Draining Himalayan glacial lakes before they burst

Transcription

1 Destructive V/ater: Water-Caused Natural Disasters, their Abatement and Control (Proceedings of the Conference held at Anaheim, California, June 1996). IAHS Publ. no. 239, Draining Himalayan glacial lakes before they burst RICHARD KATTELMANN Sierra Nevada Aquatic Research Lab, StarRoute 1, Box 198, Mammoth Lakes, California 93546, USA TEIJI WATANABE Graduate School of Environmental Earth Science, Hokkaido University, Sapporo 060, Japan Abstract Many catastrophic floods have occurred in the Himalaya when a dam of moraine and/or ice failed suddenly and released massive amounts of water that had been stored in a glacial lake. Such outburst floods are far more destructive than floods generated by rainfall or snowmelt. As a means of reducing the risk of catastrophe from some of these lakes, a simple siphon technique has been proposed to reduce the water level. Although this approach appears to be relatively effective, practical, inexpensive and safe, there is substantial risk of seismic events, ice calving, rapid incision of the outlet channel, and piping through the terminal moraine. Artificial drainage of glacial lakes presents a rare opportunity to actively reduce the potential for a natural disaster caused by destructive water. INTRODUCTION Glacial lakes are a common feature of the Himalaya in general and the Khumbu Himal region of eastern Nepal in particular (Ives, 1986). Some of these glacial lakes are naturally unstable because of the conditions that contain them. They occur where water is impounded by glacial ice and/or moraines. There are many types of these lakes, ranging from meltwater ponds on the surface of glaciers to large lakes in side valleys dammed by a glacier in the main valley. Water trapped behind moraines forms a particularly hazardous type of glacial lake. When glaciers retreat up valley as their lower portions melt away, a topographic depression is often formed by the moraine at the maximum extent of the glacier. If this enclosure is water tight, melt waters will accumulate in the basin until seepage or overflow limits the lake level. Warmer climates of the past 100 to 150 years have resulted in widespread glacial retreat and formation of moraine-dammed lakes in many mountain ranges (Evans & Clague, 1994). Such lakes appear to be the most common type of glacial lakes now found in Nepal (Yamada, 1993). Where the impoundment of these lakes are unstable, there are substantial risks of a sudden release of large quantities of stored water. A common trigger of outburst floods are surge waves generated by snow, ice or rock avalanches into the lake that then overtop the dam and rapidly erode the outflow channel, ultimately collapsing the moraine in the vicinity of the channel. Other failure mechanisms include seepage and piping through the moraine, slow melting of an ice core within the moraine, earthquakes, and progressive thinning of the moraine by landslides. Failure of these moraine or ice dams are very destructive events and have been

2 338 Richard Kattelmann & Teiji Watanabe documented throughout the world (e.g. Hewitt, 1982; Ives, 1986; Vuichard & Zimmermann, 1987). Outburst floods can be several times greater than floods produced by even extreme rainfall (e.g. Dhital et ai, 1993; Meon & Schwarz, 1993). Outburst floods are characterized by extraordinary peak discharges at the onset of the event and entrainment of vast quantities of rock, soil, vegetation and other debris. Observers of glacial lake outburst floods have described literal walls of water and debris m in height. The density and viscosity of such flows is much greater than water alone. The channel is subject to high shear stress, and valley slopes and terraces may be undercut, which can trigger landslides. The valley may be highly unstable for months to years after the initial event. The destructive potential of an outburst flood depends on: the total volume of water released; the magnitude and duration of release, which depend on the dam failure situation; river slope and width and variations in these characteristics along the channel; the amount and size of erodible material in the dam area and downstream; the resistance of the channel and valley sides to erosion and landslides; the location of people at the time of the event; and location of houses, structures, trails and bridges relative to the flood itself and resultant landslides (Yamada, 1993; Braun & Fiener, 1995). Attenuation of the flood wave also influences the extent of damage downstream and is affected by channel conditions, landslide activity, and entrainment of debris. Glacial lake outbursts in the Himalaya have