cier
e
FEES
300 00
ing:
oherency
land (c);
c profile
ferogram
s clearly
least, 15
ntains on
nd main
value of
elevation
ferogram
fferences
between the elevations dereived from interferograms and those
given in the topographic map of the archipelago (620 m and 606
m, respectively) do not provide any evidence for significant
changes at the tops of these islands.
Table 6. Some interferometric data of the highest tops in FJL
Object 0. ? B.,m | em K Ah, m
Peak Parnas | 22.287 56.8 195.6...1.3.17..]; 619.5
Wullerstorf | 19.907 55.9 167.8 | 3.60 | 604.0
3.4. Combined topographic-glaciological interpretation of
interferometric and stereophotographic models
Apart from the high metric feasibilities, the multi-looked INSAR
products, including amplitude image, coherency image and
fringe image (interferogram itself), contain an important
qualitative information about glacial landscapes and their
changes. Interpretation of glacial topography and main structural
zones of the glacier surface in amplitude radar images has been
earlier considered in (Kostka, Sharov 1996, a).
A coherency image results from co-registration of radar images
on the basis of a correlation procedure and represents complex
correlation coefficients in shades of gray (Lado et al. 1996). Any
decorrelation in the interferometric phase caused by temporal
changes of the Earth’s surface reduces the coherency, which
allows time-stable and variable areas to be distinguished. All
objects ungergoing spatial or/and physical changes during the
time interval between surveys, such as sea surface, areas with
high rates of glacial flow, hydrographic features, etc., are
reproduced by dark-gray values in the coherency image. Sea ice
floes in the straights of the archipelago are invisible, and the
shoreline even of small islets is well detectable (See Fig. 9, b).
With the exception of several islands with distinctive vegetation
cover, e.g. Brosch Island, all ice-free land areas are represented
with a coherence value larger than 0.6. Lines of ice divides like
those shown in Fig. 7 are also characterized by a high degree of
coherence and are clearly visible in the coherency image of 3-
4.09. 95.
A new islet with coordinates 80°36.5’N, 56°33’E, which
appeared close to Champ Island due to glacial retreat, was at first
discovered in the lab by means of such imagery. Figure 10
shows the radiometric profile taken across this islet called Radar
Island. However, the real presence of this object has yet to be
verified by field observations.
Careful comparison between optical, interferometric and
cartographic data has additionally revealed essential changes in
both ice shores and ice-free coastal areas at Hall, La Ronciere,
Payer and Wiener Neustadt islands. The joint interpretation of
radar amplitude and coherency images allowed the sea ice
attached to the ice shore to be recognized, and provided essential
evidence for the presence of several floating ice shelves on Hall
and Salisbury islands, Prince George and Wilczek lands. On the
other hand, it was proved that the steady areas of unbroken sea
ice in De Long Bay and Rhodes Channel are of temporary
character and can not be reckoned as floating ice shelves. The
areas of numerous icebergs close to the front of the largest outlet
glaciers and inland borders of outlet glaciers could be reliably
delineated. The comparison of relative rates of glacial flow was
also possible.
However, a positive interpretation of INSAR products js not
always obvious in the High Arctic due to their changeable
appearance depending on the state of the surface,
hydrometeorological conditions and INSAR imaging geometry.
Figure 11 represents, for example, two interferometric fragments
of a huge Vostock-1 ice dome situated on La Ronciere Island
with a maximum height of 431 m a.s.l. SAR images used for the
generation of interferograms were obtained under different
weather conditions with heavy Ns-As clouds, precipitation and
average cloudiness of 8 on 3-4. Sepetember 1995 (a) and steady
anticyclonic weather with a few Sc clouds and average
cloudiness of 3 on 9-10. October 1995 (b).
Narrower interferometric fringes in Fig. 11, b) are due to the
larger length of baseline. In general, the baselines of about 100-
200 meters are believed to be the most preferable for INSAR
topographic modelling in the High Arctic. The difference in
average radiometric contrast of interferograms with somewhat
higher contrast for the right fragment did not exceed several
percents, thus being negligible. However, one can see diffuse
edges of fringes (phase noise) in Fig. 1l, a), which may be
explained by the lack of coherency because of changeable
hydrometeorological conditions. Even thick clouds are invisible
in radar imagery, and additional optical images taken
simultaneously over the same area are needed for the evaluation
of weather conditions during INSAR surveys.
o e RT EL Gr SR ORR,
a0 a t Xi
Fig. 10. Radiometric profile taken across Radar Island
Fig. 11. Two interferograms of La Ronciere Island generated
from SAR data taken in September 1995 with B, & 56 m (a) and
in October 1995 with B, & 150 m (b)
Ground resolution of satellite SAR images is still significantly
lower than that of spaceborne photographic imagery, which
doesn't allow them to be separately used for accurate delineation
of small topographic objects. The identification of the highest
positions in rocky areas is also rather difficult on the basis of
SAR data. Precipitous coastlines facing away from the sensor are
practically undetectable in INSAR products. Ice borders on dry
land are better detectable in spaceborne photographs. The same
applies to the rocks croping out at the glacier surface in the
accumulation zone and meltwater ponds in the ablation zone.
The reliable interpretation of areas of pioneer vegetation in the
High Arctic could only be done by mere chance with KATE-200
multispectral photographs (Sharov 1997, b). Joint analysis of
radar data and spaceborne stereophotographs is also required in
order to detect areas of superimposed ice and to determine the
position of the snowline.
Intemational Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 209
