Document Actions

ijege-13_bs-notti-et-alii.pdf

background image
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
349
DOI: 10.4408/IJEGE.2013-06.B-33
STUDYING AND MONITORING LARGE LANDSLIDES
WITH PERSISTENT SCATTERER DATA
D
aviDe
NOTTI
(*)
, C
lauDia
MEISINA
(*)
, a
lessio
COLOMBO
(**)
,
l
uCa
LANTERI
(**)
& F
ranCesCo
ZUCCA
(*)
(*)
Università degli Studi di Pavia - Dipartimento di Scienze della Terra e dell’Ambiente - Via Ferrata, 1 - Pavia (Italy)
(**)
ARPA - Piemonte - (Italy)
INTRODUCTION
Large landslides are widespread both in the Alps
and in Apennines (M
ortara
& s
orzana
, 1987; F
orlati
et alii, 2001; a
Mbrosi
& C
rosta
, 2006); the observed
movements are generally from extremely slow to slow
(C
ruDen
& v
arnes
, 1996); they are quite regular with
some occasional acceleration. Rapid and superficial
phenomena are often associated and they result in signif-
icant direct and indirect damage. The sudden and parox-
ysmal collapse of the whole mass may have catastrophic
consequences. They are also difficult to characterize in
their boundaries and state of activity, to monitor with
traditional tools due to their extension (landslides over
0.2 km
2
) and low rates of movement, which are close to
the detection limit of traditional monitoring equipment.
In the last years the development of Persistent Scat-
terer Interferometry (PSI) methods, e.g. PSInSAR
TM
F
erretti
et alii, 2001), SqueeSAR
TM
(F
erretti
et alii,
2011), small baseline subset - SBAS (b
erarDino
et alii,
2002), SPN (a
rnauD
et alii, 2003), and coherent pixel
techniques - CPT (b
lanCo
-s
ànChez
, 2008), PSP-DIF-
SAR (Persistent Scatterers Pairs - Differential InSAR;
(C
ostantini
et alii, 2000), allowed to help the detection
and monitoring of the very slow and extremely slow
movements typical of many large landslides (C
olesan
-
ti
& W
asoWski
, 2006; F
arina
et alii, 2006; M
eisina
et
alii, 2006; h
errera
et alii, 2009; G
uzzetti
et alii, 2009;
l
auknes
et alii, 2010; Y
in
et alii, 2010).
The aims of this study are:
a) to study large landslides and to assess their state
ABSTRACT
This work is focused on very slow moving land-
slides and the new generation of Persistent Scatterers
PSI (SqueeSAR™ processing, developed by Telerile-
vamento Europa) that allows to increase the density
and the time series quality of interferometric data. The
improvement in the time series quality helps also to
understand the behaviour of some processes and to
have a best comparison with traditional monitoring
system and/or rainfall data.
The consequent aim of the research is to evaluate
the potential and the limitations of PSI data for large
landslide studying and monitoring.
Some large landslides belonging to different geo-
logical, geomorphologic and land-use contexts and
with different monitoring systems, in Western and
Ligurian Alps, Langhe Hills and a portion of North-
ern Apennines (Oltrepò Pavese), have been analyzed.
The study area is covered by 18 years of SAR data,
consisting on ERS (1992- 2001) and RADARSAT
platform (2003-2010).
The results show that the PSI analysis is useful
both on regional and local scale. At regional scale PSI
allows to improve landslide inventories. At local scale
the PSI joined with other data can help in the under-
standing landslide features and kinematics.
K
ey
words
: large landslides, Persistent Scatterers, monito-
ring, SqueeSAR
TM
, DSGSD, GIS
background image
D. NOTTI, C. MEISINA, A. COLOMBO, L. LANTERI & F. ZUCCA
350
International Conference Vajont 1963-2013. Thoughts and analyses after 50 years since the catastrophic landslide Padua, Italy - 8-10 October 2013
tectonic settings, are the causes of widespread
presence of Deep Seated Gravitational Slope De-
formations DSGSD (M
ortara
& s
orzana
, 1987)
and large complex landslides that affect many
large portion of the valleys and also some villages
and infrastructures (roads, railways). The move-
ment of these landslides is usually slow but some-
times rapid acceleration or collapse may occur
(e.g. Grange Orgiera landslide (a
rpa
p
ieMonte
,
2009)). The areas affected by large landslides are
usually also interested by more rapid landslides
like rockfalls or debris flows. The large landslides
are particularly diffused in the area, where calc-
schist formation of Penninic nappe outcrops.
The presence of many debris and rock (good scat-
terers) is favourable for the PSI analysis.
