Understanding the Chamoli flood: Cause, process, impacts, and context of rapid infrastructure development
By Arun B. Shrestha, Jakob Steiner, Santosh Nepal, Sudan B. Maharjan, Miriam Jackson, Ghulam Rasul, and Birendra Bajracharya
The Chamoli flood was not caused by a GLOF as there were no significant glacial lakes in the area
The flood was triggered by a massive rockslide just below Ronti peak, of ~22 mio m3 of rock mixed with ice and snow
The energy of the fall melted the ice creating the source of flood. This remobilized the debris and ice on the valley floor deposited by previous events, pushed the stream water and created an excessive flood wave.
A couple of days prior to this, a strong western disturbance resulted in heavy precipitation in the area, which increased the flood magnitude downstream
Comprehensive monitoring of mountain environments is recommended
Infrastructure in the flood path, particularly hydropower projects, exacerbated the impact of the flood. Infrastructure development in fragile mountain environments should consider a sustainability framework, including environmental sustainability
Disaster struck Uttarakhand’s Chamoli District on 7 February 2021, when a massive flash flood ravaged through the valleys of the Rishi Ganga, Dhauliganga and Alaknanda rivers. More than 70 people have been confirmed dead and another 134 people reported missing. The flood swept away the unfinished Tapovan Vishnugad Hydropower Project and inflicted substantial damage on the Rishi Ganga Hydropower Project. Even as rescue and relief efforts were underway, a glacial lake outburst flood (GLOF) was suggested as the reason for the flood, possibly triggered by glacier collapse. Some reports even used the term “glacier burst”, possibly referring to a GLOF. In the immediate aftermath of the event, ICIMOD decided not to comment on the cause of the flood for two reasons. Firstly, while understanding the cause is important to identify possibilities of similar hazards in this and other areas, the priority in the first few days and weeks should be rescue and relief. Secondly, as a regional knowledge institution, we felt it important to verify and corroborate all available information to come up with a credible reason for the flood. Now, two weeks after the event, we have adequate information to present our understanding of this flood event. The purpose of this report is to share our understanding of the cause and downstream propagation of the flood. The report also touches upon the cascading nature of floods, the challenges to infrastructre development, and the need to consider environmental sustainability while planning infrastructure development in fragile mountain environments.
2. General description of the area
The flood event took place in the Tapovan area of Joshimath in Chamoli District, Uttarakhand State. The region consists of high mountain ranges with steep topography, including the second highest peak in India, Nanda Devi (7816 masl; Fig. 1). The mountain ranges are made up of high grade metamorphic and volcanic rocks. The Lower Himalaya Range to the south is composed of sedimentary and low-grade metamorphic rocks. The Chamoli flood occurred in an area about 60 km northeast of where the devastating Uttarakhand flash flood occurred in 2013. The Chamoli flood took place on the Rishi Ganga River on the northern side of Nanda Ghumti peak (7050 masl). The Rishi Ganga is a tributary of the Dhauliganga River, which originates from the Raikhana Glacier (5375 masl) to the north and meets the Alaknanda River further downstream. The high-altitude area consists of glaciers and snow peaks which melt during the spring and summer season and provide meltwater to downstream areas. Because of the perennial sources of water and steep topography, hydropower plants have been constructed across many tributaries. Presently, about 3900 MW of hydroelectricity is generated in Uttarakhand, out of its estimated hydropower potential of 20,000 MW (Agrawal et al. 2017), with several under construction projects expected to add another 3200 MW.
The climate of the area is characterised by warm-wet summers (June-September) and cold-wet winters (December-February). The difference between average monthly maximum and minimum temperature is about 20 oC, which is higher in the winter season compared to summer (Fig. 2). The area receives precipitation from both westerlies and the summer monsoon, with annual average precipitation around 1000 mm; of this, winter precipitation due to western disturbances contributes 28% and summer precipitation 42%.
