GeoCommunity GeoImaging Feature
Turning the Scientifically Possible into the Operationally Practical:
RADARSAT-2 Polarimetry Applications
By Gordon C. STAPLES and John HORNSBY, RADARASAT International (2002)
RADARSAT-2, planned for a late 2003 launch, is
an advanced SAR satellite. Key features of
RADARSAT-2 are high resolution (3 m),
polarimetric modes, enhanced ground system
providing rapid satellite tasking and near-real time
data processing, improved image location
accuracy, and on-board solid state recorders. The
focus of this paper is on the RADARSAT-2
polarimetric capability. RADARSAT-2 offers
three polarimetric modes: (1) selective polarization
(dual pol) providing one co-pol channel (HH or
VV) and the corresponding cross-pol channel
(HV); (2) high resolution (3 m) single pol channel
(HH or VV); and (3) a fully polarimetric mode
(quad pol) providing both amplitude and phase.
The fully polarimetric mode is significant since
RADARSAT-2 is the first commercial satellite to
offer this mode. The commercial focus of the
RADARSAT-2 mission dictates the development of
operational applications, and ultimately the
extraction of information from the SAR data. This
paper concludes with an assessment of operational
practicality of the RADARSAT-2 polarimetric
modes.
RADARSAT-2, the second in a series of
Canadian spaceborne Synthetic Aperture Radar
(SAR) satellites, is being built by MacDonald
Dettwiler, Richmond, Canada. RADARSAT-2
builds on the heritage of the RADARSAT-1
SAR satellite, which was launched in November
1995 and continues to perform extremely well
(Srivastava et al, 2001). The RADARSAT-2
sensor is planned for launch in late 2003. Plans
are underway to build RADARSAT-3, and
operate RADARSAT-2/3 in a tandem mission.
The RADARSAT-2/3 tandem mission will focus
on interferometric applications, particularly for
the generation of Digital Elevation Models
(DEMs).
RADARSAT-2 will be a single sensor satellite,
and will have on-board a C-band SAR (5.405
GHz). Of note is the fact that while both
RADARSAT-2 and RADARSAT-3 will operate
at the same frequency, the RADARSAT-1 SAR
frequency is marginally different (5.3 GHz). The
RADARSAT-2 orbit parameters will be the same
as RADARSAT-1 thus allowing co-registration
of RADARSAT-1 and RADARSAT-2 images.
Radiometric and geometric calibration will also
be implemented permitting correlation of time
series data for applications such as long-term
change detection (Luscombe and Thomson,
2001).
RADARSAT-2 retains the same capability as
RADARSAT-1 (Luscombe, 2001), but has
advanced SAR-based Earth Observation (EO) by
offering enhanced features including:
- 3 m high resolution mode
- Selective and Polarimetry modes
- Enhanced ground system
- Routine left and right looking capability
- Increased geometric accuracy
- On-board solid state recorders
While each of the enhanced RADARSAT-2
features are deserving of individual discussion,
the focus of this paper is on the polarimetry
modes and what these modes will bring to the
end-users in terms of better information for
SAR-based applications. The next section will
outline the RADARSAT-2 polarimetry modes
and provide definitions as adopted by the
RADARSAT-2 program. Section 3 will discuss
the applicability of polarimetry for a variety of
applications, and finally Section 4 discusses how
polarimetry data can be exploited to meet user
needs.
II. RADARSAT-2 POLARIMETRY MODES
The intent here is not to outline polarimetry
theory, but to present the concepts in an intuitive
manner so that those not familiar with
polarimetry can understand the benefits of
polarimetry and the information available in
polarimetry data. Many articles are available that
discuss polarimetry theory, applications, and
provide excellent background information
(CCRS, 2001; Ulaby and Elachi, 1990).
Notwithstanding the inherent complexity of
polarimetry, polarimetry in its simplest terms
refers to the orientation of the radar wave
relative the Earth’s surface and the phase
information between polarization components.
