Hydrological Aspects of Alpine and High Mountain Areas (Proceedings of the Exeter Symposium, Juiy 1982). IAHS Publ. no. 138.

Natural dams and outburst floods of the Himalaya

KENNETH HEWITT Wilfrid Laurier University, Waterloo, Ontario Canada N2L 3C5 ABSTRACT Glacier dams and outburst floods ("jokulhlaups") have been reported in many glacierized mountain regions, and may create hazards for human populations. Specially large and dangerous examples occur where the rivers of extensive ice-free zones are blocked. This hydrological anomaly has been rare in modern times except for two areas: the southern Alaska-Yukon ranges and Karakoram Himalaya. In the Karakoram some 30 glaciers may form substantial dams on the Upper Indus and systems. Many more interfere with the flow of rivers in a potentially dangerous way. There is evidence of some 35 disastrous jokulhlaups since 1826. Rarer landslide dams have resulted in the largest dam-burst floods. The paper provides a record of known dams and related events, and identifies the glaciers involved. It indicates the role of the regional environment in the widespread potential for these glacier dams and catastrophic outbursts. Some data are given on the dimensions of past dams and the nature and impact of the flood waves. No dams were reported from the mid 1930's until 1978 when satellite imagery showed a 6 km glacier lake on the Upper Yarkand. The absence of dams in recent decades relates to a general glacier recession here. Renewed activity creates serious problems for water resource development and settlement growth that occurred in the recent, unusually favourable period.

INTRODUCTION Bodies of water ponded by glaciers are common in most glaciated regions. They can range in size from small ice-marginal pools to the enormous glacier lakes dammed by certain Pleistocene ice sheets. A particularly large and dangerous dam occurs where a glacier enters and blocks a major river valley of which it is tributary. Evidently, however, these are an anomaly within the usual climatic hydrological and geomorphic progression downstream in a drainage basin. Rare and isolated cases have been reported from the southern Andes, the Canadian Arctic islands, the Caucasus, some Central Asian ranges and also the European Alps and Norway during the Little Ice Age. In one region of the world, however, the Karakoram Himalaya and neighbouring ranges, there has been a substantial number of these main valley glacier lakes in modern times. Outbursts from a series of dams on the Upper Shyok between 1926 and 1932 brought devastating floods along more than 1200 km of the Indus. Some even larger 259 260 Kenneth Hewitt

landslide dams and outburst floods occurred here in the nineteenth century and an exceptional concentration of surging glaciers has been found (Hewitt, 1969, 1975, unpublished). Some of the latter have formed main valley ice dams. The only comparable region in terms of these hazardous hydrological phenomena occurs in the southern Alaska-Yukon ranges. The Karakoram is far less known than the latter. This paper provides a summary of what is known of ice dam occurrence, the glaciers involved, and the impact of outburst floods.

THE RECORD OF GLACIER DAMS AND OUTBURST FLOODS

There is an extensive if scattered record of natural damming in the Karakoram region, sufficient to indicate the scope of the problem, and the particular drainage basins where damming recurs (Table 1). Thirty-five destructive outburst floods have been recorded in the past two hundred years. Thirty glaciers are known to have advanced across major headwater streams of the Indus and Yarkand rivers. There is unambiguous evidence of large reservoirs ponded by eighteen of these glaciers. Meanwhile, a further thirty-seven glaciers interfere with the flow of trunk streams in a potentially dangerous way. There is geological evidence of other dams and numerous reports of glaciers across main river channels which they were not actually damming. These too may be potentially dangerous. Geographically, glacier dams in main river valleys have occurred from the far western to the far eastern parts of the Karakoram range, and in the Lesser Hindu Kush, Nanga Parbat, Haramosh, Hindu Raj, Aghil, and far northeast Hindu Kush ranges (Fig.l). These are areas where we approach maximal local relief for the Earth's land

FIG.l Distribution of glacier dams and related events in the Upper Indus Basin (after Hewitt, unpublished). TABLE 1 Historical summary of glacier dams and outburst floods for the Karakoram Himalaya and environs. (For details and sources see Hewitt, 1968 and unpublished.)