3. The source of the flood
a. Initial suggestions about the flood and counter arguments
As news and videos of the debris flow that hit the Tapovan Hydropower Project site around 10:30 IST on 7 February surfaced, initial suggestions were that a glacial lake must have burst. As the upstream areas from the location of the first videos were investigated, it quickly became clear that no lakes were present in the catchment prior to the event. Some media outlets even called it a “glacier burst”, a term not recognized by the scientific community. As the first high resolution images (taken by coincidence just as the debris flow passed the valley; see Figure 3) became available, approximately 7 hours after the event, it was clear that an avalanche, landslide or rockfall had happened approximately 22 km upstream of the hydropower site, just below Ronti peak in the Nanda Devi massif. The location of the source was clearly identifiable from the imagery (Figure 3) while no other source (flood paths, emptied lake, or avalanche scour) could be identified.
b. What we now think is the source of the flood
Based on available imagery and relying on published data, we are able to make approximate calculations of the mass movements that have taken place. We examined pre- and post-event imagery and found that a crack had formed prior to the event (Fig. 4 and 5) at the site where the rock detachment followed by a rockslide happened (Fig. 4). This failure eventually propagated along a 550 m wide crest starting at an elevation of 5500 masl reaching down to nearly 4500 masl. Analysis of elevation models pre- and post-event suggests that the scarp left by the rockslide is 150 m deep, 100 m on average and consists largely of rock and relatively little ice (Fig. 4). It is 39° steep, 1060 m long and has an area of ~350,000 m2. This results in an approximate volume of 22 mio m3, which corresponds with the DEM differencing that puts it at 25 mio m3. Relying on modelled glacier thickness (21 to 25 m for the three glacierettes in the inventory, which corresponds to typical heights of such hanging glaciers; Farinotti et al. 2019), we can estimate the fraction of rock to be 85% and ice 15% and calculate a total mass of ~52 * 109 kg. With a straight slide line of 1.6 km (5500 to 3900 masl), this results in total potential energy of 8.24 * 1014 J. This energy is converted to kinetic energy during the fall and dissipated as enough heat to melt 2.7 * 106 m3 of ice (with 335 kJ per kg of ice necessary at 0 °C). Considering that not all the mass was converted into energy during the fall, this number is likely a lot lower (Huggel et al. 2005). As Huggel et al. (2005) argue and has been conclusively shown in experiments (Arakawa 1999) and for a large co-seismic event (Eberhart-Phillips et al. 2003), fluidization can also happen simply from a very large impact on present ice, which possibly happened in this case.
During an ice avalanche in 2016, a volume of ~7.2 * 106 m3 of ice was dislodged to the west of the rockslide in 2021, taking bed rock with it along the way, resulting in a mixed ice-debris deposit of ~1.5 * 107 m3. Sentinel-2 Imagery from 2 December 2020, when the area was not snow covered yet, shows that a large part of this deposit was still present in the valley below when the event on 7 February 2021 happened (Fig. 6). In a similar case in Langtang Valley in Nepal, a co-seismic avalanche resulted in a large compound deposit of ice, snow, and debris. As debris settles on top during the fall (as in the case here), ice was protected from melt by a debris layer of several metres thickness (Fujita et al. 2017; Kargel et al. 2016). A big part of the ice body is still present in Langtang even today, nearly 6 years later. As some of this ice body was still present just after the event below Ronti peak in 2021 (Fig. 5), we can conclude that a fraction of this previous deposit (< 7.2 * 106 m3), was fluidized by the available energy. There are also reports from observers of the event of a pungent smell, suggesting water-saturated sediments were mobilised and added to the fluid content of the debris flow.
From the 200 to 400 m high surge of debris (h) on the adjacent headwalls below, we can estimate the velocity of the fall at between 60 and 90 m s-1 (Chow 1959; Evans et al. 2001, 1989; Pierson 1985). Assuming an average velocity for the resulting debris flow of ~20 m s-1 the debris flow took about 18 minutes to reach the Tapovan Hydropower Project site and was able to move ~1000 m3 s-1 of water, assuming that only 75% of the energy from the rockslide was used to fluidize the deposited ice below. This would explain to a large part, how such a strong flood wave was able to reach downstream areas without a previous body of water being present.
c. What triggered the rockslide?
i. Precedent weather conditions
A strong western disturbance passed across Kashmir and northwest India from 4 to 6 February 2021. It was fully charged with convective instability that may have contributed to the heavy precipitation. This unfortunate event occurred on 7 February. Numerical simulation of some of the attributes have been carried out which depict strong evidence of heavy precipitation contributing to high flows downstream. The analysis of wind pattern and geopotential height contours at 500hPa level indicate that the trough of an active westerly wave was passing over Kashmir and northwestern latitudes of India with a strong vorticity and convergence combination at the leading edge of the westerly. The trough of this western disturbance showed great potential of convective instability as severe Convective Available Potential Energy (CAPE) conditions were found on the rear end of the low pressure area. The numerical simulation on 4 February is presented in Figure 7 which shows heavy precipitation over that region. The western disturbance travelled with relatively slower speed and its stagnancy produced concentrated precipitation.