2. The Ligurian Alps correspond to the South-west-
ern sectors of the Alps and are characterised by
of activity through the use of SqueeSAR™ and
PSInSAR™ techniques;
b) to analyze the large landslide kinematics through
specific case histories in Alps and Apennines with
particular focus on spatial and temporal analysis
of the movement.
GEOLOGICAL AND GEOMORPHOLOGI-
CAL SETTINGS OF THE STUDY AREA
The study area (about 16000 km
2
) is located in
NW Italy and it presents very heterogeneous geologi-
cal and geomorphologic settings (Fig. 1). It is possible
to indentify four main sectors:
1. The Alps in the Northern and Western part of Pie-
monte region represent the sector most affected
by large landslides. The alpine valleys after the
last glaciations are characterized by an high relief
energy with erosion processes, that, joined with
Fig. 1 - Geological settings of the study areas. A = Argentera Massif; DM = Dora Maira; GP=Gran Paradiso Massif, SL=
Sesia-Lanzo zone, L.A = Ligurian Alps, H = Langhe Hill, O = Oltrepo Pavese/Northern Apennines
background image
STUDYING AND MONITORING LARGE LANDSLIDES WITH PERSISTENT SCATTERER DATA
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
351
In this research we have analyzed the very (less
than 1.6 m/yr) and extremely slow (less than 16 mm/
yr) large landslides using C
ruDen
& v
arnes
(1996)
velocity classification. About 1820 large landslides
were identified in the study area. Slow flows, deep
seated gravitational slope deformations (DSGSD),
roto-translational slides and complex landslides (Fig.
2), which represent the slope movements most suit-
able for PSI techniques, were selected.
THE PSI DATASET
The landslides were analyzed with PSI data elabo-
rated mainly with SqueeSAR™ technique using im-
ages of RADARSAT (2003 - 2009) and of ERS 1-2
(1992 -2000) satellites for the Oltrepo Pavese.
The SqueeSAR
TM
technique (F
erretti
et alii,
2011) extracts movement information not only from
traditional persistent scatterers (PS) like buildings,
anthropic structure or rocks but also from distributed
scatterers (DS) like sparse vegetated areas or debris
covered areas. This allowed having a high density
of SAR data in the Alpine area covered by huge de-
bris (e.g. Fig 2). The time series of SqueeSAR™ are
elaborated with a non-linear algorithm (polynomial
trends). The time series can follow acceleration and
slowing of a scatterer. This allows to better observing
the kinematic of landslides related to seasonal cycles
or extreme rainfall events.
ERS and ENVISAT data processed with PSIn-
SAR™ technique were also used in order to extend
the time span of the movement history from 1992.
the presence of flysch and limestone formations.
The area presents a fluvial-modelled landscape.
Even if the DSGSD are not very diffused there
are many large complex landslides. The vegeta-
tion coverage is wide, however some villages are
settled over slow landslides that present a gentler
slope with a high PS density (PS generally cor-
respond to buildings).
3. The Langhe hills: the area is located in Central
and Southern part of Piedmont and it is charac-
terised by monocline tertiary formations of marl,
sandstone and shale and it is mainly affected by
translational rock-block slides on gentle slopes
(10° - 20°) with dip slope stratifications. The
area affected by planar slide is usually covered
by cultivated fields or vineyards and the PS data
density is typically low. Moreover, these land-
slides present a rapid movement triggered by
strong rainfall events (l
uino
, 1999) and with PSI
techniques is possible to detect residual or pre-
failure movement.
4. The Oltrepò Pavese corresponds to the NW Ap-
ennines and it is characterised by the presence
of flysch and shale. Mainly complex landslides
(roto-translational slides and slow flows) affect
the slopes. The area is mainly covered by vegeta-
tion, vineyards or cultivated fields, however many
villages were built within landslide areas, so PSI
could detect movements even if for only a small
number of large landslides.
The landslide monitoring system net presents
a quite good distribution especially in Piedmont,
where about 300 landslides are monitored, mean-
while in Ligurian Alps and in the Oltrepo Pavese
only few sites are monitored.
Fig. 3 - PSInSAR data from ERS (1992-2000) (left) and
SqueeSAR
TM
data from Radarsat (2003-2009)
(right) density over a landslide in Western Alps
Fig. 2 - Distribution and typology of large landsides in the
study area
background image
D. NOTTI, C. MEISINA, A. COLOMBO, L. LANTERI & F. ZUCCA
352
International Conference Vajont 1963-2013. Thoughts and analyses after 50 years since the catastrophic landslide Padua, Italy - 8-10 October 2013
ments range from 30% (of PSI suitable) in the Oltrepo
Pavese to 3% of Langhe Hills.
