The toll. On 4 September Nepal’s National Disaster Risk Reduction and Management Authority, NDRRMA, counted 1,287 dead. Later estimates, carried by agencies and aggregated sources, exceed 1,350 dead and 5,000 missing in Nepal, plus a few dozen dead and about five hundred missing in Chinese Tibet. These numbers are still moving and should be read with the date beside them. On the evening of 26 August the Nepali government counted 95 dead. A week later, a thousand.
Where it began. About twenty kilometres north-east of the crossing, across the border, above the Lhende Khola, a tributary of the Bhotekoshi. Satellite images show a detachment one kilometre wide, perhaps 1.3, on the Langtang Lirung massif, between 5,000 and 5,200 metres. Ice and rock together. The mass fell more than a thousand metres to the valley floor, below 3,000, picked up more material and entered the river.

Here the reconstructions part ways. The rapid report by Integrated Research on Disaster Risk, issued on 28 August, holds that the avalanche blocked the river on the afternoon of the 25th and that the natural dam held for 18 or 19 hours before failing between 8:30 and 8:37 on the 26th; inflow from the Mala river and other tributaries would have masked the blockage from anyone downstream. The report says so plainly: it is built on secondary data, official pages and the Nepali press, not on field survey. Other analyses, among them that of the HiRISK consortium and reconstructions from imagery, place the detachment itself on the morning of the 26th, with only a brief blockage of the river. The 8:37 seismic signal fits both: in the first case it is the dam failing, in the second the avalanche itself. No one has closed the question. Manoochehr Shirzaei, a geophysicist at Virginia Tech, told Nature that the sequence remains uncertain: it may have started with a rockslide that dragged the glacier down, or with the glacier failing first.

Shirzaei looked back. With radar data from ESA’s Sentinel-1 satellites acquired between 8 January and 18 August, he saw that the glacier–rock system near the detachment point was creeping downhill at roughly ten millimetres a month, and that it accelerated in the weeks before the collapse. Ten millimetres a month alarms no one: an ordinary glacier moves tens or hundreds of metres a year. The acceleration is what counts. The last usable image was from seven days before. No one was looking at it for that.

Italy comes into this, and not by chance. At five thousand metres in the Khumbu valley, six kilometres from Everest base camp, the broadband seismic station IO.EVN has run since 2014 at the Pyramid International Laboratory-Observatory of EvK2CNR. It is operated by the Istituto Nazionale di Oceanografia e di Geofisica Sperimentale (National Institute of Oceanography and Applied Geophysics) in Trieste. It is one of the few stations in a region almost without networks, and human noise is minimal. On 27 August the INGV, Italy’s national institute of geophysics and volcanology, published the comparison between the signal of the 26th and that of a magnitude 5.3 earthquake of 15 September 2020 recorded by the same station. The earthquake has a sharp onset, amplitudes concentrated in the first seconds, energy across many frequencies. The avalanche has a gradual onset, a far longer duration, pulses in succession, energy at low frequencies: it is the mass moving, hitting the slope, rolling. Station LMD of Italy’s National Seismic Network, in the province of Florence, also recorded the event, filtered between 0.01 and 0.1 hertz. From six thousand eight hundred kilometres it looks like an earthquake. From six kilometres off Everest it does not.

It is not the first time on this river. In July 2025 a flood on the Lhende Khola and the Bhotekoshi, then attributed to rain and a glacial lake, killed at least nine people, left about twenty missing, took away the friendship bridge between Nepal and China and cut 211 megawatts from the grid. Same corridor. Thirteen months later, the same valley.
The context was written in March by ICIMOD in Kathmandu, in its report on Hindu Kush Himalaya glaciers from 1990 to 2020: 12 per cent of area lost, 9 per cent of ice reserves, losses above 20 per cent in the central Himalaya, which is here. The rate of loss has doubled since 2000. Of the region’s 63,700 glaciers, 38 are monitored, and 7 meet the standards of the World Glacier Monitoring Service. An inventory published in April in npj Natural Hazards counted 219 hanging glaciers in the Alaknanda basin of India alone: 2.39 cubic kilometres of ice, 0.74 of them precariously balanced on 33-degree slopes. An Indian outlet attributed it to Nature. It is not Nature: it is a journal of the same group, and the difference has to be written down.
How much climate weighs on this particular collapse, no one yet knows. Warming weakens glaciers and the permafrost that holds rock walls together, and that is documented. Attributing a single event takes work that has not been done. The director of Tribhuvan University’s centre for disaster studies told SciDev.Net the energy exceeded that of the Hiroshima bomb. The calculation is not published. We leave it where it is.

