When mountains fall faster than warnings travel

The three priorities that deserve urgent investment as it is time to extend monitoring beyond lakes and rivers to the unstable slopes above them
When mountains fall faster than 
warnings travel
Updated on: 
4 min read

On 26 August 2026, a mass of rock and ice broke away from Langtang Lirung in Nepal and plunged about 1,200 metres into the valley below. Within minutes, a debris-laden flood struck the Gyirong–Rasuwagadhi border crossing. Nearly 679,000 SMS alerts followed, but the border corridor had already been hit. By mid-October, Nepal’s disaster authority had counted about 1,450 dead and more than 5,700 missing and 13 hydropower projects destroyed or damaged. The disaster exposed a critical weakness in Himalayan early warning: a mountain collapse can unleash a flood before river sensors give enough warning.

Two more mountain floods followed. On 11 October, a flood swept down Nepal’s Budhi Gandaki River, damaging homes, bridges and hydropower projects. Preliminary assessments point to an ice-rock avalanche or landslide-dam breach. On 29 October, an outburst from Bhutan’s Sintaphu Lake activated early warning systems and prompted downstream alerts in Bhutan and India. The Bhutan response suggests that existing systems can work; Gyirong–Rasuwa shows their limits.

Many existing flood-warning systems in Himalayan valleys monitor glacial lakes or river levels. A rock and ice collapse, however, can generate a destructive flood within minutes, leaving river monitoring little time to warn. Sensors in or near a river can be among the first casualties, falling silent as the flood arrives.

Heavy rain played a major role in some past tragedies, such as Kedarnath in 2013 and Melamchi in 2021. Others, like Chamoli in 2021 and now Gyirong–Rasuwa 2026, began with a slab of mountain giving way. ICIMOD’s preliminary synthesis lists rapid glacier retreat, fractured rocks, earthquake damage, thawing permafrost and an unusually warm year as possible contributors. The backdrop, however, is clear: the average annual rate of Himalayan glacier ice loss during 2000–2016 was nearly twice that during 1975–2000. Further warming is expected to degrade permafrost that binds steep rock faces. We cannot wait for a perfect explanation before acting.

Mohd Farooq Azam
Mohd Farooq Azam

The task now is to extend monitoring beyond lakes and rivers to the unstable slopes above them. Three priorities deserve urgent investment.

First, watch the mountains from space. Satellite radar can detect ground movement of millimetres to centimetres. ICIMOD’s analysis found up to 40 cm of movement across the failed slope in the 48 days before the collapse, using Europe’s Sentinel-1. This was seen in hindsight, and scientists debate whether it was a true precursor. But used routinely, with artificial intelligence flagging unusual movement across thousands of slopes, radar can tell us where to look, though not when a slope will fail.

Second, listen to them. Ground-motion sensors (seismometers) around the world recorded the Langtang collapse as a seismic signal equivalent to a magnitude 5.2 earthquake. Landslides, avalanches and floods all shake the ground, and that shaking travels at kilometres per second, while a flood front moves at tens of metres per second. A sensor placed outside the flow path can detect the event early and is less likely to be destroyed. A 2021 study byCook and colleagues showed that Indian seismic stations recorded the Chamoli disaster from up to 100 kilometres away. It estimated that detection could have given about half an hour of warning near Joshimath, and a few minutes closer to the source, potentially enough to shut down power plants. What is missing is a network in the right places, software that can detect and classify such events within minutes, and agreed rules for turning that into alerts. On 26 August, the ground signal existed at 8:37, while the mobile alerts went out 38 minutes later, at 9:15. Closing that gap could buy precious time for communities such as Syabrubesi and Timure.

Third, know where to look. We have listed dangerous glacial lakes because lakes are visible. We need a systematic inventory of hanging glaciers (ice perched on steep slopes) and steep slopes possibly held together by frozen ground (permafrost) above settlements and power plants. The same fractured, high-relief geology recurs along Himalayan border catchments, so the assessment should be regional, not a patchwork of national inventories.

For those closest to the source, warnings offer little help. There, safety depends on where we build border posts, bridges and hydropower plants. Because early signals may arise on one side of a border and the people at risk live on the other. Satellite, seismic and river data must flow across borders in real time. India, which lived through Chamoli, has as much at stake as any other neighbouring country.

As governments meet at COP31 in Antalya, Himalayan countries should call for adaptation finance that reflects the realities of mountain hazards. The test is whether funding reaches exposed valleys and sustains the people, institutions and cross-border cooperation that turn warnings into action.

As temperatures rise, the Himalaya will keep surprising us. The goal is to ensure the next surprise does not find us waiting on a silent gauge.

Mohd Farooq Azam is Senior Intervention Manager - cryosphere at the International Centre for Integrated Mountain Development (ICIMOD) in Nepal