What caused the flash flood that killed almost 250 people in India’s Uttarakhand state on 7th February 2021? Glacier collapse? Glacial lake outburst? Debris flow? While hypotheses have been plentiful, the underlying evidence is thin. Based on the data now available, we take an objective look at a disaster many consider a portent of global warming.
The first stage in the analysis is to describe the phenomena that occurred in key locations, such as the dam construction site. The videos are harrowing, terrifying. But beyond the horror of the situation, they provide clues to how it happened. They show an extremely powerful flow destroying everything in its path. Although the initial surge is too turbulent and chaotic to make out its composition, once the front has passed, the body of the flow can be seen clearly enough to discern its nature. It is clearly a debris flow—a relatively homogenous mix of water and sediment, with huge blocks floating on the surface. Most debris flows are assumed to be 50% water and 50% sediment, unless they are hyperconcentrated, in which case the proportion of water is higher.
Given the spectacular height they reached—tens of metres—the flood waters must have been flowing at a rate of at least several hundred cubic metres per second. According to a local engineer (personal communication), the mudflow that hit the Swiss village of Chamoson in 2019 had a flow rate of around 200 m3/s (water + sediment). Compare the two events and it is immediately obvious that the Uttarakhand debris flow must have involved enormous quantities of water. This observation supports the initial theory of a natural dam burst, but subsequent analyses complicate this picture.
Collapse of a Hanging Glacier
Images taken by the Sentinel satellite clearly show where the event started: an area at around 5500 m on the north face of Ronti Peak (6063 m). The foot of the mountain is 1700 m lower, at around 3800 m. This area is more than 15 km from the dam construction site (alt. 2000 m). Link to the Sentinel images :
A large part of the hanging glacier has completely disappeared, but the cause is probably deeper, as the entire face appears to have been affected. In fact, the geometry of the scar suggests that the initial landslide involved between 25 and 50 million cubic metres of rock and ice.
Natural slides that are highly conducive to large-scale ground movements
Google Earth’s 3D view shows that the face is made of sedimentary rock, probably limestone. Bedding planes (boundaries between two layers of rock) are clearly visible, so the face must be inclined at a similar angle to the dip (30°) of the limestone strata. These steeply dipping beds are particularly efficient slippage planes, natural slides that are highly conducive to large-scale ground movements, especially landslides.
The geometry of the scar also supports this hypothesis. Most scars are concave, but here the scar is planar, which is the profile produced when a homogenous block of rock suddenly breaks off along a slippage plane. Even though the increased frequency of rock falls in the Alps has been linked to climate change, we cannot be certain that this was the trigger mechanism here, as here the geography is very different and the mountains are much higher than those studied in Europe.
Satellite images taken before and after the landslide show that it was preceded by warning signs. In fact, the edges of the rock scar clearly correspond to the fracture lines visible in the last image taken before the landslide occurred.
Scientists have long known that the rheology (the way matter deforms and flows) of a rock fall depends on the intensity of the event. Above a certain volume, usually taken to be a million cubic metres, landslides turn into rock avalanches, which build up energy by entraining sediment from the slope or the valley floor. As a result, they can cover large distances and mobilise huge volumes compared with the size of the initial event. Perhaps the most spectacular example of this phenomenon in Europe occurred in 1248, when a landslide of around 15 million cubic metres from Mount Granier, near Chambéry, triggered a rock avalanche involving several hundred million cubic metres of material (Panet, 2010).
The initial rock fall was able to mobilise a huge volume of sediment
Given our estimate of the rock fall’s initial volume, it was undoubtedly followed by a rock avalanche. The valley was once filled by a glacier, so it’s floor is covered by a thick layer of sediment, deposited as the glacier ebbed and flowed. This sediment meant that the initial rock fall was able to mobilise a huge volume of material and therefore pick up energy and propagate further. The damage to the valley floor is particularly telling: a strip several hundred metres wide and several hundred metres high was completely devastated. Only an aerosol could rise so high in a valley of this shape.
