English
Italiano
Forgot your password?
New user?
Document Actions
ijege-13_bs-de-agostini-et-alii.pdf
terisation (H
are a good approach to detect and characterise instability
phenomena, because the use of a multi acquisition stack
of images lets to reduce the limitations typical of SAR
Interferometry (i.e. atmospheric distortions or temporal
de-correlation). Using these multi temporal techniques
allows not only the estimation of discrete deformation
events, but also the evolution of displacements.
tion of landslide-prone areas. At local scale, DInSAR
methods are an useful tool to rapidly recognise por-
tions of unstable area with different displacement rates
or to create a twenty years long displacement velocity
dataset, where a complete field monitor is lacking.
displacement rates and directions, in order to define the
influence of tectonic setting on landslide evolution.
gion, NE Italy) were many instability phenomena are
present. The paper is focus on Prezzo landslides, which
dicarie Valley, located in the south-western sector of
the Trentino-Alto Adige region (northern Italy), with
particular reference to those phenomena inducing
a high level of risk. The study area, located in the
Southern Alps, shows a complex structural setting,
mainly influenced by the south Giudicarie Line and
its associated steep to subvertical faults WNW-dip-
ping. Tectonic setting strongly influences the evolu-
tion of slopes, it may be considered one of the main
cause of the onset of current instability phenomena.
Geological, geomorphological, structural and geo-
mechanical field investigations were conducted on
Prezzo landslide area (Giudicarie Valley) in order
to investigate the relationship between the tectonic
setting and the instability. Advanced DInSAR tech-
niques, such as Permanent Scatterers (PS) and Small
Baseline Subset (SBAS) algorithms, were applied to
ERS 1-2 (1992-2010) and ENVISAT (2003-2010)
satellite RADAR images, to detected displacement
areas related to geological structural elements which
could influence the Prezzo landslide development.
Results show that PS and SBAS data interpretation
is an helpful tools in investigating the influence of
tectonic settings on landslide processes.
with a maximum thickness in the middle sector of the
landslide. The total volume is 20 million m
network presents losses and damages. The phenom-
enon is monitored by Trento Province Geologic Serv-
ice through GPS, topographic stations, piezometers
and inclinometers. Additional field investigations
were carry out in order to better define the geologic
and geomorphologic settings of the unstable area.
DEM derived and LIDAR data were also used.
Limestone Formation is composed of flat-lying dark-grey
fine limestone, marly limestone and siltstone. This For-
mation outcrop in the north flank of landslide. The Prezzo
Limestone, is composed of black marly limestone with
Melino. It is located between an altitude of 510 and 980
m and presents one main scarp on the top, with related
morphologic structures (such as cracks and trenches)
and two minor scarps on right and left boundary (Fig.
1). The phenomenon is an old landslide, reactivated,
that involves silty deposits, with blocks of marly-cal-
careous and dolomitic rocks. The unstable area is 380
m wide and 1150 m long with average slope of 20°.
catastrophic landslides (P
ing fault, probably associated to the South Giudicarie
Line, limits the area to the North. The Triassic succes-
sion of the Prezzo area is folded forming a synclinal-
type fold. The northern sector shows a moderate (30°-
45°) SE dipping stratification, while in the southern one
the beds dip towards N-NE with medium angle (35°-
55°). The fold axis, which is located in the central part
of the slope, gently dips towards E-NE (Fig. 3).
and outside the area affected by the slope instability .
Besides the stratification, other three sets of mechani-
cal discontinuities cut the rock masses. Sets K1 and
K2 steeply (70°÷75°) dip respectively towards SW and
W, while K3 is linked to the southern stratification, but
showing a higher angle (65°÷70°) (Figg. 4a and 4b).