Year Ice dam Outburst Major Glacier/River System flood disaster

1533 X X X Upper Shyok river 1780 X X Upper Shy ok river 1818- •1840s X (Series of barriers) Upper Shyok river 1826 X X X Upper Shyok river 1833 X X Upper Shyok river 1833 X X X Yashkuk Yaz Glacier 1835 X X Sultan Chhussku Glacier (1841 Massive landslide dam and outburst, ) 1842 X X X Upper Shyok river 1844 X X X Ishkoman river 1848 X Aktash Glacier 1848 X Kichik Khumdan Glacier 18501 '?; X X X Chungphar Glacier 1850 X Aktash Glacier 1852- •1858 X (Series of barriers) Kichik Khumdan Glacier 1855 X Upper Shyok ri ver 1855 X X Upper Shyok river (185S '.-1862 Massive landslide dam and outburst, river) 1864 X Kichik Khumdan Glacier (?) 1865 X X Ishkoman ri ver 1869- •1872 X Kichik Khumdan 1870 X Karambar Glacier 1873 x(?) Batura Glacier 1879 X X Upper Shyok river 1882 X X X Upper Shyok river 1884 X X X Shimshall river 1889 X Upper Shyok river 1891- •1892 X Ishkoman ri ver 1893 X X Shimshall river 1893 X X X Ishkoman ri ver 1899 x(?) Upper Shyok river 1901 X X X Upper Shyok river (?) 1902- •1911 X Kichik Khumdan Glacier 1903 X X Kichik Khumdan Glacier 1904- •1905 X Ishkoman river 1905- •1906 X Khurdopin Glacier 1905 X Kichik Khumdan Glacier 1905 X X X Karambar Glacier 1905 X X X Khurdopin Glacier 1906 X X X Khurdopin Glacier 1907 X X Khurdopin Glacier 1909 X Whirgut Glacier (1911 Landsli de dam near ) 1916 X Ishkoman ri ver 1924- 1933 X Chong Khumdan Glacier 1925- 1927 X Khurdopin Glacier 1926 X X X Chong Khumdan Glacier 1927 X X X Kaz Yaz Glacier 1927 X X Khurdopin Glacier 1928 X X Kilik river (Hunza) 1929 X X X Chong Khumdan Glacier 1929- 1930 X Ishkoman river 1930 X Kyagar Glacier 1932 X X Chong Khumdan Glacier 1933 X X Chong Khumdan Glacier 1939 X Chong Khumdan Glacier 1953 X Kutiah Glacier 1977 (Hunza landslide

-35-45'

CONTOURS-land MORAINE iC6 ACCUMULATION FIRN LINE ZONE FIG.2 The Kyagar Dam and Glacier in July 1978.

identified using Landsat imagery in a continuing programme to monitor glacial and other natural hazards in the region. By the end of the 2 In subsequent yearsfirst , summertwo larg, ae lakflooe do fwave somse 6comin km gha dowd ndeveloped the Yarkan. d river from its Karakoram headwaters have been attributed to the bursting of the lake (personal communication from Dr Shi Yafeng). The same glacier, the Outburst floods of the Karakoram 263

Kyagar, dammed the fifty years ago, when damming recurred over several years.

THE GLACIER DAMS

It has been possible to determine the order of magnitude of the volumes of water in three main valley reservoirs; the recent Kyagar dam, the Chong Khumdan dam of 1929, and, from field features, the Biafo Gyang dam which burst some time in the eighteenth century (Tables 2, 3 and 4). Morphological information is also given for the glaciers involved. In addition we have information from the one large tributary valley lake that has been a serious hazard in recent times, that where the Khurdopin Glacier dams the Virjerab valley in Upper Shimshall. Before its outburst in 1907,the glacier that formed there was about 3.5 km long, 1.5 km in average width and some 88 m deep at the dam.

TABLE 2 Kyagar Dam and Glacier

Kyagar Dam (18 July 1978) Glacier

Length of lake 6 .5 km Orientation NNW Average width 1.0 km Max. length 21.6 km Slope of valley Width, lower floor approximately 1 in 50 ablation zone 2.5 km Est. depth at dam 120 m Terminus altitude 4 700 m a.m. s .1 8 3 Est. volume 4.6 x 10 m Highest point of Min. width ice basin 7200 m Firn line ht (1978) 5400 ma.m.s.l. barrier 1.5 km 2 Area of basin Basin area 155.0 km above barrier Permanent snow and 2 ice cover 112 km (excluding Kyagar Basin) 366.6 km Accum. zone area 67.O km Ablation zone area 45.O km Slope main ice stream: Accum. zone (6 km) 1 in 3 Ablation zone (15.6 km) 1 in 22

