1.1.5 Special Features of ASAR
|
Figure 1.17 ERS-1 image Sept. 23, 1992 Methane Emissions, Tundra, Alaska North Slope (Copyright ESA, 1992) |
ASAR has a number of special features (see the index)
1.1.5.1 Dual Polarisation
|
Figure 1.18 ERS-1 image Aug.2, 1992 Tanana Valley Alaska (Copyright ESA, 1992) |
Imaging radars can transmit horizontal (H) or
vertical (V) electric-field vectors, and
receive either horizontal or vertical return
signals, or both. The basic physical
processes responsible for the like-polarised
return are quasi-specular surface reflection
and surface or volume scattering. The
cross-polarised return is usually weaker,
and often associated with multiple
scattering due to surface roughness or
multiple volume scattering. Scattering
mechanisms and the returns from different
surfaces may also vary markedly with
incidence angle.
To illustrate the effect of polarisation,
consider the very simple model of a
vegetation canopy consisting of short
vertical scatterers over a rough
surface, as shown in the figure1.19 below:
|
Figure 1.19 Electromagnetic Wave Orientation |
Assuming that the scatterers will act as
short vertical dipoles, then incident,
horizontally polarised microwave energy will
not interact with the canopy and will
scatter from the surface underneath.
Conversely, vertically polarised microwave
energy will interact strongly with the dipoles.
ASAR provides dual-channel data. In
Alternating Polarisation Mode (AP
Mode), it provides one of three
different channel combinations:
- VV and HH
- HH and HV
- VV and VH
Dual polarisation data is
important for a wide range of applications
such as bare soil, vegetation studies,
sea ice applications, etc.
Vertical Transmit/Vertical Receive (VV) -
Like Polarisation
|
Figure 1.20 VV Polarisation |
VV polarisation, shown in figure1.20 , is the preferred
polarisation configuration in a number
of applications. For instance, in studying
the small-scale roughness of (capillary)
waves on the water surface, VV is better
than HH or cross-polarised combinations,
which means it is used extensively for
surface wind speed extraction.
Horizontal Transmit/Horizontal Receive (HH)
- like polarisation
|
Figure 1.21 HH Polarisation |
HH polarisation, shown in figure1.21 , is the preferred
polarisation configuration in a number
of applications. For instance, in the study
of soil moisture, if we ignore the crop
density differences, then the vertically
oriented crops (e.g., wheat and barley) have
improved penetration with HH, allowing the
backscatter to represent
the soil moisture regime better rather than
the crop geometry. HH is very suitable
for separating marine ice and water, since
it is less sensitive to water roughness than
VV polarisation, thus producing an
improved contrast between the two target
types. For a similar reason, HH is used for
ship detection.
Vertical Transmit/Horizontal Receive (VH) -
cross polarisation
|
Figure 1.22 VH Polarisation |
Horizontal Transmit/Vertical Receive (HV) -
Cross Polarisation
|
Figure 1.23 HV Polarisation |
Since the backscatter from water surfaces is
reduced under cross-polarised SAR
illumination/detection, using the VH or HV
technique is very suitable for detecting
targets on the water surface, which
accommodate multiple scattering necessary
for depolarisation. Such targets are,
for example: ship superstructures and
various ice deformations (ridging, fractures
and rubble). For a similar purpose, the
separation of broadleaf from grain
crops, for example, benefits from
cross-polarised SAR imaging, since depolarisation is much
stronger with the geometries of broadleaf
vegetation where multiple scattering of the
radar beam is much more likely. There
are also indications that the detection of
geological linears benefits from cross
polarisation when the look angle is acute.
For studies of bare soil, where attention
focuses on the retrieval of soil moisture
and soil roughness, the use of different
polarisations will improve the inversion
into soil parameters. Cross polarisation
provides an important improvement for soil
moisture retrieval since the radar
backscatter is less sensitive to surface
roughness, row direction, etc.
For many vegetation studies, the use of
different polarisations, in particular cross
polarisation, will improve the
discrimination between vegetation
(volume scattering) and soil (surface
scattering). In the case of forestry, the
use of cross polarisation will improve
the forest/non-forest discrimination and the
retrieval of low biomass values (forest
regeneration, regrowth, plantation). Two
examples are provided below (figure1.24 and figure1.25 ) showing how the
use of dual polarisation data can improve
the information content.
|
Figure 1.24 SIR-C data (C-band, 26.5°) over Les Landes test site, France. Left image is backscatter intensity (VV polarisation) and right image is HH/VV correlation image. (Acknowledgement: Souyris et al, 1998.) |
In figure1.24 above, the SIR-C
backscatter intensity image (left) shows
poor discrimination between vegetated
and non-vegetated areas. This is because of
large variations in the backscatter of bare
soil surfaces related to different soil
roughness and moisture conditions and is
similar for both VV and HH polarisation
images. However, using both
polarisations to produce a HH/VV correlation
image (below), it becomes possible to
discriminate between non-vegetated
(high-correlation) and vegetated
(low-correlation) areas. On this image, the
high-correlation (bright) areas correspond
to recently harvested cornfields. In
contrast, the different states of tillage
are seen to produce large variations in
backscatter intensity in the single-channel image.
figure1.25 , figure1.26 and figure1.27 below illustrates
differences between like-polarised and
cross-polarised images of urban areas. Three
different polarisation images are shown for
an area in southern Germany which was
imaged by the JPL AIRSAR during the MAC
Europe campaign in 1989. The area is 12 km
wide, and includes forests, cultivated
fields and urban areas. The two
like-polarised images are seen to be very
similar. However, on the cross-polarised
image, urban areas are seen to be much
less bright. This is because the
cross-polarised return only appears through
multiple scattering, while the urban areas
are characterised by man-made objects
that act like corner reflectors.
Three different C-band
polarisation images for an area in southern
Germany imaged by the JPL AIRSAR during the
1989 MAC Europe Campaign.
|
Figure 1.25 C-band HH polarisation |
|
Figure 1.26 C-band HV polarisation |
|
Figure 1.27 C-band VV polarisation |
It is possible to simulate Alternating
Polarisation images (VV/HH) using ERS and
Radarsat data. figure1.28 shows a combination
of ERS and Radarsat images taken one day
apart, for an area in Oxfordshire, UK.
