Cascading hazards and the limits of transboundary preparedness in the Himalayas  

The Bhote Koshi flood was not a conventional glacial lake outburst. Understanding what happened — and preparing for what comes next — will require more than any one discipline, or any one country, can offer.  

Shortly after 9am on August 26, 2026, a mass of ice, rock and debris came down into the Lende Khola, a small tributary of the Bhote Koshi on the Nepal–China border. What followed was not a flood in any ordinary sense. A surge of water, sediment and boulders swept downstream into the Trishuli, and at Galchchi, far downstream, the river rose by as much as nine metres in half an hour. As of August 29, 626 people had been confirmed dead across Nepal’s Rasuwa district and Gyirong county in Tibet, while more than 3,000 others remained missing.  

The disaster offers an important lesson in understanding the changing nature of hazards in the Himalayas. What we are seeing is not a conventional glacial lake outburst flood (GLOF) — the familiar event in which a lake held back by ice or loose moraine breaches and empties down a valley. Instead, it appears to have been a cascading mountain hazard: ice and rock break off a steep slope and fall into a river; the debris dams the river; water accumulates behind the blockage until the dam gives way, sending a moving mass of water, mud and boulders downstream. The typical warning signs — a glacial lake or heavy rainfall — are absent.  

The exact sequence of events and the trigger still need to be scientifically established. Seismic activity could have contributed, but no conclusion should be drawn without further investigation. Similar disasters have occurred elsewhere. The Chamoli disaster in the Indian Himalayas showed how the failure of ice and rock can rapidly transform into a destructive debris flow. Comparable events have been reported from the Tibetan Plateau, the Alps and the Caucasus.  

Large Himalayan populations live in narrow downstream valleys, while roads, bridges, hydropower projects and settlements are concentrated along river corridors. But exposure is not simply a question of how many people live downstream. These valleys are home to distinct languages, ritual calendars, agro-pastoral systems and place-based knowledge that can change within just a few kilometres along a river.  

When a settlement is destroyed and residents are relocated on safety grounds, what is lost cannot be captured merely by counting damaged buildings and roads. Research following the 2015 Nepal earthquake, and the subsequent avalanche in the same district as the present disaster, found that survivors did not necessarily want to relocate to safer sites; they wanted to return to the valleys of their ancestral lands and socio-cultural identity. These are what researchers describe as Non-Economic Loss and Damage. Nepal’s most recent Nationally Determined Contributions recognise such losses for the first time, yet they remain among the least documented consequences of disasters.  

Himalayan hazards also do not respect political boundaries. Nepal and China signed a memorandum of understanding on disaster risk reduction in 2019, and meteorological agencies on both sides met within hours of this event. Yet the first indication that ice had blocked a river reached Nepal’s disaster authority through the International Centre for Integrated Mountain Development (ICIMOD), a regional research body, rather than through an official bilateral channel.  

The problem, then, is not simply the absence of agreements. It is the mismatch between the speed at which institutions operate and the speed at which hazards unfold.  

Elsewhere in the Himalayas, the institutional arrangements are even thinner. India and China have no formal basin treaty for their shared rivers, relying instead on time-limited memoranda to exchange flood-season data on the Brahmaputra and Sutlej, both of which have lapsed. For the Koshi basin, which drains Tibet, Nepal and India, no basin-wide framework exists.  

There is also an asymmetry in how risk is defined. Regional assessments that map territorial disputes and conflict risks are designed to assess threats to corridors, trade routes and investment. The risks faced by people living in those same grid cells do not appear among the indicators.  

It would be a mistake, however, to view this only as a gap in scientific knowledge. The problem of science-policy communication is equally important. Transboundary assessments of the Poiqu–Bhote Koshi and Gyirong–Trishuli basins had already identified 28 glacial lakes as highly susceptible and modelled which buildings, bridges, hydropower sites and roads could be exposed. What remains unclear is the pathway through which such findings reach the institutions and stakeholders with the capacity to act.  

These hazards are also escaping the signals our systems are designed to detect. Flash-flood forecasting in the region is calibrated largely to rainfall and the monsoon calendar, and this event appears to have produced no rainfall signal at all on the Nepali side. A hazard that begins with ice detaching from a slope in clear weather, in an unmonitored catchment across a border, can remain invisible to existing systems until it is already moving downstream.  

We need to understand Himalayan hazards as chains of processes rather than isolated events. And the chain does not end with geophysical processes. It extends to what we build in these valleys — and what we choose to rebuild after every disaster.  

The Betrawati–Rasuwagadhi corridor was damaged by flooding in July 2025 and has now been damaged again, the second time in 13 months. Reconstruction that simply restores the previous alignment also restores the previous exposure. Roads, hydropower projects and tourism infrastructure have made these valleys more accessible while, in many cases, making them more fragile through the very same interventions.  

An early-warning system is itself a chain: from risk knowledge to monitoring, communication and the capacity to respond. If a warning arrives only after the water has crossed the border, as appears to have been the case here, that chain has already been broken.  

This is not an impossible problem to solve. There are precedents. After the 2021 Chamoli disaster, a team spanning remote sensing, seismology, glaciology, modelling and social science assembled within days and published a detailed reconstruction four months later. In the European Alps, the Alpine Convention’s natural hazards working group has operated as a standing, multi-country technical body since 2004. In Nepal’s western basins, community-operated flood-warning systems proved effective during the 2014 floods on the Karnali and West Rapti.  

Each example has its limitations. But they demonstrate that the scientific, institutional and community-based approaches needed to manage high-altitude and transboundary disasters already exist.  

The knowledge of scientists across Himalayan borders, officials and communities that have read these slopes for generations must be connected, interrogated and integrated. The people of the Himalayas do not need to be told that these valleys are dangerous. They have lived with that knowledge — and its consequences — for generations.  

What is needed is research and policy that cross-pollinate disciplines and transcend borders; better science to understand how cascading hazards initiate, travel and stop; communication that reaches people early enough, and in forms they can act upon; and documentation of losses in all their forms, economic and otherwise, led by the communities that bear them.  

Co-author: Nickhil Sharma, researches energy and climate justice at the School of Climate Change and Sustainability, Azim Premji University, Bengaluru.  

Figures are as of August 29, 2026, and are provisional. 

END OF ARTICLE