Share & more
Ashok Shrestha was doing what he routinely did on his journeys – eating near a riverside – when a raging wall of ice-cold water, slush, and rocks came crashing down a narrow Himalayan mountain valley and swept him away like a leaf caught in a storm.
He had driven his 12-wheel truck up from Barabise, in Nepal’s mountainous Sindhupalchok district, to collect a load from the customs point at the border with China, a trip he had been making for years.
"It swept me away and carried me downhill," he told The New Humanitarian from his hospital bed, his body marked with injuries. Shrestha swam until he reached the riverbank. A friend and another man nearby never made it out of the torrent alive.
According to preliminary geological assessments, the disaster was triggered when a vast section of glacier and surrounding bedrock, about 600 metres wide, collapsed at an elevation of about 5,200 metres, high above the Lhende Khola river on the Tibetan side of the border.
Satellite imagery suggests the mass plunged 1,200 metres to the valley floor, sending a huge cascade of ice, rock, and debris into the Lhende Khola and generating a catastrophic flood that travelled nearly 100 kilometres downstream through the Bhotekoshi river system, destroying buildings, roads, homes, and other infrastructure in its path.
The death toll from the disaster has risen above 1,000 across Nepal and Tibet, with almost 4,500 people listed as missing. Nepalese authorities have reported 987 deaths and 3,916 missing, while Chinese authorities have reported a further 16 deaths and 546 missing. The figures remain provisional as rescue and recovery operations continue in some of the worst-hit areas, which are remote and difficult to access.
Among those missing are dozens of Hindu pilgrims who had been travelling to Kailash Mansarovar, a sacred site in Tibet, many of whom were last reported at the Gyirong Port border checkpoint between China and Nepal.
Nine hydropower projects, nine bank branches, 19 road bridges and 40 kilometres of highway were destroyed or carried away, along with the Gyirong Port complex — Nepal’s principal overland gateway for trade with China. More than 300 vehicles waiting at the checkpoint were lost.
The disaster is already among the deadliest in Nepal’s modern history, with its confirmed death toll approaching that of the devastating 1993 floods, which killed 1,336 people.
The victims were largely people who lived and worked in the corridor by necessity, not choice: border and customs staff, hydropower and bank workers, and truck drivers moving goods through the Gyirong crossing, which suffered catastrophic destruction.
In several areas, bulldozers were being used to recover human and animal remains as a strong stench hung over the scene.
This wasn’t simply some freak occurrence. The same area has suffered repeated flooding and glacial disasters over the past decade, including two significant floods in July 2025, just 13 months before the latest catastrophe.
In July 2016, a moraine-dammed lake called Gongbatongsha, in Tibet’s Poiqu basin, burst after a slope failure and sent a flood into the Bhotekoshi basin, damaging the Arniko Highway and the Upper Bhotekoshi hydropower plant. Researchers found that the flood intensified dramatically as it picked up debris downstream – a process strikingly similar to the way the latest disaster gathered force.
Then, on 8 July 2025, the sudden drainage of a supraglacial lake in Tibet sent a flood down the Lhende Khola, sweeping away the Friendship Bridge at Rasuwagadhi and killing at least nine people. Just three weeks later, on 30 July, the Bhotekoshi flooded again, damaging roads and infrastructure in the same corridor.
Each event may have had its own trigger, but together they point to a region exposed to recurring and well-documented hazards. The question is not simply whether the latest disaster could have been predicted, but whether enough had been learned from the previous ones to warn people in time.
A failed cross-border warning system
The disaster exposed a weakness on both sides of Nepal’s warning system: It had no real-time information about the hazard developing in Tibet, while the monitoring equipment it had installed inside Nepal was destroyed before it could send an alert.
Automatic flood-monitoring stations have been installed along the upper Bhotekoshi river to detect rising water levels and send alerts before floods reach populated areas downstream. But on 26 August, the surge came so quickly that the stations were destroyed, apparently before they could transmit any warning.
“None of the stations was able to send us an alert,” Binod Parajuli, a senior divisional hydrologist at Nepal’s Flood Forecasting Division, told the Kathmandu Post.
His division learned of the flood only at 8:56am, when Rasuwa’s district administration and the national disaster authority relayed reports from people on the ground – about half an hour after the water hit Timure, a settlement near the Nepal-China border. The division issued its own warning downstream within four minutes. By then, however, the flood had already swallowed large parts of Rasuwa.
“We failed,” Parajuli said.
He believes the warning still saved lives further downstream, but said the failure exposed a deeper flaw: Nepal’s sensors could not outrun a flood originating across an international border. A warning system that monitors only what happens inside Nepal, he argued, can never be enough for a hazard that begins somewhere else.