In this example a large number of
agricultural fields are seen to have a blue
colour which is indicative of a high
backscatter in HH polarisation compared with
VV. Since these fields are all cereal
fields, this is a good indication of the
value of of alternating
polarisation images for improving
crop classification. Over the remainder of
the image, urban areas, woodland and
grassland all have grey tones indicating no
significant differences in HH and VV backscatter.
|
Figure 1.28 Simulated Alternating Polarisation image (VV&HH) of an area in Oxfordshire, UK. (Red & Green: ERS 26/5/97. Blue: Radarsat 27/5/97). Acknowledgement: Remote Sensing Applications Consultants, UK |
There is considerable interest in the
Alternating Polarisation Mode for sea ice
applications. From current research results
using ERS and RADARSAT data, it is still
not clear whether VV or HH polarisation is
generally better for mapping sea ice. One of
the current problems using either ERS or
RADARSAT data at low incidence angles is
that ice/water discrimination can sometimes
be poor. Alternating polarisation, HH
and VV data, will give improved ice
edge/water discrimination. Cross
polarisation data is expected to be
particularly useful for mapping ice
topography (ridging, rubble), and is also
likely to give improved ice type discrimination.
Figure1.29 below, showing ERS
and RADARSAT images from the Arctic acquired
less than 2 hours apart, provides a very
striking example of the differences that can
occur between VV and HH polarisation images,
although in this case it should be noted
that the radar viewing directions were
virtually opposite. Over much of the area
one can see large signature reversals
between the VV and HH images. Differences
show up primarily over open water. The wind
speeds recorded on board the icebreaker
Oden, 55 km away from the scene centre, were
8 and 14 m/s respectively. The incidence
angle varies between 21 and 26 degrees
over the ERS-2 sub-image and between 29 and
34 degrees over the RADARSAT sub-image. The
different VV and HH responses to wind
roughening can be used for wind
determination, and the sea ice differences
can help in classifying ice properties.
|
Figure 1.29 Almost simultaneous ERS and RADARSAT images covering 80 x 80 sq km centred around 82°N 12°E. The ERS image was acquired from a descending orbit at 12:88 and the RADARSAT (which has been rotated by 90 degrees) from an ascending orbit at 14:33, both on 19 September 1996. The images have been averaged to the same pixel spacing and intensity stretched. (Acknowledgement: J. Askne and A. Li. Chalmers Univ. of Technology, Sweden.) |
The EMAC-95 airborne radar experiment
demonstrated the value of dual polarised data
for discriminating sea ice types. Figure1.30 shows an EMISAR C-band
co-polarisation ratio VV/HH image covering
Baltic Sea Ice (Dierking et al., 1997
Ref. [1.5 ]
). The green and yellow areas on
this image have co-polarisation ratios
larger than 1, and correspond to level ice and
thin ice/open water. The largest values (green)
are associated with smooth ice surfaces. A
low co-polarisation ratio around 1 (blue) is
observed for highly deformed areas and ridges,
where the return radar signal is dominated
by coherent (specular) scattering.
|
Figure 1.30 Copolarisation ratio VV/HH image for the Baltic Sea Ice site imaged by EMISAR during EMAC-95. (Acknowledgement: Dierking et al., 1997). |
Another example illustrating the value of the
ratio VV/HH for discriminating ice/no ice is
provided in figure1.31 , which shows HH, VV
and VV/HH ratio images of a mixed pack ice
and open water scene in the Gulf of St.
Lawrence, imaged by the SIR-C radar.
Discrimination of ice/water is complicated
by incidence angle and wind conditions,
and is not always distinguishable with
either HH or VV polarisations. However,
since the ratio of VV to HH backscatter is
larger than 1 for open water, but close to 1
for pack ice, the sea ice is seen to be much
darker on the VV/HH ratio image, independent
of incidence angle or wind conditions.
|
Figure 1.31 Dual polarisation images (VV, HH and VV/HH ratio) from SIR-C for the Gulf of St. Lawrence. (Acknowledgement: L. Gray, CCRS, Canada.) |
Over the oceans, the backscattering signal is
stronger with VV than with HH polarisation.
Experimental results indicate that oceanic
features such as internal waves, fronts and
sea floor topography tend to appear somewhat
better with HH than with VV polarisation. Figure1.32 , RADARSAT HH
polarisation, shows internal waves in the
Straits of Gibraltar particularly well.
In contrast, sea surface imprints of
atmospheric features (in particular,
convective cells) appear to be more
visible with VV polarisation than with HH polarisation.
|
Figure 1.32 Internal wave packet seen on a RADARSAT image of the Straits of Gibraltar. (Acknowledgement: Space Dept., DERA, UK: Data copyright Canadian Space Agency.) |
There are many excellent examples of
atmospheric phenomena seen with VV
polarisation ERS images, such as those
illustrated in figure1.33 below.
|
Figure 1.33 ERS-1 SAR VV image (100 km x 100 km) of the Mediterranean Sea north of the Strait of Messina acquired on September 8, 1992. (Acknowledgement: W. Alpers, Univ. Hamburg. Germany.) |
To the north-west of Gioia there are surface
manifestations of a katabatic wind (bright
area). Furthermore, between the island of
Stromboli and the Sicilian coast there is a
granular pattern, which is interpreted as
sea surface "imprints" of
atmospheric convective cells. This
cellular structure is destroyed in the
vicinity of the Sicilian coast by the
katabatic wind blowing from the
mountains onto the sea. In the lower section
of the image, an oceanic internal wave train
can be delineated propagating southwards in
the Strait of Messina.
The ASAR Alternating Polarisation Mode is
therefore of strong interest for ocean
studies. Simultaneous dual polarisation
images will allow discrimination between
similar signatures of oceanic/atmospheric
features (e.g., fronts, internal waves). The
most favourable can be chosen for
detection of oceanic/atmospheric features or
for special applications. Wind vector
retrieval from SAR images and the tuning of
imaging models (e.g., in bathymetric
assessment systems) becomes easier if
backscatter variations with HH and VV
polarisations are known.
References
ASAR
Science Advisory Group, Editor R.A.Harris,
European Space Agency 1998, "ASAR
Science and Applications", ESA SP-1225
Dierking W., Askne J. & Pettersson M.I.,
1997, "Baltic Sea Ice Observations
during EMAC-95 using Multi-frequency
Scatterometry and EMISAR Datal., Workshop
Proceedings" EMAC 94/95, "Final
Results", ESA WPP-136,
September 1997.
Le Toan
T., Smacchia P., Souyris J. C., Beaudoin A.,
Merdas M., Wooding M., & Lichtenegger
J., 1994, "On the Retrieval of Soil
Moisture from ERS-1 SAR Data",
Proceedings of the Second ERS-1 Symposium
"Space at the Service of our
Environment", ESA SP-361 Vol. II, pp
883 to 888, January 1994.