Precipitation data derived from Global Precipitation Measurement (GPM) Mission suggests that there were continuous precipitation events from 3 to 5 February which resulted in approximately 58 mm of cumulative precipitation (Fig. 10). Most of the precipitation in the high altitude areas of Chamoli would have occurred as snowfall.
ii. Climate change
Maximum temperature in the Chamoli area has increased at the rate of 0.032 oC per year between 1980 and 2018 which is statistically significant at 99.9% confidence level, compared to minimum temperature which has increased at 0.024 oC per year at 90% confidence level (Fig. 8).The trends were analysed using Mann–Kendall (MK) and Sen slope estimation (Kendall 1975; Mann 1945; Sen 1968). Furthermore, January 2021 was the warmest January on record in Uttarakhand for six decades. Note also the dramatic reduction in snow cover between 6 and 10 February as shown in Figure 4. While a hazard event like the flood at Chamoli cannot be directly attributed to climatic changes, the increased thaw-freeze cycle of permafrost could have partially contributed to the event.
iii. Other factors
At the same headwall, a large ice avalanche was previously released somewhere between 19 September and 9 October 2016, which deposited ~1.5 * 107 m3 of ice and more bedrock in the valley below (Figure 9). The resulting destabilization of the rock due to the lack of ice cover (glacial debuttressing, stress-release fracturing), and increased exposure to solar radiation and hence an increased freeze thaw cycle, in combination with a large snowfall event preceding the event of 7 February 2021 and rapid melt water production, may have favoured the fracturing of rock. This can however not explain the depth of the fracture (~150 m), which must have evolved over a longer period of time. Fracture zones at the runout of the rockslide visible before the event suggest that such detachments have happened at the same location previously. Permafrost thaw and frost cracking has been used to explain increased rockfall activity in the Alps (Deline et al. 2015; Gruber and Haeberli 2007); however, that generally only applies for the first ~10 m of bedrock.
As the debris moved downstream, it pushed running river water ahead of it, including water from small ponding structures along the river path. Besides, the water stored in the 60 metre long diversion dam of the Rishi Ganga Hydropower Project, including in its desilting chamber, should also have contributed to the flood water which was seen before the debris flow in the videos posted on social media.
4. Cascading impacts and environmental sustainability
The event and related debris flow/flood caused damage to four hydropower projects along the Rishi Ganga, Dhauliganga and Alaknanda river path (Table 1). The Rishi Ganga Hydropower Project (13.2 MW) near Raini village, located 14 km downstream from the impact site (Figure 10), was the first to be hit by the debris after the rockslide. The unfinished Tapovan Vishnugad Hydropower Project (520 MW) (Fig. 11), 8 km downsteam from Rishi Ganga Hydropower Project, was the second hydropower plant hit by the flood. The diversion dam of this run-of-the river type project faced massive damage from sedimentation and the dam was filled with debris, which can be seen in the remote sensing images taken before and after the event (Fig. 10 and 11).