received increasing outside attention as they have threatened more people and property in the past two decades. Damage to infrastructure such as bridges, roads, trails, and hydroelectric facilities has generated much of the current interest in these floods. Feasibility studies for hydroelectric projects must give more thorough consideration to the natural processes operating upstream (Meon & Schwarz, 1993; Kattelmann, 1994). More than a dozen outburst events have been documented in Nepal between 1964 and 1991 (Yamada, 1993). An outburst flood in the Khumbu Himal in 1977 was recognized in the record of a stream gauging station 60 km downstream. This flood started from a series of lakes on and below the Nare Glacier in the Mingbo Valley west of the summit of Ama Dablam (Fushimi et ai, 1985). The river channel and terraces were eroded for tens of kilometres downstream. Another outburst flood in the Khumbu Himal in 1985 received much international attention because it destroyed a nearly completed small hydroelectric project valued at about US$4 million. This flood has been thoroughly researched and described in detail (Vuichard & Zimmermann, 1987; Ives, 1986). Above the village of Thame, a lake called Dig Tsho occupied the moraine-enclosed basin created by the retreat of the Langmoche glacier. About 5 x 10 6 m 3 water was stored in the lake when a large ice avalanche impacted the lake and generated a wave about 5 m high that overtopped the moraine dam (Vuichard & Zimmermann, 1987). The rapid erosion of the dam allowed the lake to drain in 4-6 h with a peak discharge of about 2000 my 1. Surveys and mass balance estimates indicated that about 2.6 x 10 6 m 3 of material were eroded and redeposited within the first 25 km (Vuichard & Zimmermann, 1987). More than two dozen homes and a dozen bridges were destroyed along with several hectares of agricultural fields. Six years after the Dig Tsho outburst, a glacial lake flood occurred in the Rowaling Valley, just 15 km west of Dig Tsho. This outburst of Chubung lake on

3 Draining Himalayan glacial lakes before they burst 339 the Ripimo Shar Glacier in 1991 was relatively small (perhaps x 10 6 m 3 in volume) but served as another reminder of the danger. This flood initiated several landslides, eroded the river channel and damaged a few houses in Beding, the principal village in the Rowaling Valley. Glacial lakes are also recognized as a significant hazard in the eastern Himalaya of Bhutan (Gansser, 1970). The most recent glacial lake outburst flood in the Himalaya occurred in this area in October Small villages close to the outburst source on the Pho Chhu in the Lunana area were flooded and incurred some damage, but more than 20 deaths and serious property damage occurred 86 km downstream in the village of Punakha at 1250 m (Watanabe & Rothacher, 1996). After the Mingbo, Langmoche and Chubung outburst floods, residents of the Ro waling and Khumbu valleys were well aware of the hazard posed by glacial lakes. By 1992, members of the Sherpa community were expressing concern about the largest glacial lakes in both the Rowaling and Khumbu areas. In May 1992, representatives of almost every household in Beding petitioned the embassies of the major donor nations to Nepal to help with the threat from Tsho Rolpa, the giant glacial lake just 10 km above Beding. The Sherpas of Rowaling and Summit Trekking Nepal in Kathmandu have been active in publicizing the hazard. In 1993, a nongovernmental organization was formed by Sherpa people in Solu Khumbu and Kathmandu to publicize the hazard about the lake on Imja Glacier, which threatens the Sherpa heartland, and solicit ideas, technical assistance, and financial support to address the problem. This "Save the Imja Valley Committee" is following the progress of the efforts in Rowaling and is trying to identify appropriate measures for Imja Glacier Lake. The committee is actively seeking technical and financial assistance. This paper attempts to describe the problem to a broader audience of scientists and engineers in hopes of generating creative approaches to this hazard. DANGEROUS GLACIAL LAKES OF NEPAL The Water and Energy Commission Secretariat of Nepal has identified five hazardous glacial lakes to monitor for signs of increasing risk of an outburst (Mool, 1993). Two of these lakes are discussed in this paper, but several lakes in Nepal require attention. The giant Tsho Rolpa in the Rowaling valley, west of Khumbu, has been recognized as a very dangerous lake for several years (Damen, 1992; Mool, 1993; Yamada et al., 1996). This lake has increased in volume about 40 times in two decades and now contains almost 100 x 10 6 m 3 of water (Damen, 1992). The