Watching the Himalaya: Can Science Detect the Next Disaster?
Nature Insights Interview: Jakob Steiner
Nature Insights Desk
We know from news coverage and social media that the event involved a slope failure and a debris flow that moved through the valley, gathering sediment, rocks, and water and destroying roads, settlements, and other infrastructure. We would like to go beyond what is already known and understand the event from a more technical and forward-looking perspective. Could you reconstruct the initial event that triggered the slope failure and debris flow, so our readers can understand what happened from an expert's perspective?
Jakob Steiner
You used the right terminology, which is important because there has been considerable confusion around this event. That is not unusual after a disaster, especially when many experts are speaking at once and geopolitical tensions are also shaping the discussion. There has also been misinformation, including misleading or AI-generated videos circulating on social media and later appearing on television.
The key point is that this was a slope failure, or more specifically, a rock failure. At the beginning, much of the attention focused on the glacier, but the glacier is secondary to the main process. The mountain slope failed, and a glacier happened to be sitting on top of it. The slope might still have failed without the glacier, although the resulting debris flow would probably have been different because there would have been less ice and therefore less additional water in the moving mass.
The physical force of the collapse transformed the ice very quickly as it fell over a large vertical distance, and the moving mass also picked up water along the way. That distinction matters when we discuss climate change and future risk: the primary failure was in the mountain itself, not in a glacier or a glacial lake. The slope describes the topography; the rock was the material that failed and carried the ice with it.
The failure occurred at about 5,200 metres elevation. There is still discussion about exactly how many millions of cubic metres of material came down. Another concern is that more material could fail in the future because so much was removed and destabilised during this event.
The very large elevation difference was also crucial. Material fell from high on the mountain toward a valley at roughly 3,000 metres, with peaks of around 7,000 metres nearby. That generated enormous kinetic energy. Some early reporting suggested that an earthquake had occurred, but there was no earthquake. The movement of the mass itself was strong enough to be recorded as a seismic signal.
Similar rock failures can occur in the Alps, the Rocky Mountains, or the Chittagong Hill Tracts, but the Himalayan setting is different because the elevation differences are so extreme and there can be large volumes of loose material available in the valley. In a lower-relief landscape, a failure may not travel nearly as far.
There is also important recent history in this valley system. In 2025, in a neighbouring valley, a glacial lake outburst flood drained a large lake, caused a major flood, killed more than 20 people near the border, damaged infrastructure, and deposited large amounts of sediment. Much of that loose sediment remained in the landscape. When this year's flow arrived, that material was remobilised, which helped the debris flow grow as it moved downstream. The flow ultimately travelled for more than 150 kilometres, which is extreme.
Nature Insights Desk
There was another similar event in Sikkim in 2023. How did that event compare with this one?
Jakob Steiner
The Sikkim event was different because it involved a lake. In terms of the resulting flood and its impacts, there are similarities, although the Sikkim event was smaller than what we have seen in Nepal and was still devastating.
It was also politically significant because of hydropower development in the area. Sikkim already had an active civil society and a stronger political disconnect with New Delhi than Uttarakhand. After the Nepal event, civil society groups again went to New Delhi and argued that hydropower development should be reconsidered, especially because projects have continued even after the South Lhonak Lake and Teesta flood in Sikkim. Their argument is essentially: we have already seen what happened here, and now we have seen what happened in Nepal; this can happen again.
Nature Insights Desk
Hydropower dams can disrupt the natural flow of rivers and may increase risk when they are poorly planned or managed. Why do Himalayan countries continue to build hydropower projects in these mountain regions? How significant is hydropower for their national energy supply, and how do governments weigh the benefits against the risks?
Jakob Steiner
Governments do consider these risks, but hydropower is extremely important to countries such as Nepal. Almost all of the electricity Nepal uses comes from hydropower. Solar exists, but at a much smaller scale. Nepal is therefore highly dependent on hydropower.
Mountains make hydropower comparatively easy to generate because of the elevation gradient. Water flowing downhill can drive turbines and produce electricity, and for Nepal this has been an affordable domestic energy source. For many years, foreign consultants also encouraged Nepal to develop hydropower as a form of green energy rather than rely on imported fossil-fuel-based energy. The argument was to use its own resources, become more self-sufficient, and also export electricity.
