English
Italiano
Forgot your password?
New user?
Document Actions
ijege-13_bs-loew-strauhal.pdf
such as the Downie Slide (i
In addition, detailed hydrogeological investigations
have been carried out in hazardous rock-slopes along
populated fjords (Aknes Slide in western Norway;
e.g. G
Italy; e.g. P
of cataclastic breccia or fine-grained gouge, either
stretching over the entire valley slopes between river
bottom and crest line, or only occurring in the lower
slope sector (Fig. 1). The compartments between these
active sliding planes show diverse kinematic behav-
ior, ranging from extensional faulting to compres-
sive thrusting and buckling. Fine-grained shear and
damage zones composed of clay gouge typically act
as boundaries for pore pressure compartments within
which selected fractures form preferential groundwa-
ter pathways. Piezometric levels from isolated bore-
hole sections are remarkably complex and document
both deep as well as shallow and artesian pore pres-
sures at the level of the major sliding surfaces (Fig. 1).
In addition, some slides show the occurrence of deep
unsaturated zones below perched aquifers bounded by
near surface shear zones. Pore pressure transients are
slides, with special focus on saturated and unsaturated
flow in variably fractured rock masses. The study
includes a discussion of critical hydraulic borehole
observations in translational rockslides from British
Columbia, Norway and the Italian Alps and explains
these observations with a generic numerical rockslide
model. Most key observations, such as piezomet-
ric pore pressure levels at the base of translational
rockslides, the location of seepage faces, pressure
compartments, perched groundwater flows, and deep
groundwater tables can be successfully explained.
Open questions relate to the magnitudes of negative
pore pressures in the unsaturated zones and the im-
pacts of strong heterogeneity in hydraulic conductiv-
ity fields. The impacts of suction and seepage forces
on slope stability are discussed.
analysis of pore pressure measurements in brittle
rock-slopes composed of fractured crystalline rock.
Based on the experiences made at Vajont, most of
these borehole-based investigations have been car-
have a deep, basal sliding surface (200-250 m), a bulge
at the toe, and override Quaternary deposits (glacial
till, sand and gravel). Slope and sliding planes dip at
Downie Slide with 18°, and at Little Chief and Dutch-
man’s Ridge with 28° to 36°. The topography surround-
ing Dutchman’s Ridge Slide is not planar, but controlled
by two deep incisions, intersecting in the middle part
of the slope (Figure 1), where the rockslide terminates
as well. Downie and Dutchman’s Ridge Slides were
drained with 2400 and 800 m long drainage galleries lo-
cated below or within the sliding mass from which large
numbers of drainage holes (13’700 and 17’000 drilling
meters) were drilled into the sliding rock mass.
hole investigations were restricted to fewer boreholes [6
at Aknes, 4 at Rosone], where pore pressures were pri-
marily recorded in long open borehole intervals (B
towards the fjord Storfjorden. Sliding occurs along folia-
tion parallel cataclastic shear zones in granitic to dioritic
orthogneisses daylighting along subordinate low-angle
thrust surfaces at different elevations above the shoreline
(G
is of similar depth as the main sliding plane (at about
50 m depth). This leads to substantial drainage of the
sliding mass towards this lateral incision, as supported
by tracer tests with fluorescent dyes (F
shows a typical slope inclination (31°) and sliding planes
at 50 and 120 m depth with similar orientation (P
zones in Augengneiss. The morphology is complex, as
the sliding zone terminates at a ridge separating the Orco
and Piantonetto valleys. The Hochmais-Atemkopf rock-
slide in the Gepatsch Valley (Austria) has been studied
intensively (Z
data include structure, geomechanical and displacement
data but not borehole based hydraulic measurements.
