Science1 publisher3 min readPublished
Nepal's flood warning reached Dhunge Bazaar 33 minutes after the avalanche and five before the water
Nepal police warned Dhunge Bazaar by loudspeaker about five minutes before the August 26 debris flow that killed at least 1,400 people struck the town. One scientist calls the avalanche likely impossible to forecast, so the time to win back is the 38 minutes before the water came, and 33 of them had passed before the warning.
The Scientist · Science desk

What happened
- Shopkeeper Prakash Pyakurel needed minutes to decode the muffled call, and his family was still arguing about leaving when the water arrived at 9:15.
- Cryosphere scientist Ashim Sattar says forecasting an avalanche of this size is likely impossible because such events are too rare to yield much data.
- Only 21 of some 40,000 glacial lakes across the Himalaya, the more forecastable hazard, have monitoring stations on the ground.
Compiled by The ScientistSomething wrong?How this is made
Why it matters
- decision A louder speaker at 9:10 would have recovered none of the 33 minutes already gone, so upstream detection and relay have the stronger claim on warning budgets than town sirens alone.
- constraint Gauges sited in the river channel can be wiped out by the flow they exist to report, so a network built only on them can fail at the very first step.
- exposure With one glacial lake in about 1,900 watched by ground stations, towns below the rest have no station signal even for the hazard scientists can see weeks ahead.
Most of the time Dhunge Bazaar had was used up before the warning began. The rock-ice avalanche broke loose at 8:37 a.m. local time [1], and the water reached Prakash Pyakurel's street at 9:15 [2]. The window was about 38 minutes [1]. The police loudspeaker sounded around 9:10 [3], roughly 33 minutes into it [2].
The first links in the chain failed before the last one did. The flow destroyed four river gauge sensors before they could signal an alarm [4]. Local authorities learned of it by word of mouth, then scrambled to send SMS alerts and to reach riverbank towns with handheld megaphones [5]. Pyakurel said the police officer who made the announcement was himself swept away by the floodwaters [6].
The last link failed as well. Pyakurel, a 43-year-old shopkeeper [20], said the loudspeaker's words were "impossible to hear" over the noise of the town [3]. He needed a couple of minutes to work out that it was an evacuation call. He lost more time closing the shop. At home, a few hundred meters from the Trishuli, his family argued over whether to head for higher ground [7]. They ended up on the fourth floor of a house that stayed standing, and survived more than an hour of the flood [8]. "We had lost all hope," he said [9].
Moving the alert to before the avalanche looks out of reach. Ashim Sattar, a cryosphere scientist at the Indian Institute of Technology in Bhubaneswar, calls predicting an event like this "likely impossible," because events of this size are rare and leave little data [10]. Satellite images from weeks earlier showed cracks opening. "There are probably hundreds of cracks opening in the Himalaya," said Christian Huggel, a geographer at the University of Zurich [11]. The fracture zone of a rock-ice avalanche is hidden beneath the surface, so the volume that will break off is hard to know [12].
Glacial lake outbursts are the hazard that can be seen coming. "If glacial lakes are forming rapidly, scientists can detect that a few weeks before possible outburst," Huggel said [13]. Only 21 of some 40,000 Himalayan glacial lakes are monitored by stations on the ground [14], about one lake in 1,900 [3]. Switzerland offers a smaller comparison. A local hazard observer assigned to watch the mountains above Blatten recognized the possibility of the May 2025 ice and rock avalanche that destroyed the village, an event a tenth the size of Nepal's or smaller [15]. Huggel said that because of the steep slopes and valleys, "everything is much bigger in the Himalaya" [19].
Survivors and scientists told Science News that fragile warning systems and the lack of coordinated community response probably raised the death toll [16]. That is a judgement, and a plausible one. At least 1,400 people died and more than 5,000 were still missing as of September 24 [17]. The report does not estimate how many lives a working alert would have saved. The timeline above is also one survivor's, in one town, on a stretch of river almost 100 kilometers long where many people downriver had a few minutes of warning or less [18].
I think the evidence still sets a clear priority. The loudspeaker failed, but a louder one at 9:10 would have recovered none of the 33 minutes already gone [2]. Those minutes were lost to sensors that died in the flow and to news that travelled by word of mouth [4][5].
What to watch
- An official timeline of when Nepal's authorities first learned of the flow and when SMS alerts went out, which would show where the 33 minutes went.
- Whether the destroyed Trishuli gauges are replaced with sensors sited to survive a debris flow, or paired with detection farther upstream.
- The final death toll measured against the more than 5,000 people still missing as of September 24.