Vous êtes sur la page 1sur 13

Entrainment of debris in rock avalanches:

An analysis of a long run-out mechanism

Oldrich Hungr†
Department of Earth and Ocean Sciences, University of British Columbia, 6339 Stores Road, Vancouver, British Columbia V6T 1Z4,
Canada
S.G. Evans
Department of Earth Sciences, University of Waterloo, 200 University Avenue West, Waterloo, Ontario N2L 3G1, Canada

ABSTRACT INTRODUCTION initial failure. In such cases, the flow of lique-


fied soil unquestionably adds to the mobility
Many rock avalanches entrain and liquefy The 1903 Frank Slide of southern Alberta, of the event as a whole. Such landslides have a
saturated soil from their paths. Evidence for a rock avalanche of 36 × 106 m3, was the most transitional character and deserve special iden-
this includes mud displaced from the mar- lethal landslide in the history of both Canada tity both in name and in terms of descriptive
gins of rock avalanche deposits, substrate and the United States, destroying a part of and analytical treatment. Their characteristics
material smeared along the base of deposits, the town of Frank, with 73 fatalities. Most of also shed light on the behavior of other rock
extrusion of liquefied soil upward through the damage in Frank was not, however, due to avalanches, in which debris entrainment is less
the deposits, and increases of volume. A impact or burial by rock debris. Instead, homes prominent, but nevertheless present.
hypothesis first suggested in 1881 and since and other buildings were impacted by a lateral
reinforced by several authors suggests that outflow of mud, the liquefied alluvium from the CONCEPTS AND TERMINOLOGY
entrainment of substrate material increases floodplain of the Old Man River, expelled from
mobility. Although the process has been the western margin of the landslide (McConnell When a rockslide mass disintegrates and
discussed in the literature for more than and Brock, 1904). fragments in the process of becoming a rock
100 years, few detailed and quantitative Liquefied substrate can play a dominant avalanche, an initial volume increase occurs.
descriptions exist. The main purpose of this role in rock avalanche motion. Its entrainment A few estimates of the volume increase exist in
paper is to describe two recent cases from serves to increase the volume of the landslide, the literature, ranging from 7% to 26% (Hungr,
British Columbia, Canada, where rockslides but may also lead to a change in the character 1981, p. 306). However, such field estimates are
entrained substrate on a very large scale, of material in the basal part of the moving mass, not reliable, as they require accurate measure-
influencing the character of the events. enhancing mobility. This is the oldest among ment of both source and deposit volumes. In all
Estimated volume balance curves, based numerous hypotheses that attempt to explain the the examples used in this paper, it is assumed
on detailed field mapping, are provided for high mobility of rock avalanches, having been that fragmentation produces a volume increase
both cases. Dynamic analyses are carried first proposed by Buss and Heim in 1881 (see of ~25%. This is at the center of the typical
out using a numerical model and using the below). Abele (1974, 1997), Sassa (1988) and range of measured porosities of loosely placed
same set of rheological parameters. The Voight and Sousa (1994) are among its later pro- well-graded crushed rock, which is 18%–35%
mechanism of material entrainment and ponents. The underlying process of liquefaction (Sherard et al., 1963, p. 657).
displacement is discussed. The data suggest by rapid undrained loading was documented by It is of interest to note that such a volume
that rapid rock failures entraining very large Hutchinson and Bhandari (1971) for earth flows increase negates nearly all possibility of a fluid
quantities of saturated substrate material and Sassa (1985) for debris flows. However, few pore pressure existing in the fragmented rock
represent a special type of landslide, tran- quantitative descriptions of relevant case his- itself, at least in the initial stages of motion.
sitional between rock avalanche and debris tories exist to substantiate this theory for rock In-place water content of most rock masses is
avalanche. Many rock avalanches can thus avalanches. In general, it is difficult to assess the negligible. The large pore space newly created
be seen as end members of a continuum of role of basal liquefaction in field studies, as the by fragmentation must thus be essentially dry,
phenomena involving rock failure followed surface of rock avalanche deposits typically dis- until sufficient time passes to allow water from
by interaction with saturated substrate. plays dry, coarse rock fragments. However, mud some source to flow into it. On a typical rock
is often observed around the deposit margins avalanche path, however, significant water can
Keywords: rock avalanche, debris avalanche, (e.g., Buss and Heim, 1881; Cruden and Hungr, come only in association with saturated soil
dynamic analysis, runout, entrainment, Brit- 1986), and a lubricating layer of such material entrained from the path downslope of the toe of
ish Columbia. may well remain concealed beneath the coarse the rupture surface.
debris. The situation is more transparent where During or following fragmentation, further
the amount of debris entrained along the path is volume increase occurs by entrainment of sub-

E-mail: ohungr@eos.ubc.ca. very large, relative to the volume of rock in the strate material, partly or completely liquefied by

GSA Bulletin; September/October 2004; v. 116; no. 9/10; p. 1240–1252; doi: 10.1130/B25362.1; 12 figures; Data Repository item 2004126.

For permission to copy, contact editing@geosociety.org


1240 © 2004 Geological Society of America
ENTRAINMENT OF DEBRIS IN ROCK AVALANCHES

