Introduction

Periodic or occasional release of large amounts of stored water in a catastrophic outburst flood is widely referred to as a jokulhlaup (Iceland), a debacle (French), an aluvión (South America), or a Glacial Lake Outburst Flood (GLOF) (Himalaya). A jokulhlaup is an outburst which may be associated with volcanic activity, a debacle is an outburst but from a proglacial lake, an aluvión is a catastrophic flood of liquid mud, irrespective of its cause, generally transporting large boulders, and a GLOF is a catastrophic discharge of water under pressure from a glacier. GLOF events are severe geomorphological hazards and their floodwaters can wreak havoc on all human structures located on their path. Much of the damage created during GLOF events is associated with the large amounts of debris that accompany the floodwaters. Damage to settlements and farmland can take place at very great distances from the outburst source, for example in Pakistan, damage occurred 1,300 km from the outburst source (WECS 1987b).
 

Causes of Lake Creation

Global warming

There is growing concern that human activities may change the climate of the globe. Past and continuing emissions of carbon dioxide (CO2) and other gases will cause the temperature of the Earth’s surface to increase—this is popularly termed ‘global warming’ or the ‘greenhouse effect’. The ‘greenhouse effect’ gives an extra temperature rise.
 

Glacier retreat

An important factor in the formation of glacial lakes is the rising global temperature (‘greenhouse effect’), which causes glacial retreat in many mountain regions.

During the so-called ‘Little Ice Age’ (AD 1550–1850), many glaciers were longer than today. Moraines formed in front of the glaciers at that time nowadays block the lakes. Glaciation and interglaciation are natural processes that have occurred several times during the last 10,000 years.

As a general rule, it can be said that glaciers in the Himalayas have retreated about 1 km since the Little Ice Age, a situation that provides a large space for retaining melt water, leading to the formation of moraine-dammed lakes (LIGG/WECS/NEA 1988).
 

Röthlisberger and Geyh (1985) conclude in their study on ‘glacier variations in Himalaya and Karakorum’ that a rapid retreat of nearly all glaciers with small oscillation was found in the period from 1860/1900–1980.
 

Causes of glacial lake water level rise

The causes of rise in water level in the glacial lake dammed by moraines that endanger the lake to reach breaching point are given below.

  • Rapid change in climatic conditions that increase solar radiation causing rapid melting of glacier ice and snow with or without the retreat of the glacier.

  • Intensive precipitation events

  • Decrease in sufficient seepage across the moraine to balance the inflow because of sedimentation of silt from the glacier runoff, enhanced by the dust flow into the lake.

  • Blocking of ice conduits by sedimentation or by enhanced plastic ice flow in the case of a glacial advance.

  • Thick layer of glacial ice (dead ice) weighed down by sediment below the lake bottom which stops subsurface infiltration or seepage from the lake bottom.

  • Shrinking of the glacier tongue higher up, causing melt water that previously left the glacier somewhere outside the moraine, where it may have continued underground through talus, not to follow the path of the glacier.

  • Blocking of an outlet by an advancing tributary glacier.

  • Landslide at the inner part of the moraine wall, or from slopes above the lake level

  • Melting of ice from an ice-core moraine wall.

  • Melting of ice due to subterranean thermal activities (volcanogenic, tectonic).

  • Inter-basin sub-surface flow of water from one lake to another due to height difference and availability of flow path.

Bursting Mechanisms

Different triggering mechanisms of GLOF events depend on the nature of the damming materials, the position of the lake, the volume of the water, the nature and position of the associated mother glacier, physical and topographical conditions, and other physical conditions of the surroundings.
 

Mechanism of ice core-dammed lake failure

Ice-core dammed (glacier-dammed) lakes drain mainly in two ways.

  • through or underneath the ice

  • over the ice

Initiation of opening within or under the ice dam (glacier) occurs in six ways.

  • Flotation of the ice dam (a lake can only be drained sub-glacially if it can lift the damming ice barrier sufficiently for the water to find its way underneath).