THE GLACIERS

There is little or no basic glaciological data on movement, nourishment, thickness, or thermal characteristics for any of the glaciers that have formed major ice dams in the past hundred years, and only fair or poor topographical maps. It will be a useful beginning, however, to list locational information and such morpho­ logical features as can be determined for these glaciers (Table 5). 264 Kenneth Hewitt

TABLE 3 Chong Khumdan Dam and Glacier (see Hewitt, unpublished)

Glacier Dam (1929) Chong Khumdan Glacier

Length of lake 16 km Orientation E Average width 1.6 km Max. length 20 km Slope of valley Width, lower floor 1 in 130 ablation zone 2.5 km Depth at dam 120 m Terminus, altitude 4715 m a.m.,s. .1 Volume 1.5 x 10s m3 Highest point on Width of ice basin 7530 m a.m..s, .1 ice barrier 2.4 km Firn line 5250 m a.m.,s .1 Area of basin Basin area 140 km2 above barrier 25,500 km2 Water supply: summer discharge Chip Chap river 5.1 x 106 m3/day Rates of rise of lake, August 0.3-0.45 m/day

TABLE 4 Biafo Dam and Glacier (Hewitt, 1964)

Biafo Historic Dam (late Biafo Gyang eighteenth century)

Length of lake 15 km 4 Orientation SSE Average width 2-3 km Max. length 62 km Slope of valley Width, lower floor ablation zone 3 km approximately 1 in lOO Terminus altitude 3109 m a.m.s.1 Est. depth at Highest point of dam 150-200 m basin 7280 m a.m.s.1 Min. width of Firn line height ice barrier 2.5 km main stream(1961) 4725 m Area of basin Permanent snow above barrier 3500 km and ice cover 520 km2 Est. capacity Accumulation zone of lake 3.3 x 10- main glacier 310 km2 Ablation zone, main glacier 121 km2 Slope, main ice Stream: Accumulation zone 1 in 10 Ablation zone 1 in 30

High relief, steep average fall and large climatic gradient between upper and lower reaches promote vigorously active ice masses in the region. That is reflected in those few for which we have movement data. All except two of the glaciers known to have formed dams flow in northerly and easterly directions (Fig.. 3) . This Outburst floods of the Karakoram 265 TABLE 5 Some characteristics of damming glaciers of the Upper Indus and Yarkand drainage systems

Length Altitudes (m) (km) Terminus Firn line Highest Glacier Indus river system point type*

Eastern Karakoram Shyok river 1. Chong Khumdan 21 4715 5250 7530 FK 2, Kichik Khumdan 11 4630 5250 7530 FK 3. Aq Tash (Aktash) 8 4570 5250 6740 FK 4. Sultan Chhussku 8 - - - FK

Central Karakoram Braldu- 5. Biafo Gyang 68 3160 4500 7250 FS 6. Khurdopin 47 3250 4400 7760 FS 7. Yazghil 31 3200 - 7900 L 8. Malungutti Yaz 23 2900 - 7900 L Hi s par river 9. Barpu (Hopar) 33.8 2520 4600 7460 L(Surging) lO.Kutiah L

Western Karakoram ll.Ghulkin 19 2500 - 7600 L 12.Batura 58 2460 4450 7708 FK 13 .Pas u 23.5 2500 4600 7600 FK Chapursan river 14.Yashkuk Yaz 24 3550 4700 7150 L 15 .Kaz Yaz,' (Besk-i-yeng) 11 3660 4600 6690 FK 16.Pekhin 13 3050 5300 6420 FK 17 .Wirghut 7 3350 - 6130 L IS.Chillinji (West) 11 3530 4730 6130 L 19.Chatteboi (South) 13 3630 4570 5950 FK 20.Shuinj 9 39 30 - 5800 L 21.Chillinji (East) 10 3500 - 6690 L 22.Karambar 23.5 2440 4730 7160 Darkot-Thui Gol () 23.West Ghamu Bar 7 3660 6400 Tarshing river (Nanga Parbat) 24.Chungphar (Tarshing) 13 2920 6830 25.Bazhin 11 3200 8129

