Quest to unravel the cause of the Seti flash flood, 5 May 2012

   TwitCount

On Saturday 5 May 2012, flash flooding along the Seti River in the Kaski district of northwestern Nepal resulted in the death or disappearance of at least 72 people, and caused great loss of property, including homes, businesses, crops, livestock, and valuable infrastructure. 

On 9 May, a team of experts from ICIMOD, joined by Bal Krishna Panthi, Under Secretary, Ministry of Home Affairs and Chief of the National Emergency Operation Centre, travelled to the flood-affected area to make a preliminary assessment about its source and possible causes. The team observed the evidence for the high intensity of the flash flood, which had been higher than 30 m at some locations. Local people indicated that it was likely that many additional people were missing besides those formally listed, including workers who had come to this area from other parts of the country and had not been reported. It was not possible to trek beyond Sadal village because the trail bridges had been washed out by the flood. Further, the large amount of sediment deposited by the flood inhibited walking beyond this point. It was also learnt that after about 6-7 km from Sadal the river flows through a narrow gorge and there is no trail beyond that point. The team noted that in contrast with other flash floods in the mountains, this flood had not left heavy deposits of rock debris along the flood path. Instead, the team found a heavy deposit of fine sediment.

On 20 May a team of ICIMOD experts joined Prof. Jeffrey Kargel from the University of Arizona, USA, in an aerial reconnaissance of the Seti headwater area. Several sorties were made along the Seti River valley and the Seti headwaters. 

Unfortunately, to date the cause of the flash flood remains unclear, but important clues have emerged. Key among the clues is evidence that rockfall activity into the Seti River gorge had taken place in the years and weeks prior to the disaster; an unusually large rock/ice/snow avalanche/landslide occurred in the minutes before the disaster; material from the avalanche/landslide, including some airborne material and possibly some ground-flowing material, reached as far as the Seti River gorge. These dynamic events appear to be related in some way to the initiation of the disaster, but the precise mechanisms remain unclear.

Several hypotheses, some interrelated, have appeared from different sources on the cause of the flash flood. Here we present those hypotheses and a general progression of hypothesis development, including our arguments to reject, support, or modify them.

  1. Glacial lake outburst flood (GLOF): immediately after the flash flood, glacial lake outburst was indicated as the cause. This hypothesis came mainly from the non-scientific community. As ICIMOD has been working on glacial lake inventory, we quickly checked the database and confirmed that significant glacial lakes did not exist in this area. Thus a flash flood of this magnitude cannot be due to any sort of normal GLOF. 

  2. Rockfall dam outburst flood (RDOF): Later on 5 May, the Nepal Army and Nepal Police made a helicopter flight to the flood-affected area and the upstream part of the Seti River. The ICIMOD team viewed the video clip filmed by the army. The view of the rockfall was not very good. The clip captured the upstream part of the river, where the river water would have been impounded due to the damming. We were not confident that the rockfall was the very cause of the flash flood. However, lacking any other alternate hypothesis at the time, we then saw this as the most likely cause. Study of satellite images showed that the landslide area had grown in the weeks just before the event and also had experienced activity in the preceding years. On our helicopter flight of 20 May, we were able to assess this area closely. Considering the relatively small size of the rockfall and the remnant dam (Figure 1), the rather narrow river valley (gorge), and the steep longitudinal profile of the stream, we conclude that impoundment of a large water body by the rockfall is not possible. The rockfall could have played some role, i.e., partial obstruction and contribution of flood waters, or transient obstruction during the flood and contribution to the pulsed nature of the flood outburst; but the rockfall could not have been the sole cause of the flash flood in the absence of other processes. 