Nepal has, for example, been exporting 40 megawatts to Bangladesh. For Bangladesh that may not sound like a large amount, but for Nepal it is significant. Even a relatively small change in supply can matter when a power system is already under pressure, and additional load-shedding can quickly affect businesses and households.
On the Chinese side, hydropower development is much larger than one megawatt; we are talking about gigawatt-scale projects. But it is important not to link that automatically to this particular Nepal disaster. There were claims in some Indian reporting that the flood resulted from a dam breach on the Chinese side. That was not the case. The recent event in Nepal did not involve a dam break, and China had not built a dam at the location of the failure.
The wider political context helps explain why that narrative appeared. China is developing very large hydropower infrastructure, including major projects on the upper Brahmaputra near Arunachal Pradesh. The Brahmaputra is politically sensitive for China, India, and Bangladesh, and relations among these countries are not always good. That broader tension can easily enter the interpretation of a disaster, even when the specific event is not caused by a dam.
There are hydropower cascades under development elsewhere in the wider upper catchments, including around the Kosi, Arun, and Dudh Kosi systems, but those should not be confused with the river involved in this particular event. Hydropower development is certainly expanding on the Chinese side, partly because China is developing border regions and needs electricity for new infrastructure and wider electrification.
That transition also connects to electric mobility. Nepal has been able to shift rapidly toward electric vehicles because it has relatively cheap electricity. Many vehicles are imported from India and China. For Nepal, reducing dependence on imported fossil fuels is also strategically important because India has historically had leverage through fuel supply. After the 2015 earthquake, when Nepal moved politically closer to China, disruptions to fuel supply from India created long queues and major public frustration. That experience reinforced Nepal's desire for energy self-sufficiency.
India faces its own pressure to reduce coal and fossil-fuel use and is also investing heavily in hydropower. That helps explain why projects continue in places such as Uttarakhand and Sikkim even after deadly disasters and strong local opposition. In some areas, communities had already protested against dams because they considered the valleys unstable. The 2011 Sikkim earthquake, for example, reinforced those concerns.
The latest disaster is likely to force Nepal to reconsider some decisions because the economic and energy impacts are very large. Around 10% of its energy production was reportedly wiped out in a single day. That is a major shock for the economy. At the same time, hydropower projects involve powerful financial interests, so changing the development model is not simple.
Nature Insights Desk
Nepal has raised the issue of loss and damage. Given the possible connection with climate change, how likely is it that losses from this disaster could qualify for support through the Loss and Damage Fund?
Jakob Steiner
It is difficult for several reasons. We have been discussing this directly with policymakers in Nepal, including the Finance Minister, and we also briefed the Prime Minister ahead of his speech at the United Nations General Assembly, where the argument for greater loss-and-damage support is likely to be important.
The first problem is practical: the Loss and Damage Fund does not currently have very much money relative to the scale of need. Political willingness in the Global North to provide large amounts of finance is also limited, and political changes in donor countries can make international cooperation even more difficult.
The second issue is attribution. Nepal understandably wants to frame the disaster through climate change, but for this specific event it is very difficult to prove that climate change was the sole cause. A slope failed. Similar large failures occurred long before human-caused climate change, and we can see evidence of ancient events in the valley, including very large boulders deposited hundreds of years ago.
That means this event could, in principle, have happened without today's changing climate. However, the mountain environment is changing in ways that can make such failures more likely or more severe. The challenge is demonstrating that statistically because events at this scale are still rare. We have Chamoli in 2021, Sikkim, this Nepal event, and other major disasters such as Kedarnath in 2013, but that is still a limited sample for strong statistical attribution.
Even so, there are many indications that climate change is affecting mountain conditions. The likelihood that warming, glacier retreat, permafrost change, rainfall patterns, or increased meltwater helped precondition the slope is real. I cannot say climate change caused the event, but it may have influenced the conditions and possibly amplified the outcome.
That does not remove Nepal's responsibility to prepare. One question is whether it was wise to build hydropower infrastructure in a valley with a known history of floods. The area had already experienced serious flooding, yet preparedness remained limited. The companies developing hydropower - including Chinese, Korean, Indian, and other investors - also have responsibilities. They conduct due diligence, they are responsible for their workers, and they should account for known hazards before investing.