development of the lake on the Imja Glacier has been well described by Yamada (1993), Yamada & Sharma (1993) and Watanabe et al. (1994, 1995). Photographs and personal accounts indicate that there was no evidence of a lake on Imja Glacier before the mid 1960s, if not later (Yamada & Sharma, 1993; Watanabe et al., 1994). A survey of the lake in April 1992 including depth measurements determined it was about 0.6 km 2 in area, had an average depth of almost 50 m, and contained about 28 x 10 6 m 3 of water (Yamada, 1993; Yamada & Sharma, 1993). A large volume of dead ice mantled with debris downstream from the lake formed the principal dam, which is about m across. Surveys of this spillway in 1989 and 1994 show

4 340 Richard Kattelmann & Teiji Watanabe that the ice is melting rapidly and the morphology of this part of the glacier is highly dynamic (Watanabe et al, 1994; Watanabe et al., 1995). Some seepage was noted exiting from the southwest corner of the terminal moraine in December The long extent of the dead ice and terminal moraine could lessen the potential for a very sudden release of most of the water in the lake. Presumably, several hours would be required for deep incision of hundreds of meters of ice and moraine. Detailed surveys of the micro-topography of terminal moraine are needed for better assessment of potential for rapid regressive erosion. If these lakes were located in remote, unpopulated valleys, an outburst would constitute relatively little hazard. However, the Imja valley and Dudh Kosi valley downstream are perhaps the most heavily populated and travelled areas of comparable elevation in Nepal. The Sherpa people have lived in this region for hundreds of years and established villages and agriculture wherever the topography would permit (Stevens, 1993). Their lives and livelihoods are at risk from these floods. REDUCING THE POTENTIAL FOR OUTBURST FLOODS The basic approaches to reducing the hazard of glacial lake outburst floods include getting out of the way while letting the natural event take its course, strengthening the dam and providing a reinforced outlet control structure, and artificially reducing the water level of the lake. The last course has been tried in a few instances with varying degrees of success. Where the geology, access and technology permit, tunnels have been built through solid rock into glacial lakes in Norway and Switzerland. Intentional breaching of a dam in Peru failed with catastrophic consequences when newly unsupported ice gave way and created a large surge wave (Lliboutry et al, 1977). Creation of an artificial spillway allows water to quickly drain from the lake. However, there is substantial risk of rapid erosion of the spillway and uncontrolled release of water (Grabs & Hanisch, 1993). Explosive excavation of an outflow channel intended to gradually incise upto a 20 m depth, as lake water was released successfully, emptied 7 x 10 6 m 3 of water in 2 days at Bogatyr Lake in the Zailyskiy Alatau of now Kazakhstan (Nurkadilov et al, 1986). Other techniques are reviewed by Mochalov & Stephanov (1986), Damen (1992), and Grabs & Hanisch (1993). Siphons carry the least risk of inducing catastrophic failure and are appropriate for remote installations. Theoretically, lake levels could be lowered about 5 m with a simple siphon under Himalayan conditions (Grabs & Hanisch, 1993). Appropriate technology for such a siphon system could consist of a series of independent pipes that are small enough for individual sections to transported by porter (e.g. 6 m lengths of 15 cm diameter plastic pipe). A design with valves on both ends and the high point along with a portable pump for filling has been proposed by Grabs & Hanisch (1993). The number of siphons in operation would be determined by the desired rate of lowering, which must account for the rate of inflow during the summer monsoon and melt season. Excessively rapid lowering of the water level could induce failure of the lateral moraines or calving of any floating glacial ice at the upstream end of the lake (Grabs & Hanisch, 1993). Collapse of large volumes of moraine or ice into the lake could generate a surge wave leading to erosion of the