What remains is a problem of shape. A warning system measures rain and river level. Here it had not rained, and the river, until 8:30, was normal. The chain began twenty kilometres higher, at five thousand metres, on a slope moving a centimetre a month. ICIMOD has said it for years: mountain hazards can no longer be assessed one at a time. A system that searches satellite radar for accelerating slopes, and then computes what happens downstream if they fail, does not exist today at regional scale. The data are there. Sentinel-1 passes over. The station below Everest is listening. What is missing is someone to look.



Comparison between the signal of a magnitude 5.3 earthquake of 15 September 2020 (grey) and that of the avalanche of 26 August 2026 (red), recorded by seismic station IO.EVN at the Pyramid International Laboratory-Observatory of EvK2CNR, with their spectrograms. The earthquake is impulsive and short; the avalanche is gradual, long and low-frequency. The figure is reproduced unaltered.
INGVterremoti, Istituto Nazionale di Geofisica e Vulcanologia; data from station IO.EVN, Istituto Nazionale di Oceanografia e di Geofisica Sperimentale — Licenza Creative Commons Attribuzione – Non opere derivate 4.0 Internazionale.

References
Adhikari B. R., Marahatta D., Thakuri R. et al., “Rapid Analysis Report: Scientific Anatomy of the 26 August 2026 Bhotekoshi Flash Flood”, Integrated Research on Disaster Risk, Institute of Himalayan Risk Reduction, National Society for Earthquake Technology, 28 August 2026.
Basu M., “Satellite images before Nepal disaster showed warning signs”, Nature, News Q&A, 2 September 2026. doi: 10.1038/d41586-026-02746-4
INGVterremoti, “26 agosto 2026: la frana in Nepal e l’analisi dei sismogrammi”, Istituto Nazionale di Geofisica e Vulcanologia, 27 August 2026.
Pettenati F., Sandron D., Verza G., Plasencia Linares M. P., Percacci R., Bistia K., Cravos C., Rebez A., “The Seismic Station IO.EVN at Pyramid EvK2cnr (Everest): 10 Yr of Activity”, Seismological Research Letters, 96, 1733–1746. doi: 10.1785/0220240111
International Centre for Integrated Mountain Development, “Changing Dynamics of Glaciers in the Hindu Kush Himalaya Region from 1990 to 2020” and “HKH Glacier Outlook 2026: Insights from 50 Years of Himalayan Glacier Monitoring”, Kathmandu, 21 March 2026.
Krishnan N., Sattar A. et al., “Basin-scale inventory and exposure assessment of hanging glaciers, Central Himalaya”, npj Natural Hazards, 3, 44, April 2026, open access. doi: 10.1038/s44304-026-00205-8
United States Geological Survey, event page “M 5.2 Landslide – 55 km NW of Kodāri, Nepal”, 26 August 2026.
Asian Disaster Reduction Center, “2026 Bhote Koshi Debris Flow and Flash Flood (Nepal)”, report ADRC_FL_NPL_Rasuwa_20260826, August 2026.
The Kathmandu Post, “Update: Rasuwa flood death toll climbs to 95 as dozens remain missing”, 26 August 2026; “Nepal flood death toll reaches 1,287”, 4 September 2026.
All India Radio News, “Nepal–China Miteri bridge washed away in floods”, 9 July 2025.
Ahmad T., Raina S., “Nepal flood disaster exposes cascading threats in Himalayas”, SciDev.Net, 2 September 2026 (used as an index of sources).