rapid undrained loading (Hutchinson and Bhan- observed in each segment of the path of a par- of the 1970 event increased to 53 × 106 m3 as a
dari, 1971). Rock avalanche substrate is gener- ticular description. There are at present no theo- result of the incorporation of ice, glacial till, and
ally saturated in temperate climates, as proven retical or even empirical means of predicting colluvium from the path of the debris avalanche,
by the often-observed presence of liquid mud in yield rates a priori. Records of yield rates from suggesting an ER of 2.0. The extreme discrep-
or near the debris (see below). Given the rapid known events are useful to establish precedent ancy between these various accounts illustrates
motion and large volume of rock avalanches, for empirical predictions. the difficulty of making volume estimates in the
one can speculate that even incompletely satu- field.
rated and moderately coarse-grained soil can DEBRIS ENTRAINMENT IN ROCK A 1987 rockslide of 6 × 106 m3 from Cerro
liquefy under the intense undrained loading AVALANCHES Rabicano in Región Metropolitana, Chile,
imparted by masses of fragmented rock (cf. disintegrated and produced a debris avalanche
Sassa, 1985; Dawson et al., 1998). Impact Loading of Colluvium of a total volume of 15 × 106 m3, by entraining
To quantify the entrainment process, an snow, ice, and valley infill deposits (ER = 1.0).
Entrainment Ratio (ER) can be defined as the The 1939 landslide at Fidaz, Switzerland, The rockslide-debris avalanche traveled 17 km
ratio between the volume of debris entrained began as a 1 × 105 m3 rock failure from the head along the Estero Parraguire River on an average
from the path and the expanded volume of rock scarp of the prehistoric Flims landslide, but slope of 4.5°. It formed a short-lived natural dam
fragments produced by the initial rock failure: grew to a total volume of 4 × 105 m3 (ER = 2.2) at the confluence with Río Colorado. Breach of
by expanding during fragmentation and entrain- the dam produced a debris flood along the larger
VEntrained VE
ER = = , (1) ing a part of the colluvial apron surrounding stream, which continued to a total travel dis-
VFragmented VR (1 + FF ) the source cliff (Niederer, 1941). In 1953, a 1 tance of 57 km, descending a vertical distance
× 104 m3 block of rock detached from a cliff at of ~3400 m (Hauser, 2002).
where VE is the volume of the entrained material, Modalen, Norway, as described by Kolderup Another case for which data exist is the earth-
VR is the volume of the initial rockslide and FF is (1955) and entrained talus, producing a flow of quake-triggered 1984 Mount Ontake debris
the fractional amount of volume expansion due 1.15 × 105 m3 (ER = 8.2). In Brazil, a rock slab avalanche. The volume of the initial failure is
to fragmentation (0.25). The total volume of the with a volume of 6000 m3 fell from a granite known to be 34 × 106 m3 (Oyagi, 1987). Oyagi
landslide deposits equals VR(1+FF) + VE. Scott rock slope, bounced off a lower rock slope, and also noted that the rock avalanche entrained
(1988) used a similar index, which he referred to impacted on talus. This mobilized a large debris quantities of residual, colluvial, and alluvial
as the Bulking Factor, for volcanic lahars. avalanche that destroyed a clinic, causing the soil from the valleys through which it passed.
Hungr et al. (2001) proposed that the term death of ~30 people (Barros et al., 1988). (The Detailed mapping of the deposits (Endo et al.,
“rock slide–debris avalanche” be used to final volume was not reported in this case.) 1989) showed that the volume of the deposits is
describe landslides that begin by the failure of 56 × 106 m3, an ER of ~0.3 after a 25% increase
a rock slope and proceed to entrain large quan- Mobilization of Glacial and Residual Soils of the initial failure volume due to fragmenta-
tities of debris (talus, colluvium, residual soil, tion is taken into account.
glacial drift, alluvium, peat, or other materials) Devastating rock avalanches occurred on
from their path. Since the presence of small the west side of the north peak of Nevados Interaction with Alluvium
quantities of entrainment is often difficult to Huascarán, Peru, in 1962 and 1970. In January
detect, it is suggested that the composite term be 1962, ~2.5–3 × 106 m3 of glacier ice and grano- Abele (1974; 1997) was one of the first
applied only to those events where ER exceeds diorite broke from Huascarán’s ice cap–covered workers to suggest a link between long travel
0.25. This definition contrasts with earlier uses northern summit. After falling 1000 m down an distance of rock avalanches and interaction with
of the term “rock slide–debris avalanche” (e.g., almost vertical slope, this mass struck Glacier valley fills, on the basis of detailed field map-
Varnes, 1978). However, it is advantageous, as 511 and incorporated more glacier ice and ping of prehistoric rock avalanche deposits in
it clearly points out the important role of mate- large volumes of lateral moraine. In its passage the Alps. He proposed a mechanism whereby
rial entrainment. Landslides involving flow-like through a series of steep-sided ravines, more a combined movement of a rockslide mass rid-
motion of fragmented rock with only modest material was added to the debris avalanche ing on water-saturated silt, sand, and gravel can
entrainment can be described by the well- until it spread out on a fan where the town of increase both run-out distance and the spreading
established term “rock avalanches,” defined by Ranrahirca lay. The total deposited volume was of the debris.
Hungr et al. (2001). estimated to be 13 × 106 m3 (Morales, 1966), Field observers of historical rock avalanches
A numerical model for simulating the over four times the original detached volume have often remarked on the presence of a splash
dynamic behavior of landslides entraining (ER = 2.8). zone around the margins of the debris. Heim
mass was presented by Hungr (1995). Use of In May 1970, another landslide was trig- (1932) noted a spritzzone of fine, liquid soil
the model requires that the initial rock failure gered by an earthquake from the same source surrounding the distal and lateral margins of
volume and the rate of debris entrainment be on Huascarán. Considerable uncertainty sur- the Elm Slide debris, contrasting with the steep,
known. The latter can be expressed as a yield rounds the volume of the initial detachment. well-defined edge of the rock debris sheet from
rate (Yi), the volume entrained from the path Plafker and Ericksen (1978) estimated that 50 beneath which it had evidently been expelled.
and incorporated into the moving avalanche per × 106 m3 of material was involved in the initial Similar “splash” margins were described in
unit length of the path, in units of m3/m (Hungr fall (5 × 106 m3 of ice), but did not mention any detail by Cruden and Hungr (1986) surrounding
et al., 1984). Obviously, the yield rate depends entrainment of morainal soil. Lliboutry (1975) the Frank Slide debris. The surface of the Frank
on many factors, especially density, gradation estimated an initial volume of 9 × 106 m3. Ghig- Slide debris deposit at the foot of the proximal
and degree of saturation of the path substrate, lino Antúnez (1970) estimated a total of 14 × slope is several meters below the original bed of
the slope angle, and the current mass of the 106 m3 (5 × 106 m3 rock and 9 × 106 m3 ice). the Old Man River, showing deep erosion of the
avalanche. A specific yield rate value may be According to his data, the total deposit volume valley deposits (McConnell and Brock, 1904).

Geological Society of America Bulletin, September/October 2004 1241


HUNGR and EVANS

Evans et al. (1994) found alluvial gravels and space was infilled by deposits. We attempted to of a foliation-parallel joint and a stress-relief
wood debris which had been entrained from extrapolate the shape of the original ground sur- joint sub-parallel to the original cliff face. A
a river floodplain and transported onto a rock face as observed at the path margins. The esti- wedge-shaped block of rock, having an esti-
shelf more than 600 m above the original valley mated spot thicknesses were used to estimate mated volume of ~7.5 × 104 m3 detached along
floor by the runup of the prehistoric Avalanche the average thickness, with the use of subjective the two discontinuities and collapsed toward
Lake rock avalanche in the Northwest Territo- judgment. The volumes of the rockslide source the valley. Assuming a fragmentation volume
ries, Canada. The 1964 Hope Slide in southern areas were estimated by reconstructing smooth increase of 25%, the initial rockslide produced
British Columbia, the 1987 Val Pola Slide in contours across the scars. ~9.4 × 104 m3 of broken rock.
northern Italy, and the 1985 North Nahanni Corrections were applied to the estimated Below the source area, as shown in the lower
Slide in Canada sent flow slides, composed thicknesses in order to maintain overall volume part of Figure 2, is a broad, sloping bench,
largely of liquefied valley fills, away from their balance between erosion and deposition. This 300 m wide, at an average slope of 26°, most
lateral margins (Mathews and McTaggart, 1978, procedure amounts to averaging two indepen- of which was originally covered by forested
Govi, 1989, Evans et al., 1987). dent estimates (deposition and erosion), thus coarse talus and glacial drift consisting of silty
improving the accuracy. The overall accuracy of sand. Rockslide blocks covered this bench over
TWO RECENT CASE HISTORIES the volume estimates reported below is probably a width of 210 m and at an estimated average
in the range of ±25%. This error estimate is itself thickness of 2.5 m in the proximal part and 1 m
Field Estimation of Volumes subjective, but it is based on a consideration of in the distal part of the bench. It is estimated
the possible limits of variation. For example, that ~87000 m3 of rock debris deposited on
An important aspect of the fieldwork while the deposit thickness of the central part of the bench, corresponding to 92% of the total
described here was the estimation of volumetric the Nomash deposit was estimated as 5 m, it is volume of the rockslide. As it came to a stop,
balance of the landslides, which is required to unlikely to have been more than 6 m (otherwise the mass of fragments mobilized some of the
derive yield rates. Areal extent of the source, more substantial damming of the river would sandy substrate near the crest of the bench. This
path, and deposit was well constrained, as the occur) or less than 4 m (large boulders could is apparent from the total removal of trees on the
authors are in possession of post-event maps not be transported, original topography would slide path, from evident mixing of the substrate
at 1:2000 (Eagle Pass) and 1:2500 (Nomash be better preserved). Similar arguments can be material with the rock blocks, and from eroded
River) scale. Thus, area estimates contain errors advanced concerning the eroded slope in the small side scarps bordering parts of the path
of no more than a few percent. A major potential distal part of the Eagle Pass Slide. Admittedly, in this area. The volume of soil removed from
source of error, however, derives from the diffi- the low accuracy of the field estimates of thick- the bench area was estimated by assuming an
culty of determining depths of material removal ness casts some doubt on the quantitative results average erosion thickness of 0.25 m over the
and deposition. Successive photogrammetric reported below. In the authors’ opinion, it is not area of the path. It must be stressed that this is
mapping before and after the event would sufficient to invalidate the qualitative conclu- an average value; the erosion thickness varied
have improved the situation, but only partially. sions regarding the important effects of material between 0 and ~2 m over the area, as estimated
Such mapping would not determine the depth entrainment on landslide motion. In any case, by inspection of exposures at the slide margins.
of disturbance in areas where deposits replaced there is presently no practicable method of The averaging of such spot estimates was made
previously eroded material. More important, improving such estimates. subjectively.
both landslides were characterized by surface The estimated remaining 6000 m3 of rock
elevation changes of <5 m, within the resolution Eagle Pass Slide debris, together with a larger quantity of mobi-
margin of photogrammetry in forested regions. lized drift, flowed over the crest of the bench
Field observations were conducted by tra- Eagle Pass is the deepest gap penetrating and down the main valley slope, 37° steep and
versing the entire extent of the path of both the Monashee Mountains 12 km west of Rev- 500 m long (Fig. 3). This slope is bedrock-con-
landslides on foot. The Eagle Pass ground elstoke, central British Columbia (Fig. 1). Both trolled, but was originally covered by a thin,
investigation was supported by a helicopter. the Canadian Pacific mainline track and the forested veneer of colluvium derived from the
Measurements of the dimensions of various fea- Trans-Canada Highway traverse the 550-m- glacial drift existing farther up the slope. The
tures were made with the help of standard field high pass, between mountain ridges reaching landslide stripped the veneer from a path 100 m
equipment, including measuring tape, compass, to more than 2000 m in elevation. A landslide wide, removing practically all of the tree cover.
clinometer, and a laser range finder. The depths occurred in May 1999 during the snowmelt In calculating the volume of debris generated
of depletion and deposition were estimated in season on the northern slope of the pass directly from this part of the path, an average thickness
the field by observing freshly eroded scarps, above the western end of Clanwilliam Lake of the veneer of 0.3 m was estimated. The soil
damaged trees, changes in drainage patterns (118°22′30″W, 50°58′00″N). material is nonplastic and consists of 5%–15%
(particularly ponding of water at the deposit The Monashee Mountains are composed of silt, 30%–50% sand and 35%–60% gravel (per-
margin), and apparent modification of recogniz- gneiss belonging to the Lower Paleozoic Mona- centages by weight, based on two samples). The
able landforms. Considerable subjective judg- shee Complex, forming a part of the Omineca mass of debris, rock, and timber flowed over a
ment had to be exercised in order to estimate Belt (Johnson, 1990). The source of the 1999 cliff at the toe of the slope and landed in Clan-
the average depths of deposition or erosion in slide is a 100-m-high cliff, originally over 60° william Lake, where a deposit with an area of
various reaches of the landslide path. In some steep at the face, standing at elevation 1020 m ~1.6 × 104 m2 built up over the surface of spring
cases, direct estimates of erosion depth or a.s.l., 470 m above the lake and the valley floor. ice. The total volume of deposits on the bench
deposit thickness at singular points were possi- The foliation of the cliff dips at ~35° to the and in the lake was ~1.2 × 105 m3. Within sev-
ble, thanks to exposed side scarps, buried trees, southwest, oblique to the south-trending fall line eral weeks, the nonorganic part of the debris dis-
ponded water, and similar evidence. Elsewhere, of the slope. It appears that the rupture surface appeared below the lake surface. The Canadian
this was not possible, especially where eroded of the initial failure resulted from a combination Pacific rail line escaped serious damage only