There is also a social question. People have lived in these valleys for generations. Many communities have Tibetan cultural and linguistic roots and have long adapted to mountain hazards. It is not enough for the government simply to say that everyone living in a dangerous area should move. We face the same problem if we tell people in the Sundarbans or the Chittagong Hill Tracts that they chose the wrong place to live. These are people's homes, histories, and livelihoods. Relocation may sometimes be necessary, but it cannot be treated as a simple technical solution.
Responsibility also extends beyond Nepal. International development banks and foreign investors finance many of these projects. If a bank provides millions of dollars for hydropower, it also has a responsibility to say when a project is too risky. If a bank approves financing, the government has a powerful incentive to proceed.
For example, the Asian Development Bank has been considering very large hydropower investments in Nepal, including projects discussed for decades. After an event like this, risk assessments need to be revisited. If previous assessments did not consider a hazard of this magnitude, the bank and its board should ask whether the assumptions were adequate. Climate finance and development finance are not automatically beneficial simply because a project is labelled green. The underlying risks, location, governance, and accountability still need scrutiny.
Nature Insights Desk
It seems difficult to identify a single immediate trigger. Some reporting suggests that several preconditions were already in place before the slope failed. Is that a fair way to understand the event?
Jakob Steiner
Exactly. Those are the right words: there were preconditions rather than one simple, isolated trigger.
Nature Insights Desk
So the slope failure may have resulted from a combination of preconditions, including climatic changes, human decisions, and natural processes. Would it be inaccurate to attribute the event solely to climate change, government responsibility, or human intervention?
Jakob Steiner
A number of factors were involved, and responsibility is distributed across different actors. Scientists also have to reflect on whether we communicated the existing danger clearly enough. The media have a responsibility to report accurately as well. There are natural, climatic, institutional, and communication dimensions to the event, so reducing it to one cause would be misleading.
Nature Insights Desk
From your research and perspective, what were the main factors that may have contributed to the slope failure?
Jakob Steiner
Let us start with the non-climatic factors. Mountains are geologically active: they move slowly, they erode, and they fracture. The Himalaya is still uplifting, so cracking and slope adjustment are natural parts of the mountain system.
Nature Insights Desk
So the mountain will continue to change over time.
Jakob Steiner
Exactly. Earthquakes are another factor. In 2015, a major earthquake strongly affected this mountain region. Around 300 people died on the other side of the mountain when a large mass of ice fell onto a village. On a geological timescale, 2015 was very recent. The earthquake may have dislodged or weakened parts of the mountain that remained in place for years before eventually failing. A slope does not necessarily collapse immediately after it is destabilized.
On the climate side, we have much clearer evidence of change in the ground itself. Permafrost - ground that remains frozen for long periods - is warming. Ten or fifteen years ago, we often assumed that ground at around 5,000 metres stayed below zero degrees almost continuously. Measurements from this mountain region since 2014 show that this is no longer always the case. The atmosphere has warmed, and the ground has warmed with it.
As frozen ground warms, it begins to thaw and becomes more mechanically unstable. Instead of remaining a consistently frozen block, it moves through repeated freeze-thaw cycles. Temperatures drop below zero in winter and rise above zero in summer. Water enters cracks, freezes, expands, thaws, and freezes again. Repeated cycles gradually weaken rock and soil. The process is similar to repeatedly freezing and thawing a material until its structure begins to break down.
We know there was a crack in the slope before the failure. It was visible weeks or months before the collapse, and it may have been developing for much longer. As cracks widen, water can enter them. If that water freezes, it can exert additional pressure and further weaken the rock. Eventually the slope may reach a point where it can no longer remain stable. This is one part of the event that carries a clear climate-change signal because warming strongly affects freeze-thaw and permafrost conditions.
A second factor is glacier retreat. Our colleagues from Nepal have mapped the glacier since the 1990s using satellite imagery. It was more than half a kilometre longer around 1990 than it is now. As the glacier retreats, rock that was previously covered and partly protected by ice becomes exposed directly to the atmosphere. It then experiences temperature changes, rainfall, and other weathering processes more intensely.
Snow cover also matters. In the past, snow remained on the mountain for longer periods and covered the glacier for longer, reducing melt. With shorter-lasting snow cover and warmer conditions, more meltwater can be generated. That water can enter fractures and become another piece of the destabilisation process.