It is also interesting to consider the percentage of
landslides with movement in the Alpine area classified
for lithology (Tab. 2): calc-schist formations present
a high percentage of “active” landslides (45%), how-
ever carbonate rocks and serpentine have about 20
- 25% of landslides with movement. This is quite in
agreement with literature (F
orlati
et alii, 1995): the
landslides involving lithology with a ductile behav-
iour (calc-schist) are characterized by slow continu-
ous deformations with gentle slope morphology com-
pared with landslides on brittle-behaviour lithology
(limestone or serpentine). By observing the spatial
distributions of landslides activity it is possible to see
the areas with high concentration of active large land-
slides: the Ligurian-Piedmont domain between Maira
and Susa Valley, the area near Simplon Pass and the
area near Gran Paradiso massif (Fig. 1).
The analysis of the type of PS/DS shows that in
the Alpine area the scatterers are mostly represented
by debris, or rock. The analysis made for each sin-
gle lithology (Tab. 3) shows that the areas covered
by calch-schist formation present a great number of
building targets. This can be explained with the gentle
slopes associated to this lithology that allowed the de-
velopment of some villages (e.g. Sauze D’Oulx, Susa
Valley) over large DSGSD. In the Langhe hills and
Oltrepo Pavese the majority of PS/DS are buildings or
anthropic structures.
ANALYSIS AT REGIONAL SCALE
In the study area the PSI data were compared with
the italian national landslide inventory (IFFI, ISPRA,
2008), in order to check how many large landslides
have useful SAR data.
Only the landslides with a certain number of PS/
DS (at least 3) and with a density greater than 30 PS/
km
2
were selected. These values were chosen on the
basis of empirical experience, there is not a fixed rule,
but generally the more SAR data is dense and distrib-
uted the more reliable the results became. We con-
sidered also the landslides that have an “anomalous
area” interpreted as landslides. The anomalous areas
(M
eisina
et alii, 2008) are clusters of PS/DS that show
significant movements.
It is possible to see (Fig. 4) that in the Alps about
the 60% of large landslides are suitable for PSI analy-
sis while in the Ligurian Alps the percentage decreas-
es to 48%. In the Langhe Hills and Apennine areas the
percentage of landslides suitable for PSI analysis is
considerably lower (about 20%).
In order to discriminate moving/non-moving phe-
nomena as well as the state of activity of large land-
slides we applied the threshold of +/-2 mm/yr meas-
ured along the LOS (Vlos) (M
eisina
et alii, 2008) and
related to the precision technique. The active land-
slides represent the 30% of PSI suitable landslides
in Alps and Apennines, the 20% in the Ligurian Alps
and the 5% in the Langhe hills (Tab. 1). If we applied
the threshold of -5 mm/yr of velocity projected along
the slope (Vslope) proposed by C
iGna
et alii (2012)
the results are quite similar: the landslides with move-
T
ab. 1 - Active large landslides with different thresholds
Tab . 2 - Percentage of active landslides versus lithology
for the Alps
Fig. 4 - Percentage of large landslides suitable for
PSI analysis in the different areas of study. The
number inside the bars is the percentage of active
and not-active landslides using the 5 mm/yr Vs-
lope threshold
background image
STUDYING AND MONITORING LARGE LANDSLIDES WITH PERSISTENT SCATTERER DATA
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
353
Three case histories will be described in this
work. In two cases (Brenvetto and Rosone) PS/DS
data can be compared with other monitoring instru-
ments while in the Alpe Baranca DSGSD PS/DS are
the only monitoring system available.
BRENVETTO LANDSLIDE
The landslide of Brenvetto is located in Soana
stream basin at the east boundary of Gran Paradiso
massif. The landslide is geologically located in the
Pennine Nappe, between Sesia-Lanzo zone and Gran
Paradiso internal massif (Fig. 5). It is characterized
by calc-schist and serpentine formation bedrock. The
area is covered also by moraine quaternary deposits
(C
arraro
et alii, 1995). The main element at risk is
the road that connects the high Soana Valley with
Orco valley and then the Torino plain.
The upper sector of the slope is affected by large
DSGSD that does not show any particular evidence
of movement except in the central part of the defor-
mation, where a complex landslide was detected (Fig.
6). The complex landslide shows clear evidences of
movement like trenches, scarps, a strongly fractured
It is important to remind that the landslides state
of activity assessed at regional scale is only indicative
due to the well known limitations of PSI techniques.
