ORI GIN AL PA PER
Causes of catastrophic failure of Tam Pokhari morainedam in the Mt. Everest region
Rabindra Osti • Tara Nidhi Bhattarai • Katsuhito Miyake
Received: 8 August 2010 / Accepted: 11 January 2011 / Published online: 22 January 2011� Springer Science+Business Media B.V. 2011
Abstract The moraine dam of the Tam Pokhari glacial lake breached on 3 September
1998 and caused a catastrophic flood in the downstream areas. To learn from the event, a
field survey was conducted. The survey team found that a landslide, which is considered to
be responsible for the outburst flood, occurred in the northeast-facing slope of the moraine
dam. The dam internal structure played a crucial role in forming a landslide that triggered
the excess overflow and finally the breach of the dam. The internal structure of the dam
was made of alternating layers of finer and coarser sediments inclining at 30� downstream
and layers are truncated in the upslope direction by a huge pile of unconsolidated and
structureless moraine materials. Since the upstream slope angle of the dam i.e., 40� is
larger than the angle of repose i.e. 35� of sediments, the increased pore water pressure in
the dam triggered a landslide. The rainfall and seismological activities of that particular
day, which hit the record high, were crucial in triggering the failure. It is estimated that the
dam’s north and northeast-facing slopes completely slid involving about 30,000 m3 of
sediment mass of unconsolidated moraine materials above the shear plane. A slope stability
analysis was also performed. The calculated safety factor was 0.85, and the calculated slip
circle agreed with the shear plane marked in the dam. About 18 million cubic metres of
water was swiftly released due to the sudden breach of the moraine dam.
Keywords Climatic change � Mountain hazard � Moraine dam formation and failure �Slope stability � Landslides � Himalaya
R. Osti (&) � K. MiyakeInternational Centre for Water Hazard and Risk Management, Public Works Research Institute,Minamihara 1-6, Tsukuba 305-8516, Japane-mail: [email protected]
T. N. BhattaraiDepartment of Geology, Tribhuvan University, Tri-Chandra Campus, Ghantaghar, Kathmandu, Nepale-mail: [email protected]
123
Nat Hazards (2011) 58:1209–1223DOI 10.1007/s11069-011-9723-x
1 Introduction
The high mountain regions around the globe are geologically weak and known as global
spot of natural hazards. The frequency and magnitude of such mountain hazards have been
significantly influenced by climatic change (IPCC 2007; UNEP 2007). According to the
reports (IPCC 2007; UNEP 2007), glaciers in the region are rapidly melting due to climatic
change at almost double the rate reported in the early 1970s. Melting process including
sub-arial melting, mechanical fracture, preferential exploitation of structures within buried
ice bodies etc. enlarge existing lakes or creates new lakes (Benn et al. 2000; Richardson
and Reynolds 2000; Watanabe et al. 2009). Many glacial lakes are dammed by uncon-
solidated moraine materials. Such moraine dams can breach for various reasons (Post and
Mayo 1971; RGSL 2003; Hambrey and Alean 2004; Clague 2003; Korup and Tweed 2007)
and in various modes (Costa and Schuster 1988; Grabs and Hanisch 1993; Wahl 1998) and
pose immediate threats of flash floods to downstream communities. Different triggering
factors are proposed in the aforementioned publications to explain possible moraine-dam
breaches in different geological settings; however, it is very challenging to verify rec-
ommended criteria for each event based on field observations mainly due to a lack of
relevant data and difficulties in conducting field investigation in a high-altitude environ-
ment. Further, it is still not clear whether mentioned reasons and modes of dam failure are
the all that represent the context. Indeed, there are only a few scientific reports on GLOF
events, and other available discussions are also primarily focused on historical accounts of
past events, satellite observations, and early warning and possible control measures.
Therefore, there is a knowledge gap, despite the fact that GLOF issues are very much
highlighted in recent years. Before coming to any conclusion on prediction and mitigation
tools, many cases need to be analysed particularly to understand the moraine dam failure
mechanism and GLOF development process. In this paper, a case study of Tam Pokhari
Glacial lake outburst flood in Nepal will be discussed.