  • Pressure deformation (plastic yielding of the ice dam due to a hydrostatic pressure difference between the lake water and the adjacent less dense ice of the dam; outward progression of cracks or crevasses under shear stress due to a combination of glacier flow and high hydrostatic pressure).

  • Melting of a tunnel through or under the ice

  • Drainage associated with tectonic activity

  • Water overflowing the ice dam generally along the lower margin

  • Sub-glacial melting by volcanic heat

The bursting mechanism for ice core-dammed lakes can be highly complex and involve most or some of the above-stated hypothesis. Marcus (1960) considered ice core-dammed bursting as a set of interdependent processes rather than one hypothesis.

A landslide adjacent to the lake and/or subsequent partial abrasion on ice may lead to overtopping as the water flows over, the glacier retreats, and the lake fills rapidly, which may subsequently result in the draining of ice core moraine-dammed lakes.
 

Mechanisms of moraine-dammed lake failure

Moraine-dammed lakes are generally drained by rapid incision of the sediment barrier by outpouring waters. Once incision begins, the hustling water flowing through the outlet can accelerate erosion and enlargement of the outlet, setting off a catastrophic positive feedback process resulting in the rapid release of huge amounts of sediment-laden water (Figure 9.1). The onset of rapid incision of the barrier can be triggered by waves generated by glacier calving or ice avalanching, or by an increase in water level associated with glacial advance (examples include an ice avalanche from Langmoche Glacier on 4 August 1985 and another on 3 September 1998 from Sabai Glacier).

Dam failure can occur for the following reasons:

  • melting ice core within the moraine dam,

  • rock and/or ice avalanche into a dammed lake,

  • settlement and/or piping within the moraine dam,

  • sub-glacial drainage, and

  • engineering works.


Figure 9.1: Peak discharge from breached moraine-dammed lakes can
be estimated from an empirical relationship developed by Costa (1985)

Melting ice-core

The melting of impervious ice core within a moraine dam may result in the lowering of the effective height of the dam, thus allowing lake water to drain over the residual ice core. As the discharge increases with the melting of the ice core, greater amounts of water filter through the moraine, carrying fine materials. Eventually, the resulting regressive erosion of the moraine dam leads to its ultimate failure.
 

Overtopping by displacement waves

Lake water is displaced by the sudden influx of rock and/or ice avalanche debris. The resultant waves overtop the freeboard of the dam causing regressive and eventual failure.

 

Settlement and/or piping

Earthquake shocks can cause settlement of the moraine. This reduces the dam freeboard to a point that the lake water drains over the moraine and causes regressive erosion and eventual failure.

 

Sub-glacial drainage

A receding glacier with a terminus grounded within a proglacial lake can have its volume reduced without its ice front receding up-valley. When the volume of melt water within the lake increases to a point that the formerly grounded glacier floats, an instantaneous sub-glacial drainage occurs. Such drainage can destroy any moraine dam, allowing the lake to discharge until the glacier loses its buoyancy and grounds again.

 

Engineering works

One of the main difficulties in changing water levels or dam structures artificially is that this can unintentionally trigger a catastrophic discharge event. For example, in Peru in 1953, during the artificial lowering of the water level, an earth slide caused 12m high displacement waves, which poured into a trench, excavated as part of the engineering works and almost led to the total failure of the moraine dam.

 

Surge Propagation

As GLOFs pose severe threats to humans and man-made structures, it is important to make accurate estimates of the likely magnitude of future floods. Several methods have been devised to predict peak discharges, which are the most erosive and destructive phases of floods. The surge propagation hydrograph depends upon the type of GLOF event, i.e. from moraine-dammed lake or from ice-dammed lake (Figure 9.2). The duration of a surge wave from an ice-dammed lake may last for days to even weeks, while from a moraine-dammed lake the duration is shorter, minutes to hours. The peak discharge from the moraine-dammed lake is usually higher than those from ice-dammed lakes.

Figure 9.2: Difference in release hydrograph between
moraine- and ice-dammed lakes (WECS 1987A)
 
 

The following methods have been proposed for estimation of peak discharges.