Detailed geomorphological and historical analyses for
each landslide are necessary in order to confirm the
state of activity and to update the landslide inventory.
ANALYSIS AT LOCAL SCALE
At local scale it was possible to make further anal-
ysis on PS data. In particular we analyze time series,
the distribution of the movement related with geomor-
phology and the components of the velocity.
The large landslides with significant movement
and with good monitoring data are mostly concen-
trated in the Alps.
The Piemonte Region (Alps and Langhe hills) has
a widespread monitoring system of landslides so it is
possible to compare the capacity of PSI techniques
versus the other type of monitoring instruments. In
the alpine area PS/DS can monitor a larger number of
landslides (59%) than traditional monitoring system
(9%), while in the Langhe hills the landslides moni-
tored are almost the same: 19% with traditional moni-
toring systems and 24% with PSI data. The landsides
with both PS and traditional monitoring systems are
only 7%-8% of total.
Fig. 5 - Geological settings of Rosone and Brenvetto land-
slides
Fig. 6 - Brenvetto Landslides. GPS data (2004-2011) and
SqueeSAR
TM
data (2003-2009)
Tab. 3 - Percentage of PS/DS on buildings with RADAR-
SAT SqueeSAR
TM
data
background image
D. NOTTI, C. MEISINA, A. COLOMBO, L. LANTERI & F. ZUCCA
354
International Conference Vajont 1963-2013. Thoughts and analyses after 50 years since the catastrophic landslide Padua, Italy - 8-10 October 2013
rock-mass and many debris flows derived from rock
fall and rock mass disaggregation.
In the lower sector of the slope widespread shallow
landslides, debris flows can be noticed and at the toe
of the slope erosion of Soana stream contributes to the
general instability (Fig. 6). The landslide is monitored
by ARPA Piemonte with 5 GPS at manual reading since
2004. The measurements made once at year showed a
rate of movement that is up to 110 mm/yr in the central
sector of the complex landslide. The SqueeSAR
TM
data
derived from RADARSAT (2003-2009) ascending ge-
ometry show a very good numbers (>20) and spatial
distribution of PS/DS on complex landslide. The rate
and the spatial distribution of movement registered by
PS/DS is almost the same of GPS measurements.
The ERS PSInSAR
TM
data (1992-2000) present a
low density of PS and they were not considered for the
analysis.
The direction of the movement (from ESE to SE) is
quite parallel with LOS ascending direction and about
the 80% of the movement can be detected. It is not
possible to resolve the north/south component of the
movement due to the few descending data available.
The PS/DS data suggested potential enlargement of
the complex landslide boundaries to include also some
downslope sectors and to update the state of activity:
now the landslide is classified as active. The time
series of PS and GPS (Fig. 7) data show a linear and
constant trend of deformation that seems not influenced
by rainfall or snow melting. However a small accelera-
tion of the movement was detected by GPS, and also by
some PS in 2008/2009. This can be related to the rainy
period started in May 2008. Due to the high velocity
registered it is possible that some errors related to phase
unwrapping (movement greater than λ/4 between two
consecutive acquisitions) should be occurred
ROSONE LANDSLIDES
The Rosone landslide is located in the NW Italian
Alps in the Orco Valley and it is classified as DSGSD.
The landslide affects the penstock of the near hy-
droelectric central and in case of collapse may create a
dam in the Orco river and an interruption of the road
that connects the upper part of the valley with the plain.
For these reasons the landslide was well studied by
many authors (F
orlati
et alii, 2001; p
isani
, 2010; R
aM
-
asCo
et alii, 1989) that provide to asses the geomorpho-
logical, hydrogeological and mechanical settings and
they try to model the possible evolution.
The Orco Valley is located in the central part of
the Gran Paradiso Massif. This complex belongs to
the Upper Pennine Units (Pennine Nappe System) and
it consists of a composite crystalline basement and a
Permo-Liassic cover (Fig. 5).
Rosone landslide is modelled on the Augen
Gneiss Complex. The geological-structural configu-
ration of the studied area is relatively simple: granite
and augen gneiss crop out.
The rock mass is characterized by several alpine
Fig. 8 - Rosone Landslide. PS Time series (2003-2009
RADARSAT descending data) of the sectors A and
B compared with 6 months (6m) cumulated rain-
fall (2000-2009)
Fig. 7 - Brenvetto Landslide PS (RADARSAT ascend-
ing data) and GPS time series compared with 6
months cumulated rainfall
background image
STUDYING AND MONITORING LARGE LANDSLIDES WITH PERSISTENT SCATTERER DATA
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
355
debris. The sliding surface can be identified 46 m depth.
