-
3d GeoPhysicAl modellinG of The vAJonT lAndslider. francese, m.
Giorgi, G. böhm OGS - Istituto Nazionale di Oceanografia e di
Geofisica Sperimentale, Trieste, Italy
Introduction and Motivations. The 1963 Vajont landslide, mostly
because of its size and of the catastrophic effects, has been
studied for several tens of years by many different authors (Selli
and Trevisan, 1964; Rossi and Semenza, 1965; Martinis, 1978;
Hendron and Patton, 1985; Semenza and Ghirotti, 2000). The majority
of the studies focused on the triggering mechanisms (Kilburn and
Petley, 2003) and on the post-failure geology. A comprehensive
review is given by Semenza and Ghirotti (2005). Although the
collapse has been largely studied some of the factors controlling
the dynamic of the movement are still not completely clear. Among
the major issues there is the high velocity of the sliding mass
itself and the movement almost as a unitary rock block that caused
such an unexpected large wave.
A geophysical parameterization of the landslide body could
result in a better insight in both in the understanding of the
geometry settings and the topology of the collapsed units and in a
better estimate of the changes of the elastic properties caused by
the collapse. This last information could be used as a vital
constrain in the recently developed collapse model. Very few
geophysical data are presently available for the landslide body.
Some measurements of the P- and S- wave velocity were undertaken on
the northern scarp of the Monte Toc before the collapse but with
conflicting results and no clear indications about the rock quality
(Caloi and Spadea, 1960, 1961). After the collapse, on behalf of
the Court of Belluno, was carried out a seismic campaign with
P-wave velocity and borehole sonic measurements on the landslide
body (Morelli and Carabelli, 1965).
A new and comprehensive geophysical investigation, based on 2D
and 3D seismic and resistivity imaging was then undertaken since
the year 2011 and it’s still in progress. This geophysical
experiment was designed and conducted integrating the
re-interpreted geological (Bistacchi et al., 2013) and structural
data (Massironi et al., 2013). A novel series of borehole
stratigraphies made recently available by ENEL were also
incorporated in the geophysical modelling. Finally some accurate
surface models obtained processing aerial and terrestrial laser
data were extremely useful to constrain the inversion of
resistivity and seismic data and to properly account for the
deformation of the electrical field caused by the rough
topography.
A geophysical profile collected along the rock wall below Casso
on the other side of the Vajont valley was used a reference for the
geophysical response of the geological units involved in the
landslide. The geophysical images of the landslide body, especially
in the near surface, showed a very good correlation with the
post-failure geology (Rossi and Semenza, 1965).
An initial correlation between lithology and physical parameters
has been proposed for the various geological units embedded in the
landslide mass. This 3D physical model of the landslide introduces
a series of new constrains for an accurate numerical simulation of
the landslide kinematics.
Geological setting. Geology in the Vajont valley is comprised of
a Jurassic-Cretaceous carbonate sequence (Carloni and Mazzanti,
1974; Semenza, 1965; Martinis, 1978). The thickness of the various
formations, at the landslide scale, could be considered roughly
constant. The base of the Jurassic sequence is marked by a massive
(Vajont Formation) limestone (350-400 m) overtopped by a layered
cherty (10-40 m) limestone (Fonzaso Formation) and followed by the
nodular limestone of the Ammonitico Rosso Formation (15 m). The
Cretaceous sequence is comprised of the Soccher Limestone (200-250
m) and of the layered marly limestones and marls of the Scaglia
Rossa Formation (about 300 m). The upper part of the Fonzaso
Formation and the Soccher Limestone Formation were involved in the
landslide. Rossi and Semenza (1965), analyzing the post-failure
accumulation, mapped six different lithological members indicated
with letters from a to f (from the older to the youngest).
190
GNGTS 2013 SeSSione 3.3
131218 - OGS.Atti.32_vol.3.27.indd 190 04/11/13 10.39
-
Some very thin clayey interbeds were identified in the upper
part of the Fonzaso Formation (unit a’) and indicated by Hendron
and Patton (1985) as the stratigraphic level where the initial
sliding happened. There are just few outcrops of this unit as
during the failure the unit was dragged and crunched along the
sliding surface. The bottom level of unit a’’ is represented by the
Ammonitico Rosso Formation (Rossi and Semenza, 1965) while the body
of unit a’’ is comprised of an alternation of layered cherty
limestone and marly limestones. Unit b is represented by a thin
conglomerate layer and it is very important because it represents
an isochron surface and it’s a marker level within the landslide.
