Document Actions

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

background image
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
73
DOI: 10.4408/IJEGE.2013-06.B-06
IMPACT OF LARGE LANDSLIDES, MITIGATION MEASURES
J
ean
F. SCHneIDeR, F
abIan
e. GRUbeR & M
aRtIn
MeRGILI
University of Natural Resources and Life Sciences (BOKU), Vienna, Peter-Jordan-Str. 70, 1190 Vienna, Austria
lake outburst floods. Whilst much work has been done
on glacial lake outburst hazards (R
ICHaRDSon
& R
ey
-
noLDS
, 2000), the present paper focuses on landslide-
dammed lakes.
Landslides are common geomorphic processes in
high mountain regions such as Central Asia. Whilst
the direct impact of such phenomena on mountain
communities is obvious, many landslides are only
the starting point of process chains. The formation of
landslide-dammed lakes is often a highly significant
secondary effect (C
LaGUe
& e
vanS
, 1994; C
aSaGLI
&
e
RMInI
, 1999). Such landslide dams may fail suddenly
due to impact waves, internal or retrogressive erosion,
resulting in potentially destructive flood waves down-
stream. Lakes may also drain stepwise or continuous-
ly, others persist for a long period of time. C
oSta
&
S
CHUSteR
(1988, 1991) and S
CHUSteR
& e
vanS
(2011)
ABSTRACT
Besides existing landslide-dammed lakes there is
evidence of former cases in the high-mountain areas
of Europe, Asia and America. In the Holocene, large
landslides have repeatedly dammed lakes. Numer-
ous prehistoric, historic and recent cases are evident
where the dams could not resist the pressure of the im-
pounding water. The result were flood waves charac-
terized by particularly high peak discharges and long
travel distances, leading to disasters where interfering
with populated lands downstream. Even though most
dam failures occur in the early phase after formation,
lakes may also drain suddenly at later stages. Case
studies from Central Asia and Northern Pakistan are
employed in order to exemplify the involved phenom-
ena regarding dam formation, outburst mechanisms
and flood wave propagation. A particular focus is put
on discussing the options for disaster risk reduction
and hazard mitigation.
K
ey
words
: Attabad, computer modelling, flood wave, Hat-
tian Bala, lake outburst flood, landslide dam, Siachen-Ga-
yari, spillway
INTRODUCTION
Natural dams of various types retain lakes in
many areas of the world. C
oSta
& S
CHUSteR
(1988)
point out that mainly landslide-dammed lakes, glacier
ice-dammed lakes and late neo-glacial moraine dams
are prone to fail and to produce potentially destructive
Fig. 1 - Percentage of failed dams plotted against the time
after formation (after S
chuSter
& e
vanS
, 2011)
background image
J.F. SCHNEIDER, F.E. GRUBER & M. MERGILI
74
International Conference on Vajont - 1963-2013 - Thoughts and analyses after 50 years since the catastrophic landslide Padua, Italy - 8-10 October 2013
and has not yet reached the dam crest. Even though
S
CHUSteR
& a
LFoRD
(2004) list several possible fail-
ure mechanisms, there is still disagreement upon the
level of hazard emanating from Lake Sarez. The dam
is rated rather stable due to the consolidated structure,
huge dimensions and the existence of a preferential,
non-eroding flow path through the dam (I
SCHUk
, 2004,
2011). However, there is a creeping rock mass head-
ing into the lake which, in case of sudden acceleration,
could trigger an impact wave and consequent overtop-
ping of the dam (R
ISLey
et alii, 2006). That rock mass,
the dam and the Murghab River downstream are mon-
itored and a flood early warning system was installed.
LAKE SHIVA
Lake Shiva has a maximum length of approx. 9
km and is located in the Shugnan District of north-
east Afghanistan, at a distance of approx. 16 km from
the south-west of Khorog, capital city of Gorno-Bada-
khshan, Tajikistan (see Fig. 2). It is impounded behind
a natural composite dam across the valley of Arakht, a
tributary of the river Panjwhich, in that region, consti-
tutes the border between Afghanistan and Tajikistan.
A preliminary geological hazard assessment of
the lake and dam was conducted by means of helicop-
ter survey, satellite imagery interpretation and ground
check in summer 2011. No immediate hazard of sud-
den drainage was detected, but a partial collapse of the
dam due to retrogressive and piping erosion cannot be
have shown that most dam failures occur in the first
few months after the landslide event. Afterwards, the
dam is usually consolidated so far that outburst floods
become less likely (Fig. 1). However, impact waves
triggered by mass movements into the lake may occur
a very long time after dam formation.
Whilst geomorphic evidence indicates the exist-
ence and sudden drainage of landslide-dammed lakes
in earlier stages of the Holocene, very recent cases
have illustrated the huge challenge such phenomena
pose for the population and the authorities of (possi-
bly) affected areas.
The present paper is understood as a contribution to
the understanding of the dynamics of landslide-dammed
lakes and the challenges for risk mitigation in order to
minimize future losses. The lessons learned from three
recent cases from northern Pakistan (Hattian Bala, At-
tabad and Siachen-Gayari) are combined with histori-
cal and geomorphologic evidence from past events in
Tajikistan and eastern Afghanistan. Fig. 2 shows the
geographic location of the cases discussed below.