Sector C. It is in the bottom part of the slope and
it is the most weathered sector, the rock mass is com-
pletely disarticulated. The depth of sliding surface is 39
m. This sector is also affected by rock fall and debris
flow. In the 1953 a strong reactivation of the movement
caused severe damages to Bertodasco village and the
inhabitants were evacuated.
The landslide monitoring started in the sixties and
now a wide system of monitoring data is installed:
inclinometers, extensometers, GPS, optical measure-
ments and from 1922 also from SAR data (Tab. 4).
On this site ERS 1-2 data (1992-2001) elaborated with
PSInSAR techniques and RADARSAT data (2003-
2009) elaborated with SqueeSAR
TM
are available.
The PSI analysis it is possible only in the sectors
A and B where a good density of data is available
especially with RADARSAT data. The sector C has
a wide forest coverage and too fast movement and
ductile deformation phases. The main schistosity is af-
fected by the periclinal orientation of the massif, which,
in this area, displays an average dip direction of about
150°, with 35° dip (r
eGione
p
ieMonte
, 1996). The brit-
tle structural attitude is defined by three main disconti-
nuity systems: a KS system parallel to main schistosity
(SR) and two sub-vertical lineaments corresponding to
E-W and N-S striking normal fault conjugate system.
The landslide of Rosone is located on a slope af-
fected by large DSGSD but only the sector of Rosone/
Bertodasco has high deformation and instability. The
western sector of DSGSD (Ronchi village) presents less
evidence of deformations and the movement is more re-
lated to shallow debris and colluvium instability.
Considering the most active part of the DSGSD,
the Bertodasco area, the geomorphological and the
structural analysis joined with monitoring data allowed
to detect three main sectors with different rock mass
evolution and movement (Fig. 9).
Sector A. It corresponds to the upper part of
slope. This is the less weathered rock mass, how-
ever many trenches and scarps can be detected. The
movement is generally weak and sliding surface is
between 30 and 75 m.
Sector B. It is in the central part of deformations.
The rock mass is rather weathered with completely dis-
articulated rock block, scarps and the presence of many
Tab . 4 - Velocity registered by the different monitoring system
Fig . 9 - Rosone geomorphological map and PS data (SqueeSAR
TM
, RADARSAT Ascending)
background image
D. NOTTI, C. MEISINA, A. COLOMBO, L. LANTERI & F. ZUCCA
356
International Conference Vajont 1963-2013. Thoughts and analyses after 50 years since the catastrophic landslide Padua, Italy - 8-10 October 2013
density of data especially for ascending geometry,
however the few PS on the landslide confirm the
same rate of velocity of RADARSAT.
The projection of the velocity of PS data was made
on the mean dip direction schistosity of the rock mass
and from the results of GPS measurements. The direc-
tion of the movement is between SE and SSE (150°-
170°). With this parameters ascending geometries can
detected about the 60 % of real movement.
The combination of ascending and descending ge-
ometries allowed to resolve the vertical and E-W com-
ponent of the movement. The results (up to 6 mm/yr
toward East, and 10 mm/yr to down) are in a very good
agreement with the other monitoring instruments .
Finally on the C sector of the Bertodasco landslide
some corner reflectors will be installed in the next fu-
ture. Thanks to these artificial scatterers the satellite
Cosmo-SkyMed will acquire SAR data for civil protec-
tion application.
ALPE BARANCA DSGSD
The Alpe Baranca DSGSD is located in the upper
Mastallone stream basin near the village of Fobello.
The PS/DS data is the only monitoring system avail-
able on this landslide.
Excepted an alpine hut there are no anthropic el-
ements at risk however the landslide is interesting
because after October 2000 flood a large fracture ap-
peared in the upper part of the slope as evidence of a
great acceleration of the movement of the whole mass.
only 1 PS is present.
- The sector A presents weak movements (5 mm/
yr) in the eastern part and moderate movements (up
to 10 mm/yr) in the western part that affect the pen-
stock. The movements progressively increase from
upslope to downslope.
- The sector B is characterized by moderate move-
ments from 10 to 20 mm/yr. It is possible to see that a
scarp borders a sub-sector with major movement.
- The sector C does not have significant PS data
due to the vegetation coverage and the high movements
that probably cause unwrapping problems. However,
the only PS in the sector C recorded a velocity of 25
mm/yr. The GPS and optical measurements show a ve-
locity from 45 up to 130 mm/yr.
The time series of PS RADARSAT data (Fig. 8)
show a small decrease of the movement in 2005-2007,
probably related to a drier period. The movement can
be considered as constant and linear. Other monitor-
ing systems show some acceleration after October
2000 flood especially in the sector “C” but no PS data
are available for this period.
