A massive collapse of ice and rock, a debris avalanche, a river blocked, and then a catastrophic flood. Reconstructing the dynamics of the disaster also means understanding how climate change can profoundly transform the stability of high mountain environments—without resorting to oversimplification.
On 26 August 2026, shortly before nine in the morning, a sudden catastrophe swept through mountain valleys along the border between Nepal and Tibet. Within minutes, an enormous mass of water, mud, ice, rock and debris surged through the Lhende Khola, Bhote Koshi and Trishuli river systems, destroying villages, roads, bridges, border infrastructure and hydropower facilities. The flood arrived with such force that those caught in its path had almost no time to react. In some locations, according to Nepalese authorities, the level of the Trishuli rose by as much as nine metres in around half an hour. In the days that followed, the human toll rose dramatically—and was still being updated as I wrote—while thousands of people remained missing and rescue operations continued amid vast accumulations of debris and the risk of further collapses.
In the first few hours, however, understanding what had actually happened was far from straightforward.
A seismic signal initially led to reports of a magnitude 4.4 earthquake. The explanation seemed intuitive: a Himalayan region, a seismic event, a massive landslide and then a flood. But that reconstruction was soon called into question. Subsequent analyses indicated that the recorded signal had not been produced by a tectonic earthquake that caused the mountain to collapse. Instead, the movement of the enormous mass of rock and ice itself generated a seismic signal of extraordinary intensity.
The US Geological Survey subsequently identified the event as a rapid slope collapse involving a glacier, triggering a massive debris avalanche and a sudden flood.
This correction is far more than a technical detail. It fundamentally changes how we need to interpret the catastrophe.
The sequence, as far as we can currently reconstruct it, was not: earthquake → landslide → flood.
The earthquake—or rather, the signal initially interpreted as one—was a consequence of the collapse. The more likely sequence is:
> collapse of a high-mountain sector involving both rock and ice
> debris avalanche and transformation of the material into an extremely mobile flow
> temporary blockage of the watercourse, followed by the catastrophic release of the accumulated water.
Even the phrase that rapidly spread through media outlets around the world—"a glacier collapsed"—risks being incomplete.
The first satellite images showed the detachment of a large mass of ice at high altitude near Langtang Lirung. Reuters reconstructed how the material fell around 1,200 metres towards the valley floor, generating an avalanche of ice and rock that subsequently developed into an enormous debris flow.
But later analyses of satellite imagery added another crucial element. According to a reconstruction reported by the Associated Press, it may not have been ice alone that collapsed: the failure may also have involved the underlying—or immediately adjacent—bedrock.
In other words, what most likely occurred was not simply the detachment of an unstable mass of glacier ice. We may instead be looking at a massive slope collapse in which rock and ice failed together.
The distinction is crucial because it changes the question that needs to be investigated: why might a section of mountain, composed of rock, ice and probably other materials, have lost its stability?
And this is where the relationship with climate change becomes both interesting and far more complex than early interpretations might suggest.
At present, there is no attribution study allowing us to state that climate change directly caused the collapse of 26 August. The immediate causes of the failure still need to be reconstructed in detail: the geology of the slope, the presence of fractures, thermal conditions, the possible presence of permafrost, water infiltration, glacier evolution, previous precipitation, snow conditions and other potential local factors. The USGS itself stresses that its reconstruction remains preliminary and may be updated as new data become available.
To say today that "climate change caused the disaster" would therefore mean going beyond the available evidence.
But it would be equally wrong to conclude that, because we cannot directly attribute this individual collapse to global warming, climate has no relevance at all.
Climate change may not be the immediate trigger of an event while at the same time profoundly altering the physical system in which that event becomes possible. A rock face may be gradually preparing for collapse over years or decades and then fail following rainfall, water infiltration, a small internal movement or another apparently contingent factor. The scientific challenge, therefore, is not simply to identify "the cause" on the day of the collapse, but to understand how the conditions determining the stability of the mountain have changed over time.
To understand this point, we need to widen our perspective and look at Nepal as a whole.
Just days before the Nepal disaster, Asha Bhatta and Ashish Devkota published a systematic review that is particularly useful for understanding Nepal's water cycle: Impact of climate change in Nepal across sectors from water-centric perspective: A systematic review, published in August 2026 in H2Open Journal. Using the PRISMA protocol, the study synthesises 122 scientific papers examining the relationship between climate change, water, and Nepal's economic and social sectors.
Its greatest strength lies precisely in its refusal to view the problem in fragments.
