Geoscience Reference
In-Depth Information
that GRACE TWS assimilation can lead to better understanding of the hydrological cycle
in remote regions of the globe where ground-based observation collection is difficult, if not
impossible. This information could ultimately lead to improved freshwater resource
management as well as reduced uncertainty in river discharge.
3.3 Microwave Radiative Transfer Models for Radiance Data Assimilation
It is well established for atmospheric data assimilation systems that the assimilation of
satellite radiance observations is preferable to the assimilation of geophysical retrievals
(Eyre et al.
1993
; Joiner and Dee
2000
). The former approach incorporates the radiative
transfer model into the assimilation system and thereby avoids inconsistencies in the use of
ancillary data between the assimilation system and the (pre-processed) geophysical retri-
evals. For land data assimilation, however, the vast majority of publications assimilate
geophysical retrievals (Lahoz and De Lannoy
2013
). In this section, we discuss the
development of forward radiative transfer models (RTMs) that convert land surface model
variables into microwave brightness temperatures. The first example presents such a model
for warm-season microwave brightness temperatures (Sect.
3.3.1
). The second example
introduces a neural network approach to predict microwave brightness temperatures over
snow-covered land (Sect.
3.3.2
).
3.3.1 Warm-Season, L-Band Radiative Transfer Modeling
Global observations of brightness temperatures (Tb) at L-band (1.4 GHz) are available
from the SMOS mission, and similar Tb observations are expected from the planned Soil
Moisture Active Passive (SMAP; Entekhabi et al.
2010
) mission. In preparation for the
assimilation of Tb observations from SMOS and SMAP, De Lannoy et al. (
2013
) added a
physically based, warm-season microwave RTM to the GEOS-5 Catchment model. The
RTM is based on the commonly used, zero-order ''tau-omega'' approach that accounts for
microwave emission by the soil and the vegetation canopy as well as attenuation by the
vegetation. While the RTM is based on sound physical principles, determining the required
parameter values for the microwave roughness, scattering albedo, and vegetation optical
depth on a global scale is a serious challenge.
De Lannoy et al. (
2013
) collected three different sets of the literature values for the
L-band RTM parameters. ''Lit1'' refers to parameters that are proposed for the future
SMAP radiometer retrieval product, ''Lit2'' are parameters collected from the literature
studies using the L-band Microwave Emission of the Biosphere model (Wigneron et al.
2007
) and related models, and ''Lit3'' is the same as Lit2 except that the microwave
roughness parameter is set to values used for SMOS monitoring in the European Centre for
Medium-Range Weather Forecasts (ECMWF). The three sets of parameters are illustrated
in Fig.
6
, which shows the resulting microwave roughness (h), vegetation opacity (s), and
scattering albedo (x) by vegetation class. As can be seen from the figure, there are large
differences in h, s, and x between the three sets of the literature values. These differences
translate into climatological differences in the simulated brightness temperatures.
For example, Fig.
7
a-c shows the differences between 1-year mean (July 1, 2010-July
1, 2011) model simulations (using the three different literature-based sets of RTM
parameters) and SMOS observations for H-polarized Tb at 42.5 incidence angle. Modeled
brightness temperatures are at 36 km resolution, commensurate with the resolution of the
SMOS observations. Brightness temperatures are screened for frozen soil conditions, snow
on the ground, heavy precipitation, proximity to open water surfaces, and radio-frequency