5 Draining Himalayan glacial lakes before they burst 341 impoundment. Siphons have been successfully installed in a lake in the Peruvian Andes near Huaraz where two siphons with a combined capacity of 0.5 my 1 lowered the water level by almost 1 m per month (Reynolds, 1992, 1994). However, this project was prematurely halted by political strife. Investigation of the potential impacts of an outburst and options for reducing the hazard at Tsho Rolpa in the Rowaling Valley (Damen, 1992) led to the involvement of the Wavin company of The Netherlands that donated 500 m of 16 cm diameter plastic pipe for an experimental siphon system. In July 1995, the first siphon was installed with a capacity of 0.17 m 3 s" 1. An outflow of about 5 m 3 s" 1 would be necessary to lower the level by 3 m in one year, without considering additions from rainfall or ice melt (Damen, 1992). The use of siphons at Imja Glacier Lake has been endorsed by the Save the Imja Valley Committee, and funding is being sought to begin installation. A principal difficulty with the physical situation at Imja Glacier is the long expanse of dead ice between the lake and valley floor beyond the terminus. A continuous siphon would need to be about m long. However, separate siphons between pools in the outlet channel may offer some efficiencies. As the volume of water siphoned from the lake increases, the outlet channel may need to be protected against accelerated erosion. Reinforced plastic mats (i.e. hypalon), wire rope or jute matting may prove to be useful in maintaining a channel. Engineering of an actual drainage system will require careful surveys of the outlet channel and the flexibility to adapt to the dynamic nature of the glacier. Another possible approach to lowering the lake level might be cutting sections of ice in the winter and hauling them over the moraine at a point where it had been excavated manually and with explosives. Where the valley floor beyond the glacier is higher than the lake level, as is the case in the upper part of Imja Glacier Lake, the moraine can be safely removed. As the water level is lowered, the spillway must also be excavated. Use of high pressure water jets might have some application in such excavations where an upslope water source is accessible. Where there risk of large waves from glacier calving as the lake level drops, controlled break-up of smaller pieces of the floating glacier with explosives could be investigated. Obviously, any interventions would be very difficult and expensive. However, the investment in making the downstream valleys safe for habitation and travel should be approached in the manner of a large construction project such as a road or hydroelectric facility. Any operations must be viewed as potentially risky. The possibility of dam failure during lake-lowering activities by initiation of rapid channel incision, glacier calving, mass failure of oversteepened lateral moraines, piping or earthquakes must be considered. The risk of initiating a failure by action must be weighed against the risk of eventual natural failure. If an engineering approach to lowering the water level proves to be infeasible, relocation of houses, businesses, bridges and trails might be contemplated. Lake levels need only be lowered to a point where dam failure is unlikely or where such failure would not cause catastrophic damage downstream. Once the dam is lowered to the level of a reasonable volume of water and a relatively stable spillway exists, then the prospects for a severe flood have been minimized because possibility of refilling the lake has been eliminated. As an initial step, a radio warning system should be installed as soon as possible. Flood warning systems are a potentially valuable interim means of reducing hazards

6 342 Richard Kattelmann & Teiji Watanabe to people. Any warning system must be carefully designed to ensure proper operation at the critical moment and include significant redundancy to avoid failure. Radio communications failed to provide adequate warning to downstream residents during the 1994 outburst flood in Bhutan because the receiving stations were not manned during the night of the flood (Watanabe & Rothacher, 1996). CONCLUSIONS Controlled drainage of dangerous glacial lakes presents a rare opportunity to actively reduce a natural hazard instead of just respond to the eventual damage. Engineering studies should be accelerated to enable operational lake lowering to begin as soon as possible. Warning systems using observers and reliable radio communications are necessary until the risk of an outburst is reduced. We hope the international aid and development community will act on this opportunity at Tsho Rolpa and Imja Glacier Lake to avoid impending tragedies. Any ideas and suggestions regarding technical, logistical, and financial support are most welcome. Acknowledgements We thank W. E. Grabs for informing us about the potential utility of siphons as a means of hazard reduction. M. Damen, T. Yamada, P. K. Mool and T. Kadota provided critical information about Tsho Rolpa and Imja Glacier Lake. The Save the Imja Valley Committee can be reached in care of J. P. Lama Sherpa, GFAS Pvt. Ltd., PO