The ERS data (1992-2001) present a very low
Fig. 11 - Alpe Baranca DSGSD: original and “unwrapped”
Time series (2003-2009) compared with and 6 mon-
ths cumulated rainfall (2000-2009). Note the high
peak of rainfall occurred in winter 2000-2001
Fig. 10 - Alpe Baranca geological settings
background image
STUDYING AND MONITORING LARGE LANDSLIDES WITH PERSISTENT SCATTERER DATA
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
357
tor moved of some meters and at the edge of upper
sector a large scarp appeared in 2001 spring after
snow melting (Fig. 12). The scarp shows a dis-
placement up to 6 m in the central part.
The ERS data (1992-2000) (pre-failure) identified
a movement from -5 up to -15 mm/yr measured along
descending LOS. The RADARSAT data during the pe-
riod 2003-2009 (post-failure) measured the same rate
of velocity (average Vlos -13 mm/yr). The analysis of
time series allowed also to discover a phase unwrap-
ping problems in 2004 caused by a gap of some im-
ages. With the correction of this unwrapping problem
the average Vlos can be estimated to -20 mm/yr.
Considering an average slope of 34° and a slope
orientation of 205°N azimuth the average velocity
projected along the slope is about -50 mm/yr for the
period 2003-2009.
This area is geologically located inside the Ses-
ia-Lanzo zone, with bedrock of diorite-kinzigite and
mica-schist. The main tectonic lineament are orient-
ed SW-NE (Fig. 10).
The DSGSD is located on a slope with SW orien-
tation from 2300 to 1600 m a.s.l.
The deformation presents three main sectors
(M
aFFeo
& z
anottelli
, 2008):
- The upper sector with a steep slope is characterized
by weak weathered bedrock but with great evidence
of deformations like double ridge and counter slope
areas affected by rockfall. The PS/DS data do not
show any particular movement in this sector.
- The central sector is the area where bedrock is
mostly disarticulated with many scarps, counter
slope, bulged profiles and many debris deriving
from rockfall. After the 2000 flood this entire sec-
Fig. 12 - Alpe Baranca DSGSD. Main Geomorphological elements and RADARSAT descending data
background image
D. NOTTI, C. MEISINA, A. COLOMBO, L. LANTERI & F. ZUCCA
358
International Conference Vajont 1963-2013. Thoughts and analyses after 50 years since the catastrophic landslide Padua, Italy - 8-10 October 2013
By combining ascending and descending ge-
ometries it was possible to extract East-West and
vertical component of the velocity. The vertical com-
ponent presents greater values than E-W component
(up to 10 toward West), the North-South component
cannot be calculated but is probably greater than E-W
due to SSW slope orientation.
The analysis of time series do not show any partic-
ular trend and the movement can be considered linear
(Fig. 11). It is possible to appreciate however that in
the first years of measurements (2003-2005) the veloc-
ity had higher values (-25/-27 mm/yr, along LOS) then
slowed to -16/-18 mm/yr. This trend is compatible with
deceleration after the events of 2000-2001.
- The third sector is located at the toe of the slope
and it is probably not affected by DSGSD. In this sec-
tor we have the presence of talus cone and debris deriv-
ing from rockfall. A large fan deriving from rapid flow
accumulation is located in the SE sector of the slope.
If the velocity and the slope profile are compared (Fig.
13) a quite good match between geomorphology and
velocity detected by SAR data can be appreciated.
CONCLUSION
The analysis of large (> 0.2 km
2
) slow moving
landslides (DSGSD, complex slides, slow flow and
roto-traslational slides) with PSI data showed good
results, considering the traditional limitations of PSI
technique, both on regional and local scale.
- At regional scale, PSI data allowed to study and
post-monitoring a wide number of large landslides,
particularly in the Alps where the new processing tech-
niques like SqueeSAR allow a high density of targets in
the area covered by talus and debris. In this area more
than 50 % of large landslides have useful PS/DS data.
The other monitoring systems allow to cover only the 9
% of landslides. It is important to remark however that
traditional monitoring system is installed on the most
interesting and dangerous landslides, while PSI moni-
toring depends on natural settings. On the other sectors
(Langhe and Oltrepo Hills) the percentage of landslides
that can be studied and post-monitored with PSI data is
sensible less; however the analysis is useful because the
scatterers are represented by buildings that are usually
built directly on landslides.
