Since 2022, the Stimson Center has been working on a project that brings together researchers, disaster-management authorities, and partners in Nepal to develop new approaches for monitoring fast-changing hazards and improving preparedness as climate change reshapes the Himalayan risk landscape. Learn more about the Early Warning Systems for Cascading Disasters in Nepal project.

What Happened 

On the morning of August 26, 2026, a glacial collapse occurred in central Nepal on the north side of the Langtang range near the border of China, setting in motion what is now the deadliest extreme flow disaster on record in Nepal. This event was the result of a complex chain of cascading hazards. To begin with, a large mass of bedrock and glacial ice with an estimated volume of almost 100 million cubic meters detached from the mountainside at an elevation of around 5200 masl, triggering a rock-ice avalanche flow that descended over 2000 meters down a steep slope directly into the Lhende River on the border with China. When this avalanche hit the river level, it sloshed against the far bank and turned west and southwest, a wall of material that pushed the river ahead of it, gathering more and more water and eroding the river banks as it gathered force. Earlier theories focused on the ways this flow would have created a massive debris deposit that temporarily blocked the river, gathering water behind it; but scientific consensus is now that this damming effect was limited and the flow was largely continuous, traveling at an average of ~160 kilometers per hour (~100 mph) downstream toward the Nepal-China border. 

This flood wave first hit the Chinese border facilities at Rasuwagadhi just 7.5 minutes after the cascade began (this moment of impact has been tragically captured on CCTV cameras), causing intense destruction and casualties, before it flowed across the border into Nepal. The flood then ran down the Bhotekoshi/Trishuli River corridor, swallowing a significant portion of settlements through Rasuwa and Nuwakot districts and recruiting sediment, rock, and other materials along the way. The flood caused significant impacts, damage, and casualties along most of the Upper Trishuli River course, and bodies were recovered almost all the way to the plains of Nepal on the India border. A one-meter flood pulse was recorded all the way down the Nepal-India border. 

At the time of writing, at 4 pm ET on August 28, over 579 deaths have been confirmed, and the death toll is expected to continue to rise as hundreds of people remain unaccounted for. A total of over 1,924 people are still reported missing (most of them in the most highly impacted areas upstream), including almost 517 foreign tourists — many of them Indian pilgrims using this overland Nepal-China crossing to visit Mount Kailash in China. Scores of security personnel, customs and immigration staff, and employees at hydropower projects along the river course were also impacted. Search and rescue operations remain ongoing in a devastated and largely impassable landscape. 

The destruction along the corridor is close to total in many places. The border facilities for both Nepal and China at Rasuwagadhi were wiped away — the massive Chinese gate (again, featured on the CCTV footage), customs facilities on the Nepal side, and the main bridge linking Nepal and China (which was also swept away by a smaller glacial flood just a year ago in July 2025). (Read more on that prior event in these two issue briefs written in July 2025 and early August 2026 respectively, available here and here.)   

Once the flood swept into Nepal, it immediately washed away the borderland town of Timure (also called Sedang by locals) and surrounding settlements on the river, as well as the 111 MW Rasuwagadhi hydropower project (built by a Nepali company working with Chinese contractors) and the Timure dry port where transboundary cargo is staged. Timure Bazaar, which has grown rapidly in the past decade as this area became a boom town on Nepal’s busiest trans-Himalayan trade route, was destroyed – with hundreds of locals and workers missing. The town of Syaphrubesi, a market hub for the region located downstream at the confluence with the Langtang River, was similarly devastated as the flow swirled around, carving off parts of the mountainside and flooding settlements on both sides of the river. The old town on the east side of the river and the new main street, with dozens of local businesses and hotels filled with tourists on the west, were both wiped off the map.  

As the debris flood continued downstream, it swept away all of the dams and bridges on the main corridor. Nineteen bridges and roughly 40 kilometers of highway were damaged, and over 430 MW of power generation capacity is offline (over 10% of the national capacity). Smaller settlements like Mailung were heavily damaged, leaving people stranded and hydropower workers trapped in tunnels.  

As the flood entered Nuwakot District, it caused extensive damage and killed many more in towns like Betrawati and Trishuli Bajaar. Video of the water rushing through this area, this far downstream, moving giant boulders and pulling multi-story buildings into the river is surreal – as an event of this scale was both unexpected and unprecedented for these communities. Damage from the flood was recorded past Galchhi, where the Trishuli River turns west, over 100 kilometers from its origin on the China side. 