1) Clague and Mathews formula

Clague and Mathews (1973) were the first to show the relationship between the volume of water released from ice-dammed lakes and peak flood discharges.

Qmax = 75(V0*10–6)0.67

where

Qmax = peak flood discharge (m3 s-1)

V0 = total volume of water drained out from lake (m3)

The above relationship was later modified by Costa (1988) as the peak discharge yielded from the equation was higher than that measured for Flood Lake in British Columbia that occurred in August 1979:

Qmax = 113(V0*10–6)0.64

Later Desloges et al. (1989) proposed:

Qmax=17V0*19(0–6)0.64

This method of discharge prediction is not based on any physical mechanism, but seems to give reasonable results.

2) Mean versus maximum discharge method

If the volume of water released by a flood and the flood duration are known, the mean and peak discharges can be calculated. Generally the flood duration will not be known in advance. Hence, this method cannot be used to determine the magnitude of future floods. Observations of several outburst floods in North America, Iceland, and Scandinavia have shown that peak discharges are between two to six times higher than the mean discharge for the whole event.

3) Slope area method

This method is based on measured physical parameters such as dimensions and slope of channel during peak flood conditions from direct observations or geomorphological evidence.

Qmax = vA

The peak velocity is calculated by the Gauckler–Manning formula (Williams 1988)

v = r 0.67 S 0.50/n

where

v = peak velocity

S = bed slope for a 100m channel reach

n = Manning’s roughness coefficient

r = hydraulic radius of the channel

r = A/p

where

A = cross-sectional area of the channel

p = perimeter of the channel under water

For sediment floored channels, bed roughness is mainly a function of bed material, particle size, and bed form or shape and can be estimated from:

n = 0.038D 0.167

where

D = average intermediate axis of the largest particles on the channel floor.

Desloges et al. (1989) compared the results from all the three methods for a jokulhlaup from the ice-dammed Ape Lake, British Columbia. All the methods gave comparable results.

  • The Clague and Mathews method gave a calculated peak discharge of 1,680 ± 380 m3 s-1.

  • The mean versus maximum discharge method gave 1,080–3,240 m3 s–1.

  • The slope area method gave 1,534 and 1,155 m3 s–1 at a distance of 1 and 12 km from the outlet respectively.

These general relationships are useful for determining the order of magnitude of initial release that may propagate down the system. However, to predict the magnitude of future floods, the first method should be applied, because volume of lake water can be estimated in advance.

 

Attenuation of a peak discharge of 15,000–20,000m3 s–1 has been reported for the Sun Koshi River in Tibet within a distance of 50 km (XuDaoming 1985) (Figure 9.11). The propagation of surge waves can be numerically modelled using the dam-break flood-forecasting model.
 

Sediment Processes During a Glacial Lake Outburst Flood

During a GLOF, the flow velocity and discharge are exceptionally high and it becomes practically impossible to carry out any measurement. Field observations after a GLOF event have shown a much higher sediment concentration of rivers than before the GLOF event (Electrowatt Engineering Service Ltd 1982; WECS 1995a). WECS (1995a) calculated the volume of scoured sediment as 22.5*104 m3 after the Chubung GLOF in 1991. A high concentration of 350,000 mg–1 during a GLOF in the Indus River at Darband in 1962 is reported by Hewitt (1985). Hypothetical illustrations showing discharge and variation in sediment concentration (WECS 1987a) are shown in Figure 9.3.
 

The total sediment load is generally accepted as the wash load, which moves through a river system and finally deposits in deltas. In Nepal, no measurements have been taken of total sediment during GLOF events, however, rough estimates of total load during torrents can be made assuming a high sediment concentration (WECS 1987b). During a GLOF event, stones the size of small houses can be easily moved (WECS 1987b). The relationship between flow velocity and particle diameter can also be used to calculate the size of boulders that can be moved during such events.
 