Climate change does not act separately on glaciers, rivers, agriculture and communities. All of these systems are connected through water.
The review shows, first and foremost, a clear rise in temperatures, particularly pronounced at high elevations. Nepal's Himalayan regions are experiencing warming trends generally stronger than those observed in the Terai lowlands, while minimum and winter temperatures are, in many cases, rising faster than maximum and summer temperatures. The transformation also affects extremes, not just climatic averages.
The review explicitly describes an increasing complexity in Nepal's hydroclimatic conditions, in which floods, droughts and even compound events can follow one another or interact.
This matters for understanding the Langtang disaster because Nepal is a country in which the atmosphere, glaciers, snow, rivers, sediments and mountains are all changing simultaneously. And when they change together, risks can change as well.
When we think about the warming of major mountain ranges, the first image that comes to mind is that of a retreating glacier. And rightly so. But inside and beneath high mountain slopes there is another form of ice, far less visible but equally important for landscape stability: permafrost.
Permafrost is ground, sediment or rock that remains at or below zero degrees Celsius for at least two consecutive years. In mountain environments, it can also occur inside fractures and discontinuities in rock.
This does not mean, as is sometimes claimed, that permafrost is simply "the glue holding mountains together". Reality is more complex. But ice within fractures can contribute to the mechanical properties of a rock mass, and its degradation can alter the balance between the forces that keep a slope stable and those that promote movement.
A landmark 2019 review by Annette Patton, Sara Rathburn and Denny Capps, published in Geomorphology, synthesises this issue clearly:
permafrost thaw can increase the susceptibility of slopes to landslides by altering the geotechnical properties of materials, reducing cohesion and increasing hydrological connectivity. In other words, thawing does not merely change the temperature of a mountain: it can change the way water circulates within it and the way masses of soil and rock withstand stress.
This mechanism is particularly important because a mountain can deteriorate gradually without anything dramatic being visible from the outside. Fractures may progressively widen. Water may penetrate deeper. Freeze-thaw processes can alter materials. Small rockfalls may change stress distribution. The loss of surface ice can expose rock to new thermal conditions.
Then, at some point, the system crosses a threshold. And that is when a slope collapse can occur.
In recent years, scientific research has begun to gather increasingly strong evidence that warming can alter landslide activity in high mountain environments.
One of the most interesting studies was published in 2024 in Nature Geoscience. Markus Stoffel and his colleagues reconstructed a century of rockfall activity in the Swiss Alps using, among other evidence, the growth rings of 375 trees damaged by falling rocks. Their results show a significant correlation between rockfall activity and summer temperatures at both annual and multi-decadal scales. After the mid-1980s, activity reached unprecedented levels within the historical record examined. The authors link this shift to warming and permafrost degradation, while stressing the complexity of the factors capable of triggering any individual event.
The conclusion is not: it gets warmer, therefore the mountain collapses.
The conclusion is: warming can progressively alter the conditions governing slope stability and increase the likelihood of rock masses being released.
A study published just a few weeks ago in Earth Surface Dynamics provides an even more revealing example. Focusing on the collapse of Platteikogel in the Austrian Alps, the research reconstructs how the loss of small bodies of ice attached to mountain slopes—known as ice aprons—and the warming of permafrost may contribute to the destabilisation of a rock slope. The authors combine geomorphological observations, reconstructions of thermal evolution and mechanical models of slope stability.
It is the study's final conclusion that makes it particularly useful to our discussion.
The Platteikogel collapse cannot be explained simply by saying that warming reduced the strength of ice inside fractures. The system is more complex. The loss of ice aprons promotes new thermal regimes, can increase rockfall activity and allow greater water infiltration. This, in turn, can increase hydrostatic pressure within critical zones of the slope, contributing to destabilisation. The authors therefore describe a system of feedbacks between the cryosphere, hydrology and mountain mechanics.
This is an important lesson for Nepal as well. Climate change does not necessarily have to "break" a mountain directly. It can gradually change the physical system that keeps it in equilibrium.
In 2026, Chengbin Zou, John Jansen and other researchers published a highly relevant study in National Science Review focusing on the Eastern Himalayan Syntaxis, one of the most dynamic and unstable regions in the entire Himalayan system.
The authors compiled an inventory of 8,607 climatically related landslides observed between 1987 and 2020. Their most striking finding is that, above 3,000 metres, the annual volume of landslides increased by around fivefold over three decades. The distribution of these phenomena also appears to be shifting towards higher elevations, following glacier retreat, permafrost thaw and changes in precipitation, which increasingly falls as rain rather than snow.