EVIDENCE OF FORMER EVENTS
LANDSLIDE DAMS IN THE TAJIK AND AFGHAN
PAMIR
LAKE SAREZ
The highest natural dam known today, the Usoi
Dam, has remained stable for more than 100 years
now. It was formed by an earthquake-triggered land-
slide in 1911, blocking the Murghab Valley in the
Tajik Pamir (see Fig. 2). Up to 600 m high, it im-
pounded Lake Sarez, now 60 km long with a volume
of 17 km³ (S
CHUSteR
& a
LFoRD
, 2004). Since seep-
age through the dam almost offsets the inflow into the
lake, the lake level rises only approx. 0.2 m per year
Fig. 2 Case studies discussed in the article
Fig. 3 - Lake Shiva composite dam, view to the W. The de-
pression spring (circular lake, diameter approx.
200 m) and headwater of Arakht torrent in the
foreground; to the left of the lake the unstable part
of the downstream side of the dam, with a band of
much smaller springs, conspicuous because of spots
of green vegetation. Moraine deposit in the centre,
rock slide deposit in the right part of the photo-
graph, the source of the rock slide is farther right
background image
IMPACT OF LARGE LANDSLIDES, MITIGATION MEASURES
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
75
cular lake (Fig. 3).
However, apart from the large spring with, sta-
ble piping, several additional springs have appeared
on the downstream slope of the dam to the south of
the main spring. This indicates seepage through the
comparatively impermeable till constituting the main
part of the dam. This has already led to retrogressive
erosion of a currently rather small part of the dam af-
fected by the seepage, and it is obvious that repeated
slumping is taking place and retrogressive erosion is
active (see Fig. 3). In case these instabilities continue,
the dam - in the long run - could be weakened by ero-
sion of its narrowest part, and also be undermined by
concentrated seepage (piping).
Furthermore, S
HRoDeR
& W
eIHS
(2010) describe
the dam site as situated at the crossing of two active
tectonic lines, which could trigger local earthquakes.
PASOR - GHUDARA SYSTEM
Whilst Lake Sarez and Lake Shiva are prominent
examples of still rather intact natural dams, the Pamir
also bears a lot of evidence for failed dams. Several
valleys were temporarily blocked by huge, predomi-
nantly coarse-grained deposits. Fine sediments up-
stream indicate lakes that have disappeared either by
sedimentation or by sudden or stepwise dam failure.
Even though the origin of these deposits is not undis-
puted and some may represent Pleistocenic moraines,
many of them are identified as landslide deposits. One
such example is located in the upper Bartang Valley
(Central Pamir, Fig. 4a). The Pasor landslide dam, ap-
prox. 300 m high, blocked the valley and impounded
the up to 8 km long Ghudara Lake. The age of the lake
sediments was determined as ≤4000 years using Op-
tically Stimulated Luminescence (OSL). After accu-
mulating several tens of metres of lake sediments, the
lake drained in stages and the sediments were deeply
eroded. The narrow gorge through the landslide de-
posit may have been blocked several times. Down-
stream alluvial deposits, which are partly eroded leav-
ing only remnant large blocks, indicate at least one
powerful outburst flood (see Fig. 4b).
DASHT-SULAYMAN SYSTEM
The Panj Valley, forming the border between
Tajikistan and Afghanistan (Wakhan Corridor), is
partly blocked by the deposit of a debris avalanche
at Dasht-Sulayman, upstream the town of Ishkashim
completely ruled out. Subsequent flood wave model-
ling of two partial collapse scenarios have shown that
in either case, in addition to all the villages on the way,
the city of Khorog may be affected.
The dam retaining Lake Shiva is 1.6 km wide. Fol-
lowing S
HRoDeR
& W
eIHS
(2010) it is composed of the
material of at least three landslides and a rock glacier.
According to an assessment carried out by the authors
the main part consists of a morainic dam, representing
at least two late-glacial stages. This morainic dam, in
its northern part, interdigitates with the deposit of a
landslide which rushed down from the slopes to the
north, presumably while the landscape had to accom-
modate the loss of its ice cover at the end of the last
ice age. Perhaps a partial breach of the morainic dam
did occur back then but if so, the gorge is now covered
by landslide debris.
The dam has since consolidated, and obviously
has never been overtopped by the impounded water
in the shape it is now (after the possible first break-
out). Instead, all the water supplied by the catchment
is travelling through the dam by virtue of seepage, and
for the most part is channelled through the permeable
material supplied by the landslide. On the downstream
side of the dam, part of the seeping water emerges in a
stable depression spring, thereby creating a small cir-
Fig. 4 (a) Pasor landslide (background) and eroded
lake sediments in the foreground. The lake level
dropped in several stages. (b) Partly eroded lake
outburst sediments directly downstream of the
dam showing typical outburst flood structures
background image
J.F. SCHNEIDER, F.E. GRUBER & M. MERGILI
76
International Conference on Vajont - 1963-2013 - Thoughts and analyses after 50 years since the catastrophic landslide Padua, Italy - 8-10 October 2013
early June the barrier was overtopped and breached,
leading to a tremendous flood wave down the river.
The meagre historical records of this huge flood wave
were compiled by M
aSon
(1929). Approx. 3-5 billion
m³ of water were released in less than 24 hours, 2
million m³ of solids were eroded within a short time.
430 km downstream, the wave front was described
as “a wall of water, mud and rocks” which still had
a height of approx. 25 m (S
HRoDeR
, 1998; n
eSpak
,
2010). A Sikh army that had camped upstream of At-
tock was hit by the wave, with at least 500 casualties.