RADARSAT-1 is horizontally polarized
meaning the radar wave (the electric component
of radar wave) is horizontal to the Earth’s
surface. In contrast, the ERS SAR sensor was
vertically polarized, implying the radar wave was
vertical to the Earth’s surface. Spaceborne SAR
sensors such as RADARSAT-2, ENVISAT, and
the Shuttle Imaging Radar have the capability to
send and receive data in both horizontal (HH)
and vertical (VV) polarizations. Both the HH
and VV polarization configurations are referred
to as co-pol modes. A second mode, the cross-
pol mode, combines horizontal send with vertical
receive (HV) or vice-versa (VH). Polarization
configurations are shown in Fig, 1. As a rule,
the law of reciprocity applies and HV ? VH
(Ulaby and Elachi, 1990). The co-pol and the
cross-pol modes are amplitude data, and in the
format familiar to SAR-data users (e.g.
RADARSAT-1 Path Image product).
A unique feature of RADARSAT-2 is the
availability of polarimetry data, meaning that
both the amplitude and the phase information are
available. The amplitude information is the
same as discussed above, but the phase
information is likely new to most SAR users and
rather non-intuitive. In its simplest term, phase
can be thought of as the travel time for the SAR
signal: the travel time is the two-way time
between the sensors and the Earth, and includes
any propagation delays as a result of surface or
volume scattering. It is the propagation delays
and the scattering properties of the HH and VV
polarization configurations that make
polarimetry data so powerful. In effect, SAR
interferometry exploited the phase information,
not by the time delay between, for example, the
HH and VV modes at the same time
(RADARSAT-2 case), but the time delay
between the HH and HH mode at different times
(RADARSAT-1 case).
The RADARSAT-2 program has adopted
conventional terms to define the polarimetry
modes, which are consistent with accepted
definitions. The two polarimetry terms are
Selective Polarization and Polarimetry.
Selective Polarization implies the availability of
only amplitude data. For example, amplitude
data may be HH, VV, or HV imagery. Other
terms for Selective Polarization include dual,
alternating, and multi-polarization. The second
term, Polarimetry implies the availability of both
amplitude and interchannel phase information.
The amplitude information is the same as the
Selective Polarization case, but adds phase
information, such as the co-phase term
(i.e. ?HH-VV). Other terms include quadrature
polarimetry (“quad-pol”) and fully polarimetric.
The RADARSAT-2 modes are shown in Table 1.
Fig. 1. RADARSAT-1, ERS, and RADARSAT-2 polarization configurations.

Table 1. RADARSAT-2 modes. Beam mode name, swath width, swath coverage, and nominal resolution.
III. RADARSAT-2 POLARIMETRY
APPLICATIONS
RADARSAT-2 will offer a rich source of data
for a variety of applications. As with any
spaceborne sensor, there are trade-offs, which for
SAR sensors are mainly spatial coverage and
resolution. Typically high resolution is available
at the expense of reduced spatial coverage, and
increased spatial coverage at the expense of
coarser resolution. These trade-offs will apply to
RADARSAT-2, although mitigated by the
availability of left and right looking SAR.
The availability of polarimetry data adds
additional imaging choice, with trade-offs
dependent on application needs. Referring to
Table 1., the polarimetric trade-offs are reduced
for the RADARSAT-1 Selective Polarization
modes, since there is no impact on swath width
or resolution as a function of polarization choice.
By default, data products are one co-pol channel
(HH or VV) and the cross-pol channel (HV).
The only impact will be on the ground segment
due to the extra channel of data which adds to
the processing and delivery (if electronic) time.
Depending on application needs, a decision is
required between the use of Selective
Polarization or Polarimetry modes. The
Polarimetry mode offers a choice of resolutions
(25 m and 10 m nominal), four channels of data
(HH+VV+HV+VH), and both co-pol and cross-
pol phase information, subject to the caveats that
as aforementioned HV ? VH, and the cross-pol
phase information does not provide as much
information as the co-pol phase. From an
applications perspective, a key parameter of the
Polarimetry mode is the 25 km swath.
Therefore, the end user must decide, depending
on the application requirements, whether the
Selective Polarization mode or the Polarimetry
mode will provide the information needs.
