiance when the snow layering structure and other
state variables, such as grain-size correlation length, density,
snow depth, liquid water, and temperature, are accurately rep-
resented. They also showed that the accuracy of RTMs particu-
larly at horizontal polarization (h-pol) is highly sensitive to the
stratigraphic representation of the snowpack and particularly to
dense snow layers, such as melt–refreeze crusts.
Durand et al. [10] subsequently demonstrated the potential
of snow RA using the simple snow-atmosphere-soil transfer
model (SAST) [11] SM coupled to MEMLS. SAST is a simple
three-layer energy-balance scheme. An improvement of the
simulation of snow depth was achieved through the use of an
ensemble Kalman smoother scheme at this local scale for a
weeklong study [10]. However, limitations of SAST (three lay-
ers and no representation of melt-refreeze crusts) were evident.
Indeed, only Tbs at vertical polarization (v-pol) could be assim-
ilated due to the fact that SAST does not adequately represent
the vertical stratigraphy of the snowpack. This is inherently
suboptimal, as only half of the available information could be
assimilated; the h-pol channels were ignored. Assimilation of
0196-2892/$26.00 © 2011 IEEE