Khumbu in Solukhumbu District, Nepal, also known as the Everest region, is one of the
highest altitude regions in the Himalayas, which consists of large number of glaciers and
glacial lakes and is considered a highly GLOF prone area. The Tam Pokhari glacial lake,
locally known as Sabai Tsho, is one of many glacial lakes in the region and located at
27�440N, 86�500E at an altitude of 4,400 m (Fig. 1). A small stream, the Inkhu River,
originates from the lake. The moraine dam of Tam Pokhari breached on 3 September 1998
and resulted in a catastrophic glacial lake outburst flood, which is widely known as glacial
lake outburst flood (GLOF). The event killed a few people and damaged million dollars
worth of property (Osti and Egashira 2009). This is one of such events that have not been
well reported despite the fact that it is one of the serious GLOF events in Nepal (WWF
2005).
Geologically, the region belongs to the ‘‘Higher Himalayan Zone’’ extending
throughout the Himalayas. The Main Central Thrust (MCT), i.e., a low angle reverse fault,
and the South Tibetan Detachment System (STDS), i.e., a normal fault system, mark their
southern and northern boundaries, respectively. The zone consists of a 10–12-km-thick
succession of high-grade metamorphic rocks, such as gneisses, migmatites, schiests,
quartzites and marbles (Gansser 1964). These rocks have developed a steep, rugged
topography throughout the Himalayas.
The region is also characterized by humid climate with snowfall in winter (November–
April) and south-west monsoon in summer (June–September), which together dominates
the regional hydrology. The main type of summer monsoon precipitation is drizzle with
sporadic, more intense events. Rainfall usually starts around 16:00 and continues until
1210 Nat Hazards (2011) 58:1209–1223
123
around 6:00 with its peak around mid night (Bollasina et al. 2002; Ueno et al. 2001).
According to Ueno et al. (2001), 79% of the annual rainfall in the region was observed in
the monsoon season during the period from 1995 to 1999 (observation station located at
86�430E, 27�490N, 3,833 m asl). The region’s daily low and high temperatures for the same
period fluctuated between -11 and 11�C, and January and August were the coldest and
warmest months, respectively.
The main objective of this study was to identify triggering factors and explain a
breaching mechanism of the Tam Pokhari moraine dam aiming at improving the under-
standing and techniques of GLOF disaster assessment, prediction and mitigation. This was
possible only through integrated field observation and practical assessment of the event
Fig. 1 Locations of glacial lakes(circled) in the Inkhu River basinand topographical map of thesurrounding area (red circle is aTam Pokhari lake and bluecircles are other glacial lakes)
Nat Hazards (2011) 58:1209–1223 1211
123
based on field evidences. This paper describes observational evidence of the Tam Pokhari
moraine-dam failure and discusses triggering factors and a dam failure mechanism.
2 Methodology
A field survey including geological investigation was conducted during the late summer,
2008. The drawdown depth of the Tam Pokhari glacial lake was measured in reference to
the permanent water mark observed on the vertical slope located in the northwest part of
the lake. The lake width and length were measured by using an inclinometer and mea-
surement tape. The surface inspection of the moraine dam involved on-the-around
observations of grain size and sediment composition, geometry of the breached part, in-
and out-ward slopes of the dam and other geometric features. On-site sediment size
samplings were done by referencing a measurement tape laid on the exposed surfaces of
the dam. The locations and elevations of observation points were identified with the Global
Positioning System (GPS) and GPSmap hand tool kits. Dam cross sections were measured
at several locations to capture representative dam structures. Rainfall data recorded at three
nearby stations were also analysed to understand any possible influence of rainfall in
triggering the dam failure. Since the region is highly prone to earthquakes, relevant his-
torical seismic data of the region was additionally used to check whether seismic activities
could have had any impact on the event. The surface discharge of the lake was measured at
about 500 m downstream from the lake by using the float (distance travelled by the cork
divided by total time required to travel) method. The discharge measured during field
survey in September indicates roughly the outflow condition before the dam breach. It was
difficult to measure discharge at the outlet of the lake due to the existence of boulders and
an excess amount of seepage flow. However, the discharge measured at 500 m is also
indicative only as it might have been influenced by secondary discharges such as incoming
and outgoing seepage flows. Group and individual interviews were conducted with local
people of Thangnag Village, which is located approximately 1 km downstream from the
lake, especially to cross-check the observations.