A reconnaissance video shows that the lake is still there
How far did the avalanche travel and how did it turn into a mudflow ? These questions are difficult to answer, as the satellite pictures are not detailed enough to clearly show what happened (Sentinel’s orthoimages have a resolution of only 10 m, compared with 20 cm for the images on the French geographical survey website geoportail.gouv.fr.) and the helicopter video footage is too limited. The lake outburst theory, involving a natural lake between the Ronti Gad—the river below Ronti Peak—and the Rishi Ganga—the main river—is contradicted by a reconnaissance video showing the lake is still there. In fact, this lake represents a new danger. Most importantly, the video shows that the debris flow was still several tens of metres high, perhaps more, when it reached the confluence at the end of the Ronti Valley.
The satellite pictures also suggest a large degree of continuity in the flow’s deposits between the foot of Ronti peak and the confluence. Another satellite photo taken during the event shows that the aerosol travelled at least as far as the confluence, so it most probably occurred as a single event, with the rock avalanche gradually turning into a debris flow and then a hyperconcentrated flow. For the moment, these are still hypotheses and further analyses are needed to determine whether or not the flow was continuous. In fact, numerous questions remain unanswered. For example, what was the role played by the glacier ice which the rock fall stripped off the face and injected into the rock avalanche ? This we still don’t know, but similar processes may have been involved in forming the mudflows that followed the Pizzo Cengalo (Switzerland) rock fall in 2017 (Walter et al., 2020).
The event in Uttarakhand was clearly of an exceptional magnitude but we do not have sufficient data to confidently untangle the complex series of processes involved (glacial, gravitational, rheological). A lot of grey areas need to be clarified in order to explain the chain of events.
Because these phenomena are complex, they are difficult to describe and even harder to foresee, especially in high mountain areas which are not monitored as intensively as the Alps. Hence, in order to predict these phenomena, research must consider them in their totality and take into account their intrinsic complexity.
Could Similar Phenomena Occur in the Alps ?
The most recent event in Europe that comes to mind is the huge rock fall on Pizzo Cengalo in 2017. Although it involved three million cubic metres of material and was followed by mudflows (Baer et al., 2017), it was a much smaller event than the one in Uttarakhand.
On a longer timescale, the Alps have witnessed numerous complex disasters (rock falls, natural dam bursts, glacier floods) over the centuries: the collapse of Mont Granier in 1248 (Pachoud, 1991); the Gietroz glacial outburst flood, which reached Martigny; landslides at Le Dérochoir, near Passy, which have dammed the River Arve on several occasions (Amelot, 2005), and in the Oisans (L’Hutereau, 2020). The largest known rockslide in the Alps occurred at Flims, in eastern Switzerland, around 10,000 years ago (Ivy-Ochs et al., 2009) and its effects can still be seen today.
One thing this far-from-comprehensive list shows is that lower mountain ranges are just as vulnerable as the high mountains. And there is no reason to think we are prepared for the possibility of another major event, given the limited resources available for predicting disasters and the lack of resilience and acceptance among politicians and, especially, among the people who live in these areas.
Risk assessments prior to building new infrastructure in the (high) mountains should not be seen as academic exercises; they are indispensable for avoiding enormous economic and, most importantly, human costs in the future
Greater steps must also be taken to understand complex phenomena. Risk management is still too fragmented and too often based on models whose uncertainties we don’t understand, as was demonstrated by the response to the damage caused by storm Alex in the Maritime Alps. Therefore, it is important to be able to contextualise the risks in each area studied in order to better identify cascade effects. The appearance of warning signs prior to the Uttarakhand rock fall suggests that better detection—possibly by using deep learning to develop automated methods—could be used to predict large-scale ground movements and thereby prevent further disasters.
Finally, it is worth noting that most of the people killed in Uttarakhand were working on the new dam. Without them we may never have heard of the disaster. Indeed, an increase in the intensity and frequency of this type of event is not necessarily a problem in itself; the real problem is the increase in human exposure to the dangers these events pose. As such, risk assessments prior to building new infrastructure in the (high) mountains should not be seen as academic exercises; they are indispensable for avoiding enormous economic and, most importantly, human costs in the future.
To conclude, here is a video from 2013—a prescient indication of the devastation a landslide and the succeeding chain of events could cause in Uttarakhand. What can I say?