Yarkand river system

Eastern Karakoram Shaksgam river 26.Gasherbrum 20 4350 5350 8047 FK 2 7.Urdok 23 4370 5350 8068 L 28 .Staghar 4430 5200 6460 FK 29 .Singhi 24 4550 5400 7750 FK 30. Kyagar 22 4700 5400 7220 FK ' L = Lawinen or "Turkestan type" glacier; FK = Firnkessel or Firn Cauldron type; FS = Firnstrom or Firn Stream type; (see Hewitt, unpublished, Chapter 11). 266 Kenneth Hewitt

180 FIG. 3 Orientation of the main ice streams of damming glaciers of the Karakoram region. (0° is north).

suggests that aspect is a major factor in the potential for damming, as in other characteristics of these high, sub-tropical glaciers. Mason, the last scholar to make an extensive study of Karakoram dams (Mason, 1935) believed they formed only with an "accidental" or abnormally rapid ice advance. He was not aware of rates of flow for these glaciers, or the role of winter conditions, when termini may advance many metres, reactivate dead ice, and have little or no river flow to contend with (Hewitt, 1967, unpublished). Nevertheless, some ice dams have been associated with surging glaciers. Between 1975 and 1978 one can detect little change in the frontal position of the Kyagar Glacier. However, between 1977 and 1978 there was a sharp change in the geometry of the right flank of the snout, as though the easternmost ice stream became more active, extending across the Shaksgam Valley to make the ice barrier impassable. While dams of at least four glaciers have repeatedly re-formed over periods as long as two decades, the lakes rarely last more than two summers without an episode of complete draining.

OUTBURST FLOODS Breaching of a dam creates a sudden short-term increase in discharge downstream. If breaching is catastrophic, as so often in the Karakoram region, the impact of the flood wave can far outweigh that of other high flows. In part, this can involve a concentration of flow in the upper reaches of the rivers well in excess of weather-produced extremes of runoff. The 1929 outburst flood of Chong Khumdan Glacier was monitored from near the glacier over more than 1500 km downstream (Gunn, 1930; Mason et al., 1930). Along with some comparative observations for the smaller 1932 outburst, (Mason, 1932) this gives a unique record of flood-wave behaviour on the Upper Indus (Table 6). Gunn, (1930) estimated the reservoir to have contained almost 8 3 5 3 13.5 x 10 m (1.1 million acre feet) . Some 3 x lO m of ice were Outburst floods in the Karakoram 267

TABLE 6 Progress of the 1929 outburst-flood on the Upper Indus (after Gunn, 1930; Mason et al., 1930)

Location Distance Maximum Rise Duration Rate of from dam flood rise to peak of wave travel -1 (km) (m) (h) (h) (km h .

Sasir 16 26.0 4.0 40 8.3 Khalsar 217 19.2 2.0 10 20.0 Skardu 499 7.6 3.5 28 22.0 Par tab P. 719 13.7 8.0 60 13.2 Bunji 731 10.6 5.9 50 ) Chi las 803 16.1 4.0 40 ) 18.8 Tarbela 1120 7.0 12.0 50 ) At took 1194 8.1 17.0 70 5.9

also carried with the flood and stranded on large blocks in the valley below the dam. If loss to channel storage and seepage is somewhat greater than gains from inflows below the dam, the complete draining in 48 h suggests an average discharge between Sasir Brangsa and Khalsar in the region of 7100 m3s_1 (250000 cfs). In the steeply rising and falling main flood peak however water discharges in excess of 22 650m3s_1 (800000 cfs) are indicated. That equals the largest discharges measured for the entire Upper Indus at Attock. The Upper Syok drains less than two per cent of the basin, and most of its area is arid. The mode of dam failure is critical to the size and shape of flood wave. All we know of the 1926, 1929 and 1932 Khumdan outbursts is that breaching began through subglacial tunnels, but then carried away the entire thickness of ice above. Gunn (1930) describes the breach of 1929 as having: "... burst along a curving line from the near right-hand bank of the lake on the northern side through the highest portion of the dam, nearly to the left bank of the river on the south side. The cut ... was about 400 ft £l20 m] wide (and 500 ft + [lSO m +] deep) and the ice stood vertically on either side. The lowest-portion of the dam along the cliffs was unaffected." This contrasts with most examples discussed in the literature, in which the sudden closing of the outlet when the lake is partially emptied is said to produce another characteristic feature of "jokulhlaups" (see Maag, 1969; Young, 1980). The flood peak was highest at Sasir Brangsa, but the 1929 wave showed remarkable recuperative power in the Indus gorges, below Skardu. The effect is the same as in the well-known Johnstown, Pennsylvania flood disaster of 1889. As the flood waters gather in, then leave intermontane basins such as at Skardu, they re-enact a pseudo-dam break.