The two-dimensional numerical programme namely GeoStudio developed by GEO-
SLOPE International Ltd. in Canada was used to roughly assess the stability of moraine
dam in a given geological and hydrological conditions. GeoStudio is useful to solve a
wide variety of practical geotechnical problems and can be used to simulate both sat-
urated and unsaturated ground-water flows under steady-state and transient conditions
(GEO-SLOPE International Ltd. 2010). The SEEP/W module of GeoStudio can be used
to simulate, in particular, the pore-water pressure changes inside the dam body due to the
accumulation of rainfall or the surge from the lake. However, the SLOPE/W module of
GeoStudio was only used to quantify the stability of the slope at a given pore water
pressure condition, conceptually and quantitatively represented by the factor of safety. In
SLOPE/W, a wide variation of trial slip surfaces can be specified with a defined grid of
circle centres and a range of defined radii. In the modelling process, the surface areas can
be specified on the top and on the toe of the dam, where the trial slip surfaces will enter
and exit respectively. The number of entries and exits can also be specified as the
number of increments along these two lines. The inbuilt Morgenstern-Price Method,
which uses two factors of safety equations: one with respect to moment equilibrium and
another with respect to horizontal force equilibrium, was used to access the stability i.e.
the factor of safety of the slope.
1212 Nat Hazards (2011) 58:1209–1223
123
3 Site inspection
Tam Pokhari is a kidney-shaped lake, 1 km long at the surface along the centreline (north–
south) and 350 m wide at its widest point (east–west). The lake is formed within an end
moraine composed mainly of loose and unconsolidated boulders, gravels and sand with a
subordinate amount of silt and clay. No ice component was found within the sediment mass
of both outer and inner parts of the dam. Except the northern side, where massive gneiss is
exposed and forms a steep cliff, all the remaining lake sides are made up of loose moraine
materials laid on the top of the rocky slopes. Since the altitude of the region is above
4,500 m, there is no vegetation cover in the dam and surrounding areas. The end moraine
dam at the outlet is about 100 m high and takes a curvilinear shape. Its top width is
estimated to be 50 m with its upstream and downstream faces inclining at 38 and 25� on
average, respectively (Fig. 2). The frontal or lake-facing slope of the dam is steeper (38�)
than the angle of repose of similar loose moraine materials (30�–35�). These are common
characteristics of unstable moraine dams (Costa and Schuster 1988), and it suggests that
moraine dams can easily deteriorate due to excess rainfall or earthquakes without obvious
surface manifestations. The steep upstream slope of the dam was maintained by the
pressure exerted by propagating glaciers.
Figure 3 shows the upstream facing pictures of the existing lake. As shown in Fig. 3a,
the lake receives melt water from the existing glacier (the Sabai Tsho glacier), which snout
is located about 150 m upslope to the north from the lake head. In the space between the
glacier tongue and the lake is a steep and narrow cliff composed of massive gneiss.
The present outlet, the origin of the Inkhu Khola, lies at the end of the lake and exactly at
Fig. 2 Geometrical feature of Tam Pokhari moraine dam
Nat Hazards (2011) 58:1209–1223 1213
123
the dam breached section (Fig. 3b). The outlet discharge is estimated at 1.4 m3/s, i.e., the
discharge measured at 500 m downstream from the outlet. This is not the time for annual
peak discharge due to the snow melt, and this discharge is far less than the total capacity of
the outlet.
The breached portion of the dam has a trapezoidal shaped opening that has an average
width and height of 60 and 50 m, respectively (Figs. 2c, 3b). The existing outlet (trape-
zoidal opening) is marked with two distinct banks cross-sectioned along the dam alignment
(Fig. 4). The dam materials in the west (right) bank, which is the inner part of the bend in
Fig. 3 Lake level down-draw (50 m), damaged moraine dam and debris plunged into the lake (camerafacing upstream (a) and downstream (b))
1214 Nat Hazards (2011) 58:1209–1223
123
the kidney-shaped lake alignment, are comparatively finer than those in the east (left) bank
as clearly seen in the photographs (Fig. 4). Gravels and boulders are predominant in the
left bank while such materials are less populated in the internal structure of the right bank.
This difference in sediment composition, especially in the internal structure of the left and
right banks, resulted from variation in deposition by glacier as the dam was formed at the
end of the curvilinear alignment of the glacier path. As a result of such variation in
sediment compositions, the left-side slope of breached section remains steeper (more than
60�) and continuous, resembling a part of the U-shaped valley, whereas the right-side slope
is gentler (less than 40�) and irregular (Fig. 4). In the pocket (lake facing side) of the dam
on either side of the outlet at the present lake level, fine materials mixed with gravels are
seen predominantly deposited nearby while boulders are in the lake body and spread up to
30 m from the foot of the dam (Fig. 3a). The sediment size distribution of the deposited
sediment mass at the mouth of the lake is illustrated in Fig. 5.
