Geology Reference
In-Depth Information
ice classification, and ocean wave products. All products
were presented on Eulerian (i.e., regularly spaced) grids.
However, the first research study to generate age distri-
bution from Lagrangian ice motion using this system is
presented in
Kowk et al
. [1995].
In 1998 the successor of the GPS, called the Radarsat
Geophysical Processor System (RGPS), was developed
also by JPL and installed in the ASF. Description of the
system and a full account of its development history
are presented in
Kwok
[1998]. This operational system
has been dedicated to process Radarsat sea ice images of
the Arctic, which are acquired within the field of view
of the ground receiving station at the ASF. The system
handles a huge volume of ScanSAR data, which started
in November 1996. About 6400 scenes are handled per
year, corresponding to about 20 scenes per day. It produces
nearly complete coverage of the Arctic every 6 days.
Review of its applications is presented in
Lindsay and
Stern
[2003] and
Kwok
[2010].
The RGPS improves upon the GPS in various ways.
Instead of 3 products it produced 16 products. The main
product remains the ice motion, but other products
include forms of ice deformation, ice thickness, ice age,
backscatter histograms, and ice classification in addition
to ocean wave characteristics. A list of the products with
their specifications is presented in
Kwok et al
. [1999].
Only ice motion is relevant to this section. Derived ice
deformation is addressed briefly in section 9.1. Another
improvement of the RGPS is the production of ice
motion maps on a Lagrangian (i.e., irregular) grid in
addition to the Eulerian grid motion products. The differ-
ence is explained also in section 9.1. It should be noted,
however, that the RGPS generates its high‐resolution
products for periods of up to 6 months during fall/winter/
spring seasons, tracked every 3-6 days [
Kwok and
Cunningham
, 2002]. The product is irregular in both time
and space because it depends on the availability of suita-
ble pairs of Radarsat images.
Generation of motion fields can be accomplished
using Eulerian or Lagrangian representation of velocity.
The difference between the two representations of ice
motion is illustrated in Figure 10.48. In the Eulerian rep-
resentation the velocity is generated at a fixed position (it
can be a grid point) in the first image. That position must
have an identifiable feature (e.g., an ice floe or aggregates
of floes as shown in Figure 10.45). The MCC technique
is applied to locate the feature in the second image. This
is repeated for a number of features to locate their posi-
tions in the second image. The output is a snapshot of
displacement or velocity vectors from a single pair of
images as shown in Figure 10.48. An instructive
animation that explains how an Eulerian grid map of
ice motion can be generated between two dates using
the MCC technique is shown on the website of GlobICE
http://www.globice.info/Section.php?pid=26.
The Eulerian
approach is more suited for studies of large‐scale circula-
tion patterns as well as regional and basin‐scale ice
advection that do not require details of ice deformation
[
Kwok
, 2010]
In the Lagrangian representation the grid elements (not
just the center points of the selected features) are tracked
over time (not just between two successive images). This
can be done by tracking the trajectory of the corners of a
given cell between sequential pairs of images. The cell can
have four or more corners. This allows the user to detect
the ice deformation (e.g., ice ridging) during the freezing
season. Figure 10.48 shows a Lagrangian representation
of a deformed grid. A new visualization tool was incor-
porated in the RGPS to handle the format of the
Lagrangian data. An example of evolution of a deforma-
tion field is presented in Figure 9.4.
Kwok et al
. [1995] have shown the advantage of using
products from the Lagrangian representation of ice
motion to estimate the ice age (or type). This approach
is complementary to the use of the backscatter signa-
ture as it avoids the complications of the signature
ambiguity of ice types. The study is the first attempt to
estimate ice thickness distribution in leads using linear
kinematics feature (LKF) maps (for a definition see
section 9.1). Changes to the area of the cells lead to
inferences about the creation or disintegration of young
ice. Consequently, the thickness of new and young ice is
computed using the local surface air temperature
history and a simple freezing degree‐day parameteriza-
tion.
Kwok
[2002] presents an attempt to retrieve the ice
concentration in winter using the LKF maps as well.
Openings along fracture zones provide an unambigu-
ous measure of the location of open water within the ice
cover. The author states that “the estimate is unique in
that it does not depend on the calibration of the sensor
or a thorough physical understanding of the ice signa-
tures (p. 25-1).”
Lindsay and Stern
[2003] describe how
monthly ice deformation products are derived from the
changes in the cell area available from the Lagrangian
representation of ice displacement. This includes the
three spatial gradients in the velocity field: the diver-
gence, vorticity, and shear (section 9.1).
It should be mentioned that a similar system developed
by ESA, called GlobICE, also employs the Lagrangian
tracking capability and has been used for the analysis of
ENVISAT ASAR data. The system is part of the Data
User Element of ESA's Earth Observation Envelope
program. It produces monthly averaged ice motion in
the Arctic, monthly averaged ice deformation, and sea
ice mass flux through selected gateways. All products
are geared to support the World Climate Research
Programme project with validated SAR‐driven sea ice
motion, deformation, and flux products.