1.1.5.2 Selectable Incidence Angles
|
Figure 1.34 ERS-1 image, Wien Austria, Acquired at Fucino (I), 4 Jan 1993 (red) 17 August 1992 (green) 4 May 1992 (blue). Processed by I-PAF (Copyright ESA 1993) |
The incidence angle is defined as the angle
formed by the radar
beam and a line
perpendicular to the surface at the point of
incidence. Microwave interactions
with the surface are complex, and different
scattering mechanisms may occur in
different angular regions. Returns due to
surface scattering are normally strong at
low incidence angles and decrease with
increasing incidence angle, with a slower
rate of decrease for rougher surfaces.
Returns due to volume scattering from a
heterogeneous medium with low dielectric
constant tend to be more uniform for all
incidence angles. Thus, radar backscatter has an angular
dependence, and there is potential for
choosing optimum configurations for
different applications.
The incidence angle range for each of the Swath positions and the
slightly narrower range of incidence
angles for Wide Swath and Global
Monitoring Modes are shown in table 1.3 below:
|
Table 1.3 Specifications for ASAR Image Mode Swaths (for satellite altitude of 786 km).
|
Image Swath
|
Swath Width(km)
|
Ground, position from
nadir (km)
|
Incidence Angle Range
|
Worst Case Noise
Equivalent Sigma Zero
|
IS1 |
105 |
187 - 292 |
15.0 - 22.9 |
-20.4 |
IS2 |
105 |
242 - 347 |
19.2 - 26.7 |
-20.6 |
IS3 |
82 |
337 - 419 |
26.0 - 31.4 |
-20.6 |
IS4 |
88 |
412 - 500 |
31.0 - 36.3 |
-19.4 |
IS5 |
64 |
490 - 555 |
35.8 - 39.4 |
-20.2 |
IS6 |
70 |
550 - 620 |
39.1 - 42.8 |
-22.0 |
IS7 |
56 |
615 - 671 |
42.5 - 45.2 |
-21.9 |
One significant advantage of higher incidence
angles is that terrain distortion is
reduced. This is well illustrated by a
comparison of ERS-2 and RADARSAT images
of the Zillertal region in the Austrian
Alps, shown in figure1.35 below. This region
includes narrow valleys which range from 600
m at Mayrhofen, up to 3500 m on the highest
peaks. Up to about 1900 m, the slopes
are partly forested, with alpine vegetation
(grass, sedge, etc.), rocks and moraines at
higher levels. The two images were
obtained within a few days of one another,
with a very similar viewing direction.
Looking first at the ERS-2 image,
obtained with incidence angles of 24°
to 26° from near to far range, one sees
extreme terrain distortion in the form of
severe foreshortening and layover (
brightening) of slopes facing the radar,
combined with significant lengthening of the
slopes facing away from the radar. In
contrast, these distortions are seen to be
much less in the RADARSAT image where the
incidence angle varies between 41°
and 44° from near to far range on this
image extract. One clear benefit of the
higher incidence angle is the extra
information which can be observed on the
bright steep slopes facing the radar, and
this has been found to greatly improve the
value of the image for classification of
surface classes such as moraine, bare soil
and vegetation types.
|
Figure 1.35 Terrain distortion effects on SAR images obtained with different incidence angles: Zillertal Region, Austrian Alps. Area covered is approx. 36 km x 40 km. (Acknowledgement: H. Rott, University of Innsbruck.) |
With a range of different
incidence angles available, it becomes
possible to select optimum angles for
different applications, or to use
acquisitions from two separate passes for
multi-angle analysis. In the context of
vegetation and soil applications, there
are some general points which can be made,
based on previous research results:
- For soil moisture and soil roughness
studies, the combination of different
incidence angles is of interest, with
the condition that there be short
time intervals between acquisitions.
- For agriculture, the use of particular
incidence angles will improve selective
observation of vegetation (high
incidence angles) or underlying
soil (low incidence angles).
- For forestry, the use of low incidence
angles enhances the sensitivity to
biomass, whereas the use of high
incidence angles enhances the
discrimination of forest types through
interaction with forest structure.
Figure1.36 below illustrates
the importance of the incidence angle in
isolating the radar response due to the
vegetation canopy from that of the
underlying soil. Each curve represents the
simulated response, with C-band W
polarisation, of a soybean canopy for
varying gravimetric soil moisture ranging
from 20% to 30% (from bottom to top). Whilst
the backscatter at 20° incidence angle
is still sensitive to underlying soil
conditions, that at 40° is stable and
invariant with respect to soil moisture.
|
Figure 1.36 Simulated backscatter for a soybean canopy showing increased sensitivity to soil moisture at low incidence angles. (Acknowledgement: Nghiem et al, 1993.) |
Figure1.37 below provides an
excellent example of how vegetation mapping
can be enhanced by using high incidence
angle data. This pair of RADARSAT images
shows discrimination of forest clearcuts in
Whitecourt, Alberta, an active logging
area in the foothills of the Rocky
Mountains. On the first image, acquired at
an incidence angle of 20° to
27° there is poor contrast between the
clearcuts and the forest. On the second
image, which was acquired at a much larger
incidence angle of 43° to 46°,
the dark tones of the clearcut areas
contrast strongly with the brighter returns
from the surrounding forest.
|
Figure 1.37 RADARSAT images acquired at different incidence angles (a. 20° to 27°. b. 43° to 46°), showing Forest Clearcuts in Alberta, Canada. (Acknowledgement: L. Gray, CCRS, Canada.) |
Images acquired with different
incidence angles may be used in combination
to improve land cover discrimination, but
since each image has to be acquired on a
different day, any composite image will also
include a temporal change component.
Images acquired with different
incidence angles may be used in combination
to improve land cover discrimination, but
since each image has to be acquired on a
different day, any composite image will also
include a temporal change component.
Figure1.38 below provides an
illustration of the use of multiple
incidence angles to improve land cover
discrimination for an area near Oxford, UK.
|
Figure 1.38 Multiple incidence angle image of Oxfordshire area, UK. Composite of Radarsat images: Blue: 23° - 23/3/97, Green: 37° - 13/3/97, Red: 43° - 3/3/97. (Acknowledgement: Remote Sensing Application Consultants, UK) |
In this case, 3 Radarsat images
taken within a period of 10 days have been
combined (Blue - 23° 23 rd March 97,
Green - 37° 13 th March 97, Red -
43° 3 rd March 97). Most of the
coloured areas on the image, indicative
of backscatter differences related
to incidence angle, are bare soil fields,
while grassland, woodland and urban areas
tend to have grey tones, showing a
similar backscatter at the different
incidence angles. In the northern half of
the area, which has clay soils, practically
all bare soil fields have a blue colour,
indicating higher backscatter at the lowest
incidence angle, as one would expect. In the
southern half of the area which has
chalk soils, some of the bare soil fields
also have blue colours, but some of the
fields coloured red are also bare soil
fields and this seems something of an
anomaly. Possible explanations are that
these fields have marked differences in soil
roughness, or possibly that cultivation
changes took place during the period over
which the 3 images were acquired.