The actual volume of water discharged by the Great
Indus Flood is unknown and estimates vary similar to
those regarding lake length, but M
aSon
(1929) gives
dimensions of the barrier that would indicate a vol-
ume of more than one cubic kilometre. The release of
this volume within one day would indicate an aver-
age discharge of over 14,000 m³/s, but the height of
the flood wave at Attock suggests an initial discharge
several times higher. C
oRnWeLL
& H
aMIDULLaH
(1992)
point out that the estimation of the peak discharge var-
ies from 56,630 m³/s (H
eWItt
, 1964) to 509,000 m³/s
(S
HRoDeR
et alii, 1991). D
eLaney
a
nD
e
vanS
(2011)
calculated a peak discharge of approx. 114,000 m³/s.
In 1858, another massive slope failure (Ghamm-
essar landslide, 125 million m³) occurred just down-
stream of Attabad, close to the village of Sulmanabad.
Its extensively eroded toe still contains boulders 20–
30 m in size (S
HRoDeR
, 1998). The landslide blocked
the Hunza River and impounded a lake, D
eLaney
&
e
vanS
(2011) estimate a lake volume of 0.8 km³. In
August 1858 the dam was overtopped and the result-
ing erosion, more than 300 m deep, led to retrogres-
sive slumping of the toe of the landslide – on which
the town of Sulmanabad is now located – into the
river. M
aSon
(1929) attributes a 20 m flood wave at
Attock to the Ghammessar slope failure. According to
n
eSpak
(2010) the flood led to a wave height of 16.5
m at Attock, the flood hydrograph adding up to a vol-
ume of 1.85 billion m³. This second Great Indus Flood
destroyed several villages and forts downstream of the
dam but the population was warned, remembering the
1841 great flood (D
eLaney
& e
vanS
, 2011).
Additionally to the Ghammessar slope failure,
S
HRoDeR
(1998) describe several other slope failures
in the proximity of Attabad, for instance the older
and the younger Serat slope failures just opposite of
the 2010 Attabad landslide (see next chapter). An es-
(see Fig. 2). The origin of the landslide consisting of
black slate is located in the mountain range south of
the Panj River. Lake outburst sediments and residual
boulders indicate a temporary blockage of the valley
followed by sudden drainage, but historical evidence
is missing. Ruins of a medieval fortification are situ-
ated on an older eastern lobe of the debris (Fig. 5).
Detailed interviews in the near village uncovered
legends about a "big flood long time ago". Such evi-
dence is certainly purely speculative but can anyway
give some indication of former events. More detailed
geomorphological work is required in order to better
understand this and other prehistoric events.
HISTORICAL LAKE OUTBURST FLO-
ODS IN NORTHERN PAKISTAN
The Hunza and Indus Valleys of northern Paki-
stan, deeply incised and seismically active (S
HRoDeR
,
1998), have a particular history of landslide-dammed
lakes (H
eWItt
, 1982, 1998, 2011) with specific cases
in 1841, 1858, 1962, 1974 and 2010.
The largest event is documented from the Indus
valley near the Nanga Parbat. In December 1840 or
January 1841 a giant earthquake-triggered landslide
dammed the Indus River in the vicinity of Raikot
Bridge. The exact location of this natural dam is still
disputed, but it impounded a temporary lake with a
length estimated between 30 (S
HRoDeR
, 1998), 57 (D
e
-
Laney
& e
vanS
, 2011) and 64 km (M
aSon
,1929). In
Fig. 5 - Debris avalanche at Dasht-Sulayman upstream
Ishkashim in the Wakhan Corridor, Afghani-
stan. The Panj River eroded the lobe reaching
Tajikistan, emptying a temporary lake impounded
by the dam. A thin cover of lake sediments exists
in the cultivated area on the lower left side and
the ruins mentioned in the text are situated on the
upper left of the photo. Washed out boulders cover
the orographic right side of Panj River as rem-
nants of the lake outburst flood. The eroded cliff
on the orographic left side is approx. 50 m high
background image
IMPACT OF LARGE LANDSLIDES, MITIGATION MEASURES
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
77
CASE STUDY 1: HATTIAN BALA LAN-
DSLIDE
On October 8, 2005, a magnitude 7.6 earthquake
struck Kashmir in northern Pakistan and caused many
casualties as well as severe damage. Several mass
movements were triggered. The Hattian Bala landslide,
located at a tributary of Jehlum River southeast of Mu-
zaffarabad (see Fig. 2 for location), was reactivated,
resulting in a rock avalanche with a volume of approx.
65 million m³ (D
UnnInG
et alii, 2007; S
CHneIDeR
, 2009;
Fig. 6). Consisting of sand-, silt- and mudstones of the
Murree Formation, it destroyed a small village and sev-
eral farms An area of 1.8 km² was directly affected by
the landslide, the deposit formed a dam with an area of
0.9 km² impounding Karli and Tung rivers and creating
two lakes (Fig. 7a, b). Based on the geometry of the em-
bankment, the maximum volume of the lakes, i.e. when
the water level equals the elevation of the lowest sad-
dle of the dam crest, was calculated: Karli (or Zalzal)
Lake, the larger of the two, would grow to a volume
of approx. 61.7 million m³, Tung (or Bani Hafiz) Lake
to 3.6 million m³ (S
CHneIDeR
, 2009). The portion of the
deposit impounding Karli Lake had a maximum depth
of 230 m (D
UnnInG
et alii, 2007) to 350 m (S
CHneIDeR
,
2009). Large sandstone blocks were stabilizing the sur-
face of the orographic right (distal) portion of the dam.