As far concerning the state of activity of large
landslides, it is possible to observe that in calc-schist
formations of the Pennine Nappe of Western Alps
there is larger percentage of active landslides com-
pared to other brittle lithologies. This result confirms
Fig. 13 - Alpe Baranca DSGSD: Topographic profile from M
affeo
& Z
anotelli
(2008) compared with Vlos descending, East-
West and Vertical velocity of RADARSAT descending data (2003-2009)
background image
STUDYING AND MONITORING LARGE LANDSLIDES WITH PERSISTENT SCATTERER DATA
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
359
ment, while the large landslides, which are related to
the main DSGSD, are active and potentially dangerous.
On the opposite for the Alpe Baranca DSGSD the
role of PSI coupled with geomorphological analysis was
fundamental because it was the only available monitor-
ing system. ERS (1992-2000) and RADARSAT (2003-
2009) data measured a strong acceleration (some me-
ters) after October 2000 flood and a fracture appeared
in the upper part of the slope. PS time series confirm the
movement of this sector before and after the paroxysm
event but with rate of movement constant and relatively
low. This can be an evidence that a good PS coverage al-
lows to detect landslides characterized by extreme slow
movement that sometime can have potentially hazard-
ous acceleration when extreme trigger factors occur.
ACKNOWLEDGMENTS
The authors are grate to: TeleRilevamento Eu-
ropa (TRE) that processed the SAR data; ARPA Pie-
monte that provides information for local site, SAR
and other monitoring system data; Regione Lombar-
dia and Regione Liguria for the supply.
literature study on DSGSD (F
orlati
et alii, 1995)
based on geomorphological approach.
- At local scale the PSI data were useful to increase
the knowledge about landslide kinematics and to inte-
grate information coming from other monitoring systems.
The Brenvetto landslide, in the Soana Valley, is
an example of good integration of PSI and GPS data;
both spatial and temporal distributions of the move-
ment (time series) agree and well match the geomor-
phological evidence.
The Rosone Landslides is one of the most studied
landslides in western Alps, due to catastrophic conse-
quence of a collapse that may affect the penstock. In this
case the PSI data confirm the already dense and hetero-
geneous monitoring system network. Due to the benefits
in density and for the easiness in data acquisition and
storage over large areas, it was decided to integrate and
largely replace the old monitoring system with a regular
PSI analysis (every year) in order to follow the evolu-
tion of the phenomenon for civil protection purposes.
Generally the monitoring data showed that the
DSGSD affecting a slope are generally without move-
REFERENCES
ARPA p
ieMonte
(SC22-SC15-SC05) & r
eGione
p
ieMonte
(s
ettore
p
rotezione
C
ivile
e
s
isteMa
a
ntiCenDi
b
osChivi
a.I.B.) (2009) -
La frana di Grange Orgiera nel Comune di Sampeyre (CN).
a
Mbrosi
C. & C
rosta
G.b. (2002) - Large sackung along major tectonic features in Central Italian Alps. Engineering Geology, 83
(1-3): 183-200.
C
iGna
, F., b
ianChini
, s. & C
asaGli
n. (2012) - How to assess landslide activity and intensity with Persistent Scatterer Interferometry
(PSI): the PSI-based matrix approach. Landslides, 0: 1-19.
C
arraro
F., F
orno
M. G. & b
oCCa
p. C. (1995) - Fenomeni gravitativi nell'alta Val Soana (Torino). Mem. Soc. Geol. It., 50
:
45-58.
C
olesanti
C & W
asoWski
J. (2006) - Investigating landslides with space-borne Synthetic Aperture Radar (SAR) interferometry. ENG.
GEOL.: 88: 173-199,
C
ostantini
M., i
oDiCe
a., M
aGnapane
l. & p
ietranera
l. (2000) - Monitoring terrain movements by means of sparse SAR differential
interferometric measurements. In Proc. IGARSS 2000, Honolulu, USA, 3225-3227.
C
ruDen
D. M. & v
arnes
D.J (1996) - Landslide types and processes. In: t
urner
a.k. & s
Chuster
r.l. (
eDs
.). Landslides, investigation
and mitigation. Special Report 247, Transportation Research Board, National Research Council; National Academy Press,
Washington, D.C.: 36-75.
F
arina
p., C
oloMbo
D., F
uMaGalii
a., M
arks
, F. & M
oretti
s. (2006) - Permanent Scatterers for landslide investigations: outcomes
from the ESA-SLAM project. Engineering Geology, 88 (3-4): 200-217.,
F
erretti
a., p
rati
C. & r
oCCa
F. (2001) - Permanent Scatterers in SAR interferometry. IEEE T Geosci Remote, 39 (1): 8-20.
F
erretti
a., F
uMaGalli
a., n
ovali
F., p
rati
C., r
oCCa
F. & r
uCCi
a. (2011) - A new algorithm for processing Interferometric data-
stacks: SqueeSAR. IEEE T Geosci Remote, 49 (9): 3460-3470
F
orlati
F., G
ioDa
G. & s
Cavia
C. (2001) - Finite element analysis of a deep seated slope deformation. In printing: Rock Mech. and
Rock Eng.