As observed on the right slope of the breached portion (Fig. 6), the dam is made up of
alternating layers of finer and coarser sediments. These sediment layers incline at the same
angle (about 30�), parallel to the downstream-facing slope of the moraine dam. The longer
axis of gravels and boulders found in these layers are also parallel to the inclination of the
sediment layers. These sediment layers are truncated in the upslope direction by a huge pile
of unconsolidated moraine materials in which boulders are randomly oriented and a
Fig. 4 Difference in sediment composition in the left and right banks of dam breached section
Fig. 5 Grain size distribution of deposited sediment mass at the lake mouth
Nat Hazards (2011) 58:1209–1223 1215
123
layering structure is also lacking. An inclined thin layer (dip amount 45� due northeast) of
clayey (crushed) materials marks the contact between the edges of underlying inclined
sediment layers and the overlying unconsolidated moraine materials. The clayey materials
consist of a very few gravels with their longer axis aligned parallel to the 45-degree
inclination of the clay layer. It is to note that the layering structure of dam extended up to
the upstream face of the dam in the past, but the structure was damaged along with the
downward movement of the upstream-facing slope (a mass slide without detachment from
the dam), whose stability became critical after glacier melt. These features of the upstream-
facing part of the dam suggest the previous movement of sediment mass within the dam
and further indicate the fact that the contact plane could have been the share plane of the
landslide.
4 Hydro-meteorological conditions
In order to analyse rainfall trends, 28-year rainfall data (1974–2002) observed at the
Chaurikharka Weather Station in Lukla, the nearest available permanent weather station
monitored by the Department of Hydrology and Meteorology of Nepal, were examined.
Another set of data is also available for 5 years (1994–1999) from short-term weather
monitoring conducted by Ueno et al. (2001) at Syangboche (27�490N, 86�430E,), located
closer to Tam Pokhari than Chaurikharka. The trend of rainfall during the peak monsoonal
months (July, August and September) at the Chaurikharka station for the period of
1974–2002 indicates that there was no significant change in the 3-month total rainfall
(Fig. 7). However, August monthly rainfall was noticeably in an increasing trend, while
July and September monthly rainfalls were in a decreasing trend over the same period
(Fig. 7). The data from Ueno and his colleagues clearly shows that the rainfall in the area
was exceptionally large in August 1998. The sharp increase in rainfall in the month of
August was also confirmed based on the data set from the Chaurikharka Station (Fig. 7).
Furthermore, at Chaurikharka and also at another nearest weather station in Chepuwa
(25�900N, 87�250E), the daily rainfall was also found to be the highest during 29 August–
2 September 1998 (Fig. 8). Therefore, the August monthly rainfall at the nearest
Fig. 6 Internal structure of the Tam Pokhari moraine dam
1216 Nat Hazards (2011) 58:1209–1223
123
observatory surpasses the historical high from 1974 to 2002, while the 3 days rainfall from
29 August to 2 September 1998 was the highest in 1998.
5 Seismological condition
It is also important to note that the study area is highly prone to earthquake. The earthquake
history of the region (USGS 2009), particularly the marked epicentres within 300 km
radial distance from the lake, indicates that about twenty high-magnitude earthquake
events (over 4 Richter Scale) were recorded during the last week of August to the first
week of September 1998 (Table 1), which is extremely high in the earthquake history of
the region. The first biggest trigger of magnitude 5.9 was recorded on 25 August, 1998
followed by 11 consequent earthquake events on the same day. The other events of
magnitude over 5 were also recorded on 28 and 30 August followed by two other triggers
of magnitude over 4 on 30 August, 1998. There are eight recorded earthquake events in
total during 1973–2009, which have epicentres within 15 km aerial distance from the lake,
and four of them were recorded on the day of the dam breach. Referring to the trend of
occurrences of earthquake events during that time period, there is also a chance that several
low-magnitude (less than magnitude 4 i.e. besides four major events of the day) earth-
quakes could have hit the area on 3 September 1998 prior to the dam breach. The events
with magnitude less than 4 are not documented in the report (USGS 2009). In fact, a report
by Sponsortrek (2008) quoted the local people as well as the government sources and
claimed that three earth quakes hit this region on September 2nd and 3rd and the one that
took place at 5.50 am (about 10 min prior to the dam breach) on 3 September had a
magnitude of 3.77 on the Richter Scale. Although none of the recorded timings of the
Fig. 7 Observed monthly rainfall at Chaurikharka Hydrological station
Fig. 8 Observed 6 days daily rainfall at Chepuwa and Chaurikharka (after Ueno et al. 2001)
Nat Hazards (2011) 58:1209–1223 1217
123
earthquakes exactly coincides with the timing of Tam Pokhari moraine dam failure i.e.