The Rasuwa corridor, where the impacts were the most extreme, has been one of Nepal’s most dynamic development frontiers and, simultaneously, one of its most vulnerable. Trade routes, border facilities, and Chinese-backed hydropower expanded rapidly over the past decade, and tourism to Langtang and Kailash had recovered after the COVID-19 pandemic. Around 50,000 people live in Rasuwa district, including a large Tamang population, and many others are drawn in by the border economy. People here have lived with risk for centuries, and they have adapted. But the risks are changing; the level of exposure that communities face today is unprecedented. On a personal level, seeing images of these towns and places disappearing was deeply shocking for me. I have spent weeks and months in all of these places through Rasuwa while doing research and post-disaster response work after the 2015 earthquake. The homes and businesses of many friends and contacts were also swept away. I have tried to reach out to many contacts since the event; some have responded, but some are still out of touch. Tracking a glacier-related avalanche on the backside of the Langtang Lirung mountain is also all too familiar for me – having personally survived a massive glacier avalanche on the south side of that very same mountain during the 2015 Gorkha earthquake. Put succinctly, as I still lack the words, none of this is abstract to me, and my heart breaks for Rasuwa again.  

A Cascade of Hazards, Not a Rainfall-Triggered Event 

This disaster was the result of a complex chain of cascading and compounding hazards – rather than a single event. This was a process where different factors interacted in succession, creating something greater and more destructive than the sum of its parts. 

To briefly summarize: It began with a high-elevation slope failure, in this case a fracture of glacial materials and bedrock collapse, which generated an enormous amount of force and energy, recruited rock, sediment, and pulverized ice, and gained water content as it descended. In the Himalaya, these cascading flows accelerate sharply, and in the steep, canalized canyons of the Trishuli, the wave of water and debris remains focused and gathers energy until it can disperse on wider floodplains much farther downstream.   

Post-event aerial photograph of the source area and the deposit where the avalanche met
the Lhende River in China – shared by Chinese authorities and published in HIRISK Report,
August 28.

Critically, this was not a rainfall-triggered event. Despite occurring during the monsoon season, the disaster unfolded on a clear day. Scientists are working to assess the relationship between high-elevation temperatures, suspecting that climate change may have played a role in destabilizing the source area by weakening the glacier and bedrock around it. Other extreme flow events have been connected to climate change’s impacts on the mountain cryosphere, related to the thawing of permafrost or the weakening of glaciers and other ice forms, for example. The driver of this kind of disaster is not climate change acting alone but climate change intersecting with other processes. Glacier-related hazards are increasing as the mountain cryosphere warms; in the Himalaya, they overlap with monsoon-triggered failures and with seismic risk — the 2015 Gorkha earthquake triggered a catastrophic avalanche on the other side of this same mountain, in the Langtang Valley. It is the interaction that produces the heightened risk, and the interaction that existing monitoring frameworks are least equipped to see. 

Before and After imagery of the area where the glacial collapse occurred, showing the flowpath to the Lhende River. (Created by Prof. Hu Kaiheng, Institute of Mountain Hazards and Environment (IMHE), China; shared with Stimson team member Manish Raj Gouli).

Attribution is complex, and this is still being analyzed by world-leading experts, but it seems fair to say that a massive glacial collapse serves as yet another reminder that we need to rethink the ways we assess climate risk. Critically, looking at cascading and compounding disasters shows us how different kinds of hazard processes become entangled, creating a higher level of overall risk. 

Sadly, extreme flow events like this are becoming more familiar globally, especially in the Himalaya, and their frequency and intensity seem to be increasing. A list of the most destructive events in this genre includes: the 2021 Chamoli disaster in Uttarakhand (which also happened on a blue-sky day), the 2021 Melamchi disaster in Nepal, the 2023 South Lhonak GLOF in Sikkim, and the 2025 glacial collapse in Blatten, Switzerland. In Nepal, this is the deadliest and most destructive flow yet recorded, and its transboundary character makes it harder to assess and respond to, but it is, sadly, part of an overall pattern. Many of these events were initiated in places no one would have expected, highlighting the emergent quality of these risks and the scale of the unknown. 

Importantly, a different kind of glacial flood event occurred in this exact watershed last year on July 8, 2025. In this case, a large supraglacial lake that formed atop the Purepu Glacier drained rapidly into the same Lhende River, destroying the same border bridge at Rasuwagadhi, damaging some of the same hydropower projects, and killing 19 people. This event was shocking at the time, though on a smaller scale compared to the recent disaster, highlighting both the increasing need to monitor supraglacial lakes (which Chinese and Nepali authorities are now doing) and for better coordination (which is still not as robust as it could be). This warning signal remains important, and coupled with the August 2026 disaster (a different kind of hazard, importantly), it highlights the urgency to assess these changing risks. 