As highlighted by the IPCC 2019 report (Hock et al. 2019), the mountainous regions are exposed to many cryosphere-related hazards. The frequency, magnitude and areas of these hazards are projected to change as the cryosphere continues to decline. The escalation of cascading hazards to a cascading disaster is a common phenomenon observed in the Hindu Kush Himalaya region (Cutter 2018; Vaidya et al. 2019). One of the prominent recent examples is the Uttarakhand flood of 2013, which started with heavy rainfall and caused a chain of events including landslides, flash floods, and the Chorabari lake outburst and debris flow, which killed more than 6,000 people and damaged roads, bridges, and buildings (Allen et al. 2016; Ray et al. 2016). Similar hazard events or in combination with other geophysical processes can damage several hydropower stations, which can be further exacerbated with future floods in the context of global climate change (Nie et al. 2021). Hock et al (2019) also suggest that snow avalanches involving wet snow even in winter will occur more frequently in the mountainous regions. Nie et al. (2021) reported 105 existing hydropower projects (HP) (≥ 40 MW) with an installed capacity of 37 GW, 61 projects (≥40 MW) currently under construction (39 GW) and 890 projects (≥10 MW) in various stages of planning (242 GW) in the Karakoram-Himalaya region. Most of the existing hydropower projects were built in the past two to three decades, mainly starting from the downstream sections. Now these projects are gradually moving upstream where the exposure to mountain hazards is high, the chances of multiple hazards happening in combination and occurring more frequently, and cascading effects can create compounding impacts on the system. Hydropower projects are particularly at risk because of the proximity of their infrastructure (such as diversion dams/reservoirs) to the river network where water-related hazards occur (Kumar and Katoch 2016). Many hydropower projects have been damaged by events like the Chamoli flood. For example, the Dig Tsho 1985 GLOF event in the Everest region, the 2015 earthquake, the 2014 Jure landslides, the 2013 Uttarkhand flood, and the 2016 Bhote Koshi GLOF (Nepal) damaged hydropower plants in Nepal and India (Vaidya et al. 2021). Conversely, hydropower infrastructure also impacts the local environment, causing changes in natural flow regimes and environmental flows, alteration of aquatic ecosystems, and deterioration of water quality, among others.
5. Conclusion and recommendations
The rockslide-triggered flash flood in Chamoli is one of many possible hazards in the HKH mountains. Mountain hazards like glacial lake outburst floods, torrential floods, debris flows, landslides, and avalanches, especially caused by the coupling of avalanches, glacier movement, snow melt, and extreme precipitation are common in this region. While this event cannot be directly attributed to climate change, it is well known that climate change can lead to increase in the frequency and severity of mountain hazards (Krishnan et al. 2019; Vaidya et al. 2019; Hock et al. 2019). It is necessary to carry out quantitative studies on the status of mountains, understand their formation mechanism, and monitor dynamic processes in order to have advance knowledge of impending hazard events and improve preparedness. These should be done through ground based research, analysis of geospatial information, and modelling. All these need sustained investments from national agencies including establishment of environmental monitoring, analysis and information systems. Collaborative efforts between institutions within the region and with international institutions can help in building robust systems and capacity within the region. The HKH is a multi-hazard environment. Often these hazards are of a cascading nature with multiple hazards interconnected with a primary hazard trigger and a chain of secondary and tertiary hazards. Human interference in the mountain environment is rapidly increasing. Mountain settlements are increasing in size and land use patterns are changing. Infrastructure such as roads and hydropower projects are rapidly penetrating mountain landscapes. The interplay between natural hazards with human settlements and infrastructure is an important aspect, which can significantly escalate the impacts of event like the Chamoli flood. Disaster risk management therefore needs to incorporate a multi-hazard risk assessment approach. In the aftermath of recent disaster events, the role of infrastructure, especially hydropower and its interplay with natural hazards has emerged as a topic of strong debate. These events have raised the question: Is hydropower a boon or bane? With the need to green the energy sector and the challenges with solar and wind energy, hydropower seemed to be a viable option. However, hydropower development faces multiple challenges. Apart from financial and technical challenges, it faces strong environmental and social challenges. On the environmental front, hydropower development impacts environmental flows, water quality, and the health of aquatic and terrestrial ecosystems. At the same time the physical environment poses many challenges to hydropower development and sustainability. Climate change related flow variations, extreme events, erosion and sedimentation, and GLOF/LDOFs, are some of the environmental challenges to hydropower. A comprehensive sustainability framework considering financial, environmental and social sustainability can help make hydropower a viable energy option. Vaidya et al. (2021) argue that for the sustainability of hydropower in the HKH region, environmental threats need to be minimized by mitigating risk through both structural (e.g. erosion protection work) and non-structural measures (e.g. operating rules). Besides this, mitigating the risk of climate change and flow variability is of paramount importance for future energy security for which a better understanding of future climate projections and water availability is needed. That understanding can be reflected in the design and location consideration of future hydropower projects in the region.