After detailed investigations, several measures to
mitigate the hazard related to a possible dam failure
were initiated (S
CHneIDeR
, 2009). Besides the installa-
tion of a monitoring system and the design of hazard
maps and evacuation plans, it was decided to limit
the water level of the lakes and to ensure controlled
overflow by excavating reinforced spillways for Kar-
li Lake and Tung Lake, with lengths of 425 m and
130 m, respectively. For Karli Lake the spillway was
not built not over the saddle, representing the natu-
ral drainage path, but over the centre of the dam in
order to avoid a destabilization of the adjacent slope.
The spillway was partly completed in June 2006, but
not reinforced. Its depth was 10 m and the clast size
varied from sand to cobbles beneath a relatively thin,
coarse bouldery surface layer (D
UnnInG
et alii, 2007).
The level of Karli Lake reached the spillway at the end
of March 2007. In June 2007 the spillway appeared
stable, but was not lined. Seepage was observed in the
lower, unchanneled section.
Due to the increasing consolidation of the dam,
S
CHneIDeR
(2009) rated the probability of a failure as
timated volume of 1.5 million m³ of the younger Serat
landslide was eroded by the river, presumably by an
outburst flood. In October 1962 a mass movement was
released from the scarp of the Ghammessar landslide,
killing six men of the Public Works Department of the
Government of Pakistan and impounding water up to
Gulmit. When the lake drained suddenly after several
months, it undercut remnants of the 1858 landslide,
causing several houses of the village of Sulmanabad to
drop into the Hunza River. However, they had not been
occupied as the population was aware of the instability
of that slope. n
eSpak
(2010) reports of another Hunza
River blockage in the area of Gulmit in 1974: this dam
failed due to overtopping a few months later, the result-
ing flood caused some minor damage.
The 2010 Attabad landslide and formation of Hun-
za Lake, with its far-reaching consequences for the re-
gion, is a further entry in the long list of events in that
area. It is discussed in detail in the next chapter.
CASE STUDIES OF RECENT EVENTS
Several landslide-dammed lakes have formed in
the last 50 years. Examples include the events of Aini
(Tajikistan, 1964), Mayunmarca (Peru, 1974), Val Pola
(Italy, 1987), Tsatichu (Bhutan, 2003), Hattian Bala
(Pakistan, 2005), Tangjiashan (China, 2008), Attabad
(Pakistan, 2010) and Siachen-Gayari (Pakistan 2012).
Some of these dams have failed. The cases of the failed
dam of Hattian Bala as well as of Attabad and Siachen-
Gayari where the dams were still intact in March 2013
shall be discussed in detail
Fig. 6 - (a) Proximal base of the Hattian Bala rock ava-
lanche, helicopter at the location of the future
spillway channel. Karli Lake is situated at the
left of the photograph. (b) Sandstone block of the
Murree formation in the silt-to claystone landslide
debris matrix. (c) Cemented debris of a former
landslide at the same location
background image
J.F. SCHNEIDER, F.E. GRUBER & M. MERGILI
78
International Conference on Vajont - 1963-2013 - Thoughts and analyses after 50 years since the catastrophic landslide Padua, Italy - 8-10 October 2013
unlikely if the artificial spillway would be lined with
sandstone blocks in order to avoid erosion. However,
he also pointed out the scenario of impact waves trig-
gered by the sudden acceleration of active slumps ob-
served on the margins of Karli Lake, especially on the
orographic left side (see Fig. 7b).
In February 2010, a process chain including spill
over of the dam of Karli Lake and deep retrogressive
erosion of the spillway occurred (Fig. 7c and Fig. 8).
The following debris flow led to severe damage and
one fatality downstream. k
onaGaI
& S
attaR
(2012)
conclude that the breach can be attributed to the hy-
drologic situation (moderate rainfall after dry condi-
tions) in combination with a deteriorating dam body
due to. weak weathering resistance of the material.
Some landslides observed close to the lake were most
likely caused by slope destabilization due to the sud-
denly lowered lake level and the resulting changes in
pore water pressure. However, further investigations
are required in order to fully understand the process
chain that occurred here.
CASE STUDY 2: ATTABAD DAM
The two Great Indus Floods of 1841 and 1858,
as well as smaller events such as GLOFs, increased
the awareness of the local population which is gener-
ally prepared against natural hazards. The village of
Attabad registered cracks and slides over a period of
several years. On January 4, 2010 a new 45 million
m³ rock slide occurred on the orographic right side of
the Hunza gorge, destroying part of the village of At-
tabad (see Fig. 2 for location). The landslide occurred
in a tectonically very active region on a local fault just
north of the Main Boundary Thrust and was certainly
prepared by seismic destabilization. However, no obvi-
ous trigger for the rock slide is evident as the weather
preceding the event was cold and dry and no signifi-
cant seismic activity was measured. The area of west-
ern Attabad had been declared a high hazard area for a
large-scale failure some years earlier and was therefore
evacuated at the time of the event.