So the possible causes range from long-term geological adjustment and the legacy of earthquakes to permafrost thaw, freeze-thaw cycles, glacier retreat, changing snow cover, and increased meltwater. We cannot say that one of these alone caused the collapse, but together they can weaken a mountain over time.
We see smaller slope failures frequently, although most are not noticed because they do not affect settlements or infrastructure. There is clearer evidence that landslides are increasing in parts of the region, including in relation to changing rainfall patterns. What we do not yet have is enough evidence to say that extremely large failures of this exact type are becoming more frequent, because the events are too rare. The same caution applies to glacial lake events. We also have an observation bias: today satellites, phones, and social media allow us to record many events that would have gone undocumented a century ago.
Nature Insights Desk
Rising temperatures are contributing to glacier melt, and this is mainly linked to the broader warming trend. Is that correct?
Jakob Steiner
Yes.
Nature Insights Desk
Does this also increase future risks as glaciers retreat and previously ice-covered rock becomes exposed?
Jakob Steiner
Exactly. There is very strong evidence that glacier retreat and changing mountain conditions are already occurring and, in many places, accelerating. Monitoring remains limited in some locations, but remote sensing provides increasingly strong evidence of these long-term changes and the hazards associated with newly exposed and destabilising terrain.
Nature Insights Desk
We cannot reverse climate change in a year or two, and events of this scale cannot necessarily be prevented. What can be done to reduce loss of life and damage? Nepal already has river-monitoring systems, yet this event was not anticipated in time. Is it possible to identify warning signs early enough, and what kind of monitoring and warning system is needed?
Jakob Steiner
The first point is that we cannot prevent a mountain of this scale from failing. If the slope collapses, we cannot simply install a wall or a net strong enough to stop it. The realistic objective is to reduce exposure, improve warning, and be prepared to act.
There are two levels of warning. The first is rapid warning after a failure begins. When this mountain collapsed at about 8:37 in the morning, we later saw that the event had been recorded by seismic sensors because the impact was large enough to resemble an earthquake signal. From the initial collapse to the flood reaching the first infrastructure near the China-Nepal border, there were only about seven minutes.
Seven minutes is extremely short. Someone has to detect the signal, determine that it is a mass movement rather than an earthquake, understand what it may produce, decide whether to issue an alert, and then communicate that alert. False warnings are also dangerous because repeated false alarms can erode trust. For locations immediately below a high-risk slope, there may simply not be enough time for a conventional warning system to save everyone.
However, people farther downstream could have been warned. We know of individuals who understood that a flood was coming and tried to alert others. A school teacher with hundreds of students reportedly received information, recognised the danger, and acted. Workers also tried to warn people inside hydropower tunnels. Informal warning networks existed through phone calls, WhatsApp, relatives, construction sites, and people physically observing the valley. Communities have always had ways to pass warnings downstream, even before mobile phones.
The problem was that these warnings were not integrated into a systematic official system. Once the flood reached the border and its scale became clear, there should have been a way to trigger a central alert that quickly reached downstream communities. A 'red button' concept is simple, but the difficult part is establishing who is authorised to use it, what evidence is required, and how to minimise false alarms.
A rapid alert could not have prevented every death or saved infrastructure already in the flow path, but it could have reduced loss of life farther downstream and given families time to move themselves and, where possible, their livestock. For many households, losing animals can destroy their livelihood even if family members survive.
The monitoring system also has to be designed for the hazard. A sensor placed directly in the river may simply be washed away. Seismic sensors, satellite observations, upstream visual reports, and other independent signals can provide additional information. But technology alone is not enough. Agencies need people who are watching the data, a clear chain of command, and a communication system capable of reaching the population immediately. In this case, SMS alerts were reportedly sent only after the flood had already arrived in some areas, which was too late.
Mass mobile alerts are technically possible. Many countries can now send disaster warnings to all phones connected to a network. But early warning only works if it leads to early action. If you tell someone a flood will arrive in five minutes, they also need to know what to do: move uphill, move away from the river, avoid a particular slope, or follow another route. A warning without clear action can create confusion.