  3. Rock/ice/snow avalanche: The first eye-witness evidence of the event was by Captain Alexander Maximov, an ultra-light airplane pilot for Avia Club Nepal. On 5 May Captain Maximov, while flying over the mid-section of the Seti River canyon, noticed a big avalanche on the south slope of Annapurna IV, near the headwaters of the river several kilometres upstream from where he was flying. Wing-mounted cameras captured the avalanche as it took place. The avalanche cloud, unlike that of a regular snow avalanche, was brown in colour and extremely large (Figure 2a). He immediately turned around and flew back to Pokhara airport. On his way he saw a big wave of flash flood rushing down the Seti River. He immediately informed the aviation tower in Pokhara and the news was then disseminated by the tower to concerned agencies and local FM radio stations. It is believed that his action helped reduce the loss of human lives significantly in the lower part of the river valley. Ken Noguchi, a mountaineer and environmentalist from Japan, was the first to propose a massive avalanche from the south face of Annapurna IV as the cause of the flash flood. However, Mr Noguchi did not explain the exact mechanism that created the flood. On 23 May, David Petley, University of Durham, UK, posted a blog supporting a similar idea, but instead of ice avalanche, he suggested that a large rock mass from the south slope of Mount Annapurna collapsed. This blog showed the slope of Annapurna IV that had collapsed. It referred to a satellite image of the rockfall area posted by the United States National Aeronautics and Space Administration (NASA). The image showed a large brown area supposedly due to the perturbation by the rock avalanche/landslide of the thick lacustrine deposit created in past geologic times (Figure 2b). The brown colour of the cloud was most likely due to the brown sediment derived from those ancient sediments. An image of the same area before the event showed the area under snow cover (Figure 2c). Another proponent of the ice/snow/rock avalanche hypothesis is Mr Hidetami Oi, representing the Nepal-Japan Friendship Association for Water Induced Disaster Prevention (NFAD). Mr Oi has also made an aerial survey of Seti headwater area. Both Prof. Petley and Mr Oi ascertained that the debris of the ice/snow/rock collapse ran far enough to reach the river channel. 

  4. Landslide/avalanche cloud fallout into the rockfall-dammed gorge: Satellite image analysis performed by Jeffrey Kargel and Gregory Leonard (University of Arizona), supplemented by helicopter-borne aerial study, supports the runout of a component of the avalanche/landslide into the upper gorge area. The landslide has a proximal deposit that was mainly due to ground-based flow of avalanche/landslide debris, and a more distal deposit that was at least partly and probably mainly airborne in its delivery. Pictures taken by the ICIMOD/University of Arizona team during our helicopter surveillance of the area clearly show that debris reached the river channel (Figure 2d). Although the boulder landslide part of this mass movement clearly did not reach the channel, fine-grained debris – and a lot of it – clearly did reach the channel. We also found fallen trees on the opposite side of the stream just above the top of the upper part of the gorge area (Figure 2f). The direction of the fallen trees was away from the stream, indicating that the impact came from the direction of the stream. The fallen trees were showered by fine-grained debris, as were the slopes above the gorge and leading into the gorge. One may only surmise that fine-grained debris also entered the gorge. The treefall and fine-grained debris were apparently induced by dust-laden slope winds as the avalanche cloud settled. However, volumetrically, this mechanism still seems to fail to account for the delivery of enough water volume for the flood.

  5. Outflow from underground karst storage: The prelude of this hypothesis is the enormous amount of water discharged in the Seti river during the flash flood event. A rough calculation done by Prof. Kargel based on a video of the flood posted on YouTube suggested that the volume of water released during the event could be between 250,000 (for the first flood pulse only) and a few million cubic metres (for the entire flood sequence involving many flood pulses). The main problem of the preceding hypotheses is the lack of evidence of a sufficient source of water. The commonality of this Hypothesis 5 with Hypothesis 2 is that the rockfall into the gorge did indeed occur (multiple episodes in the years and weeks preceding the disaster, according to satellite image analysis by David Petley and by Greg Leonard and Jeffrey Kargel). The commonality with Hypotheses 3 and 4 is that the giant ice/snow/rock avalanche indeed occurred. This is proven by the picture from Captain Maximov of Avia Club Nepal and the NASA satellite imagery. Furthermore the global and national seismic centres measured significant surface waves typical of large landslides. According to Mr Som Nath Sapkota, Department of Mines and Geology, Nepal, the tremor was a magnitude 4 event; according to some translations to energy, such a tremor would equate to about 63 GJ of released energy. Any such calculations must be considered approximations at best, but they give an idea of what initiated the process. The shock itself was not the sole cause of the flash flood, but it might have triggered the flash flood. The Seti gorge is characterized by karst features with many cavern-like structures under the surface (we do not know the extent of cavern development; it may be limited to slot canyon development, or there could be large caverns). These types of caves and caverns are found in abundance in and around the city of Pokhara (Figure 3), but in a completely different rock formation and tectonic setting. However, in this hypothesis, low-grade metamorphosed carbonate rocks may also exist in the gorge area and may have been subject to karst forming erosion. We note that geologic mapping is still not complete in the upper Seti River gorge area, but the geomorphology is consistent with karst formation. The voids were likely filled with water accumulated over a long period of time. According to this hypothesis, a plug (e.g., collapsed underground rock wall) somewhere was stopping the water outflow; it was suddenly released because of the big tremor created by the ice/rockfall, and the trapped water gushed out creating the flash flood. The fine nature of the sediment deposited by the flood also indicates that the source water had accumulated gradually. The only way to prove (or disprove) this hypothesis is to explore the gorge of the Seti, which is a very difficult task that can only be undertaken by highly trained rock climbers and cave explorers.
Among the hypotheses presented here, only the last one can really explain the large amount of water. To prove (or disprove) this hypothesis it is necessary to explore inside the Seti gorge. While this is not impossible, it is a mightily difficult mountaineering task. 

