Geology Reference
In-Depth Information
of using a single‐channel SAR for ice classification. This
was recognized as a major obstacle in developing an
automated operational ice classification system. More
details on this point are presented in section 8.1.1.
The value of visual interpretation of the ERS SAR
images for operational ice monitoring was confirmed but
with some critical limitations in a few application demon-
stration programs [e.g.,
Hakansson et al
., 1995;
Shokr
et al
., 1996; and
Johannessen et al
. 1996). Some of these
studies recommended the use of ERS SAR images as an
auxiliary source of information in ice monitoring pro-
grams. Retrieval of ice motion was developed using
ERS‐1 and ERS‐2 at ASF but with limited success due to
the narrow swath of the satellite (100 km) and the infre-
quent revisits over the same site (the repeat cycle was 35
days).
Dokken
[2000] compiled a number of research
summaries of ERS SAR projects conducted under the
auspices of ESA, based on documentation provided by
ESA and the Canadian Space Agency (CSA).
In February 1992 the National Space Development
Agency of Japan (NASDA), which became JAXA in 2003,
launched its first space‐borne SAR sensor onboard the
Japanese Earth Resources Satellite (JERS‐1). It featured
an L‐band (23.5 cm wavelength) operating at HH polari-
zation. This relatively long wavelength allowed more pen-
etration through the ice; therefore the received backscatter
carried more information about the ice subsurface proper-
ties. In essence, the longer wavelength of the L‐band can
be more appropriate than the C‐band in detecting ridges,
rubble fields, and brash ice. Moreover, the relatively shal-
low incidence angle of JERS‐1 (about 35° at the middle
of the swath) made it more capable of detecting ridges
and other ice surface deformation.
Dierking and Busche
[2006] conducted a useful comparison between the L‐
band JERS‐1 and the C‐band ERS‐1 and demonstrated
that the images of both sensors complement each other,
resulting in a more detailed view of the sea ice cover state.
The CSA launched its first SAR system onboard the
Radarsat‐1 satellite in November 1995. This was followed
by the launch of Radarsat‐2 in December 2007. Both sat-
ellites carried a C‐band (5.6 cm wavelength) SAR system.
A prime objective of Radarsat‐1 was monitoring sea ice in
the polar regions and in the Canadian waters routinely.
This objective has been extended to Radarsat‐2, but the
range of applications of the data from both satellites has
significantly expanded over the years. Radarsat‐1 was
developed and operated by CSA while Radarsat‐2
was developed and operated by MacDonald Dettwiler
and Associates (MDA) Geospatial Services in Vancouver,
Canada, under contracts from CSA. Radarsat‐1 was
decommissioned in March 2013 due to technical anomaly
after it surpassed its originally designed 5 year life expec-
tancy by 13 years. At the time of this writing Radarsat‐2
is still operating. Radarsat‐1 had seven imaging modes
that offered different viewing geometry and spatial
resolution. The standard mode had seven beams, each one
covering 100 km swath at different viewing angle widths
with 25 m spatial resolution and 12.5 m pixel spacing.
However, the mode that was used most frequently (and
still being used by Radarsat‐2) is the ScanSAR‐Wide.
This has nominal swath of 500 km and nominal resolu-
tion of 100 m. More modes have been added to Radarsat‐2
as shown in Figure 7.8, but the ScanSAR mode is still the
preferred mode for ice monitoring. Details of Radarsat‐2
modes and technical specifications are available in
Foxe
et al.
[2004] and
MacDonald Dettwiler and Associates
[2009].
Initial SAR systems on Seasat, ERS‐1, ERS‐2, JERS‐1,
and Radarsat‐1 were single channel, transmitting and
receiving signals in the same polarization. While many
studies identified the potential and limitations of single‐
channel SAR for sea ice applications, a remarkable devel-
opment was achieved by adding the multipolarization
sensors to the space‐borne SAR system (definitions,
notations, and more information on radar polarization
are introduced in section 7.6.2.2).
The first system that featured multipolarization data
acquisition was the Advanced SAR (ASAR) onboard the
European satellite ENVISAT, launched on 1 March 2002.
ASAR was also a C‐band sensor operated in one of five
selective “polarization” modes: two single co‐polariza-
tion HH or VV and three alternating co‐ and cross‐polar-
ization modes: HH and HV, VV and VH, or HH and VV.
These were known as alternating polarization (AP)
acquisitions. The data were available from the standard
beam modes only (i.e. not from the ScanSAR wide mode).
Data from the AP mode were used in several studies to
evaluate their potential for mapping sea ice types.
Scheuchl et al
. [2004] evaluated the higher information
contents from this mode for its potential use in the
Canadian ice monitoring program. It confirmed its utility
but recommended some treatment of the noise level,
which is higher from the cross‐polarization channel (com-
pared to co‐polarization channels) and varies across the
swath. The satellite was suddenly lost on 8 April, 2012
after it exceeded its life expectancy and acquired valuable
data on sea ice during its 10 years of operation. Radarsat‐2
has also the capability of acquiring dual polarization in
two modes: HH and HV or VV and VH. These modes are
available from the ScanSAR acquisitions yet only limited
research studies have been conducted to explore the
potential of these data in retrieving sea ice parameters.
Dierking and Pedersen
[2012] presented an overview on
the achievements made in sea ice monitoring by using
ASAR on ENVISAT during its lifetime. They empha-
sized the utility of the dual polarization through visual
analysis of the images. Figure 7.9 is reproduced from their
study. It shows ASAR dual‐polarization images (HH and
HV) over sea ice near the northeast of Nordaustlandet/
Svalbard. They concluded that cross‐polarization data
(HV) is less sensitive to the incidence angle compared to