Separating Signal from Noise 

It would be wrong to read this as a failure of Nepal’s early warning systems. The source area where this tragic event started sits in a genuine blind spot — a narrow sliver of Nepali territory that drains north into China, falling between the scientific attention long focused on the south side of the Langtang range and the Chinese monitoring effort that intensified after last year’s flood. Nothing about this site, one of many glaciers in the range and the region, would have drawn attention beforehand. And the architecture for Nepal’s monitoring systems, like most, was built around gauge and station networks designed for monsoon flooding and glacial lake outbursts. Unfortunately, a sudden catastrophic flow of this speed outruns it; there is simply no time to share an alert 

The constraint is not scarcity of data so much as sparsity, and the underlying difficulty is scale: The Himalayan region offers more slopes capable of failing at this magnitude, across greater vertical relief, than anywhere else on earth. Of roughly a dozen extreme flow events in the region in the past ten years, only one — the 2023 South Lhonak outburst in Sikkim — originated in a place already known to be a risk. Every other one of these events also came from somewhere no one was watching. 

So, the work is to narrow the search – to scan for needles in haystacks using different approaches and then to monitor places where risks are identified. That means treating each event as instruction — the 2024 Thame GLOF, for example, taught us that small glacial lakes can grow fast enough to matter; the 2025 flood in this river showed that supraglacial ponds can coalesce and drain catastrophically. And then this week’s event highlights how an rock-ice avalanche into an unpopulated gorge can kill people a hundred kilometers downstream, which will inform future efforts to model similar events. Once potential hazard chains are identified, the next step is then returning to the map to find where comparable chains might be stacking up.  

The next crucial step is to initiate a monitoring program and to use various tools to try to detect changes – to pick up on other signals amid the noise. For Stimson, this means layering systems rather than perfecting one: the existing network of weather stations and river gauges is a foundation; satellite monitoring can help routinely trace changes in remote areas, a network of seismic stations can serve as a distributed filter capable of triangulating to locate a failure within seconds of its onset, and, critically, developing localized monitoring networks of humans paying close attention to changes and evolving risks is critical.  

As I have pointed out many times before, for example here, the observations of people attuned to notice changes in the landscape, like herders, mountaineers, and fishermen, are a critical component of monitoring and early warning systems. Residents who notice a crack in the mountainside, a glacial lake filling up, or a moving glacier can flag hotspots and risks long before an instrument does. Nepal’s government has invested seriously in getting alerts to people, down to local-language voice messages; the harder problem is knowing what to warn them about, and when. But two-way dialogue is critical to ensure better monitoring and better communication when an alert or warning needs to get out to locals. 

Critically, no configuration will catch everything, and rapid verification and alerting (through proper authorities) is a challenge. But even a few minutes of warning may prove the difference between an evacuation and a recovery operation. 

What the Stimson Nepal Early Warning Team Is Doing 

Our team moved into continuous operation within hours of this event, working across Kathmandu and Washington, D.C. time zones alongside the National Disaster Risk Reduction and Management Authority (NDRRMA) and the divisions of the Department of Hydrology and Meteorology (DHM). One of our colleagues in Nepal, team member Manish Raj Gouli, has been embedded at NDRRMA since the event, working closely with their technical team on impact assessment, hazard analysis, and information management. We’ve also been working alongside other experts within Nepal and the international scientific community to identify and understand the processes that shaped this event, while helping assess and map impacts. 

Screenshot of the Stimson team working virtually between Nepal and the U.S. and analyzing Planet Labs satellite imagery to assess the source and pathway of the disaster in the immediate aftermath of the disaster on August 26^th. 

The immediate priority was attribution — establishing what actually happened, fast enough to inform both response and residual-risk warnings. Working with other experts to assess satellite imagery from Sentinel-1 and Landsat and available Planet Labs imagery, we identified the avalanche source area, ruled out an outburst from a three-lake supraglacial cluster we had assessed last year, and confirmed the landslide-dam mechanism. Working with Government of Nepal colleagues, we analyzed the morphology of the flood flowpath and helped NDRRMA create early maps and imagery of impacted hotspots along the path of the disaster. Stimson then tasked commercial synthetic aperture radar provider ICEYE US to collect focused high-resolution imagery over the source and blockage sites. Our team also helped source high-resolution Planet Labs imagery focused on areas of concern in the flowpath (see below).  