6.15 am in the morning of 3 September 1998 as quoted by the local people, the record of
that particular period clearly shows the evidence that the earthquake is one of the factors to
destabilize the moraine dam structures prior to the breach.
6 Result and discussion
The observation results suggest that the surface area of the lake reduced by 44% due to the
1998 dam breach event. The total water volume discharged during the event was
approximately 18 9 106 m3, which was calculated based on the observed drawdown depth
Table 1 Earthquake events recorded within 300 km distance from the Tam Pokhari Glacial Lake during 15August to 15 September 1998 (USGS 2009)
No Date Time Location Magnitude
(UTC) Latitude Longitude (Richter Scale)
1 August 24, 1998 7:32:21 30.001 87.919 5.0
2 August 25, 1998 7:41:40 30.079 88.109 5.9
3 August 25, 1998 7:59:00 30.024 88.188 4.6
4 August 25, 1998 8:01:45 30.061 88.172 4.1
5 August 25, 1998 8:13:04 30.172 88.134 3.9
6 August 25, 1998 9:43:06 29.998 88.099 4.1
7 August 25, 1998 9:56:44 29.831 87.910 4.4
8 August 25, 1998 10:25:06 29.983 88.103 4.4
9 August 25, 1998 12:29:45 30.022 88.060 4.6
10 August 25, 1998 12:43:04 29.961 88.091 4.4
11 August 25, 1998 13:39:49 29.926 88.124 3.8
12 August 25, 1998 13:50:06 30.099 88.054 4.1
13 August 25, 1998 15:16:02 29.976 88.105 4.4
14 August 28, 1998 22:01:55 30.191 88.150 5.0
15 August 30, 1998 3:37:48 30.044 88.077 5.0
16 August 30, 1998 3:49:48 29.904 87.989 4.0
17 August 30, 1998 4:11:35 30.034 88.081 4.6
18a September 3, 1998 18:15:56 27.85 86.941 5.6
19 September 3, 1998 18:22:19 27.973 87.040 –
20a September 3, 1998 18:51:38 27.65 86.802 4.4
21a September 3, 1998 21:00:53 27.729 86.880 4.1
22a September 3, 1998 23:02:28 27.6 86.733 4.6
23 September 4, 1998 0:36:17 27.455 86.464 4.1
24a September 4, 1998 1:10:18 27.777 86.793 4.2
25 September 4, 1998 21:16:13 27.651 86.720
26a September 6, 1998 21:35:49 27.859 86.895 4.6
27 September 10, 1998 22:57:16 27.199 88.341 4.7
28a September 12, 1998 5:08:52 27.852 86.900 4.2
a Epicentre located within 15 km radius from the Tam Pokhari Glacial Lake
1218 Nat Hazards (2011) 58:1209–1223
123
(50 m) and the difference in surface area. According to local people, the overflow from the
lake continued for several hours, while the maximum flow lasted for about half an hour.
Snow avalanches are often considered to have caused many historical GLOF events in
the region. Some researchers (e.g. Vuichard and Zimmermann 1987) have also pointed out
a fact that the glacial lake had breached in the past due to snow or ice avalanches in the
region. In the case of Tam Pokhari Lake, a snow avalanche was considered as the sole
cause of dam failure (Dwivedi et al. 2000). A clear evidence of snow avalanche (not
necessarily on the same day of the dam breach) is also the snout of the existing Sabai Tsho
glacier, located in the northern part of the lake, which has shifted about 150 m upslope than
it was in the past (Figs. 3b, 9). However, if a huge mass of snow or Ice avalanche had
fallen into the lake from a glacier, it would not have been able to cause much impact
because it would have rather formed a pile of snow or ice mass at the glacier foot where the
lake is narrow and shallow. It is also to note that the lake shape does not permit any wave
originating from the lake-head to propagate in a single direction; such a wave usually
reduces its strength due to to-and-fro motion between east and west banks. Therefore, it is
possible to say that a snow or ice avalanche plunged into the lake, displaced the lake water
and generated surges that propagated up to the dam and raised the water level for a while,
but this cannot be the direct cause of the dam breach. Lacking any evidence of impact on
the eastern and western sides of the lake also does not permit to justify the occurrence of
huge surge induced by a snow or ice avalanche. Additionally, the left and right sides of the
lake are steep (more than 60�) and stable mountains (Fig. 3), where a huge snow mass may
not be able to retain for a long period; therefore, chances of snow avalanche generation
from either side slope is also minimum. Similarly, there were also no signs of the
occurrence of rock avalanches from the side slopes around the lake, which indicates that
the dam breach was not caused by a rock avalanche, either.