That is why impact-based forecasting is important. Communication has to reflect local conditions. What language do people speak? Can they read an SMS? Do they have mobile connectivity? Do they understand where the safe area is? A technically perfect warning can fail if it does not match the social reality of the community.
The second level is longer-term monitoring before a collapse. We now know, looking back at satellite imagery, that the slope had been changing for weeks or months before the event. If we had known which mountain to watch, we might have identified the change earlier and treated the area as high risk during the dangerous part of the season.
That still would not allow us to predict the exact day or hour of failure. But we do know that hazard levels vary seasonally. Warm periods, high meltwater availability, and intense rainfall can increase risk. That information can guide temporary restrictions, evacuation planning, tourism advice, and closer monitoring. The goal is not to claim that we can predict every collapse; it is to identify where risk is rising and prepare before the final failure occurs.
Nature Insights Desk
Did the local authorities have any indication before the event that conditions in this area were changing?
Jakob Steiner
They could potentially have identified changes, but they did not have a system that continuously highlighted this slope. After the event, looking back at satellite imagery, the signs are easier to see. The images existed before the collapse, but no agency has enough people to examine every satellite image of every mountain all the time. Most slopes do not fail, and the Himalayan region is vast.
What we need is a system that directs human attention toward places where something unusual is happening. This is where automated analysis and AI can help. A computer can repeatedly scan imagery and flag changes that deserve closer examination.
We already use this approach for glacial lakes, where the problem is easier. A model can compare satellite images from different dates, map lakes, and identify lakes that are growing or changing shape unusually quickly. Instead of asking an expert to manually inspect every lake, the system can flag a small number as 'red' and tell analysts where to look more carefully. We are working with the Nepal government on approaches like this.
The final decision still requires expert judgement. A computer may exaggerate a change or mistake something else for a hazard. Its role is to narrow the search and say: pay attention here.
For this Nepal event, the challenge was much harder because there was no large lake; there was a crack developing in a mountain slope. Cracks form in many places and most do not produce catastrophic failures. Identifying which crack matters is therefore difficult. Still, the event gives us clues about what indicators we should monitor in the future.
The same principle can be applied to landslides in places such as the Chittagong Hill Tracts. Intense rain often triggers landslides, but dangerous slopes may already have been moving slowly in the preceding months. Remote sensing can identify areas with unusual ground movement, allowing agencies to focus monitoring and preparedness on those locations.
The remaining challenge is institutional capacity. Disaster-management authorities need staff who understand remote sensing, can run and interpret these systems, and can connect technical results to warnings and action. Technically, we are moving in that direction; institutionally, there is still work to do.
Nature Insights Desk
From what you are saying, local knowledge is essential to impact-based forecasting and early action. In places such as the Chittagong Hill Tracts, people need warnings that reflect the local language, terrain, and social context so they can act quickly. Should local communities therefore be directly involved in both understanding risks and designing warning systems?
Jakob Steiner
Yes, and local communities should not only receive information; we should also ask what they already know. People who live with these hazards often understand local conditions better than outside experts. Early warning therefore has to be a two-way exchange.
An agency in Dhaka may have satellite data and forecasting technology, but it may not understand the local terrain, language, or everyday warning signs. A community may have detailed local knowledge but lack access to remote sensing or regional forecasts. The strongest system combines both.
That requires coordination between national agencies, local government, and communities. It is still a major challenge because higher-level officials do not always listen to local people, while local communities may distrust instructions from distant authorities. These tensions become even more complicated across national borders.
Nature Insights Desk
Ignoring local and Indigenous knowledge can therefore make the situation more dangerous.
Jakob Steiner
Exactly. Ignoring local knowledge increases mistrust. Once communities stop trusting government warnings, even a technically good early-warning system becomes much less effective.
After a disaster, that mistrust can become stronger because people ask why no warning or preparedness existed, especially if the same area had already experienced flooding. Disaster management is therefore not only a technical problem; it is also a social and governance problem.
Nature Insights Desk
Many natural and climate-related hazards in the Hindu Kush Himalaya have transboundary consequences. Bangladesh lies downstream of the Himalayan river systems. Could an event of this kind directly affect Bangladesh? If so, what kinds of impacts should Bangladesh prepare for, and what precautions would be most important?