At the bottom of the valley, lake sediment pre-
sumably originating from the 1858 landslide dam
lake was mobilized through undrained loading and
possibly through liquefaction of clay, overtopping the
rock avalanche deposit and leading to two secondary
mudflows (n
eSpak
, 2010). One of them propagated
upstream for a distance of approx. 1.5 km, the other
Fig. 7 History of the Hattian Bala landslide and the re-
lated lakes: (a) Situation before the landslide. (b)
Hattian Bala landslide, Karli Lake and Tung Lake
before the breach of the dam. (c) Situation after
the breach of the dam with remnants of Karli Lake
and Tung Lake
Fig. 8 Eroded breach in the debris deposit of the Hat-
tian Bala landslide below the artificial spillway.
Note the landslide scars in the background and
the former level of Karli Lake marked by the snow
line. Photo taken on February 10, 2010 by Dr.
Kausar from the Geological Survey of Pakistan
background image
IMPACT OF LARGE LANDSLIDES, MITIGATION MEASURES
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
79
Following the saddle, the slope of the downstream face
of the dam is 35° (n
eSpak
, 2010).
The grain size of the dam material ranges from
clay and silt to sand, gravel and large boulders. A
large amount of black clay with high organic content
was observed in and on the deposit as well as up- and
downstream of the dam (remnants of the secondary
mudflows). Laboratory tests of the lacustrine sediment
mobilized by the landslide showed a plasticity limit of
21-22% and a liquid limit of 28%. The dam is partly
covered by an up to 0.5 m thick layer of fine rock pow-
der. The main part of the dam (the actual rock slide
deposit) is gneiss, with intrusions of pegmatite and
aplite. Whilst finer material dominates the area around
the saddle, the coarser material and large boulders have
accumulated at the orographic left (distal) side of the
dam. The large boulders are not confined to the top of
the embankment, but exist also inside the dam, highly
contributing to its stability.
Seepage through the dam developed after about 2
months (Fig. 9) and then increased in a nonlinear way.
At that point, internal erosion by seepage was consid-
ered a potential failure mechanism.
In order to decrease the overall volume of the
lake and to regulate the future flow over the dam, the
National Disaster Management Authority of Pakistan
(NDMA) oversaw the construction of an artificial
spillway at the saddle of the embankment. The result
was a narrow channel with a bottom width of about
1 m and a depth of 14 m, mainly in the silty clay of
the lake deposit.
Figure 10 illustrates the temporal development
of Hunza Lake, monitored by the NDMA. Due to
the morphology of the valley, which broadens farther
upstream of the dam, the filling rate of the lake was
initially high and then decreased with time. A slight
increase in the filling rate took place in spring 2010
due to snow and glacier melt. In the night of May 28
to May 29, 2010 the dam was overtopped and drained
through the constructed spillway. At that point the lake
level at the spillway was 111.41 m above the original
valley bottom. Overflow increased slowly at first, still
allowing for an increase in lake water level of up to 50
cm per day. The spillway underwent retrogressive ero-
sion with almost no basal down cutting. On June 5, in-
flow and outflow of the lake reached a balance at a lake
level of 115.21 m above the original valley bottom. In
May 2011, erosion at the outflow of the lake was still
travelled 3 km downstream. It hit the settlement of
Sarat, claiming 19 lives. 141 houses became uninhab-
itable (p
etLey
et alii, 2010).
As a consequence of the Attabad event, a huge de-
bris deposit in the valley blocked the Hunza River. The
length of the embankment along the river is approx. 2
km, the width up to 400 m.
As it is the case for most landslide dams, the high-
est point (210 m above the old valley bottom) is situ-
ated at the distal part of the deposit. This is where a
large amount of the landslide material accumulated.
The saddle is located close to the proximal northern
slope of the valley, 126 m above the original riverbed.
Fig. 9 Rock slide dam of Attabad, looking W. The rock
slide originated from the upper right side of the
photograph. The photo was taken on May 26,
2010, a few days before overflow started. Note the
erosion channel from seeped water and the bound-
ary between the dark lacustrine deposits and the
brighter mass of gneiss rock slide debris
Fig. 10 - Attabad rock slide and the temporal development
of Hunza Lake: (a) and (e) Situation before the
Attabad rock slide. (b), (c), (f) and (g) Growth
of the lake prior to overflow (d) lake extent after
overflow. (h) Situation after overflow with drain-
age through the spillway. The white dashed line
shows the extent of the lake on July 7, 2010, the
black dashed line shows the extent of the Attabad
rock slide. North is up
background image
J.F. SCHNEIDER, F.E. GRUBER & M. MERGILI
80
International Conference on Vajont - 1963-2013 - Thoughts and analyses after 50 years since the catastrophic landslide Padua, Italy - 8-10 October 2013
which covered an area of more than 1 km² to a depth of
up to sixty metres. Large blocks were embedded in a
concrete-like matrix of ice and crushed rock with grain
sizes down to silt fraction (Fig. 12, Fig. 13).
Several smaller avalanches from adjacent moun-
tains followed after the main slide. Even though, due to
their limited extent, they caused no additional damage,
these slides hampered the search operations. The com-
pacted debris cone impounded a lake with a surface
area that over time increased to 25 ha. Excavating a
drainage channel was necessary in order to reduce the
hazard of an outburst flood putting the rescue works as
well as the population and infrastructure downstream
at risk (Fig. 14). The excavation efforts were success-
ful and an outburst flood was avoided.