Ship detection with ERS data was limited to a
certain extent by the steep incidence
angles. As illustrated in figure1.39 , RADARSAT has now
clearly demonstrated the benefits of higher
incidence angle data for the detection of
ocean-going trawlers (typically, 55 m long)
and RADARSAT images are already used
pre-operationally for monitoring fishing
activity in the Barents Sea. The pattern of
trawlers seen on this image shows a marked
concentration in International Waters along
the boundary with Norwegian Waters. Several
of the outer ASAR standard beams will be
capable of detecting trawlers, although
in a rather narrow swath. Also,
cross-polarised images from the Alternating
Polarisation Mode should further improve
detection capability at steeper incidence angles.
|
Figure 1.39 RADARSAT image showing fishing vessels in the Barents Sea, similar to what will be possible with ASARs higher incidence angles (26 km scene width). (Data copyright Canadian Space Agency.) |
Although higher incidence angles are
preferable for ship detection, wide swath
and ScanSAR images, such as that shown in figure1.40 below, can be used
across most of the incidence angle range,
giving excellent wide area coverage.
|
Figure 1.40 RADARSAT ScanSAR image of the English Channel and North Sea showing ship/oil rig detections. The enlarged inset shows a cluster of oil rigs in the North Sea. (Acknowledgement: Space Dept., DERA, UK; Data copyright Canadian Space Agency.) |
For a further discussion of this subject see
the section entitled "Ocean Applications." 1.1.6.2.4.
1.1.5.3 Wide Area Coverage and Frequency of Coverage
|
Figure 1.41 ERS-1 SAR image, Adelaide Island Antarctica Nov. 3, 1991 (Copyright ESA 1992) |
ASAR low-resolution images
provided by the Wide Swath and Global
Monitoring Modes open up new possibilities
for applications requiring large area
coverage and/or more frequent revisit. Both
modes will provide 105 km swath coverage for
applications where higher resolution is necessary
(better than a 5-day frequency).
ASAR Wide Swath is aimed
primarily at sea ice and other oceanographic
applications, where there is a special
interest in obtaining a wide area view with
high temporal frequency. (See Figure in the section
entitled "Geophysical Coverage" to
view a graphic portrayal of the comparative
swaths for the ENVISAT instruments).
figure1.42 below shows a RADARSAT wide swath image of the Gulf of St.
Lawrence on which ice types are seen in
various shades of light grey, in contrast
with water which has the darkest tones. Such
images are now used routinely by the
Canadian Coast Guard for ice breaker
operations and routing of ships in the Gulf.
|
Figure 1.42 Sea Ice Monitoring using RADARSAT ScanSAR data in the Gulf of St. Lawrence, Canada, March 6, 1996. Swath coverage is 450 km, with 250 m pixel spacing. (RADARSAT Data Copyright Canadian Space Agency/Agence spatiale canadienne 1996. Received by the Canada Centre for Remote Sensing. Processed and distributed by RADARSAT International. Imagery enhanced and interpreted by CCRS.) |
The ASAR Global Monitoring images, with low
data rates, promise to be particularly
valuable for sea ice mapping over extensive
areas. figure1.43 below shows a
simulation of an ASAR Global Monitoring Mode
image for regional ice reconnaissance.
The simulation is based on a RADARSAT
ScanSAR wide, C(HH), image of the southern
Beaufort Sea. The original 100 m
resolution, 500 km image, which was acquired
on October 11, 1996, shows the Mackenzie
delta and Tuktoyuktuk peninsula to the
south, the new ice and open water in the
lead just north of the shore-fast ice, and
the pack ice to the north occupy most of the
image. The simulation was obtained by
reducing the swath from 500 km to 400 km,
then sub-sampling and smoothing the image to
simulate both the spatial resolution (1
km) and the radiometric resolution of the
ASAR Global Monitoring Mode. The simulation
shows the potential value of the
resulting product. Although the transition
from land to shore-fast ice is not distinct,
there is still sufficient detail in the ice
imagery to recognise areas of slightly
lower pack ice concentration (to the east),
and to recognise the westward drift of the
pack ice just north of the shore-fast
ice. This is consistent with the normal
clockwise ice movement in the Beaufort Sea gyre.
|
Figure 1.43 ASAR Global Monitoring Mode simulation using RADARSAT ScanSAR wide data. (Acknowledgement: L. Gray, CCRS.) |
The frequent large area coverage which ASAR
is able to provide is also important for
monitoring ice sheets. Of course, ERS
already provides frequent revisits of
polar regions and the value of this has been
well demonstrated in a study of the collapse
of the northern Larsen Ice Shelf,
Antarctica (Rott et al, 1996). Figure1.44 below shows two
ERS-1 SAR images of the Larsen Ice Shelf.
The first (left image) is a strip of three
100 km x 100 km images from a descending
pass on January 30, 1995, in which the
ice sheet can be seen to be breaking up.
Looking at the second (right) image, which
was taken just five days earlier, one
sees the ice shelf still largely intact.
Such images provide valuable information on
the timing and rates of change, in this
case illustrating the extremely rapid
disintegration of the ice shelf over a few
days at the end of January 1995. ASAR will
be able to provide daily coverage of
such phenomena in the polar regions and
another important advantage will be the
on-board storage capability. For
Antarctica, the O' Higgins ERS
receiving station currently operates only
for two 5-week periods per year.
|
Figure 1.44 ERS SAR images of the Larsen Ice Shelf, Antarctica. The left image is a strip of 3 standard 100 km x 100 km images (descending pass) taken on January 30, 1995, showing break-up of the ice shelf. The right image is another strip of 100 km images (ascending pass) taken on January 25, 1995, when the ice shelf can be seen to be largely intact. (Acknowledgement: H. Rott, University of Innsbruck, Austria.) |
Over the land, interests focus on the
potential of low-resolution SAR data for
soil moisture and vegetation monitoring.
Previous work carried out over land using ERS
Wind Scatterometer data has already shown
that low-resolution measurements, in this
case 50 km spatial resolution, can
provide useful information concerning
vegetation dynamics and freeze/thaw on a
continental and regional scale (Wismann
& Boehnke, 1994). Figure1.45 shows how the ERS
Wind Scatterometer has been used to
monitor seasonal variations over the African
continent. Besides the scatterometer image
for summer 1993, the so-called
Hovmoeller diagram shows a slice through
Africa from 35°N to 35°S extending
longitudinally from 20° to 26°E.