Jakob Steiner
A mountain failure of this type is very unlikely to send a destructive surge all the way to Bangladesh at the same scale seen in the upstream valley. The river system becomes much larger downstream, so the volume produced by one mountain failure becomes small relative to the discharge of rivers such as the Padma or Meghna. You would not expect a ten-metre surge from a single mountain collapse to travel intact into Bangladesh.
That means Bangladesh does not need to prepare for this hazard as if communities had only seconds to escape a mountain-generated flood wave. The more important risks are the longer-term and cascading consequences.
Energy is one example. If Bangladesh becomes more dependent on electricity imported from Nepal, and a disaster destroys a significant part of Nepal's hydropower capacity, Bangladesh can feel that impact through the energy system even though the flood itself never reaches the country as a dangerous surge. Cascading risks can therefore cross borders through infrastructure, trade, and resource dependence as well as through rivers.
Sediment is another issue. Large upstream failures can mobilise enormous amounts of sediment. Some of that material moves downstream and can be deposited in riverbeds, including in systems connected to the Teesta, Padma, or Meghna. If a flood deposits enough sediment to raise a riverbed, even by part of a metre, a later flood of the same volume can produce a higher water level because the channel has less capacity. Bangladesh is, of course, a highly dynamic delta and much sediment continue toward the ocean, so this does not mean that every upstream event simply piles sediment onto the country. But changes in sediment transport can still influence river morphology and flood risk.
There is also an important governance dimension. During the Sikkim flood, Bangladesh played a valuable role in scientific collaboration because some of the useful downstream monitoring data available to researchers came from Bangladesh. India shared very little official data, even though the event occurred within India. That made independent analysis difficult.
There are political reasons for this. Governments may fear criticism if data reveal weaknesses in preparedness or dam management. In the Sikkim case, there was public anger about the absence of warning and questions about the operation of the Teesta III hydropower project. These sensitivities can make agencies reluctant to release information.
Bangladesh can lead by taking a more open approach. Sharing environmental and hydrological data with Bangladeshi scientists, neighbouring researchers, and international scientific bodies helps build an evidence-based understanding of regional hazards. It also strengthens the country's credibility in environmental diplomacy.
These hazards do not stop at political borders. Open data and scientific cooperation are therefore not simply academic issues; they are part of regional risk reduction.
Nature Insights Desk
In the Nepal case, there were also concerns about limited data and information sharing from China. Without assigning blame for this specific event, what responsibilities do upstream countries have when hazards involve shared transboundary rivers? Is there also a responsibility to cooperate after a disaster and share information about losses and risks?
Jakob Steiner
There is some information sharing, but it is not always transparent. In this specific event, Nepal initially criticised China for not providing a warning, but the timing and geography make that criticism difficult to sustain. The slope failure originated in Nepal, the moving mass crossed into China, and the resulting flood then moved back toward Nepal. There were only about seven minutes before the first major impacts near the border.
Nepal had access to the same broad satellite observations that China could have used. There were also very few settlements in the immediate upstream valley and limited river monitoring because the river was about to cross the border. In that sense, China could not realistically have provided a conventional early warning that would have prevented the initial disaster.
However, after the event China became relatively tight-lipped, and that lack of transparency does not help. Part of the reaction may have been driven by the blame game, but the larger regional problem remains: upstream and downstream countries often accuse one another of withholding data. India criticises China; Pakistan criticises India; Bangladesh has similar concerns with India. The same pattern can even occur within a country. In Pakistan, for example, downstream Sindh has longstanding disputes with upstream Punjab over water and information.
Blame alone does not solve the problem. What matters is building mechanisms for routine information sharing before, during, and after disasters.
There was at least one encouraging development after the Nepal flood. Nepal and China held a bilateral meeting near the border. The Nepali border post at the lower elevation had been destroyed, while the Chinese facility on higher ground was largely unaffected. They discussed relocating the border function to a safer place farther upstream, including arrangements that would require unusually close cross-border cooperation. From a physical-risk perspective, that makes sense, and geopolitically it is significant.
Environmental risk can sometimes create opportunities for cooperation because both sides recognise that the hazard does not respect the border. Scientific collaboration can play a similar role. During a period of severe India-Pakistan tension, I was at a regional remote-sensing workshop in Kathmandu with researchers from across South Asia. Even when political conflict escalated and travel became difficult, Indian and Pakistani scientists continued working together. That kind of professional exchange is important because regional hazards require knowledge to move across borders even when politics is strained.