The increase of fracturing is related to an expectable
lowering of the rock masses quality of all the forma-
tions from the areas outside to those inside the landslide
body. Geomechanical surveys have also provided two
mation and Wengen Formation are present in the lower
part of landslide and dipping towards N-NE; the first one
is composed of dolostone, limestone, marlstone inter-
bedded with volcanoclastic deposits and chert nodules,
the second one is a silicoclastic-carbonatic unit made of
sandstone and marlstone interbedded with tuff layers.
the Quaternary sediments. The slopes of Giudicarie
Valley are mostly covered by glacial deposits and more
rarely by moraine. The valley bottom is covered with
large alluvial fans that rest on the flat alluvial plains.
the tectonic boundary between the Southern Alps and
the Austroalpine domain. Due to its location, oblique
to the Southern Alps chain, the Giudicarie belt shows
a very complex structural setting, mainly character-
ized by steep to subvertical WNW-dipping transpres-
sive faults and moderate dipping N to NNE-trending
reverse faults (D
the evolution of slopes, causing the onset of large scale
iour of rock masses in terms of both Joint Compressive
Strength (JCS) and Geological Strength Index (GSI)
(D
strength decreases starting from the stable areas of the
slope, to the inside of the landslide body. GSI values
highlight the weakness behaviour or rock masses in-
side the landslide body and in the areas approaching the
fold axis (Fig. 5). The presence of a rock masses highly
fractured is also confirmed by the boreholes executed
by Geological Service (data not reported in this paper).
GSI value of 35, whereas, far from it, the JCS increases
up to 70÷130 MPa and the GSI is around 45-50, both
in the northern and southern limb. More northerly, this
trend is not observed and both JCS and GSI decrease,
probably due to the proximity to the NE-SW trending
fault. This trend could not be assessed for the other for-
mations due to the lack of outcrops.
cal Geological Service, to improve the knowledge
phenomena or anthropic changes or vegetation growth.
Satellite look angle (θ) of 23.3° (for ERS and ENVISAT
images) and right side-looking acquisitions mode are
the responsible for geometric distortions effects. Layo-
ver effect is present where slopes inclination exceed the
look angle, producing a strong image distortions, which
prevent the correct signal interpretation. Shadow effect
is present in areas that are not illuminated by the radar
signal. These effects in relationship with aspect and in-
clination of slopes need to be taken in account before
starting an investigation of mountainous area, because
slope instability processes could be on area affected by
layover or shadow effects.
ments along the Line of Sight (LOS). Therefore, if the
direction of displacements is not along LOS, the satel-
lite will measure only a part of real velocity.
by European Space Agency, ESA) were used. Advanced
DInSAR techniques, Permanent Scatterers (PS) and
Small Baseline Subset (SBAS) algorithms, were applied
to ERS 1-2 (1992-2010) and ENVISAT (2003-2010) sat-
ellite RADAR images to detected displaced areas.
ver, the ERS ascending dataset has many corrupted and
unprocessable images. So the obtained displacement
velocity dataset is provided by 43 ENVISAT images
acquired in ascending mode, spanned from November
2004 to September 2010, with revisiting time of 35 days.
(122 PS were found on an area of 0.35 km
density is very low (coherence threshold = 0.6). Also
the Distributed Scatterers of the SBAS analysis are lo-
cated on the Prezzo village. Both, PS and SBAS data
distributions, show a clear identification of the unstable
area on Prezzo village (Fig. 7).
10 mm/year with a maximum of 20 mm/year for the
PS data) , whereas the green ones identify a stable area,
unstability (Fig. 6). There are 14 GPS point (some
of them since 1995 and other since 2005), 10 incli-
nometers since 2002 and 4 piezometers since 2004.