1242 Geological Society of America Bulletin, September/October 2004


ENTRAINMENT OF DEBRIS IN ROCK AVALANCHES

shown in Figure 4C. The maximum volume in


motion was 9.4 × 104 m3 of fragmented rock,
passing the toe of the source cliff. Deposition
of rock fragments on the bench, partly offset by
erosion, reduced the volume in motion to ~2 ×
104 m3. Entrainment of glacial and colluvial
material on the slope produced a second peak of
some 3.5 × 104 m3, which was deposited in the
lake at an estimated average deposit thickness of
2.2 m. The volume estimate of material passing
the crest of the bench could be quite inaccurate,
as it is within the error margin of the initial failure
volume. Nevertheless, the qualitative description
of the sequence of events given above serves to
substantiate the basic shape of the mass balance
curve. The fahrböschung (“travel angle,” being
the vertical angle between the crown of the initial
failure and the toe of the deposits) of the land-
slide is ~31° and the ER is 0.3.
Velocity of the landslide was roughly esti-
mated at one location near the crest of the bench
at the left margin of the path, where slight super-
elevation of the debris occurred along a curv-
ing trajectory controlled by topography. The
velocity of the slide was estimated applying the
forced vortex equation for flow through bends
(e.g., Chow, 1957):

u = Rg tan β , (2)

Here, u is the estimated mean flow velocity,


R is the radius of the central streamline of the
flow, derived from a map or from field observa-
tions, g is the gravity acceleration and β is the
transverse angle between trim lines defining
the boundary of the flow path. The conditions
at a distance of 500 m from the crown of the
rockslide suggested a speed of ~8 m/s. Higher
velocities undoubtedly occurred on the steep
slope of the lower path, although no direct esti-
mate is possible.

Nomash River Slide

The Nomash River is a small stream in the


western part of the Insular Mountains, some
20 km inland from the outer coast of Vancouver
Island (Fig. 1). The geographic coordinates of
the site are 126°42′00″W and 49°59′00″N. The
Figure 1. Location plan.
bedrock of the area consists of crystalline lime-
stone with basaltic intrusions, belonging to the
Upper Triassic Karmutsen Formation. A land-
because a short rock tunnel extends precisely The yield rate estimates have been derived from slide occurred in the headwaters of the Nomash
across the width of the slide path. field estimates of the depth and area of the River on April 25 or 26, 1999, during spring
The approximate volume balance of the event various entrainment and deposition segments, snow melt. The slide originated at the crest of
is reconstructed in Figure 4. Figure 4A is the as described in the preceding paragraphs. Error a 430-m-high southwest wall of a U-shaped
slope profile, showing the rockslide source, the bars indicate the probable range of depth esti- glacial valley. It moved across the valley floor,
bench with the coarse deposits, and the main mation errors (±25%). changed direction by 90° and moved northwest
slope segment of the path. Figure 4B is a plot Integrating the yield rate along the length for 1.2 km, following the channel of the river
of the estimated yield rate, Yi, in units of m3/m. of the path produces the mass balance curves and covering its floodplain.

Geological Society of America Bulletin, September/October 2004 1243


HUNGR and EVANS

The source of the initiating rockslide is a


mountain slope angled at 50° to the northeast
and composed of a sequence of white marble
interbedded with basaltic sills (Fig. 5). The sill
contacts dip at shallow angles to the southeast,
perpendicular to the slide motion. The concave
rupture surface consisted of a series of intersect-
ing joints, including some stress-relief joints
sub-parallel with the slope face and several ran-
domly oriented joints and shears. The detached
mass was roughly tabular in shape, tapering to
the base and ~3 × 105 m3 in volume. Applying
a fragmentation volume increase factor of 25%,
the slide produced 3.75 × 105 m3 of fragments,
which collapsed into the valley, perpendicular to
the Nomash River.
The valley bottom is mantled by a ground
moraine consisting of silty sand. Glacial till
underlies the narrow floodplain of the Nomash
River and extends up the sides of the valley
to a height of ~100 m above the valley floor.
Fine-grained colluvial aprons rise another 80 m
higher in certain parts of the valley slopes. Figure 2. Eagle Pass rockslide–debris avalanche: frontal view of the source area. The rock-
The floodplain is typically 50–100 m wide, slide scar is ~50 m wide. (Photo, D. Ayotte).
composed of shallow sand and gravel deposits
and organic matter. The depth of these alluvial
deposits is not known, but it is probably limited
to several meters.
The rockslide descended from an upper bed-
rock slope, overriding a colluvial apron and till-
covered slope, angled at 30° (Fig. 6). The col-
luvium and glacial till at location A in Figure 6
were scoured 100–150 m wide to a maximum
depth of 8 m, leaving prominent side-scarps
along the sides of the slide path, visible in Fig-
ure 7. A field survey of the area indicated that
~360,000 m3 of material were eroded from the
slope and incorporated into the moving mass.
Based on two samples, the eroded material con-
tained <2% clay, 14%–35% silt, 25%–44% sand
and 39%–41% gravel. The mass then crossed
the floodplain of Nomash River and deflected
from the right bank in a wide curve (Location B,
Fig. 6; Fig. 7). As a result of rebound from the
first curve, the flow elevated to a lesser height
on the left bank (Location C, Fig. 6), before
continuing to move down the Nomash River
valley. The width of debris deposits gradually
decreased along the valley. The flow terminated
at Location D, 2270 m in plan distance and
560 m below the crown of the initial slide. The
fahrböschung of the landslide was 13.8°.
The debris appears to be spread as a thin
veneer throughout the deposition area. Evidence
for this is the lack of damming of the Nomash
River at the upstream margin of the debris
(Location E, Fig. 6) and the general smooth-
ness of the debris surface, which mimics the
U-shaped profile of the valley as observed both Figure 3. Eagle Pass, vertical aerial view of the main path of the debris avalanche, below the
downstream and upstream of the slide site. The bench. (Aerial photo, D. Ayotte).