Nature Insights Desk
There are many dams on transboundary rivers across the region. To what extent can these dams contribute to, amplify, or reduce the impacts of future floods and similar disasters?
Jakob Steiner
Dams can either reduce or increase flood risk depending on how they are designed and operated. If a reservoir has sufficient empty storage before a flood arrives, it can capture part of the flood wave. But if a reservoir is already full or is poorly managed, a major flood can create additional danger. The Sikkim experience illustrates that problem.
Hydropower projects therefore need to be designed and operated with extreme events in mind. Downstream countries also need timely information about the condition and operation of upstream dams so they can understand the additional risk.
The issue is not limited to sudden floods. Dams and diversions also affect long-term water availability, river flow, sediment, fish, aquatic plants, and water temperature. Those changes can have far-reaching consequences downstream.
That is why river treaties and formal information-sharing mechanisms matter. The Ganges agreement between India and Bangladesh and the Indus arrangements in South Asia are examples of why institutional mechanisms are important. In other parts of the region, including some China-Nepal basins, cooperation is more informal and has not always worked well. Transparency about what is happening in shared rivers is essential.
Nature Insights Desk
I would like to return to the slope failure. The debris flow became larger as it picked up sediment, rocks, and boulders. But where did the large volume of water in the flood come from? There was no major rainfall at the time, so was the water mainly from glacier ice and the river, or were other sources involved?
Jakob Steiner
That is one of the remaining scientific puzzle pieces. We saw something similar in Chamoli, Uttarakhand, in 2021: a relatively small glacier and a large rock mass collapsed, there was no obvious lake, yet a very large flood was produced.
The glacier contributes water, but the ice alone does not explain the whole flood volume. One source is the water already flowing in the river. The debris flow travels much faster than the normal river flow, so it effectively gathers and pushes the river water ahead of it. In the upper catchment, discharge may initially be only a few cubic metres per second, but farther downstream the river is carrying much more water. As the moving mass travels along the channel, it incorporates that water.
A second source is water stored in the soil and shallow ground. Even if it was not raining heavily on the day of the disaster, preceding rainfall can leave the soil saturated. In August, the landscape can behave like a wet sponge. When a powerful debris flow erodes that material, water stored in the soil is released into the moving mass.
There is also hidden water stored within fractured mountain rock. From the outside, a mountain may look dry, but fractured rock contains many small voids and pathways that can hold water. When a major collapse tears through the landscape, some of that stored groundwater can also be mobilised.
So the flood water likely came from several sources: glacier ice, existing river water, saturated soil, and groundwater or other hidden storage within the mountain. We are still trying to understand the relative contribution of each source.
One way to do that is through modelling. Researchers can vary assumptions about how much water was stored in the soil, river, glacier, and subsurface, then test whether the simulated flow travels as far and behaves as the observed event did. That can help estimate which sources were most important, although there will probably always be some uncertainty.
This is why scientists initially look for a lake when a very large flood appears without major rainfall: a lake provides an obvious explanation for a sudden water volume. There have also been cases of water pockets within glaciers. Some people have suggested that possibility here, but it appears unlikely because the glacier is so steep that a large internal water pocket would probably not remain stable for long.
Several research groups are currently reconstructing the event, and more precise estimates should emerge as the modelling and measurements develop. Scientifically, explaining the water balance is one of the more difficult parts of the story.
Nature Insights Desk
Hearing directly from someone working on the geology and geography of the region will make this interview especially valuable. We plan to publish it through Nature Insights. Thank you very much for your time.
Jakob Steiner
You are most welcome.
About the Guest: Jakob Steiner is a geoscientist specializing in high-mountain hydrology, cryosphere change, mountain hazards, and climate-related risks. He is a researcher at the University of Graz, a visiting scientist at UNU-EHS/GLOMOS, a fellow of the Asian Mountain Academic Alliance, and a senior environmental engineer at hydrosolutions. Previously associated with ICIMOD, his work has focused extensively on the Hindu Kush–Himalaya region, including Nepal, Pakistan, and Bangladesh. His research connects changes in glaciers, snow and mountain water systems with downstream risks, vulnerability, water resources, and wider social and political challenges.
Comments