5. A Two-Layer Microwave Snow Emissivity Model
Satellite Data Assimilation
Banghua Yan, Fuzhong Weng, and Ding Ellen Liang
Author：
Applicable for
Affiliation： USA/NOAA/OSPO
The one-layer Microwave Land Emissivity Model (MELM) developed by Weng et
al. (2001) has been used in the NCEP Global Forecast System (GFS) through the
Joint Center for Satellite Data Assimilation (JCSDA) Community Radiative Transfer
Model (CRTM). The introduction of the MLEM into the NCEP assimilation system
has significantly increased uses of satellite microwave data over most land
conditions. However, the model displays a large uncertainty in simulating snow
emissivity especially at frequencies above 30 GHz thus resulting in a low utilization
rate of satellite data at window and surface-sensitive channels over snow surfaces in
global satellite data assimilation system. This degraded performance of the model
stems primarily from the invalidity of the dense medium theory (Tsang et al. 1984)
in computing snow optical parameters at middle and high frequencies. It is also
partially due to neglecting snow stratification in the model since satellite
observations within field-of-view (FOV) may be sensitive to snow stratification at
depths of a few centimeters from the surface especially for snows after a period of
metamorphosis. In this study, the Dense Media Radiative Transfer theory (DMRT)
with the Quasi-Crystalline Approximation (QCA) (Tsang et al. 2001 and Liang et al.
2008) simulations are used to generate a look-up table (LUT) including single
scattering-albedo, extinction coefficient, and asymmetry factor vs. grain size and
fractional volume (proportional to snow mass density). The LUT is implemented into
the MLEM to improve calculations of snow parameters. Meanwhile, the MLEM is
extended to two layers to catch stratification features of snow medium at shallow
depths (mostly two top layers), simply named to be 'two-layer snow emissivity
model' hereinafter. This consideration is especially useful for a newly falling snow
overlying aged snow. The performance of the new two-layer model with the LUT of
snow optical parameters is assessed by applying the model to a series of snowpacks
in Colorado Rocky Mountains in late winter of 2002 and 2003. Mass density, grain
size and temperature of these snowpacks at multi-depths were measured through the
NASA Cold Land Processes Field Experiment (CLPX). The model-simulated
emissivities are compared with ground emissivity retrievals from the University of
Tokyo's ground-based passive microwave radiometer (GBMR-7) observations from
18.7 to 85 GHz and as well with satellite emissivity retrievals from Advanced
Microwave Sounding Unit (AMSU)-A/B observations from 23.8 to 150 GHz.