1244 Geological Society of America Bulletin, September/October 2004


ENTRAINMENT OF DEBRIS IN ROCK AVALANCHES

den by return flow. Instead, a runup formula


derived by Hungr et al. (1984) was applied to a
projection of the center of the flow path toward
the highest point of the right-hand flow margin
at Location B. The runup formula, derived from
the conservation of momentum, gives the verti-
cal runup height, H:
−V 2
H= tan θ , (3)
G
where:
gh cos θ 
V = u cos(θ 0 − θ)1 + (4)
 2u 2 

and

G = g( tan φ cos θ − sin θ). (5)

The slope angle of the approach path is θ0


(0° in this case), and the runup slope angle is θ
(−14° at Location B; note slopes are measured
down form the horizontal). The flow depth (h) is
assumed as 10 m and g is the acceleration of grav-
ity. Because the formula is rheology-dependent,
both h, the flow depth (10 m) and φ, the friction
slope angle (6.6°), had to be determined based
on the dynamic model, described below. With
these input parameters, the velocity, u, was esti-
mated iteratively from Equation (4) as 22.5 m/s
at the approach to Location B. Additional photos
of the Eagle Pass and Nomash River landslides
can be found in the Data Repository.1

Figure 4. Eagle Pass rockslide–debris avalanche. (A) Path profile. (B) Estimated yield rate DYNAMIC ANALYSIS OF THE CASE
distribution. (C) Volume balance curve. HISTORIES

The dynamics of a rockslide–debris ava-


lanche involves momentum exchange between
bulk of the bouldery rockslide debris was found The total volume of moving material passing the failing rock mass and the material entrained
upstream of the second bend (Location C). through this point is more than 7 × 105 m3. The along the path. The numerical model “DAN”,
Beyond this point, the debris consists primarily ER is 0.9. described by Hungr (1995) can be used to
of silty sand similar to the colluvium entrained Using the forced vortex equation (Equation simulate this scenario. Although based on a one-
on the slope below the source, although consid- 2) for flow in the second curve, at Location C, dimensional integrated form of the equations of
erable alluvial gravel and organic debris are also Figure 6, yielded a velocity estimate of 12 m/s. motion, the model is capable of producing rea-
evident. Rock debris occurs in this distal part of The superelevation of the flow was barely sonably good approximations of three-dimen-
the slide deposit mainly in the form of several measurable in the final two bends upstream of sional flow problems, as shown by comparisons
large boulders, up to 6 m in diameter. Location D, indicating that the velocity there carried out by Chen and Lee (2000) and more
Debris yield rates were estimated as shown had decayed to <2 m/s. recent unpublished comparisons with the
in Figure 8B. Error bars show the estimated The vortex formula could not be applied in authors’ own three-dimensional model and with
maximum error of erosion and/or deposition the first bend, at Location B, Figure 6, because laboratory flow experiments.
depth estimates (±25%). Figure 8C shows the the left margin of the flow through the bend The model permits entrainment of material
volume balance curve resulting from these esti- could not be determined, having been overrid- along the slide path. The volume of entrainment
mates. The shape of the mass balance curve is is controlled by the debris yield rate, which must
very different from that for the Eagle Pass Slide. be specified for each path segment as an input
Only one peak occurs, near a distance of 700 m 1
variable and is based on Figures 4B and 8B.
GSA Data Repository item 2004126, supple-
from the crown of the rockslide, where the frag- mentary figures, is available on the Web at http:
The entrainment volume is scaled in proportion
mented rock was still in motion while the total //www.geosociety.org/pubs/ft2004.htm. Requests to the fractional volume passing. Thus, the full
volume of entrained debris approached its peak. may also be sent to editing@geosociety.org. specified yield rate is reached a given point only

Geological Society of America Bulletin, September/October 2004 1245


HUNGR and EVANS

once the entire volume of the moving mass has


passed. It is also possible to change the rheology
of the moving mass to include the change from
dry frictional to velocity-dependent, at the point
where saturated material is entrained.
Both case histories have been analyzed, with
topography based on detailed site maps. The
same set of rheological properties was used for
both cases. In each case, the initial rockslide was
represented by a frictional material with no pore
pressure and a constant dynamic friction angle
of 30°. At the point where entrainment of debris
began, the rheology of the moving mass was
changed to the Voellmy model (Hungr, 1995).
The two-parameter model of Voellmy (1955)
combines a frictional coefficient (f) and a veloc-
ity-dependent turbulence parameter (ξ). The
frictional coefficient (tangent of the “bulk fric-
tion angle” as defined by Hungr [1995]) is the
result of dry friction of the material, modified by
pore pressure assuming a constant ratio between
pore pressure and total normal stress on the base
of the flow. Thus, the friction coefficient may be
a small fraction of the dry friction coefficient of
the substrate soil (cf. Hungr, 1995). The resist-
ing shear stress (τ) at the base of the flow, is then
represented (in a simplified formulation) as:
u2
τ = γhf cos α + γ , (6)
ξ
where h is the flow depth, α the slope angle, γ
the unit weight of the material, and u the mean Figure 5. Nomash River rockslide–debris avalanche: view of the source area. The rockslide
velocity. Voellmy (1955) developed this model scar is ~150 m wide.
empirically for snow avalanches by combining
Coulomb frictional and Chézy formulas. The
turbulence term ξ (which is similar, although
not exactly equivalent, to the Manning “n”) dynamically with each other and with the provides good results in terms of velocity and
was intended to cover all velocity-dependent fluid to produce dispersive effective normal distribution of deposits. Details of the calibration
factors in snow avalanche motion, including stress. This would tend to expand the layer. If, process have been discussed by Hungr and Evans
turbulence of the air-snow dispersion and however, pore pressure diffusion is too slow to (1996) and Ayotte and Hungr (2000). The present
air drag on the top surface of the avalanche. permit expansion, a decrease in pore pressure back analyses differ from previous work in that
Körner (1976) showed that, perhaps by coin- will occur. According to Bagnold’s results, the two rheologies are used, dry frictional for the ini-
cidence, the model offers a good simulation of pore pressure decrease should be proportional to tial rockslide and Voellmy for the flow lubricated
velocities for rock avalanches. the square of velocity. The frictional resistance by entrained liquefied soil. It must be stressed
Successful application of the Voellmy model will increase as represented by Equation (6). A that the only adjustable parameters in the analy-
to debris flows and debris avalanches may recent evaluation of Bagnold’s (1954) experi- sis are the two Voellmy coefficients. The friction
possibly benefit from a similar coincidence ments has placed the value of the exponent 2.0 angle of the initial sliding movement (30°) is a
(e.g., Rickenmann and Koch, 1997; Ayotte and in Equation (6) in question (Hunt et al., 2002). reasonable estimate for dry dynamic shearing of
Hungr, 2000). However, the model may have While this development may eventually result in a granular material or moderately rough joint sur-
a more fundamental physical basis. Bagnold a need to modify Equation (6), good empirical faces. The locations where the rheology changes
(1954) showed that both the shear stress and the results as reported by Körner (1976), Rick- from frictional to Voellmy are also not arbitrary,
effective normal stress in a dense dispersion of enmann and Koch (1997), Hungr and Evans but correspond to the points where entrainment of
grains in fluid, rapidly sheared at constant vol- (1996), this paper, and others justify the use of saturated soil is known to occur.
ume, depend on the square of the shear strain the simple Voellmy model at least until further The Voellmy parameters for both of the pres-
rate. We may visualize a plug flow moving on experiments and back analyses are completed. ent cases have been selected by trial-and-error
a saturated basal layer of constant thickness hm. As described by Hungr (1995), the DAN as f = 0.05 and ξ = 400 m/sec2. These parameters
Initially, while the movement is slow, the resis- model uses empirical calibration to constrain both are in the range of properties used earlier in the
tance will depend on friction and pore pressure the rheological relationship and the specific resis- simulation of long-runout flow slides involv-
in the basal zone. As velocity increases, grains tance parameters used. Previous experience with ing coal mine waste (Hungr et al., 2002). The
in the shearing layer will begin interacting rock avalanches shows that the Voellmy equation low magnitude of the f coefficient implies the