Due to a concerted effort by the Pakistani Army
seconded by rescue teams from Germany, Norway,
and Switzerland, the number of recovered bodies is
increasing each day. Around 450 engineers and work-
ers, with heavy equipment, were working around the
clock whenever possible. Simultaneous efforts were
undertaken to tackle effects of intruding water on site
such as the inundation of rescue excavations, erosive
cuttings and crevasses. This water was slowing down
the pace of rescue efforts.
The Gayari camp was considered a safe place
since a 700 years old mosque at the camp site had
not been affected by geohazard processes for centu-
ries. However, with global warming, the subsequent
controlled by large boulders in the dam. Blasting ef-
forts did not significantly alter this situation (Fig. 11).
When overflow started, the lake had reached a
length of 21 km and an approximate volume of 450
million m³. In the middle of July 2010, Hunza Lake
was about 22 km long, covered an area of 12 km² and
had a volume of almost 600 million m³ (k
aRGeL
et
alii, 2010). The main reason for the further growth of
the lake after the onset of the overflow was the higher
inflow during summer.
Flooding of the area upstream of the dam led to the
inundation of 240 houses in 5 villages. 23 km of the
Karakorum Highway were destroyed. 25,000 people
living upstream of the dam suffer from lack of eco-
nomic activity and items of daily sustenance.
CASE STUDY 3: SIACHEN-GAYARI ICE/
ROCK AVALANCHE
On 7 April 2011, a snow avalanche from the Salt-
oro Range hit a northern parent glacier below Bilafond
Glacier in the Siachen Region in Jammu Kashmir, Pa-
kistan (see Fig. 2 for location). The resulting ice ava-
lanche entrained material from a lateral moraine and
overran the Gayari military camp. 139 people were
buried under the deposit of snow, ice, rock and debris
Fig. 12 - Siachen-Gayari ice/rock avalanche, looking NW:
the remaining tongue of the glacier, the capped
moraine, the ice/rock debris-cone/dam and the
beginning of impounding are visible on this pho-
tograph. The buried camp is situated on the left
side of the picture, where also the spillway was
dug out. Photo taken by Pakistan Army shortly af-
ter the event
Fig. 11 - Attabad Dam: Front Works Organization (FWO)
excavating the overflow channel in order to lower
the lake level as well as the hazard of an outburst
flood wave. (a) Coffer dam to hold lake water
back with drilling rigs for installing explosives.
(b) Blasting coffer dam to let the water erode
the artificial channel. (c) Excavators working on
deepening the channel. The work is slow despite
the heavy machinery. It can be done stepwise only
during the dry season. Note the large gneiss boul-
ders in a clayey matrix of fine debris mixed with
lake sediments, which prevent the channel being
eroded. Photos courtesy of FWO
background image
IMPACT OF LARGE LANDSLIDES, MITIGATION MEASURES
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
81
CHALLENGES FOR RISK MITIGATION
As demonstrated in the previous chapters, land-
slide-dammed lakes may drain after a few hours or
days, but may also persist for millennia. Even though
most dam failures occur in the first few months after
their formation, lakes may represent hazards several
years later (Karli Lake) or may be at least perceived
as such even after 100 years (Lake Sarez). Each stage
in the history of a dam requires specific risk mitigation
strategies, including a combination of technical and
non-technical measures.
Immediate emergency measures have to include
the construction of erosion-resistant open spillways or
drainage tunnels in order to constrain the lake water
level. Such structures reduce upstream flooding and
pressure on the dam, allowing for controlled drainage.
The finding by S
CHUSteR
(2006) that some spillways
work fine while others fail is supported by the recent
cases presented here: the Attabad spillway was still
working one and a half years after coming into opera-
tion in spite of a rather negative prognosis before. It
is often hard to predict the performance of spillways
prior to the actual overflow due to a variety of uncer-
retreat of permafrost and glaciers is putting many
settlements and activities in high mountain valleys
at risk. Therefore, places formerly deemed safe with
regard to natural hazards need to be reinvestigated. A
brief look at satellite images of Kashmir showed sev-
eral hamlets, camps and infrastructures located below
possibly hazardous glaciers or rock formations, thus
being in situations similar to that which led to the
tragedy at the Gayari Camp.
Fig. 13 - UNOSAT poster used as base for decision making. The avalanche originated in the firn area of the Saltoro Range
on the left side of the scene, entraining ice from the seracs and till from the lateral moraine on the orographic right
side of the valley. The debris cone/dam and the impounded lake are clearly visible on the post-event scene. Imagery:
Ikonos, May 4, 2012
Fig. 14 - Heavy equipment excavating human remains and
the artificial spillway. Photo courtesy of ISPR.
Note the compacted debris consisting of ice and
crushed rocks. Blasting efforts were not successful
background image
J.F. SCHNEIDER, F.E. GRUBER & M. MERGILI
82
International Conference on Vajont - 1963-2013 - Thoughts and analyses after 50 years since the catastrophic landslide Padua, Italy - 8-10 October 2013
tain parameters, particularly regarding the internal
structure of the dam.
Furthermore, well-designed spillways still have a
limited capacity and are not able to withstand powerful
impact waves triggered by mass movements into lakes.
In the case of Hattian Bala, such a wave initiated the
retrogressive erosion of the dam because of the absence
of rip rap or gabions.