The moraine dam location, lake shape, surrounding topography and outlet condition
indicate that the lake shape and volume have little changed in recent decades even if there
is an increment in glacier melting rate. The stability of lake size and shape highlights the
fact that the dam did not breach as a consequence of the increase in water volume in the
lake.
Fig. 9 Sabai Tsho Glacier before glacial lake outburst (left—Sponsortrek 2008; right—Photo by MichaelDurham dated October 2005 with permission)
Nat Hazards (2011) 58:1209–1223 1219
123
It is important to note that the continuous rainfall during 29 August to 2 September 1998
had contributed to a significant amount of infiltration into the dam. Since the dam sedi-
ments are heterogeneous, the infiltration may have caused a substantial increase in pore
water presser inside the dam, especially where finer materials are predominant. It is also to
note that summer infiltration is much higher than infiltration due to snow melt in late
winter as the ground is frozen, and snow on the ground start melting from top to bottom.
The seismic activities of that particular period had also helped weaken the soil strength
and increased the pore water pressure inside the soil mass within a short period of time.
Since the upstream slope angle of the dam (e.g. 40� on the lake side) is larger than the
angle of repose (35�) of sediments, the increased pore water pressure in the dam triggered a
landslide. Usually, a landslide is also preceded by saturation of the slope. This is com-
monly caused by heavy rainfall events or periods of extended precipitation that saturated
the upper part of the slope and raised groundwater level. Research results presented by
many researchers such as Fleming and Varnes (1991), Reid (1994), USGS (2004), Iverson
(2000) and Sassa et al. (2010) have shown that a rise in pore water pressure level above a
potential sliding surface can increase and sometimes double the potential for a landslide.
When the pore water pressure at the slip surface increases, it reduces the effective normal
stress and thus diminishes the restraining friction along the slip line. This is combined with
increased soil weight due to the added groundwater. Additionally, the seepage water a)
increased the weight of materials on a slope above their point of gravitational equilibrium,
b) increased pore pressures within a zone of weakness in the materials underlying a slope,
c) decreased the coefficient of friction on a potential sliding surface d) caused soil to
hydrate and expand increasing slope instability and e) dissolved minerals that were binding
the particles of the materials that make up the slope. The mentioned factors decreased the
overall strength of the materials making up the slope.
The geological features of the right bank of the outlet, especially the presence of a shear
plane that inclines towards the lake (Fig. 6), can be taken as evidence of the occurrence of
a landslide. The deposition of finer sediment (silty sand) overlain by coarser sediment and
deposited boulders on the lake bed around the outlet observed at the existing lake level are
an indication of a landslide occurrence on the dam. Landslide materials that plunged into
and deposited on the lake after segregation formed an inverted U-shape debris pile with a
vertex height of about 30 m from the foot of the dam (Fig. 3), as described in the previous
section of this paper. It is estimated that the dam’s north- and northeast-facing slopes
completely slid involving about 30,000 m3 of sediment mass of unconsolidated moraine
materials above the shear plane. The landslide may have led to the following
consequences:
i) The landslide reduced the height and width (thickness) of the dam, which in turn
increased the driving/shearing load and decreased the resisting/normal load in the dam. As
a consequence, the remaining portion of the dam became critically unstable and ultimately
collapsed.
ii) The snow avalanche that developed at the glacier snout plunged into the lake,
displaced the lake water first and then created dynamic waves in the lake. The surges or
seiches of any magnitude reached at the dam site and further increased in pore water
pressure, in addition to the one caused by continuous infiltration, in the dam materials. The
sudden increase in pore water pressure reduced the shear strength of the dam materials and,
consequently, triggered a landslide on the north- and northeast- facing slope of the dam
(near the outlet).
iii) The landslide mass also displaced the lake water and created back water waves.