Monthly averages of the radar intensity
are plotted for the period from 1991 to
1997. The predominant signal in the
Hovmoeller diagram is the annual
variation in radar backscatter in the
savannah region north and south of the rain
forest, and it can be seen how the pattern
of increased backscatter in 1992 is
repeated every year. The much better
resolution of the ASAR Global Monitoring
Mode will provide a significantly
improved capability for continental or
regional scale measurements.
|
Figure 1.45 ERS-1 scatterometer map of Africa and a Hovmoeller diagram for a slice through Africa from 35°N to 35°S at a longitude of 20°E. Monthly averages are plotted for 1991 to 1997. (Acknowledgement: V Wismann and H. Boehnke, WARS, Germany.) |
The availability of
approximately 5-day revisit coverage (in
Central Europe) using Image Mode promises to
be particularly important for flood
mapping. For floodplain mapping and
emergency management, a resolution of around
30 m is required, with frequent
coverage. The much improved temporal
coverage possible using ASAR is vital for
developing operational systems. In addition,
the high incidence angles and HH
polarisation capabilities of ASAR will give
better mapping of flood extent.
For applications where this
improved temporal coverage is important, the
main issue then becomes the utility of data
acquired at a wide range of different
incidence angles. Combined use of images
acquired with different incidence angles
poses a new set of challenges.
1.1.5.4 Interferometry
|
Figure 1.46 ERS-1 Interferogram, Feb. 7 2002, Bay of Naples/Vesuvius Italy (Copyright (c) 2001 by SRC SASA, original data by ESA) |
1.1.5.4.1 Principles
As was discussed in the section entitled
"Scientific
Background" 1.1.2.3. , a SAR works by
illuminating the Earth with a beam of coherent microwave
radiation, retaining both amplitude
and phase information in the radar echo
during data acquisition and subsequent
processing. This radiation can be
described by three properties:
-
Wavelength - the
distance between peaks on the wave.
-
Amplitude - the
displacement of the wave at the peak.
-
Phase - describes
the shift of the wave from some
other wave. Phase is usually
measured in angular units, like
degrees or radians.
Synthetic Aperture Radar (SAR)
interferometry exploits this coherence, using
the phase measurements to infer
differential range and range change in
two or more complex-valued SAR
images of the same surface, thereby
deriving more information about an
object than is obtainable with one
single image.
|
Figure 1.47 Phase shift |
The resulting difference of phases is a
new kind of image, called an
interferogram, which is a pattern of
fringes containing all of the
information on relative geometry. Figure1.48 below, offers
an example of such an image.
|
Figure 1.48 Radar interferogram of a portion of the Rutford ice stream in Antarctica, based on two ERS-1 images taken six days apart. The fringe pattern (colour cycle) is essentially a map of ice flow velocity, with one fringe representing 28 mm of range change along the radar line of site. (Image courtesy Jet Propulsion Laboratory, California Institute of Technology) |
For a second SAR image to provide
additional information, it must be
acquired from a different sensor
position or at a different time. The
difference between the acquisitions of
the first and second images determines
the type of interferometer that results.
Some of the most common forms are:
-
Across-track - used
primarily for topographical
information, this type utilises a
difference in across-track
position, or look angle.
-
Along-track - used
primarily for ocean currents
information and moving object
detection, this type utilises a
difference in the along-track
position, which can be achieved by a
small difference in
acquisition time, on the order of
microseconds to seconds.
-
Differential - this
method utilises a difference in
time, on the order of days to years,
and is used primarily to
observe glacier (ice field) or lava
flows, if the time difference is
within days. If the time difference
is measured in days to years, it
can be a very useful method of
observing subsidence,
seismic events, volcanic
activity, or crustal displacement.
Each of these types will be touched on
briefly below.
1.1.5.4.2 Across-track Interferometry (InSAR)
The best known application of SAR
Interferometry is the reconstruction of
the Earth topography by using different
look angles to
compare the same object. This is what is
referred to as across-track
interferometry. Across-track is also
known as the range direction,
defined as the dimension of an image
perpendicular to the line of flight of
the radar.
Consider two radar antennas, A1 and A2,
simultaneously viewing the same surface
and separated by a baseline vector
B with length
B and angle
with respect to horizontal, as
shown in figure1.49 below. A1 and
A2 may also represent a single antenna
viewing the same surface on two
separate passes.
|
Figure 1.49 Basic imaging geometry for InSAR |
A1 is located at height h
above some reference surface. The
distance between A1 and the point on
the ground being imaged is the
slant range
, while
is the distance between A2 and
the same point.
In the case of
simultaneous imaging from two
separate antennas, one antenna both
transmits and receives the radar
signal. This antenna is known as the
master. The second antenna,
know as the slave,
only receives. This method is
sometimes referred to as
single-pass interferometry.
In the case where a
single-antenna SAR system revisits
the same position and images the
same area on the ground after
several days or weeks, the
repeat-pass
interferometry method is
used. With this method, each antenna
acts as both transmitter and
receiver, as depicted in figure1.50 below.
|
Figure 1.50 InSAR Data Collection |
The phase of pixel
value in a complex SAR image depends
on the scattering mechanism in
the resolution cell, and the
distance from the antenna to the
point. If the scattering mechanism
in the two images is similar,
then the phase difference between
the two complex SAR images is
proportional to the difference in slant range
from the two antenna to the point.
The similarity in scattering
mechanism in the two images is
indicated by the correlation
coefficient of the image, which is
also called coherence in SAR
interferometry literature. Any
dissimilarity of the
scattering mechanism between the two
images, indicated by a low
coherence, results in phase noise. A
certain loss of coherence
results from the different look angles
from two antenna to the point,
and from receiver noise. Coherence
loss can also result from changes in
the surface between acquisitions, in
the case of repeat-pass interferometry.
The difference between
and
can be measured by the phase
difference between the two complex SAR
images. That is, by multiplying one
image by the complex
conjugate of the other image, an
interferogram is formed whose phase
is proportional to the range
difference to the point. The phase
of the interferogram, displayed as
an image, contains fringes that
trace the topography like
contour lines, as shown in figure above.
Interferograms are often displayed by
showing the phase difference as
colour, and the SAR amplitude as
brightness (see figure1.51 below).
|
Figure 1.51 Interferogram creation |
SAR Amplitude (0..300) +
Phase Difference (-pi..pi) =
The phase of the above interferogram
shows the topography of an imaged mountain.
The slant range difference is
proportional to the full phase or
absolute phase of the complex-valued
interferogram. However, the
measured phase values of the
interferogram can only take values
between 0 and 2
. That is, the phase is
'wrapped'. Thus, in order
to compute the slant range
difference which is needed to
compute topography, the 2
ambiguity inherent in the phase
measurements must be solved,
using techniques of 'phase-unwrapping'.
