International Archives of Photogrammetry and Remote Sensing. Vol. XXXII Part 7C2, UNISPACE III, Vienna. 1999
89
I5PR5
UNISPACE III - ISPRS/EARSeL Workshop on
“Remote Sensing for the Detection, Monitoring
and Mitigation of Natural Disasters”
2:30-5:30 pm, 22 July 1999, VIC Room B
Vienna, Austria
In the context of natural disasters the three-pass interferometry is
useful for determining slight movements on ground which may
be made at an accuracy of a fraction of a wavelength. Thus, slid
ing land surfaces ltave been determined with movements down to
about 1 cm when ground markers were present. Another applica
tion is for mapping of subsidence of the earth surface due to
water pumping or abandoned mines. Also, in these cases accu
racy’s of the order of 2 to 5 cm have been reported. Monitoring
volcanic activities may also benefit from such measurements of
variations in surface elevation.
The techniques require that the surface does not change between
the two or three passes of the satellite, i.e. the surface shall ex
hibit signal coherence. This condition is difficult to achieve when
the ground is covered by vegetation or is snow covered. Even
movements of the vegetation with the wind may be detrimental.
Also, atmospheric variations and ionospheric disturbances may
destroy the coherence of the signals. To overcome the former
difficulty night-time measurements are preferable. To ensure
surface coherence, the ERS satellites were operated at a one-day
repeat cycle during the Tandem Missions for DEM compilations.
Literature describes many applications in connection with earth
quakes for studies of the after-effect, the so-called co-seismic
deformation of tire surface. In these cases, the techniques relies
upon satellite passes at many days interval tliat might even go
beyond 400 days. An excellent discussion of this is given by
Massonet and Feigl (1998), including the limitations and the
inaccuracies of the techniques and examples of successful but
also unsuccessful cases and pitfalls, They show examples ob
tained by a two-pass technique developed by CNES using a
DEM obtained from SPOT data. The likelihood of surface coher
ence is larger in this case than in the case of three-pass interie
ro metiy.
Long-term monitoring may be made with satellite InSAR tech
niques. However, aircraft InSAR with three spaced antennas
enables real-time interferometry of small movements, for in
stance, thus overcoming the problem of surface coherence.
Applications and limitations
In the previous sections a number of applications of active mi
crowave systems have already been mentioned and limitations of
a general nature have been discussed. In tire subsequent sections
I shall briefly discuss a number of applications in the context of
natural disasters: Flooding, land slide and oil spill on the ocean.
Floods. Literature shows a number of examples of application of
radar for flood mapping and monitoring taking advantage of the
‘all-weather’ performance of the radar. Observations have in
most cases been made by satellite SAR but also airborne radar
images have been acquired. In general, SAR data are preferred
due to the fine spatial resolution but in cases of plain flodding
where a coarser resolution is acceptable SLAR systems may be
exploited.
The mapping feature relies upon the difference in backscatter
from vegetated areas and the water surface of the flood with the
aim of determining the extent of the flooded area(s) as a function
of time. However, tire water surface may be rougltened by wand
so that the otherwise good contrast between the backscatter sig
nals from tire water and tire surrounding vegetation is reduced
appreciably, so that delineation of the water area becomes uncer
tain. Thus, at a wind speed of more than 7 m/sec, delineation
nray only be made relying on uniformity of the reflection front
the water surface. Experience shows, however, that a stormy
period often is followed by a period of quiet weather conditions
so that the usefulness of the techniques becomes dependent upon
the repetition rate of the observation.
Another limitation is tire shadow effect created by trees at the
edge of or in the water body. This results in a reduced accuracy
of the mapping that depends upon the angle of incidence of the
radar signal and the height of the trees. One way of overcoming
this limitation is to ob servo the area from different view angles
which presently is only feasible with airborne radar’s. Another
way is the indirect way of mapping the radar image onto a DEM
exploiting the contour lines at far range where shadow effects are
not present. The effect may be reduced by using longer wave
lengths, L-band, for instance, that penetrate the vegetation more
easily. Combining with visual/near-infrared data may be another
approach although the likely time difference between the obser
vations (from satellites) may reduce the accuracy because of
changes in the flood situation.
Landslides. Movements of land surfaces are normally slight and
monitoring may only be done using the InSAR technique. Areas
of possible landslides can in principle be monitored from satellite
regularly down to shifts of the order of centimetres when surface
or natural markers are available. When combined with modelling
and knowledge of soil characteristics, prediction of a disastrous
landslides may be made.
The limitation of the observational techniques is related to sur
face coherence whether two-pass or three-pass interferometry is
applied. Often a DEM is involved so that two-pass interferome
try may be used increasing eh likelyhood of coherence.
Other limitations are the shadows inherent in radar measure
ments and the view angle, in the latter case because the move
ment is measured in the across-track direction only. This problem
may be partly overcome by observations at ascending and de
scending passes.
Oil spill on the ocean. What is at stake are oil spills on the
ocean in connection with ship wreckage and illegal discharge.
Wreckage often takes place near tire shore whereas the spill is