Returning surges or seiches of any magnitude may have also further enhanced the failure of
1220 Nat Hazards (2011) 58:1209–1223
123
the dam whose stability had been already in a critical state due to the increase in pore water
pressure and decrease in height and width. A huge mass of slid moraine materials still
remain deposited on the foot of the moraine dam as described. The sediment mass could
not have been fully flushed except the part at the centreline of the flow.
iv) The increased discharge at the outlet due to the increased outlet capacity further
helped the dam failure process by rapid erosion. As a result, the dam breach occurred
within a short time inducing a huge water flux, which might have lasted for about 30 min
as noticed by local people.
The stability analysis based on the SLOP/W module in the GeoStudio model results in
the landslide with a slip circle radius of 170 m (Fig. 10) and a safety factor of 0.85. This is
quite reasonable for a case of failure. The detail of acting forces (in kN) and momentum
calculated by using the Morgenstern-Price Method for slice number 13 (shaded in Fig. 10)
as an example is illustrated in Fig. 11. It has been found that the slip circle coincides with
pre-existing share plane in the dam. In this result, the change in pore water pressure due to
Fig. 10 Calculated landslide slip circle
Fig. 11 Free body and force polygon for Morgenstern-Price method
Nat Hazards (2011) 58:1209–1223 1221
123
rainfall infiltration and earthquake are not taken into account. The results agree with the
field evidence and help deduce the conclusion that the main cause of the moraine dam
breach in Tam Pokhari Lake on 3 September 1998 was due to the landslide that occurred in
the north-facing slope of the moraine dam.
7 Conclusion
Moraine dam breaches and associated GLOF disasters have been a long scientific debate
without reaching any concrete conclusions. In order to understand the Tam Pokhari
moraine dam breach mechanism, a 2-week-long field survey was conducted in the Tam
Pokhari glacial lake in the Khumbu region, Nepal. The field observation found that the
internal structure of the Tam Pokhari moraine dam was made of alternating layers of finer
and coarser sediments inclining at 30� downstream. The layering structure extended up to
the upstream face of the dam in the past, but afterwards the upstream facing part of dam
structure slid with the downward movement possibly during glacier melting process.
Therefore, the upstream part of the dam lost its internal structure and remained as huge pile
of unconsolidated moraine materials, in which boulders are randomly oriented and the
layering structure is also lacking. An inclined thin layer of clayey (crushed) materials
marks the contact between the edges of underlying inclined sediment layers and the
overlying unconsolidated moraine materials. The contact plane was a share plane of a
landslide. Noticeably, the August monthly rainfall at the nearest observatory surpasses the
historical high from 1974 to 2002, while the 3 days rainfall from 29 August to 2 September
1998 was the highest in 1998. There were over twenty recorded earthquake events in
7 days time and four earthquake events recorded within the same day of dam breach. On
3 September 1998, due to the increase in pore water pressure in the share plane resulting
from continuous and excess rainfall as well as seismic activities and surge generated
through the snow avalanche, sediment pile above the shear plane slid and generated a
landslide that plunged into the lake body. Returning surges or seiches generated during the
landslide caused further damage to the dam, whose stability had already been critical due
to the landslide. The landslide occurrence is confirmed in the field observation and
numerical simulation. As a result, the Tam Pokhari glacial lake partially emptied, gener-
ating a huge water discharge, which became a catastrophic GLOF in the downstream areas.
References
Benn DI, Wiseman S, Warren CR (2000) Rapid growth of a supraglacial lake, Ngozump Glacier, KhumbuHimal, Nepal. In: Nakawo M, Raymond CF, Fountain A (eds) Proceedings of a workshop on debris-covered glaciers, IAHS Publication 264
Bollasina M, Bertolani L, Tartari G (2002) Meteorological observations at high altitude in the KhumbuValley, Nepal Himalayas, 1994–1999. Bull Glaciol Res 19:1–11
Clague JJ (2003) Catastrophic floods caused by sudden draining of lakes in high mountains. In: Martin K, HikDS (eds) The Science highlights from symposium on ecological and earth sciences in mountain: areasstate of ecological and earth sciences in mountain areas, September 2002, Banff, Alberta, Canada, 46 p
Costa JE, Schuster RL (1988) The formation and failure of natural dams. Geol Soc Am Bull 100(7):1054–1068
Dwivedi SK, Acharya MD, Simard R (2000) The Tam Pokhari Glacier lake outburst flood of 3 September1998. J Nepal Geol Soc 22:539–546
1222 Nat Hazards (2011) 58:1209–1223
123
Fleming RW, Varnes DJ (1991) Slope movements. The heritage of engineering geology; the first hundredyears. In: Kiersch GA (ed) Boulder, Colorado, Geological Society of America, Centennial (SpecialVolume 3)
Gansser A (1964) Geology of the Himalaya. Interscience Publishers, New York, p 289GEO-SLOPE International Ltd (2010) Stability Modeling with SLOPE/W 2007 version: an engineering
methodology. GEO-SLOPE International Ltd, CalgaryGrabs WE, Hanisch J (1993) Objectives and prevention methods for glacier lake outburst floods (GLOFS).