Nepal and China laid the groundwork for closer disaster cooperation in 2019, when their governments signed a memorandum of understanding covering disaster risk reduction, emergency response, information sharing, and the exchange of hydrological data. But in practice, those warning systems have failed to cross the border.
“It is up to the government to keep responding with relief and rescue after each disaster, or to start working to prevent the loss of life in the first place.”
Similar failures have happened before, with floods originating on the Chinese side reaching Nepal without advance warning. They were not taken seriously enough, Anil Pokhrel, former chief executive of Nepal’s disaster authority, told the Kathmandu Post.
“This problem will continue … It is up to the government to keep responding with relief and rescue after each disaster, or to start working to prevent the loss of life in the first place,” he said. Nepal, he added, needs a functioning cross-border system through which warnings about floods, landslides, and other hazards originating in China can reach its disaster authorities in time to act.
There are signs that officials recognise the gap. After the latest disaster, Nepal’s Department of Hydrology and Meteorology and the China Meteorological Administration held a virtual meeting at which China agreed to notify Nepal as soon as water levels begin rising on its side of the border – restating a commitment made in 2019.
But experts say a reliable warning system needs to go beyond water levels. It must also monitor unstable ice and rock, share information across borders, and give communities enough time to evacuate.
Vulnerable Himalayas
Jeffrey Kargel, a senior scientist at the Planetary Science Institute, a US space research organisation, has spent recent days comparing the Nepal disaster with an earlier Himalayan catastrophe in Chamoli, India, where a chunk of glacier broke away from a steep mountainside, mixed with rock, and became a debris flow that destroyed two hydropower projects and killed more than 200 people.
"The process going on throughout the whole thing is really almost identical, but this one was a bigger event – two to three times larger by volume, with a correspondingly higher death toll,” he told The New Humanitarian.
What worries Kargel is not one disaster, but the pattern behind them. Rock-ice avalanches, in which rock and glacier ice collapse together and turn into fast-moving debris, were until recently considered rare.
Scientists were not systematically looking for this kind of hazard across the Himalayas, where steep, unstable terrain and rapidly changing glaciers create heightened risk. Chamoli changed that. The latest disaster, Kargel said, is yet another warning that the threat needs to be taken seriously.
“We need a new look at risk, and bring the hydropower people into the same room and ask, do they really want to just build hydropower like crazy, without even thinking about the hazards?”
“It changes the risk landscape dramatically,” Kargel said, outlining how that risk could be better understood and mitigated. “Most hazard and risk mappers would look at risk due to landslides individually – slope, geology, proximity to earthquake fault zones, and so on – and calculate hazard maps and risk maps that way. Let’s look at where the glacial lakes are, what are the chances of an outburst? Then let’s do a risk map for monsoon rainfall floods, or snowmelt floods – one process at a time.”
The other challenge is working out which parts of the Himalayas – one of the world’s youngest mountain ranges, still rising today – are most vulnerable. Its steep slopes are full of fractured rock and hanging ice, but scientists still lack a reliable way to identify where a rock-ice avalanche might occur.
“So far, nobody... has really figured out how to handle the problem of rock-ice avalanches,” Kargel said.
That uncertainty raises difficult questions for a region rapidly expanding hydropower as it shifts towards cleaner energy. Hundreds of hydropower projects are reportedly underway or planned across India, China, Pakistan, Nepal and Bhutan, many in valleys exposed to landslides, floods, and other hazards.
Environmentalists and climate activists have long warned that rapid construction in this ecologically sensitive region can amplify the damage caused by floods, landslides, and other natural hazards.
“Hydropower is a way to decarbonise the economy, [but] risk mappers need to start asking if there are some valleys with such a bad history that we should just stop developing them, and withdraw?” Kargel said.
“We need a new look at risk, and bring the hydropower people – the financial backers, political backers, engineers, geologists, electrical engineers and suppliers, the economic planners – into the same room and ask, do they really want to just build hydropower like crazy, without even thinking about the hazards?”
Those decisions are being made in a corridor where people have to live and work, often because there are few alternatives.
Mina Tamang has been frantically searching for her husband Jairam Tamang. The 33-year-old had been driving trucks between the customs point and the Chinese border for over six years.
He called Mina at around 7am on Wednesday while she was having her morning tea. He told her his truck would leave the next day, that his friends were coming, and asked her to collect some chocolates that he was sending her. They spoke for half an hour until he left for the customs office. She has not heard from him since.
“I have registered his name at the hospital and with the police,” she said. “But more than 1,000 people have reportedly registered the names of their missing relatives here. I don’t know how we are going to find him.”
Mir Wafa Rasheeq provided additional reporting from New Delhi. Edited by Andrew Gully.