Socio-economic Effects of Glacial Lake Outburst Floods

The impact of a GLOF event downstream is quite extensive in terms of damage to roads, bridges, trekking trials, villages, and agricultural lands as well as the loss of human lives and other infrastructures. The sociological impacts can be direct when human lives are lost, or indirect when the agricultural lands are converted to debris filled lands and the village has to be shifted. The records of past GLOF events show that once every three to ten years, a GLOF has occurred in Nepal with varying degrees of socioeconomic impact. Therefore, proper hazard assessment studies must be carried out in potentially problematic basins to evaluate the likely economic loss and the most appropriate method of mitigation activities.
 

The 1981 GLOF from Zhangzangbo in Tibet (China) brought a lot of destruction in Tibet (China) and Nepal. It even caused severe damage to sections of the Nepal–China Highway including the Phulping and Friendship bridges in Nepal. The road was rebuilt at a cost of US $3 million. The present road level is now above the historic 1981 GLOF.
 

The 1985 GLOF from Dig Tsho in the Dudh Koshi Basin damaged Namche hydropower station (US $1.5 million), 14 bridges, cultivated lands etc. (Vuichard and Zimmerman 1987). The hydropower plant has been rebuilt at another site. The sociological cost of lost lives and dwellings to communities was enormous. The study shows that this glacial lake is refilling again and possibly engineering a greater risk of a GLOF occurrence in the same basin. This and many more GLOF events indicate that before any major project is undertaken in the basin, in-depth cost and benefit analyses have to be carried out for deciding on the most appropriate alternative that will enable project financiers to assess their risks from a GLOF. The assessment of tangible benefits in respect to mitigation of GLOFs is, however, difficult. Reduced damage is considered a benefit and can be quantified, but the frequency of the reduced damage is difficult to ascertain due to lack of data. One cannot simply predict the timing and occurrences of GLOFs. It is extremely difficult to simulate numerically the flood level and velocities at a particular place.
 

At this stage, from brief studies of GLOFs throughout the world, it appears that there are no simple direct means of estimating the recurrence of GLOFs.
 

Brief Review of Glacial Lake Outburst Flood Events and Damage Caused

The reported GLOF events are given in Table 9.1. In 1964, the Gelhaipuco GLOF was experienced along the Arun Valley. Severe damage and heavy economic losses occurred in Chinese Territory. And, the Ayaco Lake had GLOF every year from 1968 to 1970.

 

Table 9.1: List of GLOF events that have occurred in Pumqu basin, affecting Nepal

No.

Index

Date

River basin

Lake

Source

Cause of GLOF

Losses

1

1

21 September 1964

Arun

Gelhaipco

Natangqu sub-basin

Glacier surge

Highway and 12 trucks

2

2

1968

Arun

Ayaco

Zongboxan river

Not known

Road, bridges etc

3

2

1969

Arun

Ayaco

Zongboxan river

Not known

Not known

4

2

1970

Arun

Ayaco

Zongboxan river

Not known

Not known

5

3

27 August 1982

Arun

Jinco

Yarozangbo river

Glacier surge

Livestock, farmland


Gelhaipuco

Gelhaipuco is an end moraine-dammed lake located in the headwaters of the Gelhaipu Gully (Natangqu River Basin), east of Riwo, Dinggye County, Tibet (China). Its geographic position is latitude 27º 58’ N and longitude 87º 49’ E. The lake burst abruptly due to an ice avalanche at 2 pm, on 21 September 1964. According to an investigation by Chengdu Institute of Geography of the Chinese Academy of Sciences, from the middle of March to the end of September 1964, there was a large precipitation in the Natangqu River Basin, which caused the glacier of the Natangqu River to slide (LIGG/WECS/NEA 1988). Huge amounts of ice slid into the lake resulting in the generation of a shock wave and water level increase. Finally, the lake water overflowed through the moraine dam and breached the 30m steep valley through the dam.
 

The flood, with a huge amount of debris, damaged Chentang-Riwo Highway and 12 trucks transporting timber were washed away. The debris flow rushed down to the lower reaches of the Arun (Pumqu) River of Nepal and caused heavy economic losses. Based on flood trace marks and sediment deposits on the river-bed, it was concluded that it was a turbulent debris flow with a bulk density of about 1.45 t m–3.
 