Units c, d and e are comprised of massive limestones grading to
layered marly and cherty limestones. Finally unit f represents the
top of the Soccher Formation and it is comprised of layered marly
cherty limestones.
A review of available structural data, associated with some new
field observations, has been recently completed (Massironi et al.,
2013). According to these data the E-W trending Erto syncline
(Giudici and Semenza, 1960) is further folded by a N-S trending
syncline with it’s axis elongated along the pre-landslide
Massalezza valley. The interference of these two sets is exposed on
the sliding surface and in some cases the undulations generated
steps that interrupted the continuity of the surface itself.
The ensemble of these structural features appears to be a major
constrain in the kinematics of the landslide. In particular the
association of the stratigraphy and of the N-trending bedding
planes with the curved shape of the sliding surface converging on
the Massalezza valley behave as a track for the 1963 failure.
The physical database. The success of the geophysical
experiment, in reconstructing the landslide stratigraphy and
structure, strongly depends on the possibility of measuring some
key properties of the in-situ formations. A reliable and accurate
physical reference model of the upper part of the Fonzaso Formation
and of the Soccher Limestone Formation was then constructed to
reduce the uncertainty in correlating physical properties (e.g.
velocity and resistivity) to specific lithological units. This
robust tie could be also a constrain while attempting to estimate
the degree of fracturing of the landslide mass itself. The first
indicator of physical properties is the rock coherency. Hard rocks
(HR) will be probably characterized both by high velocity and high
resistivity while soft rocks (SR) are generally low velocity and
low resistivity. In this simplified approach units a’, a’’ and f
could be classified as moderately soft rocks while units b, c, d
and e are mostly referable to moderately hard rocks.
Some of the boreholes drilled on the landslide accumulations
where also used as major constrains in assisting geophysical data
analysis, processing and interpretation. The majority of these
boreholes reached the depth of the in-situ bedrock.
The few geophysical data available for the left side of the
Vajont valley, before the failure, are represented by four seismic
profiles collected immediately after the discovery of the
paleolandslide (Caloi and Spadea, 1960, 1961). Results from a first
seismic survey indicated how the P-wave velocity in the uppermost
layers was higher than 5000 m/s. Some authors (Semenza, 1965; 2001;
Selli and Trevisan, 1964) pointed out that these values appear to
be too high also for compact limestones. A second seismic survey
(Semenza, 2001) carried out after the discovery of the perimetrical
crack, showed more realistic P-wave velocities around 2500-3000
m/s. The Court of Belluno to collect new evidences for the trail
during the preparation of the case required a new seismic survey.
The major purpose of this new survey was to collect data inside and
outside the landslide area. Seismic velocities were measured at few
locations (Morelli and Carabelli, 2005) inside and outside the
landslide area. Unfortunately the measurements on the landslide
body were carried out in boreholes mostly below the sliding surface
making this new data not directly comparable with the previous
surveys. Other measurements carried out on the Cretaceous sequence
outside the landslide show P-wave velocities ranging from 2100 m/s
to 3000 m/s. These numbers are more and less similar to the values
measured during the second seismic survey before the failure.
191
GNGTS 2013 SeSSione 3.3
131218 - OGS.Atti.32_vol.3.27.indd 191 04/11/13 10.39
-
Under a theoretical point of view (Telford et al., 1990; Dobrin
and Savit, 1988) the seismic and electrical properties of a medium
mostly depend upon density (also related to lithology and age),
porosity and fluid content. Water in the Vajont landslide body is
almost absent due to the very high permeability of the collapsed
mass. The water table in the accumulation is directly related to
the water level in the residual lake located eastern than the
landslide body. Several attempts to measure the water level in the
boreholes were made but a water table has been never detected. The
unknowns are then two: lithology and porosity (or fracturing) and
in the equation there is just one parameter (velocity and
resistivity). To reduce the uncertainty resistivity and seismic
velocities should be measured on the same formations but outside
the landslide. For this purpose a reference geophysical profile was
then collected along the rock wall below the village of Casso. On
this rock wall there is almost a full exposure of the geological
sequence involved in the landslide. Unit a’ is the only missing as
it’s covered by the talus deposits.