The basal sliding surface dips with 31° (like the mean
slope dip) and occurs at a maximum depth of 220 m. The
currently active mass is constrained to the uppermost
sliding plane - a 4 m thick shear zone - occurring at a
depth of 30-50 m and dipping about 33°.
lake level fluctuations. This paper reviews predispo-
sition and critical hydraulic observations in these 5
crystalline rockslides (chapter 2), develops a generic
hydrogeological model for translational rockslides
with variably saturated fractured rock (chapter 3),
numerically investigates the pore pressure distribu-
tions (including suction) and groundwater flows in
the saturated, as well as unsaturated zones (chapter
4) and finally discusses the potential implications of
variably saturated groundwater flow on translational
rockslide stability (chapter 5). This study is not ap-
plicable for toppling slopes or strongly rotational/
compound slides, where the internal deformation or
slide geometry causes different types of permeability
distributions. The focus of this study is mainly on brit-
tle rockslides as they have a much higher probability
for catastrophic failure than rockslides composed of
weak rocks (e.g. h
strain conditions in a vertical plane, striking nor-
mal to the mean slope orientation. However, some
of these planar slopes and rockslides are laterally
bounded by deep fluvial incisions, which might act
as local receiving streams with out-of-plane directed
flow lines and strains.
ties for hydropower projects along the Columbia River
in British Columbia - following the catastrophe of Va-
jont in 1963. First data sets from these Canadian projects
were published in the early nineties (i
man’s Ridge Slide. Later, detailed hydraulic and me-
chanical subsurface investigations were reported from
the nearby Little Chief Slide (W
Columbia River, with the main foliation and bedding
of the high-grade metamorphic rocks (granodiorites,
gneisses and shists) dipping obliquely into the slope
(Dutchman’s Ridge, Little Chief) or parallel to the slope
cant vertical head gradients, such water table measure-
ments represent complex mixtures of pressure heads,
that cannot be resolved without detailed knowledge of
the local permeability profiles. Water table measure-
ments and optical image logging carried out during
drilling operations at Aknes showed perched groundwa-
ter in the lower section of the highly fractured rockslide
mass above the major sliding plane (at 50 m depth) and
locally unsaturated conditions down to end of 200 m
deep boreholes (T
(GEO-SLOPE i
translational rockslide with similar properties as the
cases discussed above. The model represents transla-
tional rockslides with single or multiple planar sliding
surfaces oriented parallel to a linear slope with a dip
angle of 35° (Fig. 2). A basal sliding zone with a typi-
cal thickness of 5 m was defined 100 m below ground
surface and is heading up to the crest. A secondary
sliding plane with a thickness of 5 m is located at a
depth of 50 m and runs from the toe up to mid-slope
region. Homogenous materials with variable proper-
ties are distributed within the model domain, which
are called stable rock mass, sliding zone and rockslide
mass. Depending on the allocation of these materials
to the different model zones, large and small rock-
slides can be analyzed (Models B, C, D, E; compare
Fig. 5). Like in the examples shown in Fig. 1, a region
of alluvial deposits is located at the toe.
node density adjusted to the region size. For steady-
state modeling the following boundary conditions
are used: a fixed reservoir level at the toe of the
slope (total head at an altitude of 1800 m), potential
groundwater infiltration of 300 mm/a along the slope
area from the crest (altitude of 2800 m) to the toe (al-
titude of 1800 m) and potential seepage face review
at the whole slope. The bottom and right boundaries
of the model were defined as no flow boundaries.
SURE DISTRIBUTIONS
man’s Ridge and Little Chief Slides were investi-
gated by exploratory adits and monitoring boreholes
equipped with large numbers of multiple piezometers.
These piezometers consisted of 1-2 m long monitoring
sections isolated by Westbay multi-packer systems. In
addition, pore pressure profiles were recorded during
drilling using a single packer isolating the lowermost
borehole section. As shown on Figure 1 piezometric
pressures at the base of Downie Slide prior to drainage
showed artesian conditions in the upper slope portion
(500-700 m above Columbia River), shallow phreatic
groundwater levels (up to 100 m below ground sur-
face) in the mid-slope portion and seepage conditions
in the lower-most slope section above the Columbia
River. Such artesian pressures are in fact typical for
rockslides containing massive layers of weak rocks,
such as for example Campo-Vallemaggia (B
the Dutchman’s Ridge Slide prior to drainage have
been about 100 m below ground level for the upper
and middle rockslide section, and at ground surface
close to the Columbia River (Fig. 1). Deep piezometer
intervals in the middle slope section show confined
pore pressures below the basal sliding surface.