Once the absolute phase of each pixel
of the interferogram is known, the
geometry of the figure can be used
to compute the topography
z(y), if the baseline
vector (vector B in
figure1.49 ) is known.
The result is a Digital Elevation
Model (DEM) of the
observed area.
1.1.5.4.3 Along-track Interferometry
Along-track (azimuth)
interferometry uses two antennas; the
master that transmits and receives,
and the slave that receives only. Such a
system takes two images of the same
target, with a time delay that results
in an along-track difference in
position. Typically, this time delay is
between 10 microseconds (ms) and 100 ms.
If the target remains stationary
between acquisitions, the two data sets
are ideally identical (i.e. in the
absence of any system phase noise) and
the interferometric phase is zero.
However, any relative range shift of the
targets between the two images will
result a non-zero interferometric
phase. The along-track interferometric
method is most often used when detecting
relatively fast motion, such as ocean currents.
1.1.5.4.4 Differential Interferometry (D-InSAR)
The temporal separation in repeat-pass
interferometry of days, months,
or even years, can be used to advantage
for long term monitoring of geodynamic
phenomena, in which the target has
changed position at a relatively
slow pace. This would be true when
monitoring glacial or lava flows. It is
also useful for analysing the results
of single dramatic events, such as
earthquakes. If two acquisitions are
made at different times from the same
position, so there is no across-track baseline,
then the phase of the
interferogram depends only on the change
in topography between the
acquisition times. In general, a
difference in across-track, (azimuth), position of
the acquisitions also exists. In this
case, multiple acquisitions can be made
to measure the topography, and
measure the change in topography
(differential effects) over time.
The block diagram below depicts a typical
D-InSAR processor using one SAR and one
ScanSAR image, that
incorporates a Digital Elevation
Model (DEM).
|
Figure 1.52 Block diagram of a typical D-InSAR processor. |
In the 35 day repeat orbit scenario, the
wide swath capabilities of ASAR,
combined with the predictable high
reliability and repeatability of the
orbits, will allow the retrieval of
extremely useful data for differential
interferometry (D-InSAR). Another
important benefit of ASAR will be the
availability of different viewing angles
and in particular higher off-nadir
angles, that enhance the
interferometric visibility of steeper
slopes, which are otherwise in layover with the
23° incidence angle of
ERS-1/2. As for ERS, in order that
terrain can be imaged with
differential interferometry the surface
conditions should be stable enough so
that more than 30-40% of the strong
scatterers remain unchanged in two
images acquired 35 days apart.
ASAR will be able to measure very small
terrain displacements due to co-seismic
motions, subsidence, volcanic
upswelling, landslides, ice movement and
possibly oscillatory effects like earth
tides and the loading of sea tides on
the continental shelf. ( See section
entitled "Land
Applications" )
An excellent example of the detection of
surface subsidence is shown in figure1.53 below. The
image displays an exaggerated three
dimentional (3D) perspective view of the
Belridge (middle left) and Lost
Hills (lower left) Oil Fields,
California, viewed from the north-west.
Both oil fields are located in the San
Joaquin Valley. The surface
deformation derived from ERS-1 data
collected in September and November 1992
(70 days time difference), shows
subsidence of up to 6 cm.
|
Figure 1.53 Subsidence in the Belridge Oilfields, California. The colours derived from ERS-1 data collected in September and November 1992 (70 days time difference) show subsidence ranging from 1 cm (blue) to 6 cm (red-brown). The 3D view has been produced using the DEM generated from the ERS tandem pair, combined with a Radarsat image. (Acknowledgement: M. van der Kooij, Atlantis Scientific Inc.; Radarsat Data Copyright Canadian Space Agency/Agence spatiale canadienne 1996) |
Usually, D-InSAR surveys are
generated starting from two full resolution
SAR images. Yet, it is possible to combine
low-resolution, Wide Swath (WS) images
with full resolution ones to give high
quality D-InSAR images. See Low Resolution
Interferometry 1.1.5.4.6. below.
1.1.5.4.5 Coherence Evaluation
Coherence, when
associated with interferometry, is
related to phase variance
between the two SAR images. For the
purpose of processing the interferometry
data into topography of motion
information, the coherence can be a
useful tool in indicating areas of
noisy phase. For example, during phase unwrapping,
areas of noisy phase - as indicated
by a low value of coherence - can be avoided.
In addition, coherence is another useful
parameter that can be extracted from repeat-pass
interferograms. Coherence provides
information on stability over time, or
temporal stability, and is therefore
an important feature for land cover
classification. Temporal decorrelation
can be caused by such things as changes
in vegetation, freezing and thawing,
or human activities like plowing. All of
these changes are observed over periods
of days to years, whereas some
changes to water surfaces can occur in a
matter of milliseconds.
1.1.5.4.6 Low Resolution Interferometry
Pairs of complete Wide Swath (WS) or
Global Monitoring (GM)
Mode images will be unsuitable for
interferometry because, in the lower
resolution modes, the data are
sampled in bursts along the azimuth direction
(along-track). For interferometry
the sampling bursts need to be spatially
aligned in the two interfering frames.
However, there are interesting
possibilities for using low
resolution images in conjunction with
full resolution, Image Mode (IM)
images. A full resolution image with the
same incidence angle as
the subswath of the Wide Swath or
Global Monitoring Mode image is needed.
Once a worldwide archive of
Wide Swath or Global Monitoring Mode
images is built up it will be possible
to obtain interferometric pairs of
particular areas of interest within 35
days, by acquiring full resolution (IM)
data with the same imaging geometry as
the relevant portion of the low
resolution image. Because the previously
acquired burst images are of low
resolution, the quality of the fringes
will be proportionally lower than if
two full resolution images had been used.
Figure1.54 is a simulation
of co-seismic motion
retrieval using low resolution
differential interferometry. In this
example ERS SAR images for the
Landers 1992 earthquake have been used
to simulate interferograms, using an
Image Mode image together with
either Wide Swath Mode or Global
Monitoring Mode images. On these images
one fringe corresponds to half a
wavelength displacement along the
radial direction.
|
Figure 1.54 Simulation of low resolution interferograms of the Landers 1992 earthquake, using ERS SAR data. Image Mode (right), Wide Swath Mode (top left) and Global Monitoring Mode (bottom left), for an area of 48 km x 45 km. (Acknowledgement: F. Rocca, Politecnico di Milano, Italy). |
1.1.5.4.7 Conclusions
The above discussion of interferometry
only scratches the surface of a rich and
complex branch of remote sensing and the
list of references provided below
offers just a small portion of the
wealth of material dedicated to this
expanding area of research.