The utility of the RADARSAT-2 Selective
Polarization (SP) mode and Polarimetry (P)
mode for a number of applications is shown in
Table 2. The assessment was based on
application requirements and divided into three
categories: Improvement (I), Limited (L), and
Not Applicable (NA). Improvement implies
consistently better information, Limited implies
better information, and Not Applicable implies
no better information. This assessment is
relative to RADARSAT-1. It is clear from Table
2 that the SP mode has the widest application,
with the P somewhat less. For example, it has
been shown that the co-pol mode (HH) provides
ice type information, and the cross-pol mode
(HV) provides ice edge information (Scheuchl et
al, 2001).
The use of the Selective Polarization mode for a
given application will be better understood
following the availability of ENVISAT data,
with better understanding of the Polarization
mode following the launch if RADARSAT-2
(ENVISAT does not have Polarimetry-like
mode). It is important to point out that
RADARSAT-2 can be readily configured even
after launch. Subject to three main parameters,
namely resolution, swath width, and noise-
equivalent sigma-0, new modes can be defined to
meet users needs (T. Luscombe, pers. comm.)

Table 2. Use of RADARSAT-2 polarization modes by application.
Selective Polarization (SP), Polarimetry (P), Improvement (I),
Limited (L), Not Applicable (NA).
IV EXPOITING POLARIMETRY TO MEET USER
NEEDS
Polarimetry, while intrinsically challenging from
a scientific perspective, is daunting to the end
user. The commercial focus of the
RADARSAT-2 mission dictates the development
of operational applications, and ultimately the
extraction of information from the SAR data.
Polarimetry represents a steep learning curve for
even the radar knowledgeable user, and perhaps
a step-function for many new users.
To turn the scientifically possible into the
operationally possible, a number of approaches
are suggested.
* Much of the polarimetry reference material
has been written from the engineering
perspective, but many end users are
geoscientists, so polarimetry theory,
explained from the geoscience perspective is
required;
* Polarimetry “tool kits” that provide sample
data sets and functionality for simple
analysis. Of note is the observation that
through NASA’s Jet Propulsion Laboratory,
a polarimetry toolkit called Sigma-0 is
available. A similar toolkit, called the
Polarimetry Work Station is available from
the Canada Centre for Remote Sensing;
* Provide guides as the appropriate use of a
given polarization mode for a given
application. This will have to be done
following the launch of RADARSAT-2 and
data validation, but can be initiated with
ENVISAT data;
* There is a need to combine good R&D
directed toward the development of
operational applications. This type of
initiative has been fostered during the
RADARSAT-1 program where R&D
institutes such as the Canada Centre for
Remote work with both the private and the
public sector to develop applications.
REFERENCES
CCRS, Applications Potential for RADARSAT-2, J. van der
Sanden and S. Ross (Eds), 117 pp., Ottawa, Canada, 2001.
Luscombe, A., and A. Thomson, RADARSAT-2 Calibration:
Proposed Targets and Techniques, Proceedings IGARSS’01,
Australia, 2001.
Luscombe, A, RADARSAT-2 Product Specification,
MacDonald Dettwiler RN-SP-50-9786, January 2001.
Srivastava, S, B. Banik, R. Hawkins,T. Lukowski, K.
Murnaghan, RADARSAT-1 Image Quality, Proceedings
IGARSS’01, Australia, 2001.
Scheuchl, B., Caves, R., Cumming, I., and G. Staples,
Automated Sea Ice Classification using Spaceborne
Polarimetric SAR Data, In Proceedings IGARSS’01, Sydney,
Australia, July 2001.
Ulaby, F. and C. Elachi (Eds), Radar Polarimetry for
Geoscience Applications, Artech House, 364 pp., 1990.
About the Authors
This paper was presented at International Remote Sensing of Environment, Buenos Aires, Argentina, April 2002
Gordon C. STAPLES and John HORNSBY
RADARASAT International
13800 Commerce Parkway
Richmond, B.C., CANADA
Gstaples@rsi.ca
More News & Feature Articles
Nov 15, 2002
Feb 27, 2003
|