1246 Geological Society of America Bulletin, September/October 2004


ENTRAINMENT OF DEBRIS IN ROCK AVALANCHES

Figure 6. Plan of the Nomash River landslide path and deposit, showing locations mentioned in the text. The square grid spacing is 200 m
in both directions. The contour interval is 10 m.

profiles of the moving mass at 10 s intervals up


to 70 s. The thick line shows the final deposit.
The slide started as a flow of frictional mate-
rial. A yield rate of 54 m3/m was introduced on
the bench, at a distance of 420 m. It was then
reduced to 36 m3/m at 480 m in accordance with
Figure 4B. The frictional rheology was retained
on the proximal part of the bench, due to the
presence of coarse talus as the surficial material
in this area. The transition to Voellmy material
was made at a distance of 480 m, where the
surface material was finer-grained glacial and
colluvial soil. The separation of the landslide
deposit into an upper part (on the bench) and
lower part (in the lake) is well represented by
the model.
The model showed a substantial decrease in
front velocity at a distance of 500 m, near the
crest of the bench (Fig. 9B). The single field
velocity estimate at this location supports a sub-
stantial slow-down, although this estimate was
Figure 7. The middle part of the Nomash River slide path. The rockslide originated above made near the margin of the path and is there-
the upper left corner of the photo. Eroded side scarp in the colluvial apron beneath the fore probably not representative of the speed of
source area can be seen left of center, intersecting a twinned logging road. The lighter-col- the center of the flow front.
ored areas are covered by rockslide blocks. A corresponding analysis of the Nomash
Slide is shown in Figure 10. The amount of
erosion was specified according to Figure 8B.
presence of high pore pressure in the basal layer determined from the map are shown. Because The yield rate was taken as 1000 m3/m at a hori-
of the landslide. of the length of the profile, all depths (measured zontal distance of 360 m from the crown. The
The results of an analysis of the Eagle Pass normal to the path) have been exaggerated 10× material was changed from frictional to Voellmy
Slide are summarized in Figure 9. In Figure 9A, in the diagram. The initial slide profile is shown at the same location to account for the presence
the profile of the slide path with a width profile by a thick dashed line. Thin lines show the of saturated, fine-grained material susceptible

Geological Society of America Bulletin, September/October 2004 1247


HUNGR and EVANS

to rapid undrained loading. The yield rate was


reduced to 180 m3/m at 630 m. The flow veloc-
ity of the front is shown on Figure 10B, and
compared with the field velocity estimates
reported in the previous section. Depth profiles
at 10 s intervals and the final distribution of
debris are shown in Figure 10C.
In summary, the model is seen to simulate
very satisfactorily all main aspects of both cases
with the same set of input parameters. The key
to a successful simulation, however, is to change
the material rheology at that point of the path
where entrainment of saturated material occurs.
To illustrate the effect of entrainment, the
analysis of the Nomash case was repeated with
no entrainment and constant frictional rheol-
ogy (Fig. 11). This model shows a small rock
avalanche with limited displacement, building
a steep apron at the toe of the slope. The fahrbö-
schung is 33°, slightly larger than the assumed
friction angle of 30° and much larger than the
14° calculated by the alternative model. Thus, in
comparing Figures 10 and 11, the entrainment
of saturated material is seen to have a strong
effect on the runout in this example.

THE MECHANISM OF MATERIAL


ENTRAINMENT

The observations described in the previous


sections suggest a sequence of events form-
ing a rockslide–debris avalanche: In the early Figure 8. Nomash River rockslide–debris avalanche. (A) Path profile. (B) Estimated yield
stage, a rockslide of moderate volume occurs. rate distribution. (C) Mass balance curve.
Such a slide would ordinarily produce a steep
talus cone at the foot of the slope. However,
the moving mass of rock fragments reaches a A hypothetical mechanism of a full-scale distal debris. This is not surprising, given the
deposit of loose saturated material lying in the flow is shown schematically in Figure 12. In high velocity of the landslide motion and the
path (colluvium, glacial till and alluvium in the Figure 12A, the potentially liquefiable layer low permeability of the poorly sorted and rela-
present cases). It appears that the rockslide front of relatively loose, saturated soil on the valley tively fine grained soils at the site. Considering
does not immediately displace the loose soil in floor and apron is being approached by a rapidly the dominant material of the slide substrate to
its path, but overrides it partially, causing rapid moving mass of rock fragments from the upper be a poorly sorted silty sand, the coefficient of
loading and liquefaction of the saturated sub- slope. On collision, the soil partly or completely consolidation can be estimated as of the order
strate as suggested by Sassa (1985). This notion liquefies under impact loading, and its strength of 10−2 cm2/sec. Using consolidation theory
was tested by a small-scale model. A mass is reduced to a fraction of its dry frictional (Terzaghi and Peck, 1967, p. 181), a pore pres-
of rock fragments was released from a steep value. Part of it is pushed forward by the leading sure reduction of only 10% in a saturated layer
slope, so as to reach a thin deposit of clayey silt front of the rock mass, while part is overridden 50 cm thick would require ~40 min, much
saturated to just below its liquid limit. The frag- (Fig. 12B). A wave of liquid debris is projected longer than the 2 min duration of the entire
ments displaced the silt layer only slightly, but forward, overriding and entraining new sub- landslide event, as estimated by the model.
overrode it. In the final stage of the experiment, strate and carrying some of the rock fragments Mixing during rapid flow would likely further
the rock mass was moving on top of a sheet of with it. The rock mass follows behind, sliding hinder the consolidation process, making pore
mud formed from the silt substrate. The shear on a thin cushion of liquefied soil (Fig. 12C). pressure dissipation slower. Thus, no significant
strength and viscosity of the material was too Finally, the slide stops, the mud wave having dissipation of the pore pressures generated by
great at this small scale to effect much total dis- swept itself far forward into the deposition area, rapid loading should be expected.
placement; however, the mechanism of overrid- forming a forward-tapering tongue speckled Once liquefaction occurs, the flow front con-
ing and rapid loading was demonstrated. with transported boulders. All of these features tinues riding on top of a sheet of liquid mud in
It is probable that much more displacement correspond closely to field observations made at a configuration justifying the assumptions of the
of the substrate material occurs at full scale, the Nomash River site. Voellmy rheological model. This allows most
where the ratio between the applied stresses and Apparently, very little drainage or consolida- of the rock debris to be carried out consider-
the liquefied shear strength is several orders of tion of the fine-grained mud occurs during the ably beyond the deposition point that would
magnitude larger than in the laboratory model. event, as demonstrated by the fluidity of the be predicted by a dry frictional model. In the

1248 Geological Society of America Bulletin, September/October 2004


ENTRAINMENT OF DEBRIS IN ROCK AVALANCHES

Clastic dykes issue from this basal layer upward


into the overlying marble fragments. Wide-
spread clastic dykes and other intrusions or
pockets of valley deposits have been described
in major rock avalanches in the Bavarian Alps
by von Poschinger (2002). Upward extrusion of
liquid organic sand through coarse debris was
observed in the distal reaches of long-runout
flow slides from coal mine waste dumps (Hungr
et al., 2002).