Before the burst, Gelhaipuco Lake was 1.4 km in length and 0.548 sq.km in area with water reserves of about 25.45 million m3. The water level of the lake dropped by 40m after the lake burst in 1964 and released about 23.36 million m3 of water. The slope of the exposed lake bed is 0.6% and it is 0.2 km away from the glacier margin. The present condition of the lake indicates stability. But if the glacier advances forward again, the possibility of another burst cannot be ruled out.
 

The LANDSAT TM and field photographs of Gelhaipuco Lake are given in Figures 9.3, 9.4, and 9.5 respectively.
 

Figure 9.3: LANDSAT TM of 22 September 1988
(the Gelhaipuco Glacial Lake area is shown in the circle)
 
 

Figure 9.4: The field photograph (1987) of Gelhaipuco Glacial Lake
shows the lake in contact with the hanging glacier
 
 


Figure 9.5: The eroded banks of the Natangchu
(tributary of the Pumqu River in Tibet [China])
after the Gelhaipuco GLOF in 1964 (photograph 1987)


 
Ayaco

Ayaco is located at the headwaters of the Zongboxan River in the Pumqu Basin (Tibet) on the northwestern slope of Mount Everest. The geographic position of the lake is latitude 28º 21’N and longitude 86º 29’E. According to an investigation by Chengdu Institute of Geography of the Chinese Academy of Sciences, there were three burst events recorded in mid August 1968, 1969, and 1970 (LIGG/WECS/NEA 1988). A huge fan-shaped mass of debris was deposited at the confluence of the lake drainage channel and the main river course. The estimated sediment deposit is about 4.59 million m3. At present the lake is only 1.2 km long and 0.35 sq.km in area, which is much smaller than its size before the burst. The distance from the glacier to the lake is 0.5 km. If the glacier advances again, there is the possibility of another burst, but the intensity may not be as strong as during the period from 1968–1970. The flood damaged the highway and concrete bridges of Desha No.1 in Tibet (China). The damage on the Nepal side is unknown.
 

Jinco

Jinco Lake is located at the headwaters of the Yairuzangbo River of the Pumqu Basin (Tibet) and the Arun Basin in Nepal. It is an end moraine-dammed lake. The Jinco GLOF happened at 5 pm on August 27 1982 and formed a huge amount of debris flow. At 7 pm the flood peak arrived at Sar. The summer of 1982 was dry and hot. The outburst might have been the result of a strong glacier ablation that seeped melting water into the glacier bed and made it slide. The ice blocks collapsed into the lake and the generated shock wave damaged the dam, thus causing the burst.

Over 1,600 livestock were lost, about 19 hectares of cultivated field were destroyed, and the houses of eight villages were washed away. Gujing village suffered a different degree of destruction.
 

Zhangzangbo 
At mid-night, July 11, 1981, an end moraine-dammed lake located at the headwaters of Zhangzangbo Gully burst suddenly in Poiqu basin. A breach 50 m deep and 40--60m of bottom width was formed at the Little Ice Age moraine dam. The flood formed a large alluvial fan. According to Xudaoming , the largest burst discharge was about 1600 m³/s, which happened 23 minutes after the burst. The main flood lasted about 60 minutes and the burst water amount was estimated at 19 million m³. Along the valley where the debris flow passed through, erosion and sedimentation can be seen; about 4 million m³ of mixed materials joined the debris flow process. This debris flow damaged the highway between the outlet of Zhangzangbo Gully and the Sun Koshi Power Station in Nepal. It destroyed the Friendship Bridge of the China-Nepal Highway and the intake dam of Sun Koshi Hydroelectric Station, causing serious economic losses to Nepal. According to the investigation of Xudaoming et al. in 1984, before the burst, the end moraine-dammed lake was 1.7 km long and 0.643 km² in area; after the burst the length and area were reduced to 1.1 km and 0.265 km² , respectively. The water reserves of the lake were also greatly reduced.