It is now well established that SAR
interferometry provides a valid form of
geophysical measurement. The huge
archive of data acquired by the C-band sensors ERS-1 and
ERS-2 has imposed a de facto
standard for interferometry. The
Advanced Synthetic Aperture Radar (ASAR)
sensor, which is also a C-band
instrument, will provide considerably
higher flexibility compared with the two
ERS SARs. ASAR provides a choice of two
polarisations ( out of
H, VV, and VH ) as well as a variety of
imaging modes, such as different incidence angles,
standard and ScanSAR wide swath
modes. ( Refer to the other subsections
within "Special Features
of ASAR" 1.1.5. for more
information on these topics ). In
particular, since ASAR can provide
higher off-nadir angles than was
the case with the ERS-1/2 sensors,
the problems of layover will be
reduced, thereby enhancing the
interferometric visibility of steeper slopes.
1.1.5.4.8 References
ASAR Science Advisory Group, Editor
R.A.Harris, European Space Agency
1998, "ASAR Science and
Applications", ESA SP-1225
Bamler, R., P. Hartl,
"Synthetic aperture radar
interferometry", R. Bamler, P.
Hartl, Inverse Problems 14, pp.
R1-R54, 1998
Dixon, T. H, "Report of a
Workshop Held in Boulder, Colorado :
February 3-4, 1994",
prepared by the Jet Propulsion
Laboratory, California Institute of
Technology, under a contract with
the National Aeronautics and Space Administration.
Feigl, K. L., A. Sergent, and D.
Jacq, "Estimation of an
earthquake focal mechanism from
a satellite radar
interferogram": application to
the December 4, 1992 Landers
aftershock, Geophys. Res. Lett., in
press, 1994.
Geudtner, D., "The
interferometric processing of ERS-1
SAR data", European Space
Agency, Technical Translation of
DLR-FB 95-28, ESA-TT-1341, 1995.
Goldstein, R. M., H. Engelhardt, B.
Kamb, and R. M. Frolich,
"Satellite radar
interferometry for monitoring ice
sheet motion: application to an
Antarctic ice stream", Science,
262, 1525-1530, 1993.
Goldstein, R. M., H. A. Zebker, and
C. Werner, "Satellite radar
interferometry:
two-dimensional phase
unwrapping", Radio Science, 23,
713-720, 1988.
Graham, L. C., "Synthetic
interferometer radar for topographic
mapping", Proc. IEEE, 62,
763-768, 1972.
Li, F., and R. M. Goldstein,
"Studies of multibaseline
spaceborne interferometric
synthetic aperture radars",
IEEE Trans. Geosci. Remote Sensing,
28, 88-97, 1990.
Massonet, D.,
K.L.Feigl, "Radar
interferometry and its application
to changes in the earth
surface", Reviews of Geophysics
Vol. 36, Number 4, Nov. 1998, pp.441-500.
Massonnet, D., M.
Rossi, C. Carmona, F. Adragna, G.
Peltzer, K. Feigl, and T. Rabaute,
"The displacement field of the
Landers earthquake mapped by radar
interferometry", Nature, 364,
138-142, 1993.
Massonnet, D., K.
Feigl, M. Rossi, and F. Adragna,
"Radar interferometric mapping
of deformation in the year
after the Landers earthquake",
Nature, 369, 227-230, 1994.
Massonnet, D., and
K. L. Feigl, "Discriminating
geophysical phenomena in satellite
radar interferograms", Geophys.
Res. Lett., in press, 1995a.
Monti Guarnieri, A.,
Prati, C., Rocca, F., and Desnos,
Y-L, "Wide Baseline
Interferometry With Very Low
Resolution SAR Systems",
Dipartimento di Elettronica e
Informazione - Politecnico di
Milano, ESTEC ( available
on-line at: http://www.elet.polimi.it/users/dei/sections/telecom/andrea.montiguarnieri/papers/ceos/)
Peltzer, G., K.
Hudnut, and K. Feigl, "Analysis
of coseismic surface displacement
gradients using radar
interferometry: new insights into
the Landers earthquake", J.
Geophys. Res., 99, 21971-21981, 1994.
Rodriguez, E., and
J. Martin, "Theory and design
of interferometric SARs", Proc.
IEEE, 139, 147-159, 1992.
Ref. [1.22 ]
Solaas, G.A.,
"ERS-1 interferometric
baseline algorithm
verfication", ESA Tech. Note
ES-TN-DPE-OM-GS02, 69 p., 1994.
Zebker, H., and R.
Goldstein, "Topographic mapping
from interferometric synthetic
aperture radar observations",
J. Geophys. Res., 91, 4993-5001, 1986.
Zebker, H., and J.
Villasenor, "Decorrelation in
interferometric radar echoes",
IEEE Trans. Geosci. Rem. Sensing,
30, 950-959, 1992.
Zebker, H., P.
Rosen, R. Goldstein, A. Gabriel, and
C. Werner, "On the derivation
of coseismic displacement
fields using differential radar
interferometry: the Landers
earthquake", J. Geophys. Res.,
99, 19617-19634, 1994a.
Ref. [1.25 ]
Zebker, H.A., C.
Werner, P.A. Rosen, and S. Hensley,
"Accuracy of topographic maps
derived from ERS-1 interferometric
radar", IEEE Trans. Geosci.
Rem. Sens., 32, 823-836, 1994c.
1.1.5.5 Wave Spectra
|
Figure 1.55 ERS-1 SAR image, Dundas Peninsula (in the Parry Islands of northern Canada), Aug. 17, 1994 (Copyright ESA 1994) |
The exchange of energy between the ocean and
atmosphere, between the upper layers of the
ocean and the deep ocean, and transport
within the ocean, all have a role in
controlling the rate of global climate
change and the patterns of regional change.
Long continuity of measurements of sea
surface temperature, winds, topography,
geostrophic currents and ocean colour is essential.
Figure1.56 below, of the Gulf
of Gaeta, Italy, shows the first
multitemporal ERS-2/ERS-1 image acquired at
Fucino (I) and Kiruna (S), processed by
ESA/ESRIN: 578x900 pixels, 475 Kb. The ERS-1
and ERS-2 images of May 1st and 2nd
reveal a difference in soil moisture due to
changed weather conditions, shown by the
greenish land colour. The colours of the
sea correspond to different wind and current
conditions during the three acquisitions,
while black indicates calm areas at all
acquisition dates.
|
Figure 1.56 Gulf of Gaeta, Italy ERS-1 ( blue: acquired Mar 27, 1995. green acquired May 1, 1995 ) and ERS-2 (red: acquired May 2, 1995) |
ASAR products of interest to ocean scientists
include wind speed and wave spectra from the
Wave Mode. Other ASAR modes are of
interest for wind field measurement, studies
of internal waves and eddies and the detection of atmospheric
phenomena, with Wide Swath and Global
Monitoring Modes being of particular
interest because of the larger area and
more frequent coverage.