DISCUSSION AND CONCLUSIONS:


THEORIES EXPLAINING THE
MOBILITY OF ROCK AVALANCHES

It has long been noted that many rock ava-


lanches are excessively mobile, if considered
as shearing masses of dry broken rock (Heim,
1932). Furthermore, the degree of mobility
appears to increase approximately in propor-
tion to the logarithm of the volume of the event
(e.g., Abele, 1974; Scheidegger, 1973; Hsü,
1975). For many years, researchers have been
looking for an explanation of this phenomenon.
The main hypotheses advanced for this purpose
include the following:
1. Mobilization by an air cushion, overridden
and trapped beneath the mass of the rock ava-
lanche (Shreve, 1968).
2. Fluidization by similarly trapped air or by
steam generated by vaporization of groundwater
Figure 9. Eagle Pass, results of the dynamic analysis. (A) Flow profiles at 10 s intervals (all (Goguel and Pachoud, 1972; and others).
depths exaggerated 10×). (B) Velocity profiles, with a field velocity estimate indicated by 3. Fluidization by dust dispersions (Hsü, 1975).
a cross. The model used frictional rheology with a bulk friction angle of 30° at a distance 4. Rock melting or dissociation by the heat of
of 0–484 m and a Voellmy model with an f (frictional coefficient) of 0.05 and a ξ (velocity- friction (Erismann, 1979).
dependent turbulence parameter) of 400 m/s2 beyond 484 m. Entrainment yield rates were 5. “Mechanical fluidization,” understood as
approximately as shown in Figure 4B. a process of spontaneous reduction of friction
angle at high rates of shearing (e.g., Schei-
degger, 1975; Campbell, 1989; and others).
final stages, the flow front resembles a saturated low degree of mobility, as measured by its steep 6. Acoustic fluidization—reduction of the
debris avalanche. fahrböschung. The event had two discharge friction angle resulting from acoustic-fre-
The above scenario is probably universal in peaks, one corresponding to the rockslide and quency vibrations at the base of the flowing
a qualitative sense, but will have different char- the other to the debris avalanche phase. mass (Melosh, 1979).
acteristics depending on the relative quantities It is easy to visualize cases similar to the 7. Increase in areal dispersion of debris as a
of rock and liquefiable soil and the shape of the Nomash Slide, but where the quantity of the result of fragmentation (Davies and McSavenney,
path profile. Thus, in the Nomash case, nearly liquefiable material relative to the rock debris 1999).
all of the rock debris reached the liquefaction is less. Thus, the rock debris is still mobilized 8. Lubrication by liquefied saturated soil
region and moved a considerable distance. The by a cushion of mud, but the leading tongue of entrained from the slide path (Buss and Heim,
peak discharge was formed by the joint flow of debris may be small or missing altogether. In 1881; Abele, 1974, 1997; Sassa, 1985).
rock and mud near the center of the path. There some cases, the mud may be expelled sideways Many critical reviews and discussions of
was sufficient excess of liquefied material to instead of forward (e.g., the Frank Slide, men- these various mechanisms have appeared in
project a long tongue of mud ahead of the rock tioned in the Introduction). the literature (e.g., Hsü, 1975; Hungr and Mor-
deposits. In the Eagle Pass case, the proximal Part of the liquid material may be smeared genstern, 1984; Hungr, 1990; Legros, 2002).
part of the rock mass deposited on the bench, along the path while supporting coarser debris. Some of the critical arguments that have been
having been slowed down by non-liquefiable This hypothesis was suggested by Johnson advanced are reviewed as follows.
talus deposits. Only the leading edge of the (1978), who found that the Blackhawk Slide Hypotheses 1 and 2 postulate high gas pres-
rock debris reached saturated colluvium and set debris is underlain by a possibly continuous sures. If substantial average gas pressure existed
off a debris avalanche on the steep, colluvium- 1-m-thick contorted layer of sandstone debris, beneath a sheet of debris, those parts of the sheet
mantled slopes below the bench. As a result of derived from a rock unit originally located far that are thinner than average would become
this, the Eagle Pass landslide displays a rather upstream, below the toe of the rupture surface. fully fluidized and exhibit features characteristic

Geological Society of America Bulletin, September/October 2004 1249


HUNGR and EVANS

Figure 10. Nomash River,


results of the dynamic analysis.
(A) Flow profiles at 10 s inter-
vals (all depths exaggerated
10×). (B) Velocity profiles, with
field velocity estimates indicated
by crosses. (C) Depth and ero-
sion profiles at 10 s intervals.
The model used frictional rheol-
ogy with a bulk friction angle of
30° at a distance of 0–358 m and
a Voellmy model with an f (fric-
tional coefficient) of 0.05 and
a ξ (velocity-dependent turbu-
lence parameter) of 400 m/sec2
beyond 358 m. Entrainment
yield rates were approximately
as shown in Figure 8B.

of upward escape of gas such as normal grading, and Morgenstern (1984) observed flows of dry harmonic vibrations produced by the bound-
elutriation of finer particles, craters, and graded sand in a laboratory flume at speeds of up to ary conditions of the landslide movement is
fallout fans or cones. No description of such 6 m/sec. These flows were <100 mm thick and required. However, it is unclear why such a
phenomena can be found in the literature. Many moved in a plug-flow manner. Thus, shearing mechanism would favor large rockslides as
rock avalanche deposits exhibit reverse grading, was concentrated in a basal zone only a few mm opposed to the smaller ones.
which is opposite to what would result from gas thick, producing shear strain rates of more than Areal dispersion of debris due to fragmen-
fluidization (e.g., Cruden and Hungr, 1986). 1000 rad/s. As a result, the flows were highly dis- tation and expansion of the source rock mass
Rock melting (Hypothesis 4) may occur in persed, showing a bulk density reduction of as demonstrably occurs in all rock avalanches
some cases, where a rockslide is unusually thick. much as 12%–22% relative to loose static piling (Hypothesis 7). However, this hypothesis does
However, it almost certainly is not a widespread density (Hungr, 1981, p. 421). Yet, no decrease not explain increased mobility of a frictional
phenomenon, as very few examples of melted in basal friction angle was measured in these material, which manifests itself not only by
rock have been reported in the literature. experiments. Thus, contrary to the key hypoth- wider spreading, but also by longer horizontal
With regard to the mechanical fluidization esis of the mechanical fluidization concept, a displacement of the center of mass of the depos-
theory (Hypotheses 3 and 5), it is yet to be high degree of mechanical dispersion does not its. Such increased mobility has been shown for
shown experimentally that high rates of shearing appear to provide a decrease in friction. many large rock avalanches and some small
lead to a reduction in the overall friction angle of Acoustic fluidization (Hypothesis 6) is possi- ones as well. For example, in the case of the
dry granular material. High rates of shearing can ble and can be demonstrated experimentally by Frank Slide, the vertical displacement angle
cause dispersion of the material and reduction in conducting direct shear tests on a vibrating table of the center of gravity has been estimated as
bulk density, but offer no mechanical advantage (e.g., Barkan, 1962). It differs from mechanical ~21° (Cruden, 1980), much less than one would
to the shearing movement. For example, Hungr fluidization in that energy input associated with expect for shearing of dry fragmented rock.

1250 Geological Society of America Bulletin, September/October 2004


ENTRAINMENT OF DEBRIS IN ROCK AVALANCHES

lower elevations (cf. Sassa, 1988; Abele, 1997).