F
orlati
F., b
rovero
M. & C
aMpus
s. (1995) - Alcune considerazioni sulle deformazioni gravitative profonde di versanti inerenti il
territorio piemontese. Atti 2° incontro internazionale dei giovani ricercatori in geologia applicata, Peveragno (Cuneo): 75-81.
G
uzzetti
F., M
anunta
M., a
rDizzone
F., p
epe
a., C
arDinali
M. & z
eni
G. (2009) - Analysis of ground deformation detected using the
background image
D. NOTTI, C. MEISINA, A. COLOMBO, L. LANTERI & F. ZUCCA
360
International Conference Vajont 1963-2013. Thoughts and analyses after 50 years since the catastrophic landslide Padua, Italy - 8-10 October 2013
SBAS-DInSAR technique in Umbria, Central Italy. Pure and Applied Geophysics, 166: 1425-1459.
h
errera
G., G
arCía
-D
avalillo
J.C., M
ulas
J., C
ooksleY
G., M
onserrat
o. & p
anCioli
v. (2009) - Mapping and monitoring
geomorphological processes in mountainous areas using PSI data: Central Pyrenees case study. Nat Hazard Earth Sys, 9: 1587-
1598.
l
auknes
t.r., p
iYush
s
hanker
a., D
ehls
J.F., z
ebker
h.a., h
enDerson
i.h.C. & l
arsen
Y. (2010) - Detailed rockslide mapping
in northern Norway with small baseline and persistent scatterer interferometric SAR time series methods. Remote Sensing of
Environment, 114 (9): 2097-2109. ISSN 0034-4257, 10.1016/j.rse.2010.04.015
l
uino
F. (1999) - The flood and landslide event of November 4-6, 1994 in Piedmont Region (North-West Italy): causes and related
effects in Tanaro Valley. XXII General Assembly of European Geophysical Society, Vienna (Austria). 21-25 April 1997. Ed.
Elsevier Science Ltd, 24 (2): 123-129.
ISPRA (2008) - Landslides in Italy. Special Report 2008. Special Report 83/2008 ISBN: 978-88-448-0355-1
M
aFFeo
& z
anotelelli
(2008) - Movimento franoso dell’Alpe Baranca - Carta Geomorfologica. Municipality of Fobello Internal
Report. ARPA Piemonte archive.
M
eisina
C., z
uCCa
F., F
ossati
D., C
eriani
M. & a
liievi
J. (2006) - Ground deformation monitoring by using the Permanent Scatterers
Technique: the example of the Oltrepo Pavese (Lombardy, Italy). Engineering Geology, 88 (3-4): 240-259.
M
eisina
C., z
uCCa
F., n
otti
D., C
oloMbo
a., C
uCChi
a., s
avio
G., G
ianniCo
C. & b
ianChi
M. (2008) - Geological interpretation of
PSInSAR data at regional scale. Sensors, 8 (11): 7469-7492.
M
ortara
G. & s
orzana
p.F. (1987) - Fenomeni di deformazione gravitativa profonda nell'arco alpino occidentale italiano.
Considerazioni litostrutturali e morfologiche. Boll. Soc. Geol. It., 106: 303-314.
p
isani
G., C
astelli
M. & s
Cavia
C. (2010) - Hydrogeological model and hydraulic behaviour of a large landslide in the Italian Western
Alps. Nat. Hazards Earth Syst. Sci., 10; 2391-2406.
r
aMasCo
M., s
toppa
t. & s
usella
G. (1989) - La deformazione gravitativa profonda di Rosone in Valle dell'Orco. Bollettino della
Società Geologica Italiana, 108 (3): 401-408.
r
eGione
p
ieMonte
& u
niversitè
J. F
ourier
(
eDs
.) (1966) - La frana di Rosone, Valle Orco. In: Rischi generati da grandi movimenti
franosi. Programme INTERREG I Italy-France: 144-177.
Y
in
Y., z
henG
W., l
iu
Y., z
hanG
J. & l
i
X. (2010) - Integration of GPS with InSAR to monitoring of the Jiaju landslide in Sichuan,
China. Landslides, 7: 359-365.
Statistics