The authors do not argue that every individual Himalayan landslide is "caused by climate change". They argue something more important: landslide activity in Asia's great mountain ranges is changing rapidly in a changing climate.
The result is greater sediment production and mobilisation and, above all, a growing risk of what scientists call cascading mountain hazards: chains of events in which one process triggers another.
And it is difficult to imagine a better definition of what happened on 26 August.
A collapse → involving ice and rock → generating an avalanche → becoming a debris flow → accelerating down an extremely steep slope → reaching the valley in a matter of minutes and altering the river → creating a temporary blockage → suddenly releasing enormous quantities of water and sediment → transforming a geological crisis into a human catastrophe hundreds of kilometres downstream.
Another recent study, published in 2026 in Catena, helps us understand the issue further. Yuting Yang and colleagues analysed 905 landslides in the Qilian region of the Tibetan Plateau, examining the role of freeze-thaw cycles alongside the topographical characteristics of the landscape.
The findings are particularly interesting precisely because they avoid oversimplification.
Freeze-thaw cycles were not found to be the dominant factor controlling landslide occurrence. Slope steepness remains the primary topographical control. But thermal cycles are a statistically significant factor, capable of modifying the mechanical properties of the ground and subsurface hydrological conditions. The study also identifies a threshold in the frequency of annual cycles beyond which landslide susceptibility increases significantly.
The key point is that topography and thermal conditions do not operate independently.
A steep slope exposed to intense thermal fluctuations and characterised by particular hydrological conditions may respond very differently from another slope that appears, at first glance, to be similar.
Satellite imagery now suggests that the disaster began with a massive collapse involving both rock and ice. This makes it scientifically plausible to investigate the role that permafrost degradation, ice loss and hydrological changes within the slope may have played.
But plausible does not mean proven.
Determining whether permafrost played a concrete role in the Langtang collapse will require a detailed reconstruction of the slope's structure and of the conditions preceding the event. Researchers will need to understand where ice was present, how temperatures had changed, how the glacier had evolved and what hydrological conditions existed inside the mountain.
One thing, however, appears reasonably clear. As Roger Pielke Jr. forcefully argued in The Honest Broker, the disaster was preventable—not in the sense that the collapse itself could have been stopped, but in the sense that thousands of lives might have been saved through early warning.
Until January 2025, the United States funded SERVIR-Hindu Kush Himalaya, a joint USAID-NASA programme based in Kathmandu. The programme used satellite observation and geospatial analysis to reduce disaster risk. Its concrete outputs included Nepal's national monitoring system, forest-fire monitoring, flood and drought risk forecasting services, collaboration with the Nepal Red Cross on risk communication, and more than 170 public datasets covering glaciers, land cover, disasters and vulnerability.
In January 2025, the programme was cancelled following an 83% reduction in USAID funding decided by the Trump administration, while National Interest and Objectives reviews reportedly selected budget items for elimination using keyword searches that included "climate". USAID-NASA programmes dealing with flood forecasting, early warning, disaster preparedness and the mapping of dangerous glacial lakes in Nepal were identified by the search system and included in the lists of cuts.
According to Jeremy Konyndyk, who led USAID's response to Nepal's 2015 earthquake, these programmes had been specifically designed to anticipate and mitigate disasters like the one that struck in August 2026. This was not abstract research. It was operational infrastructure for disaster protection.
Even a rudimentary early-warning system could have provided—according to Himalayan civil protection experts—between five and thirty minutes of advance notice before the flood wave reached downstream villages. Thirty minutes is not enough to evacuate a city. But it can be enough to tell people in vulnerable villages to move vertically, away from the river channel and up the slopes. Hundreds of lives could have been saved.
The tragedy in Nepal therefore does not demonstrate that we should understand how climate is changing hazards in order to build prevention systems suited to these emerging risks. At the same time, we must work to reverse a trajectory that will otherwise lead to further tragedies—by reducing emissions, removing carbon from the atmosphere and finding alternatives to business as usual.
Because climate change does not alter only what we can see.
It is not simply a matter of retreating glaciers, changing river flows or warmer seasons. In high mountain environments, warming can also affect the invisible structures of the landscape. It can alter the temperature of rock. It can degrade permafrost. It can change how water behaves inside fractures. It can intensify freeze-thaw cycles. It can alter the balance between rain and snow. It can accelerate glacier retreat and expose slopes that, for centuries, had remained covered or supported by ice.
None of these processes automatically means that a mountain will collapse. But together, they can transform the system in which that mountain exists.