Small events, on the other hand, often run out on
the relatively coarse grained and dry soil mantles
forming upper slopes. Of course, this depends
on the local surficial geology and, consequently,
the observed trend is erratic. Both the Eagle
Pass and Nomash events can be considered as
exceptional cases that, although relatively small,
reached saturated material early in their paths.
Mobility of rockslide–debris avalanches and
many rock avalanches is thus related weakly to
volume, but its main control is the availability,
character, and distribution of liquefiable satu-
rated soil along the slide path.
Field estimation of the volumes of entrainment
Figure 11. Dynamic analysis of the Nomash River landslide, neglecting entrainment of satu-
and deposition is a very difficult task, which
rated debris and the consequent change in rheology. Only frictional rheology is used, with a
will not likely become easier in the foreseeable
bulk friction angle of 30° (same as in the proximal region of the model in Fig. 10). (No depth
future. Nevertheless, this difficulty is not a reason
exaggeration in this case).
to abandon the consideration of the important
process of debris entrainment. Even approximate
estimates of material balance in landslides are
Hypothesis 8, lubrication by liquefied soil, valuable and should be collected whenever pos-
is in many cases supported by direct field evi- sible, using the best available means.
dence, as reviewed in the preceding sections. In
a paper written shortly after his inspection of the ACKNOWLEDGMENTS
Elm Slide, Albert Heim expressed the opinion
R. Guthrie and C. Bunce brought the two landslide
that the base of the flowing debris was lubri- cases to the authors’ attention and provided valuable
cated by mud (Buss and Heim, 1881). Heim information. D. Ayotte assisted with the research of
appears to have abandoned this hypothesis in the case histories and S. Riussec with the laboratory
later publications (cf. Hsü, 1978). Several other model. J. Major, E. Wohl, R. Iverson, C. Campbell,
authors, however, support the hypothesis of and S. Kite provided constructive criticisms of two
earlier versions of the paper, which led to numerous
lubrication by saturated soil (e.g., Abele, 1974, improvements.
1997; Sassa, 1985, 1988; Hungr, 1990; Voight
and Sousa, 1994; Legros, 2002). REFERENCES CITED
The role of liquefied saturated substrate
Abele, G., 1974, Bergstürze in den Alpen: Munich, Wissen-
in mobilizing rockslide–debris avalanches, schaftliche Alpenvereinshefte, v. 25, p. 230.
such as the Eagle Pass and Nomash examples Abele, G., 1997, Rockslide movement supported by the
described in this article, is unambiguous, as the mobilization of groundwater-saturated valley floor
sediments: Zeischrift für Geomorphologie, v. 41,
process can be plainly observed in the distribu- p. 1–20.
Figure 12. A hypothetical mechanism of tion of the deposits and can be quantified by Ayotte, D., and Hungr, O., 2000, Calibration of a runout
prediction model for debris flows and avalanches, in
momentum transfer between a moving rock dynamic analysis as shown in the preceding Wieczorek, G.F., and Naeser, N.D., eds., Proceedings,
mass and liquefiable substrate at full scale. section. It is reasonable to assume that a similar 2nd International Conference on Debris Flows, Taipei:
(A) Rock mass approaching the substrate process takes place in many large terrestrial Balkema, Rotterdam, p. 505–514.
Bagnold, R.A., 1954, Experiments on a gravity-free disper-
layer. (B) Impact liquefaction, partial dis- rock avalanches. In many cases, there is direct sion of large solid spheres in a Newtonian fluid under
placement and overriding of the substrate. evidence of the presence of liquefied debris. shear: Proceedings of the Royal Society of London,
(C) A mud wave is projected forward. (D) Ser. A, v. 225, p. 49–63.
In other cases, however, the liquefied material Barkan, D.D., 1962, Dynamics of bases and foundations:
Rock mass and mud wave deposit. may have remained covered by the rock debris, New York, McGraw Hill, 400 p.
thus giving no clue to its presence. The vis- Barros, W.T., Amaral, C.P., Sobreira, F.G., D’Orsi, R.N.,
Maia, H.S., and Cunha, R.P., 1988, Catastrophic ava-
ibility of the liquefied material depends on the lanche at St. Genoveva clinic slope: Solos e Rochas,
relative proportion of source rock relative to the v. 11, p. 17–25.
Buss, E., and Heim, A., 1881, Der Bergsturz von Elm: Zur-
Reasonable simulation of the distribution of entrained debris and on the circumstances of the ich, Worster, 133 p.
debris for 23 rock avalanches has been obtained entrainment and transport process. In any event, Campbell, C.S., 1989, Self-lubrication for long runout land-
with a frictional model, using a bulk friction substrate liquefaction is a ubiquitous process slides: Journal of Geology, v. 97, p. 653–665.
Chen, H., and Lee, C.F., 2000, Numerical simulation of
angle of 10–20° (Hungr and Evans, 1996). that can provide a very plausible explanation of debris flows: Canadian Geotechnical Journal, v. 37,
An exhaustive critique of the above theories the high mobility of not only rockslide–debris p. 146–160, doi: 10.1139/CGJ-37-1-146.
Chow, V.T., 1957, Open Channel Hydraulics: New York,
would be a daunting task and one that is prob- avalanches, but rock avalanches as well. McGraw-Hill, 400 p.
ably not feasible at present. The purpose of the The apparent increase of mobility with vol- Cruden, D.M., 1980, Anatomy of landslides: Canadian Geo-
few arguments presented here is merely to sug- ume can also be explained. Larger events cover technical Journal, v. 17, p. 295–299.
Cruden, D.M., and Hungr, O., 1986, The debris of the Frank
gest that the problem of rock avalanche mobility a larger area and thus have a greater chance to Slide and theories of rockslide-avalanche mobility:
has not yet been conclusively solved. reach thicker, more highly saturated deposits at Canadian Journal of Earth Sciences, v. 23, p. 425–432.