Tasked high-resolution imagery from ICEYE’s synthetic aperture radar (SAR) satellite constellation.  Image: ICEYE US 

The second priority was getting usable products into decision-makers’ hands. Pre- and post-event image comparisons were delivered to NDRRMA and DHM within hours, creating information products that were used to brief the NDRRMA’s Chief Executive as well as the Home Ministry, and we helped create public messaging that was published on NDRRMA’s channels. We are coordinating with leading Nepali scientists and other internationally recognized cryosphere experts, and our team contributed to the HIRISK collective’s rapid situation report for this event (the HIRISK report was recently published on Aug 28 and is available here). As attention has now shifted to the newly formed lakes in the Lhende River and in the flowpath of the avalanche, we are working with authorities to monitor the evolution of these lakes and consider lingering risks. 

Lingering Risks, China and Nepal on High Alert 

At the time of writing, authorities in both China and Nepal are focused on new risks posed by two new lakes that have formed in the flowpath of the rock-ice avalanche that started the disaster: one right along the Lhende River and another near the glacial failure area high above in the upper catchment. Aerial imagery shared from the Chinese side shows the “barrier lake” trapped behind part of the avalanche debris deposit has not fully drained, and a second lake fed by the glacial cirque on the backside of Langtang Lirung has since been observed in satellite imagery (see below).  

Satellite imagery from Planet Labs shows two new lakes formed. Image created with surface area estimates by Stimson team member Manish Raj Gouli on August 27^th. (Imagery: Planet Labs). 

Both lakes were growing, and concerns over a renewed breach are plausible; anxiety about this possibility was high on the 27th and 28^th. Critically, several different breach scenarios are possible, ranging from a slow release of water to an outburst breach, and each would generate a different kind of flow or potential flood. That said, on the basis of the lake areas and estimated volume as well as comparable events, calculations both by our team and other experts suggest that any surge would likely be a fraction of the initial flood. While these hazards are not insignificant, it is highly unlikely that the release of water from these lakes, even in the worst-case scenario, would cause another disaster anywhere near the scale of the initial event. Authorities and scientists are concerned that smaller reactivations of the destabilized slope could occur if it rains heavily in the coming days, and will be closely observing the area.

At present, Chinese authorities are doing their best to monitor the situation and to provide precise and timely information to Nepal and the world. Chinese leadership has issued a mandate to closely monitor the situation. That said, because many Nepalis are still shocked and justifiably unsettled in the wake of this tragic event, it is difficult to control how this information is interpreted. Moreover, confusion over terms like “breach,” “burst,” and “overflowing” or “overtopping” makes public risk communication even more challenging. But current monitoring efforts by Chinese teams are both valuable and encouraging.  

As we have indicated in prior analysis, greater communication, information sharing, and collaborative analysis of transboundary risks between Nepal and China are sorely needed. There have been several other smaller trans-Himalayan flood events in recent years that have impacted Nepal – such as the Gongbatongscacuo GLOF in 2016 and the supraglacial flood last year that Stimson has covered extensively. In the wake of these past events, formal communication and coordination between government institutions remained limited. But in the immediate aftermath of this event, new channels of communication and information sharing are developing; this is both promising and necessary. This event is a major wake-up call, and authorities from the Chinese and Nepalese governments are doing their best to analyze both this disaster and the risks as they change.  

To be clear, there is almost no way that anyone could have detected this event prior – it happened in a blind spot. Despite much debate about why there was no prior warning and how alerts could have been generated, this event was extremely difficult to detect. No one, in China or Nepal or within the international scientific community monitoring such risks (including our team at Stimson), was able to identify this site as an emergent risk. No one held back any information. The speed and intensity of this event were overwhelming for all.  

All that said, there remains a need to improve systems and coordination overall because many other risks remain (eg., high-risk glacial lakes) in this transboundary basin and others. Critically, the monsoon season is not yet over, so both government authorities and partners providing technical support, like Stimson, need to remain vigilant in monitoring other sites and assessing both risks and sources of uncertainty as they continue to evolve.  

Perhaps this event will catalyze new and more effective forms of transboundary cooperation and action to enable more comprehensive monitoring and alerting systems in this transboundary basin and in others that flow into Nepal. That is what we must hope for and what is necessary given rapidly evolving patterns of disaster and climate risk in the Himalayan region.

Acknowledgement

Thank you to the Stimson Center team working to provide post-disaster analysis and technical support in Nepal: Manish Raj Gouli, Dipesh Pokharel, Jeevika Khadka, Regan Kwan, and Brian Eyler. Appreciation is also due to collaborators who helped inform this work: Dr. Basanta Raj Adhikari, Dr. Jakob Steiner, Dr. Simon Allen, Dr. Kristen Cook, and Abiral Khatri.

The Stimson Center’s work on disaster and climate risk management in Nepal is made possible by support from the Margaret A. Cargill Philanthropies and The McConnell Foundation.