. AND
on
traction
fter the
heavy
(and in
st all of
opes.
onships
e earth
sified in
on of a
, which
image
s avec
Ilitaire.
1994. A
dans le
ements
ents en
Is entre
Jées de
nes.
es pour
e facon
onnées
tion de
d data
one to
nitation
e area
| permit
1e with
than a
digital
loss of
2. INVESTIGATED TOPICS
The goal of the present study is to investigate the
relationships between physical parameters such as
reflectance and emittance from features located on the
earth surface (recorded in the different digital images)
and to investigate two particular occurrences: landslides
and flooded areas.
As far as the flooded areas are concerned, when
remotely sensed data area available, feature reflectance
and/or emittance:
e decreases when a water layer covers the ground
surface or when the soil is humid;
e increases in the red band (0.63-0.69 um) because of
the vegetation stress caused by moisture;
e changes noticeably when different temperatures (due
to thick water layer) are recorded.
In the microwave band, the water presence could also be
appreciated by estimating the surface roughness, where
water layers smooth surfaces; dielectric constant is then
heavily correlated to soil water content and
electromagnetic polarisation.
As far as the control of slope instability (landslides) is
concerned, a traditional analysis method of the problem
is constituted by a geologic surface mapping, integrated
by ground investigations and specific checks that could
include both geophysical prospecting and geognostical
drilling. In order to analyse phenomena that occur on a
regional scale, it is necessary to have photogrammetric
coverages and then to proceed with a stereoscopic
inspection of the landslides.
In particular, in order to characterise slope instability
occurrences on the ground in digital remotely sensed
images, it is necessary to point out one of the landslides'
main characteristics (at least one of the more
superficial): the presence of mud and debris in the
accumulation slope foot, such material being
characterised from an accented reflectivity in the visible
spectral band. Thus, it is possible to evaluate, from the
digital images, the landslide recurrent reflective
behaviour; the radiometric range thus isolated can then
give rise to an interpretation key when exploring other
images (or other spectral bands), where the presence
and the location of the aforesaid occurrences can be
detected.
3. THE STUDIED AREA
Figure 1 - The studied area
67
In November 1994, a flood event occurred throughout
almost all the Piedmont region, and in particular in the
southern part of the region in the Cuneo province,
between the Appennine mountains (South) and the
Padana plain (North) not far from the cities of Asti and
Alessandria (fig. 1).
4. DATA AVAILABLE
Different data sets have been used, such as:
e aSpot-2 panchromatic image acquired in June 1994;
e a Spot-2 panchromatic image acquired in November
1994;
e aSpot-2 panchromatic image acquired in April 1995;
e a Thematic Mapper multispectral (7 spectral bands)
acquired in June 1991;
e a Ers-1 radar image acquired in November 1994;
e a panchromatic photogrammetric coverage acquired
in November 1994;
e a panchromatic photogrammetric coverage acquired
in May-June 1992;
e a DEM derived from the
photogrammetric coverage.
The Spot-2 panchromatic image acquired in April 1994
has been absolutely georeferenced, while all the other
images have been relatively fitted to the previous one.
1992 panchromatic
5. EXTRACTION OF THE FLOODED AREAS
The determination of the extension of the flooded areas is
particularly important for a regional scale analysis when
it is necessary to evaluate the damages and the potential
flooding risk.
To obtain this goal a new methodology has been
implemented, and a flow chart is shown in fig. 2.
Radiometric
pre-processing
Geometric
pre-processing
Classification
i
Clivometric
model
Photogrammetric
cov. comparaison
Final
image
Figure 2 - Assessment flow chart of the flooded areas
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996