Geophysical data acquisition and processing. Geophysical data
acquisition was not straightforward due to the complex morphology
of the accumulation and of the associated complications in coupling
electrodes and geophones.
The geophysical reference profile, along the rock wall, was
comprised of a 24-channel seismic line and of a 48-electrodes
electrical line. The spacing was set to 10 m and to 5 m in the
seismic line and in the electrical line respectively. Three
component geophones were firmly tightened with the rock using
special screws while the electrodes were hammered into holes filled
with conductive medical gel. Data resulted good quality and the
tomographic inversion of both the two datasets was carried out with
a misfit lower than 5%.
The landslide body is comprised of three major lobes: the
Massalezza lobe and two separate masses (defined as the “eastern
lobe” and the “western lobe”) probably collapsed just after the
washout of the Massalezza lobe (Semenza, 1965, 2001).
Fig. 1 – Key features of the geophysical campaign on the Vajont
landslide.
192
GNGTS 2013 SeSSione 3.3
131218 - OGS.Atti.32_vol.3.27.indd 192 04/11/13 10.39
-
Seismic data were collected on the Massalezza lobe only (lobe A
in Fig. 1) using a DMT Summit modular system with more than 200
double channel A/D conversion units. Each receiving station was
equipped with a three component 10-Hz geophone to detect the
incoming signal. The three component sensors were laid out along
four lines (L100, L200, L300, L400). The average station spacing
was 10.0 m for a total of 276*3 live channels. Elastic waves were
generated and propagated into the ground using a medium size
vibrator (both in P-wave and S-wave modes). The source stations
were located along the roads bounding and crossing the Massalezza
lobe.
Recorded seismic data were generally of good quality; first
breaks, in P-wave mode, were sharp and easy to pick even at offsets
larger than 500 m while in S-wave mode the signal was slightly
lower amplitude. A total of about 80000 first arrivals were picked
and pre-processed and lately inverted using a 3D traveltime
tomography. Inversion was carried out with the CAT3D proprietary
software. Data resolution was improved using staggered grids
(Vesnaver and Bohm, 2000).
Resistivity data were collected on the two landslide lobes (A
and B in Fig. 1) using a 48-electrode Syscal R1 system, a
96-electrode Syscal Pro system and an experimental 36-electrode
wireless resistivity system manufactured by MultiPhase Technologies
LLC. A total of six ERT profiles, in Wenner and in
Wenner-Schlumberger configuration, and two ERT volumes, in
pole-dipole configuration, were collected on the Massalezza lobe.
Three additional ERT profiles and a ERT volume were collected on
the eastern lobe. Data were collected in separate sessions during
early spring and middle autumn after several days of heavy rain to
improve the coupling. The high permeability of the landslide mass
allowed for assuming the subsurface layers as homogeneously
wet.
Recorded resistivity data also resulted of good quality and just
few points needed to be removed from the dataset prior to run the
inversions. The inversion was carried out using the package ERTLAB+
that is based on a sophisticated reweight (Morelli and LaBreque,
1996) of the inversion parameters at each iteration.
Results and discussion. Unit a’’ in the reference section
appears to be low both in resistivity (0.5-1.0 KΩ·m) and P-wave
velocity (2200-2700 m/s). Unit b, is very thin and, due to the
large sensor spacing, is outside the resolution capability of both
the two techniques. Units c, d and e in the middle part of the
section are moderately compact limestones and appear show a similar
geophysical response. In these units the resistivity is fairly high
(2.5-4.5 KΩ·m) as well as the P-wave velocity (3400-3800 m/s). At
the top of the rock wall, where unit f outcrops, there is a sudden
lowering of both the resistivity (0.5-1.0 KΩ·m) and the velocity
(2300-2800 m/s).
The terrain resistivity in the landslide accumulation exhibit a
large degree of fluctuation. The majority of the values range from
0.15-0.20 KΩ·m to 3.50-4.00 KΩ·m.