GPS network reveals a deformation velocity of 3 to 6
cm/y. The inclinometers show a sliding surface depth
between a minimum of 27 m at the base of slope and
a maximum of 110 m in the middle part of landslide.
isn’t complete because of some missing or not vali-
date data in 2005, 2006, 2008 and 2009. However we
compared the displacement velocity data from DInSAR
processing with the rainfall data to find correlation be-
tween rainfalls regime and displacement of landslide.
the topographic one is associated to the relationship
between phase and topography, the displacement one
is due to the land movements (subsidence, landslides,
earthquake), the atmospheric one is due to the variation
of atmosphere characteristics, and the noise one due to
instrumental properties. Comparing two SAR images
is possible exploit the phase contribution due to land
movements, knowing the other phase components and
subtracting them from the total phase difference.
ple acquisitions stacks (large SAR temporal series). The
concept of combining InSAR information from a large
number of SAR images (thanks to repeated satellite pass-
es on the same area), allowing the solution of a deforma-
tion time series, has been introduced by F
tion of displacements. SBAS method exploits the spatially
distributed coherence, instead of estimating the coherence
exclusively on local scatterers. In fact, PS methods is fo-
cus on the analysis of point targets, instead SBAS one is
intended to the analysis of distributed targets.
decorrelation is due to changes of electro-magnetic
Sight displacements). SBAS processing shows displace-
ment velocity values similar to the PS one, although the
SBAS values are lower than the PS ones. Furthermore,
the direction of displacements is towards the direction of
the fold axis and of maximum slope (E-NE), and its sub-
stantially parallel to LOS of ascending satellite track, so
the real displacement velocity is substantially recorded.
248 has a low annual velocity rate and it is located in
the stable area on the southern part of landslide. Also
the monthly rainfalls, from 2004 to 2010 are reported
in figure 8 to describe the correlation between main
rainy periods and main displacements of the landslide.
In fact, a more in depth study of the rainfalls dataset
evidences that the main displacement periods (indi-
cate with pink arrows) follow the peaks in the rainfall
trend (blue continuous line in Fig. 8).
evolution of slopes (Fig. 2 and Fig. 3). The structural
analysis shows the role played by the syncline fold as
the main predisposing factor of the slope instability,
thanks to its E-NE trending fold axis oriented favour-
ably for wedge failure. In fact, the lower values of the
JCS and GSI index inside the landslide body and near
the syncline axis point out the great fracturing and
weakness of the rock masses (Fig. 4 and Fig. 5).The
advanced DInSAR approach allowed to improve
the identification of the unstable area at Prezzo vil-
tween field displacement and PS or SBAS data is not
easy due to the different acquisition geometry (LOS
vs. flow direction) and the different temporal coverage.
But, thanks to E-NE directions of landslide displace-
ments, essentially parallel to ascending LOS, the Prezzo
landslide is a good test area to study the applicability of
interferometric data on landslide study. It follows that
PS, SBAS and GPS numeric data slightly differs from
each other, but we notice that each methods clearly
identify the active and stable zones. Moreover because
the flow direction is determined by the syncline axis
direction, we can use only the ascending LOS velocity
data. Otherwise, both, ascending and descending im-
ages should be needed to exploit displacement vectors.
logic and structural analyses, and with field investiga-
tions (e.g. GPS displacement data), are helpful tools
in landslide identification and characterization. The
twenty years of interferometric data allow the defi-
nition of a long displacements dataset permitting to
retrace the history of land movements, also when a
are distributed only along the northern and southern
side of the landslide body allowing a first identifica-
tion of the landslide area. The PS and SBAS data are
distributed on the overall Prezzo village, permitting
a better definition of the distribution of activity. As a
consequences the integrations of PS, SBAS, GPS and
geologic/geomorphologic data well define the sharp
separation between the active and the stable zones,
improving the identification of different displacement
areas of the two landslide sectors (Fig. 7 and Fig. 9).
processing ENVISAT images, allowed to determine the
annual displacements rate and to compare it with the
GPS data (Fig. 9). In fact GPS maximum annual rate of
movement (2004-2010 time interval) is 3 cm/y, for PS
is 2,1 cm/y and for SBAS is 1 cm/y.
central area, along the fold axis, whereas the southern
sector of the landslide body is quite stable, suggesting
again an influence of the structural setting on the slope
ness of rock masses (low GSI and JCS near the fold
axis), due to the complex tectonic setting, is the most
likely the predisposing factor.
(7): 1377-1386.
tive way, that rainfall could be considered as the trig-
The issue 16 volume 01 has been published