It is very likely that not only one source of water was involved. For example, a combination of supraglacial ponds, water contained in wet snow, snow melted by conversion of gravitational potential energy to heat, water dammed in the gorge, and water contained in small caverns might explain the flood volume better than any one explanation alone, particularly if it turns out that there are no large caverns in the rocks of this area. 

More importantly, is it indispensable to find the real cause of the flash flood? Can the downstream risk be managed independently of knowledge about the cause of the flash flood? In lack of an easy answer to the real cause of the flash flood, it might be practical to concentrate on the downstream risk management for the time being, although ultimately the city of Pokhara will need to understand what happened scientifically in order to understand how future events might or might not scale up in size. 

Acknowledgements: Our great appreciation goes to Mr Alexander Maximov, Avia Club, Nepal; Mr Bal Krishna Panthi, Ministry of Home, Nepal; Mr Rajendra Singh Bhandari, Western Region Police Division, Pokhara; Mr Viktor Rana, Western Region Army Office, Faculties of Department of Geography, Prithvi Narayan Campus, Pokhara, and many people along the Seti river who provided useful information and support during our investigation.

Authors: Arun B. Shrestha (ICIMOD), Pradeep Mool (ICIMOD), Jeffrey Kargel (University of Arizona), Rajendra B. Shrestha (ICIMOD), Samjwal Bajracharya (ICIMOD), Sagar Bajracharya (ICIMOD), Deependra Tandukar (ICIMOD) 

Figure 1: Rockfall that was initially suggested to be the cause of the flash flood

Figure 1: Rockfall that was initially suggested to be the cause of the flash flood

Figure 2a: The brown avalanche cloud captured by Captain Alexander Maximov, Avia Club pilot

Figure 2a: The brown avalanche cloud captured by Captain Alexander Maximov, Avia Club pilot

Figure 2b: Pinnacles of eroded lacustrine sediment, the prominent landform in the headwater of the Seti (one of the origins of the Seti can be seen in the lower right corner)

Figure 2b: Pinnacles of eroded lacustrine sediment, the prominent landform in the headwater of the Seti (one of the origins of the Seti can be seen in the lower right corner)

Landsat 7 ETM+ image acquired on May 6, 2012 (Source: NASA Earth Observatory)

Figure 2c: Landsat 7 ETM+ image acquired on May 6, 2012
(Source: NASA Earth Observatory
)

Figure 2d: Annapurna IV where the ice/rockfall was suggested, with brown colour in the foreground due to settled particles

Figure 2d: Annapurna IV where the ice/rockfall was suggested, with brown colour in the foreground due to settled particles

Figure 2e: Close-up view of the wall that collapsed

Figure 2e: Close-up view of the wall that collapsed

Figure 2f: Collapsed trees on the opposite side of the stream (note the absence of significant debris above the fallen trees)

Figure 3a: Gupteshwor Mahadev Cave, Pokhara

Figure 3a: Gupteshwor Mahadev Cave, Pokhara

Figure 2f: Collapsed trees on the opposite side of the stream (note the absence of significant debris above the fallen trees)

Figure 3b: Mahendra Cave, Pokhara

Figure 3b: Mahendra Cave, Pokhara