SOME FLOOD-WAVE IMPACTS The significance of these floods lies especially in the exceptional risk to human communities or installations, and also in their role in 268 Kenneth Hewitt erosion and sedimentation. Over much of their course in the mountains, the recorded floods reach heights well above peak discharges from summer melting. Their dynamic character greatly magnifies their erosional competence and capacity. These two matters are of singular importance in the erosional context of the Karakoram valleys, and sediment transport into downstream reservoirs. Sediment yield from the Upper Indus Basin represents the highest known rate of regional erosion over such an area, of about one metre per thousand years (Hewitt, unpublished). Data for the tributaries where the dams occur suggest rates in excess of 1.8 m per thousand years. Nevertheless, average sediment concentration upon which rating curves are based may be several orders of magnitude lower than the highest concentrations for given discharges. These exceptional concentrations generally occur in association with flood waves (Fig.4). What is reflected here is a highly constrained sediment availability in the fluvial zone. The humid, glacial areas not only provide most of the water, but also most of the seasonal debris transported.

INDUS RIVER at DARBAND Lat. 34 24 Long. 72 48' Summer 1962

0) a) 90-

o<= =1•

c £ CD °- E5 •15 u Q

i i i I i i i i I—i—i—i—I—r 20 21 22 23 24 25 20 27 28 29 30 1 2 3 June 1962 July 1962

FIG.4 An example of exceptional sediment concentration (broken line) on the Upper Indus, related to the passage of a flood wave (solid line) (from data supplied by Water and Power Development Authority, ). (c.f.s. = 2.83 x 10~2m3s~1).

The great height and erosional energy of dam-burst flood waves especially allows them to reach and cut into the abundant lag deposits in or stranded upon, arid river terraces. Huge numbers of landslides have been reported on terraces and valley sides after the passage of damburst floods. Substantial channel widening, deepening and even changes of course have been reported. If one extrapolates the existing sediment rating curve for Darband - or Attock before the Tarbela Dam was built - the 1929 flood curve would have carried the equivalent of one average year's sediment yield. With greater concentrations of sediment and Outburst floods in the Karakoram 269 enhanced bedload this would be much higher. Moreover, the mobilization and sluicing of transportable sediment into stream channels would tend to increase total yields for some months or years after a major outburst. In the event of a phase of recurrent damming such as occurred prior to 1940, these erosional events could increase the rate of sedimentation in artificial dams on these rivers, and reduce their economic lifetimes.

REFERENCES

Gunn, J.P. (1930) Report on the Khumdan Dam and Shyok Flood of 1929. Government of Punjab Publication, Lahore. Hewitt, K. (1964) A Karakoram ice dam. Indus: J. Water and Power Devel. Ruth. (Pakistan) 5, 18-30. Hewitt, K. (1967) Ice-front sedimentation and the seasonal effect: a Himalayan example. Trans. Inst. Brit. Geogrs (42), 93-106. Hewitt, K. (1968) Records of natural damming and related events. Indus; J. Water and Power Devel. Auth. (Pakistan) 10(4). Hewitt, K. (unpublished) Studies in the Geomorphology of the Upper Indus Basin. 2 vols. PhD dissertation, University of London. Hewitt, K. (1969) Glacier surges in the Karakoram Himalaya (Central Asia). Can. J. Earth Sci. 6(4), 1009--1018. Hewitt, K. (1975) Perspective on disasters and natural hazards in the Northern Areas. Dawn Magazine. Karachi, Pakistan. January 1-12. Maag, H.V. (1969) Icedammed lakes and marginal giacjal drainage on Axel Heiberg Island. Axel Heiberg Research Report. McGill Univ. Montreal, 14 7 p. Mason, K. (1932) The Chong Khumdan Glacier, 1932. Himalayan J. 5, 128-130. Mason, K. (1935) The study of threatening glaciers. Geogr. J. 85(1) , 24-35. Mason, K., Gunn, J.P. and Todd, H.J. (1930) The Shyok flood in 1929. Himalayan J. 2, 35-47. Young, G.J. (1980) Monitoring glacier outburst floods. Nordic Hydrology. 11, 285-300.