The resistivity distribution along the profile ERT1 (Fig. 2)
appears to be quite complex and the Average values are slightly
lower as compared to the reference profile. This is somewhat
related to the fracturing and to the severe changes occurred in the
geological layers involved in the landslide. Unfortunately along
profile ERT1 there are no borehole data available to constrain the
interpretation. The deep conductive unit a’’ was utilized to define
the initial geological layout along the profile. The prominent
structure is a narrow syncline fold located in the middle portion
of the profile. The axis is probably elongated along the old
Massalezza valley. The bottom layers are probably belonging to the
a’’ unit while in the top layers there are the typical lithologies
of the c unit. Some resistive bodies could be associated with a
partly undifferentiated c-d-e lithological sequence while the
conductive unit, visible in the distance interval 50-200 m,
according to Rossi and Semenza (1965) still belongs to units c, d
and e. Possible explanation of this anomaly are the high degree of
fracturing in the landslide or the occurrence of lateral changes of
lithology moving from the right to the left side of the Vajont
valley. These folding with an apparent N-S trend of the
lithological units is also visible on the sliding surface
(Massironi et al., 2013). The preservation of the structural
settings, before and
193
GNGTS 2013 SeSSione 3.3
131218 - OGS.Atti.32_vol.3.27.indd 193 04/11/13 10.39
-
after the 1963 failure, confirm the hypothesis of a rigid
roto-translation of the collapsed mass. The large anomalous
thickness of the c-d-e sequence is probably caused by some
detachment phenomena occurred along existing or newly developed
high-angle discontinuities (Rossi and Semenza, 1965; Martinis,
1978) during the failure. The profile also show some low angle
detachment planes (see for example at the x-coordinate interval
200-300 m) that could explain the anomalous thickness of the c, d
and e units in depth.
The resistivity distribution along profile ERT5 (Fig. 2),
crossing lobe B, is more homogeneous and it’s more and less
comparable with the post-failure geological map of Rossi and
Semenza (1965). The outcrops are represented by an undifferentiated
unit a covering almost the entire lobe B (Fig. 1) and by some minor
exposures of the unit b. The general structure appears to be folded
up into an anticline with some major displacements. In this case
also the geometry of the bottom layers of lithological unit a’’ was
utilized as the prominent geophysical marker to define the
settings. In the reference section unit a’ is not exposed and hence
there are no indications about its possible geophysical response.
Carloni and Mazzanti (1964) suggest a total thickness of about
80-90 for the undifferentiated unit a with a maximum thickness of
unit a’’ around 50 m. According to several other authors (Rossi and
Semenza, 1965; Martinis, 1978; Hendron and Patton, 1985) unit a’ is
similar to unit a’’ with the sole difference of the presence of the
interbedded clays in the base layers. The two units should then
exhibit a comparable geophysical signature. The increase of the
resistivity in the deeper portion of the profile (where unit a’ was
expected) is rather difficult to explain. There are possible
explanations: the upper part of unit a’ is more resistive than
expected; more realistically unit a’’ overthrusts the resistive
terrains belonging to the c-d units in the deeper portion of
profile. Several high and low angle discontinuities are also
visible in the profile. These planes disrupt the former continuity
of the geological layers generating an ensemble of pop-up like
structures. Unfortunately on this lobe the sliding surface is not
constrained by post-landslide data and also during the failure
occurred a partial overlap of lobe B on the Massalezza lobe. The
deeper portion of the geophysical image could be then really
complex.
Fig. 2 – 2-D Resistivity inversion images along tomography ERT 1
(top) and ERT 5 (bottom) (see Fig. 1). The investigated landslide
mass could be then considered dry.
194
GNGTS 2013 SeSSione 3.3
131218 - OGS.Atti.32_vol.3.27.indd 194 04/11/13 10.39
-
The P-wave velocity field in the landslide accumulation range
from 700-1000 m/s to 3500-4000 m/s. Seismic data resulted
comparable to electrical data with a reasonable degree of
confidence.
The correspondence between resistivity and P-wave velocity has
been analysed in details along the section of profile ERT1 (Fig.