over large distances (600 m) even across major clay rich
shear zones, indicating the existence of higher conduc-
tive windows through these shear zones (i
Chief Slides, which show strongly developed pressure
compartments separated by persistent low permeability
fault zones forming the basal sliding plane as well as
steeply dipping pressure boundaries (m
of 50 and 75 m were observed across a 150 mm layer
of clay gouge or a 1.5 m long packer. Consequently,
drainage at Dutchman’s Ridge Slide required a larger
number and length of drain holes. Within pressure com-
partments of all three rockslides, large aperture fractures
locally form highly conductive preferential groundwater
pathways extending over distances of up to 600-700 m.
rock mass assuming a logarithmic normal distribution
of fracture apertures (P
zones was adopted from a laboratory test on a tectonic
shear zone with cohesive fault gouge in granodioritic
rock (F
urated hydraulic conductivity. Figure 3 illustrates these
model input curves for the base case parameter set.
functions. For these parameter variations, saturated
hydraulic conductivities were adapted to alternative
conditions of the presented case studies and the cor-
responding hydraulic conductivity curves were shifted
parallel to the base case curves of Fig. 3 (Tab. 1).
flow simulations in a homogeneous rock-slope with sta-
ble rock mass properties. The positions of the phreatic
ridge and the left hand boundary as a symmetric lake,
the right and left hand model boundaries have no
boundary impacts on the modeling results. The im-
pact of the lower model boundary should be minor,
as hydraulic conductivity of stable rock masses tend
to substantially decrease with depth below ground
surface (e.g. m
the test sites described above. The base case saturated hy-
draulic conductivities are approximately 1E-4 m/s for the
rockslide mass (P
and pressure in unsaturated parts of the slope, it is neces-
sary to define hydraulic conductivity functions (charac-
teristic curves) for the different rock masses. These func-
tions describe how hydraulic conductivity is decreased in
the unsaturated zone relative to groundwater flow in the
fully saturated zone and how hydraulic conductivity is
related to saturation, capillary pressure and pressure head.
sure boundary at an assumed air pressure of zero, and
the seepage face, which is the slope surface below the
intersection with the groundwater table, have similar
shapes but differ by about 500 kPa (50 m) below the
mountain crest. The flow lines are slightly different in
the saturated zones and substantially different in the un-
saturated zones. The indicated negative pore pressures
in the fully saturated model above the water table are
meaningless. Therefore in this paper only results from
saturated/unsaturated flow simulations are shown.
illustrated in Figs 5 and 6. In all 4 model cases the
groundwater table is deep as the ratio between infil-
tration (300 mm/a corresponding to 9.5E-9 m/s) and
saturated rock mass hydraulic conductivity is much
smaller than 1. Also, no significant seepage zone oc-
curs above the lake level, and most of the groundwater
discharge takes place at or below the lake level. In a
rockslide that occupies about half of the slope (model
B and C), most of the groundwater flow occurs at the
base of the rockslide mass, independent of the exist-
ence or absence of a low-permeability sliding zone
with clay gouge. This is due to the large contrasts in
saturated and unsaturated hydraulic conductivities
between the rockslide mass and the stable rock mass.
In the saturated zone, a large pressure differential of
about 350 kPa (35 m of water head) develops across
the low permeability sliding zone in model C. The
leads to almost complete drainage and no groundwa-
ter recharge from precipitation into the deeper aquifer,
showing a quasi-horizontal water table. Suction of up
to 400 kPa develops around the sliding zone.
are plotted in Figs 7 and 8. As shown in these figures,
the saturated zones are very sensitive to hydraulic
conductivity variations in the range of the groundwa-
ter recharge magnitude. For parameter variations 1
and 2 with a substantially reduced stable rock mass
hydraulic conductivity, the water table is close to the
crest of the mountain and generates substantial con-
fined pore pressure conditions below the sliding zone.
