ww.mers.byu.edu/long/papers/conf/IGARSS2006_Li.pdf. It's a snapshot of the page taken as our search engine crawled the Web.
The web site itself may have changed. You can check the current page or check for previous versions at the Internet Archive. Yahoo! is not affiliated with the authors of this page or responsible for its content.
WindSat Polarimetric View of Greenland
WindSat Polarimetric View of Greenland
1
Li Li,
1
Peter Gaiser and
1
Elizabeth Twarog
1
Remote Sensing Division
Naval Research Laboratory
Washington, DC 20375, USA
E-mail:
li.li@nrl.navy.mil

2
David G. Long and
3
Mary Albert
2
Center for Remote Sensing, Brigham Young University
Provo, Utah 84602, USA
3
Cold Regions Research and Engineering Laboratory
Hanover, New Hampshire 03755, USA
Abstract WindSat has systemically collected the first global
passive polarimetric data over both land and ocean at three
frequencies: 10.7, 18.7 and 37 GHz, including the brightness
temperatures at vertical and horizontal polarizations, and the
real and imaginary parts of the cross-correlation of the vertical
and horizontal polarizations. Prior to the launch of WindSat, it
was commonly believed that land polarimetric signatures at
satellite footprint scales are below instrumental noise levels and
do not have any useful geophysical information. On the contrary,
WindSat polarimetric data exhibit distinct geophysical and
observation geometry signatures, particularly over Greenland
and Antarctic where the signatures are related to snow
accumulation, melting and metamorphism. The third and fourth
Stokes parameters show well defined, large azimuth modulation
which is correlated with geophysical variations, particularly with
snow metamorphism, and has consistent seasonal variation. We
use simple empirical models to separate and quantify such
azimuthal modulations and geophysical changes. By comparing
the temporal variations of harmonic coefficients and brightness
temperature signatures in vertical and horizontal polarization
channels, we find that both volume and surface scattering have
important contributions to the polarimetric signature. Such
signatures are relatively weak in the summer, when sastrugi are
small and surface scattering is significant, and are strongest in
spring, when the sastrugi are larger and volume scattering is
important.
Keywords-polarimetric microwavr radiometry; WindSat; snow;
Greenland
I.

I
NTRODUCTION

WindSat, launched in January 2003 and currently in
operation, is the first spaceborne microwave polarimetric
radiometer to measure all four elements of Stokes vector,
namely the brightness temperatures at vertical and horizontal
polarizations, and the real and imaginary parts of the cross-
correlation of the vertical and horizontal polarizations [1].
WindSat was designed to measure ocean surface wind speeds
as well as wind directions by including the third and fourth
Stokes parameters, which are mostly related to the asymmetric
structures of the ocean surface roughness. Prior to the launch of
WindSat, it was a commonly believed that land polarimetric
signatures at satellite footprint scales would be below the
instrument noise level and would not carry any useful
geophysical information. However, on the contrary, post-
launch data processing reveals significant land signals in the
third and fourth Stokes channels, particularly over Greenland
and the Antarctic ice sheets. For example, Fig. 1 depicts the
third and fourth Stokes measurements at 10.7 GHz over the
North Hemisphere for the period of 2/1-9/2004 Although the
third Stokes shows the most coherent large scale signals over
the ocean, and it also shows a significant 0.5 to 1 K signal over
land. Over Greenland the third Stokes parameter varies
between ±10 K while the fourth Stokes parameter varies
between -10 and +20 K, which is up to ten times larger than
those observed over the ocean. In this paper, we focus our
analysis on WindSat over Greenland, charactering its
polarimetric signatures and its associated temporal and spatial
variations.
As the second largest ice sheet in the world, the Greenland
icesheet is the most environmentally sensitive Earth media,
playing a significant role in global sea level and climate
changes. Understanding this polarimetric signature, uniquely
afforded by WindSat, and its relation with the snow properties
and microstructures could have a profound impact on climate
study.
II. W
IND
S
AT
D
ATA
A
NALYSIS