Box 3776, Kathmandu, Nepal. REFERENCES Braun, M. & Fiener, P. (1995) Report on the GLOF hazard mapping project in the Imja Kliola / Dudh Kosi valley, Nepal. Snow and Glacier Hydrology Project, Nepal Department of Hydrology and Meteorology and German Technical Cooperation, Kathmandu, Nepal. Damen, M. (1992) Study of the Potential Outburst Flooding of Tsho Rolpa Glacier Lake, Rolwaling Valley, East Nepal. International Institute for Aerospace Survey and Earth Sciences, Enschede, The Netherlands. Dhital, M. R., Khamel, N. & Thapa, K. B. (1993) The Role of Extreme Weather Events, Mass Movements, and Land Use Change in Increasing Natural Hazards. International Centre for Integrated Mountain Development, Kathmandu, Nepal. Evans, S. G. & Clague, J. J. (1994) Recent climate change and catastrophic geomorphic processes in mountain environments. Geomorphol. 10, Fushimi, H., Ikegami, K., Higuchi, K. & Shankar, K. (1985) Nepal case study: catastrophic floods. In: Techniques for Prediction of Runoff from Glacierized Areas (ed. by G. J. Young), IAHS Publ. no Gansser, A. (1970) Lunana: the peaks, glaciers and lakes of northern Bhutan. The Mountain World 1968/1969. Swiss Foundation for Alpine Research, Zurich, Switzerland. Grabs, W. E. & Hanisch, J. (1993) Objectives and prevention methods for glacier lake outburst floods (GLOFS). In: Snow and Glacier Hydrology (ed. by G. J. Young) (Proc. Kathmandu Symp., November 1992), IAHS Publ. no Hewitt, K. J. (1982) Natural dams and outburst floods of the Karakorum Himalaya. In: Hydrological Aspects of Alpine and High-Mountain Areas (ed. by J. W. Glen) (Proc. Exeter Symp., July 1982), , IAHS Publ. no Ives, J. D. (1986) Glacier lake outburst floods and risk engineering in the Himalayas. Occasional Pap. no. 5. International Centre for Integrated Mountain Development, Kathmandu. Kattelmann, R. (1994) Improving the knowledge base for Himalayan water development. Water Nepal 4(1), Lliboutry, L., Morales, B. M., Pantre, A. & Schneider, B. (1977) Glaciological problems set by the control of dangerous lakes in Cordillera Blanca, Peru. Part I: Historical failures of morainic dams, their causes and prevention. J. Glaciol. 18(79), Meon, G. & Schwarz, W. (1993) Estimation of a glacier lake outburst flood and its impact on a hydro project in Nepal.

7 Draining Himalayan glacial lakes before they burst 343 In: Snow and Glacier Hydrology (ed. by G. J. Young) (Proc. Kathmandu Symp., November 1992), IAHS Publ. no Mochalov, V. P. & Stepanov, B. S. (1986) Glacial floods and methods of their control. Data of Glaciological Studies 58, Mool, P. K. (1993) Glacier lake outburst floods in Nepal. Water and Energy Commission Secretariat Bulletin 4(1), Nurkadilov, L. K., Khegai, A. Y. & Popov, N. V. (1986) Artificial draining of an outburst-dangerous lake at the foot of surging glacier. Data of Glaciological Studies 58, Reynolds, J. (1992) Geological hazards in the Cordillera Blanca, Peru. AGID News 61/62, Reynolds, J. (1994) The identification and mitigation of glacier-related hazards: examples from the Cordillera Blanca, Peru. In: Geohazards: Natural and Man-made (ed. by G. J. H. Hall, D. J. C. Laming & S. C. Scott), Chapman & Hall, London. Stevens, S. F. (1993) Claiming the High Ground: Sherpas, Subsistence, and Environmental Change in the Highest Himalaya. University of California Press, Berkeley, California, USA. Vuichard, D. & Zimmermann, M. (1987) The 1985 catastrophic drainage of a moraine-dammed lake, Khumbu Himal, Nepal: causes and consequences. Mountain Res. & Devel. 7(2), Watanabe, T., Ives, J. D. & Hammond, J. E. (1994) Rapid growth of a glacial lake in Khumbu Himal, Himalaya: prospects for a catastrophic flood. Mountain Res. & Devel. 14(4), Watanabe, T., Kameyama, S. & Sato, T. (1995) Imja glacier dead ice melt rates and changes in a supraglacial lake, , Khumbu Himal, Nepal: danger of lake drainage. Mountain Res. & Devel. 15(4), Watanabe, T. & Rothacher, D. (1996) The 1994 Lugge Tsho glacial lake outburst flood, Bhutan Himalaya. Mountain Res. & Devel. 16(1), Yamada, T. (1993) Glacier Lakes and their Outburst Floods in the Nepal Himalaya. Water and Energy Commission Secretariat, Kathmandu, Nepal. Yamada, T., Kadota, T., Kita, T., Mool, P. K., Markey, P. R. & and Joshi, S. P. (1996). Physiographic characteristics and hydrological conditions of Tsho Rolpa glacier lake, Rowaling Valley, Nepal Himalaya. Paper presented at the International Conference on Ecohydrology of High-Mountain Areas (Kathmandu, Nepal, March 1996). Yamada, T. & Sharma, C. K. (1993) Glacier lakes and outburst floods in the Nepal Himalaya. In: Snow and Glacier Hydrology (ed. by G. J. Young) (Proc. Kathmandu Symp., November 1992), IAHS Publ. no. 218.