For parameter variation 3, with an intermediate satu-
rated hydraulic conductivity of the stable rock mass
the water table below the rockslide lies at an elevation
of 2000 to 2100 m asl and again shows a perched aq-
uifer above the sliding zone and with a deep unsatu-
rated zone below in the upper part of the rockslide. A
reduction in rockslide hydraulic conductivity (param-
eter variation 2 and 3) leads to a substantial increase
in rockslide saturation and a seepage face extending
about 50 m above the lake level (Figs 7 and 8).
for rockslide model C and base case rock mass proper-
which varies by several orders of magnitude, depend-
ing on type and magnitude of gravitational and tectonic
deformation. A high rockslide hydraulic conductivity
of 1E-4 m/s leads to a narrow seepage face or one is
completely situated below of the reservoir level, as in
the case of Dutchman’s Ridge and Little Chief Slide.
A reduction of effective rockslide mass hydraulic con-
ductivity to 1E-6 m/s leads to a substantial increase in
saturation level and a seepage face reaching about 50
m in altitude. However, it is well-known that fractured
crystalline rocks strongly show heterogeneous hydrau-
lic conductivities at multiple scales (e.g. m
lic conductivity and flow porosity variations by many
orders of magnitude and preferential groundwater flow
paths with very high transmissivity. Therefore, the as-
sumption of homogeneity and isotropy applied to the
generic rockslide models is far from reality.
conditions. The characteristic curves for unsaturated
flow and pressure in the stable rock mass and sliding
zone are uncertain and will be critically investigated
in future investigations. Especially the magnitudes of
negative pore-water pressures in the unsaturated parts
of the stable rock mass might be overestimated. For
rockslide masses with a significant amount of total
displacement (> several meters) the apertures of ac-
tive fractures, i.e. fractures along which active dis-
placements are observed, are typically significantly
larger than the capillary pressure threshold of about
ter recharge rates during snow melt in spring with an as-
sumed duration of 30 days, a total recharge of 260 mm
and moderate groundwater recharge events (12+8+40
mm) during 3 summer/fall rainstorms periods of about
1 week duration (e.g. h
lower boundary of the sliding zone in the toe region of
the rockslide. For the assumed parameter values, satu-
rated pore pressures below the groundwater table slow-
ly respond to groundwater recharge from snow melt
and reach peak values about 2.5 months after the end
of the snow melt period. Also, negative pore pressures
of up to 320 kPa (suction) above the groundwater table
(monitoring points 10-13) decay in response to ground-
water recharge from snow melt, but with a longer time
shift. Summer rainstorms do not lead to any visible
pore pressure changes in the landslide toe area.
ality, it can explain most of the key observations from
the well-studied rockslides discussed in chapter 2. Pres-
sure compartments bounded by low permeability fault
gouge layers show the same order of magnitude pres-
sure differentials as observed at Dutchman´s Ridge and
Little Chief Slides. Major groundwater flows occur at
the base of the rockslide mass and drain most of the
slope recharge, as observed at the Aknes, Rosone and
Dutchman´s Ridge rockslides. Below these high con-
ductivity units deep, unsaturated zones might develop
depending on the effective hydraulic conductivity of
the stable rock mass below the rockslides.
- soil suction relationship strongly depends on the
saturation - matric suction relationship (V
slopes with critical slip surfaces located predomi-
nantly above the water table (base case parameter
set), the safety factors significantly increase with
suction and the critical slip surfaces move deeper
into the slope. In the base case parameter set nega-
tive pore pressures of several 100 kPa develop in the
sliding zone (model C and E, Figure 6) which are
expected to have similar stabilizing effects.
GEO.ZT GmbH, and FFG (COMET) for supporting
this work. We thank an anonymous reviewer for valu-
able comments, and Reto Thoeny, Lars Harald Blikra
and Andrew Watson for field work, field visits and
many fruitful discussions.
significant in intact blocks of the rockslide mass. On
the other hand seepage forces at the base of the highly
fractured siding mass are an additional component af-
fecting rockslide stability.
neous soils to unsaturated conditions. They show that
in unsaturated soils two independent stress state vari-
ables define the shear strength τ: the effective angle of
internal friction φ´ and the angle of the shear strength
- matric suction relationship φ
assumed to be independent of suction, the
pressure. At pore-water pressures above the air entry
Journal, 44: 1157-1180.
The issue 16 volume 01 has been published