Geological Society of America Bulletin, September/October 2004 1251


HUNGR and EVANS

Davies, T.R.H., and McSavenney, M.J., 1999, Runout of dry Norway, Proceedings, 7th International Symposium on Niederer, J., 1941, Der Felssturz am Flimserstein: Jahres-
granular avalanches: Canadian Geotechnical Journal, Landslides, v. 1, p. 233–238. bericht der Naturforschenden Gesellschaft Graubün-
v. 36, p. 313–320, doi: 10.1139/CGJ-36-2-313. Hungr, O., and Morgenstern, N.R., 1984, Experiments in dens: Chur, v. 77, p. 3–27 (in German).
Dawson, R.F., Morgenstern, N.R., and Stokes, A., 1998, high velocity open channel flow of granular materials: Oyagi, N., 1987, The 1984 Ontake-san landslide and its
Liquefaction flow-slides in Rocky Mountain coal mine Géotechnique, v. 34, p. 405–413. movement: Trans, Japanese Geomorphological Union,
waste dumps: Canadian Geotechnical Journal, v. 35, Hungr, O., Morgan, G.C., and Kellerhals, R., l984, Quanti- v. 8, p. 127–144.
p. 328–343, doi: 10.1139/CGJ-35-2-328. tative analysis of debris torrent hazards for design of Plafker, G., and Ericksen, G.E., 1978, Nevados Huascarán
Endo, K., Sumita, M., Machida, M., and Furuichi, M., 1989, remedial measures: Canadian Geotechnical Journal, avalanches, Peru, in Voight, B., ed., Rockslides and
The 1984 collapse and debris avalanche deposits on v. 2l, p. 663–667. avalanches: Amsterdam, Elsevier, v. 1, p. 277–314.
Ontake Volcano, Central Japan, in Latter, J.H., ed., Hungr, O., Evans, S.G., Bovis, M., and Hutchinson, J.N., von Poschinger, A., 2002, Large rockslides in the Alps: A
Volcanic Hazards; Assessment and Monitoring: Berlin, 2001, Review of the classification of landslides of the commentary on the contribution of G. Abele (1937–
Springer-Verlag, p. 211–229. flow type: Environmental and Engineering Geoscience, 1994)) and a review of some recent developments, in
Erismann, T.H., 1979, Mechanism of large landslides: Rock v. VII, p. 221–238. Evans, S.G., and DeGraff, J.V., eds., Catastrophic land-
Mechanics, v. 12, p. 15–46. Hungr, O., Dawson, R., Kent, A., Campbell, D., and Mor- slides: Effects, occurrence and mechanisms: Boulder,
Evans, S.G., Aitken, J.D., Wetmiller, R.J., and Horner, R.B., genstern, N.R., 2002, Rapid flow slides of coal mine Colorado, Geological Society of America Reviews in
1987, A rock avalanche triggered by the October 1985 waste in British Columbia, Canada, in Evans, S.G., and Engineering Geology XV, p. 237–255.
North Nahanni earthquake, District of Mackenzie, DeGraff, J.V., eds., Catastrophic landslides: Effects, Rickenmann, D., and Koch, T., 1997, Comparison of debris
N.W.T: Canadian Journal of Earth Sciences, v. 24, occurrence and mechanisms: Boulder, Colorado, Geo- flow modelling approaches: San Francisco, California,
p. 176–184. logical Society of America Reviews in Engineering Proceedings, First International Conference on Debris-
Evans, S.G., Hungr, O., and Enegren, E.G., 1994, The Geology XV, p. 191–208. Flow Hazards Mitigation, A.S.C.E., p. 576–585.
Avalanche Lake rock avalanche, Northwest Territories, Hunt, M.L., Zenit, R., Campbell, C.S., and Brennen, C.E., Sassa, K., 1985, The mechanism of debris flows: San
Canada: Description, dating and dynamics: Canadian 2002, Revisiting the 1954 suspension experiments of Francisco, California, Proceedings, XI International
Geotechnical Journal, v. 31, p. 749–768. R.A. Bagnold: Journal of Fluid Mechanics, v. 452, Conference on Soil Mechanics and Foundation Engi-
Ghiglino Antúnez, L., 1970, Alud deYungay y Ranra- p. 1–24, doi: 10.1017/S0022112001006577. neering, v. 1, p. 1173–1176.
hirca del 31 de Mayo de 1970: Revista Peruana de Hutchinson, J.N., and Bhandari, R.K., 1971, Undrained load- Sassa, K., 1988, Geotechnical model for the motion of
Andinismo y Glaciología, v. 9, p. 4–88. ing, a fundamental mechanism of mudflow and other landslides (Special lecture), Landslides, in Bonnard,
Goguel, J., and Pachoud, A., 1972, Geology and dynamics mass movements: Géotechnique, v. 21, p. 353–358. C., ed.: Proceedings, 5th International Symposium on
of the rockfall of the Granier Range which occurred Johnson, B., 1978, Blackhawk Landslide, California, in Landslides, v. 1, p. 37–56.
in November 1248: Lyon, Bulletin, Bureau de Récher- Voight, B., ed., Rockslides and avalanches: Amster- Scott, K.M., 1988, Origins, behavior and sedimentology of
ches Géologiques et Miniéres, Hydrogeologie, v. 1, dam, Elsevier, v. 1, p. 481–504. lahars and lahar-runout flows in the Toutle-Cowlitz
p. 29–38. Johnson, B.J., 1990, Geology adjacent to the western margin River system: U.S. Geological Survey Professional
Govi, M., 1989, The 1987 landslide on Mount Zandilla in of the Shuswap Metamorphic Complex: Victoria, Brit- Paper 1447-A, 76 p.
the Valtellina, Northern Italy: Landslide News, Japan ish Columbia: Mineral Resources Division, Ministry Scheidegger, A.E., 1973, On the prediction of the reach and
Landslide Society, Tokyo, v. 3, p. 1–3. of Energy, Mines and Petroleum Resources Open-File velocity of catastrophic landslides: Rock Mechanics,
Hauser, A., 2002, Rock avalanche and resulting debris flow Report, v. 1990-30, p. 16. v. 5, p. 231–236.
in Estero Parraguirre and Río Colorado, Región Metro- Kolderup, N.-H., 1955, Raset I Modalen, 14 August 1953: Scheidegger, A.E., 1975, Physical aspects of natural catastro-
politana, Chile, in Evans, S.G., and DeGraff, J.V., eds., Norsk Geologisk Tidsskrift, v. 34, p. 211–217. phes: New York, Elsevier Science Publishing Co., 289 p.
Catastrophic landslides: Effects, occurrence and mecha- Körner, H.J., 1976, Reichweite und Geschwindigkeit von Sherard, J.L., Woodward, R.J., Gizienski, S.F., and Clev-
nisms: Boulder, Colorado, Geological Society of America Bergsturzen und fleisschnee-lawinen: Rock Mechanics, enger, W.A., 1963, Earth and earth-rock dams: New
Reviews in Engineering Geology XV, p. 135–148. v. 8, p. 225–256. York, John Wiley and Sons, 722 p.
Heim, A., 1932, Landslides and human lives (Bergsturz und Legros, F., 2002, The mobility of long runout landslides: Shreve, R.L., 1968, The Blackhawk Landslide: Geological
Menschenleben): Vancouver, British Columbia, Bi- Engineering Geology, v. 63, p. 301–331. Society of America Special Paper 108, 47 p.
Tech Publishers, 195 p. Lliboutry, L., 1975, La catastrophe de Yungay, Peru: Snow Terzaghi, K., and Peck, R.B., 1967, Soil mechanics in engi-
Hsü, K.J., 1975, Catastrophic debris streams (sturzstroms) and Ice: Gentbrugge, Proceedings of the Moscow neering practice, 2nd Edition: New York, John Wiley
generated by rock falls: Geological Society of America symposium, International Association of the Science & Sons, 750 p.
Bulletin, v. 86, p. 129–140. of Hydrology Publication 104, p. 353–363. Varnes, D., and Varnes, D.J., 1978, Slope movement types
Hsü, K.J., 1978, Albert Heim, observations on landslides McConnell, R.G., and Brock, R.W., 1904, The great land- and processes, in Schuster, R.J., and Krizek, R.J., eds.,
and relevance to modern interpretations, in Voight, B., slide at Frank, Alberta, in Annual Report for the year Landslides, analysis and control: Washington, D.C.,
ed., Rockslides and avalanches: Amsterdam, Elsevier, 1902–1903: Department of the Interior, Government of Transportation Research Board, National Academy of
v. 1, p. 71–92. Canada Sessional Paper 25, p. 1–17. Sciences Special Report 176, p. 11–33.
Hungr, O., 1981, Dynamics of rock avalanches and other Mathews, W.H., and McTaggart, K.C., 1978, The Hope Voellmy, A., 1955, Über die Zerstorungskraft von Lawinen,
types of mass movements [Ph.D. thesis]: Edmonton, rock slides, British Columbia, Canada, in Voight, B., Schweiz: Bauzeitung, v. 73, p. 212–285.
University of Alberta, 500 p. ed., Rockslides and avalanches: Amsterdam, Elsevier, Voight, B., and Sousa, J., 1994, Lessons from Ontake-san:
Hungr, O., 1990, Mobility of rock avalanches: Tsukuba, v. 1, p. 259–275. A comparative analysis of debris avalanche dynamics:
Japan, Reports of the National Research Institute Melosh, H.J., 1979, Acoustic fluidization: A new geologic Engineering Geology, v. 38, p. 261–297, doi: 10.1016/
for Earth Science and Disaster Prevention, v. 46, process?: Journal of Geophysical Research, v. 84, 0013-7952(94)90042-6.
p. 11–20. p. 7513–7520.
Hungr, O., 1995, A model for the runout analysis of rapid Morales, B., 1966, The Huascarán avalanche in the Santa MANUSCRIPT RECEIVED BY THE SOCIETY 24 MARCH 2003
flow slides, debris flows and avalanches: Canadian Valley, Peru: Gentbrugge, International Symposium on REVISED MANUSCRIPT RECEIVED 29 AUGUST 2003
Geotechnical Journal, v. 32, p. 610–623. Scientific Aspects of Snow and Ice Avalanches, Davos, MANUSCRIPT ACCEPTED 9 SEPTEMBER 2003
Hungr, O., and Evans, S.G., 1996, Rock avalanche run- Switzerland, International Association of the Science
out prediction using a dynamic model: Trondheim, of Hydrology, Publication 69, p. 304–315. Printed in the USA

1252 Geological Society of America Bulletin, September/October 2004

Vous aimerez peut-être aussi