ASAR Wave Mode will provide wave spectra
derived from imagettes of minimum size (5 km
x 5 km), similar to the ERS AMI Wave Mode, spaced 100
km along-track in either HH
or VV polarisation. The position of the
imagette across-track can be
selected to be either constant or
alternating between two across-track
positions over the full swath width.
ERS Wave Mode products are based on image
spectra (wave number and direction)
estimated from SAR intensity imagettes using
standard Fourier Transform techniques.
These products are therefore symmetric
containing 180° propagation ambiguity.
Techniques involving the use of Wave
Model predictions have been developed to
solve the ambiguity problem, though this can
be subject to error when opposite or near
opposite wave components exist.
For ASAR, this problem was solved by using
the new wave product preserving the phase
and a new algorithm called "inter-look
cross spectral processing," whereby
information on the wave propagation
direction is computed from pairs of
individual look images separated in time
by a fraction of the dominant wave period.
(Engen & Johnsen, 1995.) Figure1.57 shows a simulated
ASAR Wave Mode Spectrum, with the top left
plot being the real part of the cross
spectrum (symmetric and equivalent to an
ERS product) and the top right plot the new
imaginary part (asymmetric, giving wave
propagation direction).
|
Figure 1.57 SAR ocean image cross spectrum (real and imaginary part) processed from ERS-1 data using the ENVISAT ASAR Wave Mode Cross Spectra algorithm. The corresponding directional buoy spectrum is also shown. (Acknowledgement: NORUT IT, Norway.) |
In the above example, the output from the new
algorithm is seen to correspond with the
wave direction provided by buoy
measurements, as shown in the bottom of
the figure.
Figure1.58 shows an example of
a Level 1B ASAR wave spectra, as
displayed by EnviView 1.2.1.2. . Note that the
image is anti-symmetric in this case. The
image is a polar plot of ASAR wave spectrum
data placed into discrete bins. Each wave
spectra image is composed of 864 bins
arranged in a polar plot.
- The first bin (1,1) is at the top (from
0° to 10°) of the image.
- The spectrum image is divided up into 36
sectors or 'spokes', each
10° wide. Note that the data is
redundant due to the symmetry of
the spectra, so only data from the first
180° of the spectra are provided
(sectors 1 to 18). The other half of the
spectrum display (sectors 19 to 36) can
be constructed from the first half, as
the spectrum is either symmetric or anti-symmetric.
- Each radial sector or 'spoke
'contains 24 bins, numbered from 1
at the outside edge of the spectra
display to 24 at the centre.
|
Figure 1.58 Level 1B ASAR wave spectra image. Each bin is colour-coded based on the value of the bin data mapped onto a colour scale shown to the right-hand side of the image. |
Figure1.59 below provides the
Wave Mode Cross Spectra Format
|
Figure 1.59 Wave Mode Cross Spectra Format |
The method for reconstructing the entire
cross spectrum from that stored in the cross
spectum Measurement Data Set
(MDS) is shown below in Figure1.60 .
|
Figure 1.60 Method to reconstruct full cross spectra from product cross spectra |
1.1.5.6 Simultaneous Observations
The ENVISAT Mission offers simultaneous
acquisition of ASAR and MERIS data, and this
promises to be particularly valuable for
ocean and coastal studies. With MERIS
having a swath width of 1150 km around
nadir, simultaneous data acquisition is
possible up to ASAR incidence angles of
approximately 34°, therefore including
IS1 to IS5 in Image Mode, and all but the
outer edge of the low-resolution modes. It
will be possible to use ASAR data together
with AATSR data, but
simultaneous data acquisition is restricted
to a narrow overlapping swath.
Figure1.61 provides an
illustration of the type of simultaneous SAR
and optical large area data acquisition
that will be possible. In this example a low
pressure system is seen on both a Wide Swath
Radarsat image and a NOAA AVHRR image acquired
within 30 minutes of each other. On the SAR
image one sees the differences in sea
surface roughness associated with
the depression, while on the AVHRR image
the same feature is depicted through cloud patterns.
|
Figure 1.61 A low pressure system seen on Radarsat Wide Swath (left) and NOAA AVHRR (right) images acquired on 30/3/97. Area shown on the Radarsat image is 500 km x 700 km. (Radarsat Data Copyright Canadian Space Agency/Agence spatiale Canadienne 1996. Received by the Canada Centre for Remote Sensing. Processed and distributed by Radarsat International. Imagery enhanced and interpreted by CCRS) |
Figure1.62 below shows the
combination of a SAR image and satellite
derived sea surface temperatures. The
ERS and AVHRR images were acquired
on 3 rd October 1992, off the west coast of
Norway. In the AVHRR IR image the surface
temperature decreases from nearly
14°C (white) in the coastal water to
12°C (purple) in the Atlantic water
offshore. The pattern of the sea surface
temperature field with the curvilinear
temperature fronts represents meso-scale
variability of 10 to 50 km,
characteristic of the unstable Norwegian
Coastal Current (Johannessen et al., 1994).
The ERS image, acquired 7 hours later,
contains frontal features at a scale,
configuration and orientation that are in
good agreement with those seen in
the IR image. The SAR image shows both
bright and dark radar modulations of various
width across the boundaries, which clearly
show current boundaries including meanders.
|
Figure 1.62 Comparison of a 1 km resolution AVHRR image acquired at 14:20 on 3/10/92 (white is 14°C and purple is 12°C; + denotes buoy position; land is masked in green and clouds in black) and a 100m resolution ERS-1 SAR ( Copyright ESA 1992 ) image acquired at 21:35 on 3/10/92. Both images cover the same 100 km x 300 km region off the west coast of Norway between 59°N and 62°N. (Acknowledgement: J.A. Johannessen et al., 1994). |
In figure1.63 below, the
temperature information provided by the
Advanced Very High Resolution Radiometer
(AVHRR) satellite combined with the
resolution of ERS-1 SAR data, both acquired
on February 24, 1992, aids in
calculating heat fluxes and in deriving salt
fluxes for coastal shelf processes and
global climate models. In the composite
image of St. Lawrence Island shown
below, the land mask is shown in blue.
|
Figure 1.63 Combined AVHRR and ERS-1 Image of St. Lawrence Island Alaska (image courtesy of the Alaska SAR Facility). |
References
ASAR
Science Advisory Group, Editor R.A.Harris,
European Space Agency 1998, "ASAR
Science and Applications", ESA SP-1225
Johannessen J.A., Digranes G., Esdedal H.,
Johannessen O.M., Samuel P, Browne D, &
Vachon P., 1994, "ERS-1 SAR Ocean
Feature Catalogue", ESA SP-1174.
October 1994.
|