Additional non-technical measures are always re-
quired. Such include the evacuation of people from
possibly affected zones. NDMA reported that in the
case of Attabad, 2,692 families living downstream
in the district Hunza and Gilgit were evacuated to 25
camps. Since such action is very sensitive, the extent
of the area to be evacuated has to be well defined. A
quick method to categorize the susceptibility of down-
stream areas to inundation due to an outburst flood
is to display their height above the river bed. Such
a map can easily be derived using GIS and a digital
elevation model, but it does not account for the spe-
cific characteristics of the possible flood (inundation
height, velocity, travel time). Such parameters require
the application of physically-based computer models
for the propagation of floods and debris flows. Fig-
ure 15 shows the hazard indication map for a possible
outburst flood of the Siachen-Gayari lake, based on
the assumption of outburst hydrographs, subsequent
physically-based modelling with FLO-2D and the
height above river. Also in the case of Attabad, a com-
bination of height above river and physically-based
modelling was applied in order to support the selec-
tion of evacuation areas downstream of the dam.
Figure 16 illustrates a section of the resulting
hazard indication map, giving a first impression of
potentially affected areas. It has to be strongly em-
phasized that such maps have to be interpreted with
utmost care and with the awareness of the specific
capacities and shortcomings of the input data applied
in the model as well as of the software used. The
comparison of the results from more than one type of
model is highly advisable.
Flood propagation modelling often simulates
worst-case scenarios, assuming a specific initial situ-
ation. Long-term evacuation of downstream areas is
generally neither desirable nor feasible from a socio-
economic point of view, apart from the fact that the
population of evacuation camps yearns to return to
their houses. Therefore the prediction and early rec-
ognition of specific critical situations is essential, par-
ticularly in the case of dams persisting for more than
a few weeks or months. However, such tasks have
proven to be difficult. Continuous monitoring of pos-
sible triggers of dam failures (e.g., unstable slopes or
inflow into the lake) in combination with the instal-
lation of sensors for impact waves and flooding may
be highly useful given that (1) the systems are main-
tained in an appropriate way, (2) they are connected
to an operational emergency warning system and (3)
the communities downstream are prepared and know
how to react in case. Computer models can help to
estimate travel times and therefore the period avail-
able for evacuation.
Fig. 16 Example of a hazard indication map for a possible
outburst of Hunza Lake covering a section of the
valley approx. 80 km downstream of the dam, based
on height above river and modelling with FLO-2D.
The planes of Gilgit and Danyore are formed by lake
sediments of the former Baktor dam (h
ewitt
, 2011)
Fig. 15 - Hazard indication map showing the area down-
stream of the Siachen-Gayari ice/rock avalanche
based on heights above river and FLO-2D. The
hydrograph applied as input to the modelling
of the flood wave propagation considers water
bulked with sediment and debris
background image
IMPACT OF LARGE LANDSLIDES, MITIGATION MEASURES
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
83
mitigation strategies, including monitoring, awareness
and preparedness building. Using the know-how from
former events helps to understand these actual rapid
landform change processes.
ACKNOWLEDGEMENTS
The first author of the present paper was invited
to the sites discussed in the article by SDC (Switzer-
land), FOCUS (Tajikistan), NDMA (Pakistan) and UN
OCHA. Special thanks go to General Ahmed Nadim
(NDMA), Kamran Shariff (UN OCHA), Dr. Kausar
(Geological Survey of Pakistan) and Mustafa Karim
(FOCUS).
CONCLUSIONS
Landslides are common geomorphologic process-
es in mountain areas all around the world. Sometimes
they block entire valleys and impound lakes which
may drain suddenly. Steep and narrow valleys in seis-
mically active zones (like the Pamir of Tajikistan and
the Karakoram in Northern Pakistan) are particularly
susceptible as geomorphologic and historical evidence
has shown along with more recent cases.
Even though there are no means to prevent the
formation of landslide dams, the prediction of possi-
ble dam failures remains a challenge. Adverse effects
on people, property and infrastructures can be allevi-
ated by applying a combination of appropriate risk
REFERENCES
C
aSaGLI
n. & e
RMInI
L. (1999) - Geomorphic analysis of landslide dams in the Northern Apennine. Transactions of the Japanese
Geomorphological Union, 20: 219-249.
C
LaGUe
J.J. & e
vanS
S.G. (1994) – Formation and failure of natural dams in the Canadian Cordillera. Geological Survey of Canada
Bulletin, 464: 1-35.
C
oSta
J.e. & S
CHUSteR
R.L. (1988) - The formation and failure of natural dams. Geological Society of America Bulletin, 100: 1054-
1068.
C
oSta
J.e. & S
CHUSteR
R.L. (1991) - Documented historical landslide dams from around the world. U.S. Geological Survey Open-File
Report, 91-239: 1-486.
D
eLaney
k.b. & e
vanS
S.G. (2011) - Rockslide Dams in the Northwest Himalayas (Pakistan, India) and the Adjacent Pamir Mountains
(Afghanistan, Tajikistan), Central Asia. In: e
vanS
S.G., H
eRMannS
R.L., S
tRoM
a., S
CaRaSCIa
M
UGnozza
G. (
eDS
). Natural and
artificial rockslide dams. Lecture Notes in Earth Sciences, 133: 205–242.
C
oRnWeLL
k. & H
aMIDULLaH
S. (1992) - Geomorphic evidence of catastrophic flooding along the middle Indus valley. Geological
Bulletin University of Peshawar, 25: 113-121.