In: GJ Young (ed) Proceedings of Kathmandu symposium on snow and glacier hydrology. IAHSPublication, p 218
Hambrey M, Alean J (2004) Glaciers, 2nd edn. Cambridge University Press, Cambridge, p 376IPCC-Intergovernmental Panel on Climate Change (2007) Climate change 2007: impacts, adaptation and
vulnerability, in contribution of working group II to the fourth assessment report of the intergovern-mental panel on climate change. Cambridge University Press, Cambridge
Iverson RM (2000) Landslide triggering by rain infiltration. Water Resour Res 36(7):1897–1910Korup O, Tweed F (2007) Ice, moraine, and landslide dams in mountainous terrain. Quaternary Sci Rev
26(25–28):3406–3422Osti R, Egashira S (2009) Hydrodynamic characteristics of the Tam Pokhari Glacial Lake outburst flood in
the Mt. Everest region, Nepal. Hydrol Process 23(20):2943–2955Post A, Mayo LR (1971) Glacier dammed lakes and outburst floods in Alaska. Hydrologic Investigations
Atlas HA-455. U. S. Geological Survey, Washington, D.CProgramme UNEP-UnitedNationsEnvironment (2007) Global outlook for ice and snow. Birkeland Trykkeri
Publications, NorwayReid ME (1994) A pore-pressure diffusion model for estimating landslide inducing rainfall. J Geol 102:
709–717RGSL-Reynolds Geo-Sciences Ltd (2003) Development of glacial hazard and risk minimization protocols in
rural environments—Guidelines for the management of glacial hazards and risks. RGSL, MoldRichardson SD, Reynolds JM (2000) An overview of glacial hazards in the Himalayas. Quat Int 65/66(1):
31–47Sassa K, Nagai O, Solidum R, Yamazaki Y, Ohta H (2010) An integrated model simulating the initiation
and motion of earthquake and rain induced rapid landslides and its application to the 2006 Leytelandslide. Landslides 7(3):219–236
Sponsortrek Nepal (2008) Natural disaster hitting village of Tangnag, 20 km SSE of Mount Everest anddownstream villages along the Hinku Drangkha and Dudh Kosi Rivers. http://p6.hostingprod.com/@treks.org/1998.htm (Viewed 07 January 2010)
Ueno K, Kayasta RB, Chitrakar MR, Bajracharya OR, Pokhrel AP, Fujinami H, Kadota T, Iida H, Man-andhar DP, Hattori M, Yasunari T, Nakawo M (2001) Meteorological observations during 1994–2000at the automatic weather station (GEN-AWS) in Khunbu region, Nepal Himalayas. Bull Glaciol Res18:23–30
USGS-United States Geological Survey (2004) Landslide types and processes. USGS Fact Sheet 2004-3072,July 2004
USGS-United States Geological Survey (2009) Earthquake Lists & Maps. Earthquake hazards program.http://earthquake.usgs.gov/. Accessed 10 September 2009
Vuichard D, Zimmermann M (1987) The 1985 catastrophic drainage of a moraine dammed lake, KhumbuHimal, Nepal: cause and consequences. Mount Res Develop 7:91–110
Wahl TL (1998) Prediction of embankment dam breach parameters: a literature review and needs assess-ment. Dam Safety Research Report DSO-98-004. Water Resources Research Laboratory, Bureau ofReclamation, Dam Safety Office, U.S. Department of the Interior
Watanabe T, Lamsal D, Ives JD (2009) Evaluating the growth characteristics of a glacial lake and its degreeof danger: Imja Glacier, Khumbu Himal, Nepal. Norw J Geog 63:255–267
WWF-World Wildlife Fund (2005) An overview of glaciers, glacier retreat, and subsequent impacts inNepal, India and China. WWF, Kathmandu, p 79
Nat Hazards (2011) 58:1209–1223 1223
123