According to the investigation of 1984, there had been a burst in 1964 from that same lake, but the breach was different from that in 1981 and the scope of the debris flow, burst discharge and the damage caused, was smaller. There is a cirque-hanging glacier in Zhangzangbo Gully (Glacier No. 91 (LIGG/NEA/WECS 1988)), whose area is 2.47 km² , length 2.2 km and it ends at the bank of the lake. From the 5th to the 10th of July, 1981, there was continuous hot weather. The increased glacier ablation produced a large amount of water seeping into the crevasse of the glacier tongue, which brought the glacier into a critical state and caused part of the glacier to slide. Huge amounts of ice collapsed into the lake which generated the shock wave that caused the dam burst.

Taraco 
Taraco lake is located in the Targyailing Gully at 28º17´ N, 86 º09´ E in Poiqu basin. Now it is a moraine dammed lake, 1.0 km in length, 0.224 km² in area. According to local old residents descriptions, in a night of August, 1935, the lake burst abruptly. It happened during the wheat harvest season. Nearly 100 mu of wheat field at the outlet of the gully were destroyed, and several heads of yaks were lost. A large fan-shaped area of debris flow outside the Targyailing Gully was formed at that burst. Now it is a vast expanse of stone and can not be cultivated. According to the description of local old men, there was water oozing from beneath the dam before that burst. It can be concluded that the burst was probably caused by part of the dam collapsing because of seepage. There is a cirque-hanging glacier (No. 74 (LIGG/NEA/WECS 1988)) behind the lake, with an area of 2.46 km² and 1.5 km in length. Now the terminus of the glacier is 0.3 km away from the lake. If the glacier moves forward again, it is still possible to have another burst, but the scale and damage degree would not be as high as in 1935.

Yadong 
At the boundary headwater between upper Kangma District of Yadong in Xizang (Tibet), and Sikkim there is an end moraine dammed lake named Qubixiama. Its length is 450 m, width over one hundred meters, water level 4560 m a.s.l., and the end moraine dam is 70 m higher than the riverbed. In rhe night of June 10th, 1940, the glacier behind the lake collapsed abruptly into the lake. The shock wave generated broke up the dam and induced a burst. Estimated from the water trace mark left from the burst, the discharge was as high as 3690 m3 s, and the flow velocity 7.7 m/ s. According to the witness, the burst flood subsided half hour later. This glacier lake burst caused the water level of Xiasima (where the County is sited), Yadong, to rise to 4 5m; the street was flooded and some buildings were damaged.

Jiangzi 
There is an end moraine dammed glacier lake named Sangwang at the headwaters of the Nyabgqu River at the boundary of Bhutan. Behind the lake there are two cirque – hanging glaciers. After several days of hot weather in mid July 1954, the hanging glacial tongue slided into the lake abruptly in the night of July 16. The strong shock waves generated and the huge amount of ice blocks burst the dam. Jiangzi flood as high as 4 meters surged down; a large amount of gravel carried by the flood formed deposit layers of up to 3-5 m thick on the valley plain. This sever flood and debris flow seriously damaged towns and cultivated fields.

Tanbulang debris flow of 1964 
The Tangbulang Gully is a tributary in the east side of the Niyang River, 15 km long, trending in the northeast direction. The Sichuan –Xizang Highway is located at the outlet of the valley. There are eight cirque glaciers and cirque – hanging glaciers at the headwaters of the valley; below the glacier there are over ten small moraine – dammed lakes. Among these lakes the largest is named Damenhaico. Behind this lake there is a cirque glacier with a length of 2.5 km., which terminates at the bank of the lake. During the Tanbulang debris flow of 1964, the water level rose over ten meters, and the burst discharge reached 2000 m3/s. The resulting huge debris flow flooded villages, and caused blockage of the Niyang River for 16 hours, and the highway traffic was closed for one week. This event took place at a location east of Lhasa.

Longda
In 1964, Longda glacier lake burst. The outburst flood washed out a huge amount of sediment which created a debris blockage 800 m long along the river, 200 m wide and 5 m deep on average on the Gyirongzangbo River, the source of the Trisuli River.