3). A profile was extracted from the 3D velocity model and compared
to the 2D ERT. The syncline structure below the Massalezza ditch is
visible also in the seismic data but it is not completely resolved
as in the resistivity image probably because the geophone line is
located 60 m northern of the electrode line. Right in the middle of
the P-wave section there is a high-velocity bottom layer that is
not visible in the resistivity section. The high-velocity layer is
located at a depth where the signal to noise ratio of the
resistivity data is very low. It is known that a deep resistive
layer below a conductive layer requires very large AB spacing to be
sampled by an electrical field that is forced into the conductor.
The seismic data are more reliable because there is a geophone line
exactly along the Massalezza ditch (Fig. 1) and the sources are
located on the nearby road. This high velocity layer (or high
resistivity) below unit a’’ is quite difficult to explain without
assuming the presence of a detachment plane that duplicate the
sequence.
Conclusion. The study of a large landslide based on the 2D and
3D geophysical parameterization of the involved geological units is
particularly difficult due to the expected complexity of a chaotic
accumulation. The Vajont landslide, given its large volume, was
even more complicated. The different units involved in the
landslide, due to their lithological changes were expected to have
a distinct geophysical signature. The reference section collected
along the rock wall below the village of Casso confirmed this
hypothesis. The pre-slide geological sequence from the bottom to
the top is a sort of sandwich of “conductive/low velocity” –
“resistive/high velocity” – “conductive/low velocity” layers.
The initial correlation of the geophysical images from the
landslide body with the post-failure geology confirmed the
observations from the reference section. The conductive unit a’’ is
an excellent geophysical marker to guide the interpretation. The
pre-landslide stratigraphy appears to be quite well preserved in
the shallow layers while in depth the geophysical response is
rather complex. Both the resistivity and the seismic images along
profile ERT1 highlight a syncline that is fully exposed on the
sliding surface. The geophysical images along profile ERT1 also
show a series of small-scale folds with north-south axes that are
probably pre-landslide as the stress occurred during the failure
folded the strata generating east-west axes. This mode of folding
is clearly visible in resistivity profile ERT5, collected on lobe
B.
These results are satisfactory but further investigations are
anyhow required to achieve a reliable reconstruction of the
landslide accumulation settings. Unfortunately borehole data
collected, before and after the landslide, are of limited use
because of the high lateral variability of the physical properties
occurring in the collapsed mass. This is somewhat confirmed by the
borehole stratigraphy reconstructed by mean of
micropalaeontological analysis of drilled
Fig. 3 – Comparison between 3D seismic tomography (top) and 2D
electrical resistivity tomography (bottom) response. The profiles
are oriented from W to E. The geophysical data are presented as 10m
by 10m cell scalars without interpolation. The two sections are
computed along the trace of profile ERT1.
195
GNGTS 2013 SeSSione 3.3
131218 - OGS.Atti.32_vol.3.27.indd 195 04/11/13 10.39
-
samples. Results of this last analysis suggests the existance of
several duplication of the original in-situ sequence.
Further work that will be undertaken in a very near future
includes: 1) geophysical parameterization of unit a’, 2) inversion
of the electrical data into a 3D resistivity volume of the entire
landslide, 3) correlation of the seismic and of the resistivity
responses to improve interpretation and finally 4) processing of
S-wave data in order to estimate the elastic parameters of the
landslide body.Acknowledgements. We acknowledge the Friuli
Venezia-Giulia Region for providing the funding for the project
(project 35935/2010). A special thank to Alessia Rosolen for her
personal support and to Ketty Segatti. A final thank to Giovanni
Rigatto and to Nuccio Bucceri for their assistance during
topographic measurements.referencesBistacchi, A., Massironi, M.,
Superchi, L., Zorzi, L., Francese, R., Giorgi, M., Genevois, R.,
and Chistolini, F., 2013. A 3D
Geological Model of the 1963 Vajont Landslide. Vajont 2013
Conference.Caloi, P., and Spadea, M., C., 1960. Serie di esperienze
geosismiche seguite in sponda sinistra a monte della diga del
Vajont
(dicembre 1959). SADE Technical Report. Caloi, P., and Spadea,
M., C., 1961. Indagine geosismica condotta nel mese di dicembre
1960 a monte della diga del Vajont
in sponda sinistra. SADE Technical Report. Carloni, G., C., and
Mazzanti, R., 1964. Rilevamento geologico della frana del Vajont,
In Selli R. et al., “La frana del
Vajont”, Giornale di Geologia, 2, 32/1.Castellanza, R.,
Agliardi, F., Bistacchi, A., Massironi, M., Crosta, G.B., and
Genevois, R., 2013. 3D finite-element
modelling of the Vajont landslide initiation stage. Vajont 2013
Conference.Dobrin, M.B., and Savit, C.H., 1988. Introduction to
Geophysical Prospecting – IV Edition. McGraw-Hill, 867 pp.Genevois
R., and Ghirotti, M., 2005. The 1963 Vaiont Landslide, Giornale di
Geologia Applicata 1, 41-52.Giudici, F., and Semenza, E., 1960.