A. Methodology
The objective of this study is to define the passive
microwave signature of Greenland from WindSat data. By
observing the changes in the microwave signature, one can
infer the temporal and spatial variations in the physical
properties of the ice sheet. A simple and effective way to define
the microwave signature is to construct empirical observation
model that can describe and separate different effects in the
measurements and summarize the microwave signature in a
small number of model parameters. Over Greenland, the
vertical and horizontal polarized brightness temperatures
respond mostly to the grain size, density and temperature of the
snow, ice and firn; while the third and fourth Stokes parameters
respond most strongly to the asymmetric structure of the snow-
pack, and can be strong functions of the observation geometry,
including the azimuth look angle. Therefore it is essential to
consider the azimuth modulations of the third and fourth Stokes
parameters and separate observation geometry effects from
environmental variations. Since vertical and horizontal
brightness temperatures have been well investigated in other
studies in the past, here we focus on modeling of polarimetric
signatures in the third and fourth Stokes parameters
Given the constant Earth incidence angle of WindSat
conical scanning geometry, the 3rd Stokes (U) and 4th Stokes
(V) brightness temperatures over Greenland are functions of
satellite azimuth look angle (observation geometry) and ice-
sheet characteristics [2][3]. We thus adopt a simple empirical
model for the polarimetric emissions,
This research is supported in part by the NRL Ocean and Atmospheric
Science and Technology Program and the NPOESS Integrated Program
Office.)
U.S. Government work not protected by U.S. copyright 3807


+ + + +
=


)
(
2
sin
)
sin(
)
(
2
sin
)
sin(
2
2
1
1
0
2
2
1
1
0
V
V
V
U
U
U
V
U

where is observation azimuth angle; {U
i
} and {V
i
} are the
coefficient of the azimuth modulation, and {
i
} and {
i
} are
the orientations of different harmonics. In the following, our
goal is to fit such an equation to time series of WindSat
observations for a particular location. Temporal and spatial
variations of these coefficients can provide insight about the
microwave scattering mechanism and its related geophysical
processes.
B. Data Preparation and Azimuth Modulation Analysis
The seasonal variations of the Stokes parameters at
locations in different snow zones in Greenland reflect the
transitions in surface and subsurface properties. Our focus is on
the time-series analysis of the third and fourth Stokes
parameters, since the first two Stokes parameters (vertically
and horizontally polarized brightness temperatures) are mostly
dominated by the snow dielectric properties and physical
temperatures, not observation geometry, we compare the time-
series of the third and fourth Stokes parameter harmonic
coefficients to the vertically and horizontally polarized
brightness temperatures to provide physical insights into the
cause of variations in the third and fourth Stokes parameters.
The third and fourth Stokes parameters are very sensitive to
asymmetric structures of snow media and therefore the sensor
observation geometry and are less sensitive to physical
temperatures of snow. As a result, the third and fourth Stokes
parameters have well defined dependencies on sensor azimuth
looking angle when compared with dual-polarized radiometer
data and scatterometer data [2], where it has always been a
challenging task to separate diurnal effects from azimuth
modulations. For WindSat data, the 3rd and 4th Stokes channel
data exhibits little diurnal effects throughout the year and fit
well to the empirical model. Fig. 2 illustrates the azimuthal
variation of WindSat polarimetric data extracted from the
Summit site for April 2003. All six WindSat polarimetric
channels are shown here. The measurements centered on 210
o

and 340
o
compass azimuth angle correspond to satellite
descending and ascending passes, respectively. The solid line is
the fitted second-order harmonic model. The left column
shows the third Stokes parameter while the right column shows
the fourth Stokes parameter. The top row is 10.7 GHz, the
center row is 18.7 GHz and the bottom row is 37 GHz. Clearly,
there are well-pronounced azimuth dependences on all the
channels. The WindSat data fits to the empirical model are
excellent, as depicted by the solid lines. We also fitted the
model to WindSat data from different months and obtained
similar results, which suggest that WindSat polarimetric data
can be modeled well using the second order harmonic model.
III. R
ESULTS

A. Dry-Snow Zone
In the Greenland dry-snow region, the t