D
UnnInG
S.a., M
ItCHeLL
W.a., R
oSSeR
n.J. & p
etLey
D.n. (2007) - The Hattian Bala rock avalanche and associated landslides triggered
by the Kashmir Earthquake of 8 October 2005. Engineering Geology, 93 (3-4): 130-144.
H
eWItt
k. (1964) - The Karakoram Ice Dam. Indus, 5: 18-30.
H
eWItt
K. (1982) - Natural dams and outburst floods in the Karakorum Himalaya. In: G
Len
J. W. (
eD
.). Hydrological aspects of alpine
and high-mountain areas. IAHS Publication, 138: 259–269.
H
eWItt
k. (1998) – Catastrophic landslides and their effects on the upper Indus streams, Karakorum Himalaya, northern Pakistan.
Geomorphology, 26: 47-80.
H
eWItt
k. (2011) - Rockslide Avalanche Dams on the Trans Himalayan Upper Indus Streams: A Survey of Late Quaternary Events amd
Hazard-Related Characteristics. In: e
vanS
S.G., H
eRMannS
R.L., S
tRoM
a. & S
CaRaSCIa
M
UGnozza
G. (e
DS
). Natural and artificial
rockslide dams. Lecture Notes in Earth Sciences, 133: 177-204.
I
SCHUk
a.R. (2004) - Usoy natural dam - Problem of security (Lake Sarez, Pamir Mountains, Tajikistan). NATO-ARW Security of
Natural and Artificial Rockslide Dams, June 8–13, 2004, Bishkek, Kyrgyzstan. Abstracts Volume.
I
SHUk
a.R. (2011) - Usoi rockslide dam and Lake Sarez, Pamir Mountains, Tajikistan. Natural and artificial rockslide dams. In: e
vanS
S.G., H
eRMannS
R.L., S
tRoM
a. & S
CaRaSCIa
M
UGnozza
, G. (e
DS
). Natural and artificial rockslide dams. Lecture Notes in Earth
Sciences, 133: 423-440.
k
aRGeL
J.S., L
eonaRD
G., C
RIppen
R.e., D
eLaney
k.b., e
vanS
S.G. & S
CHneIDeR
J.F. (2010) – Satellite Monitoring of Pakistan's
Rockslide-Dammed Lake Gojal. EOS, 91 (43): 394-395.
k
onaGaI
k. & S
attaR
a. (2012) - Partial breaching of Hattian Bala Landslide Dam formed in the 8th October 2005 Kashmir Earthquake,
Pakistan. Landslides, 9: 1-11.
background image
J.F. SCHNEIDER, F.E. GRUBER & M. MERGILI
84
International Conference on Vajont - 1963-2013 - Thoughts and analyses after 50 years since the catastrophic landslide Padua, Italy - 8-10 October 2013
M
aSon
(1929) - Indus Floods and Shyok Glaciers. The Himalayan Journal, 1: 10-29.
n
eSpak
(2010) - Dam Break Study of Hunza River near Atabad. Report to the National Disaster Management Authority, Pakistan.
p
etLey
D., R
oSSeR
n.J., k
aRIM
D., W
aLI
S., a
LI
n., n
aSab
n. & S
Haban
k. (2010)- Non-seismic landslide hazards along the Himalayan
arc. In: W
ILLIaMS
a.L., p
InCHeS
G.M., C
HIn
C.y., M
CMoRRan
t.J. & M
aSSey
C.I. (e
DS
) – Geologically Active. CRC Press, London:
143–154.
R
ICHaRDSon
S.D. & R
eynoLDS
J.M. (2000) - An overview of glacial hazards in the Himalayas. Quaternary International, 65/66: 31-47.
R
ISLey
J.C., W
aLDeR
J.S. & D
enLInGeR
R.p. (2006) - Usoi Dam wave overtopping and flood routing in the Bartang and Panj Rivers,
Tajikistan. Natural Hazards, 38(3): 375-390.
S
CHneIDeR
J.F. (2009) - Seismically reactivated Hattian slide in Kashmir, Northern Pakistan. Journal of Seismology, 13 (3): 387-398.
S
CHUSteR
R.L. (2006) - Risk-reduction measures for landslide dams. Italian Journal of Engineering Geology and Environment, Special
Issue I: 9-13.
S
CHUSteR
R.L. & a
LFoRD
D. (2004) -Usoi Landslide Dam and Lake Sarez, Pamir Mountains, Tajikistan. Environmental and Engineering
Geoscience, 10 (2): 151-168.
S
CHUSteR
R.L. & e
vanS
S.G. (2011) - Engineering measures for the hazard reduction of landslide dams. In: e
vanS
S.G., H
eRMannS
R.L.,
S
tRoM
a. & S
CaRaSCIa
M
UGnozza
, G. (
eDS
). Natural and Artificial Rockslide Dams. Lecture Notes in Earth Sciences, 133: 77-100.
S
HRoDeR
J.F. (1998) - Slope failure and denudation in the western Himalaya. Geomorphology, 26 (1-3): 81-105.
S
HRoDeR
J.F. & W
eIHS
b.J. (2010) - Geomorphology of the Lake Shewa landslide dam, Badakhshan, Afghanistan using remote sensing
data. Geografiska Annaler, 92A (4): 469-483.
S
HRoDeR
J.F., C
oRnWeLL
k. & k
Han
M.S. (1991) - Catastrophic breakout Floods in the western Himalaya, Pakistan. Geological Society
of America Abstracts with Programs, 23 (5).
Statistics