Studio geologico del serbatoio del Vajont. SADE Technical
Report.Hendron A.J., and Patton, F.D., 1985. The Vaiont slide, a
geotechnical analysis based on new geological observations of
the failure surface. Tech. Rep. GL-85–5, Department of the Army,
US Army Cops of Engineers, Washinghton D.C., 2 vols.
Kilburn, C.J., and Petley, D.N., 2003. Forecasting giant,
catastrophic slope collapse: lessons from Vajont, Northern Italy.
Geomorphology, 54, 1-2, 21-32.
Martinis B., 1978. Contributo alla stratigrafia dei dintorni di
Erto-Casso (Pordenone) ed alla conoscenza delle caratteristiche
strutturali e meccaniche della frana del Vajont. Memorie di Scienze
Geologiche, Università di Padova, 32, 1-33.
Massironi, M., Superchi, L., Zampieri, D., Bistacchi, A.,
Ravagnan, R., Bergamo, A., Ghirotti, M., and Genevois, R., 2013.
Geological Structures of the Vajont landslide. Vajont 2013
Conference.
Morelli, C., and Carabelli, E., 1965. Misure di velocità delle
onde elastiche longitudinali nella roccia del bacino del Vajont.
Court of Belluno. Lerici Foundation. Research 410.
Morelli, G., and LaBrecque, D. J., 1996. Robust scheme for ERT
inverse modelling. European Journal of Environmental and
Engineering Geophysics, 2, 1-14.
Müller, L., 1959. Talsperre Vaiont – 6 Geotechnischer Bericht.
SADE Technical Report.Müller, L., 1964, The rock slide in the
Vaiont valley. Felsmechanik und Ingenieur-geologie, 2,
148-212.Petley, D.N., and Petley, D.J., 2006. On the initiation of
large rockslides: perspectives from a new analysis of the
Vaiont
movement record. In Massive Rock Slope Failure. Mugnozza, G.,
Strom, A. & Hermanns, R. L. Rotterdam. NATO Science Series,
Earth and Environmental Sciences, 49, 77-84.
Rossi, D., and Semenza, E., 1965. Carte Geologiche del versante
settentrionale del M. Toc e zone limitrofe prima e dopo il fenomeno
di scivolamento del 9 Ottobre 1963. Scala 1:5000. Ist. Geol. Univ.
Ferrara, Italy.
Semenza, E., 1960. Nuovi studi tettonici nella valle del Vajont
e zone limitrofe. Acc. Naz. Lincei, Rend. Cl. Sc. MMFFNN, 8,
28/2.
Semenza, E., 1965. Sintesi degli studi geologici sulla frana del
Vaiont dal 1959 al 1964. Memorie del Museo Tridentino di Scienze
Naturali, 16, 1-52.
Semenza, E., 2001. La storia del Vaiont raccontata dal geologo
che ha scoperto la frana. Tecomproject Ed., Ferrara, Italy. 279
pp.
Semenza, E., and Ghirotti M., 2000. History of the 1963 Vaiont
Slide. The importance of the geological factors to recognise the
ancient landslide. Bulletin of Engineering Geology and the
Environment, 59, 87-97.
Telford, W., M., Geldart. L., P., and Sheriff, R., E., 1990.
Applied Geophysics – II edition. Cambridge University Press, 770
pp.
Vesnaver, A., and Bohm G., 2000. Staggered or adapted grids for
seismic tomography?, The Leading Edge, 19, 9, 944-950.
196
GNGTS 2013 SeSSione 3.3
131218 - OGS.Atti.32_vol.3.27.indd 196 04/11/13 10.39