On August 26, 2026, a massive section of glacier and rock collapsed near Langtang Lirung on the Nepal–Tibet border. The failure dropped roughly 1,200 metres into a river valley, triggering a destructive flood that travelled almost 100 kilometres downstream.
But the disaster did not end when the first flood passed.
Satellite imagery revealed that the collapse created two new lakes upstream. One formed behind unstable landslide debris, while another appeared higher near the glacier. Officials in Nepal and China were forced to monitor the new lakes closely as concerns grew over another possible sudden release.
In this video, DiscoverTube investigates what actually collapsed, why the event was initially mistaken for an earthquake, how floodwaters reportedly travelled at extraordinary speeds, and why much of the region’s monitoring network was destroyed before it could provide meaningful warning.
We also examine the newly formed barrier lakes, the temporary suspension of rescue operations, the enormous damage to roads, bridges and hydropower infrastructure, and the unanswered questions scientists are still investigating.
The first flood is over.
But the landscape that created it is still changing.
Subscribe to DiscoverTube for more investigations into extreme geology, glaciers, natural disasters and the forces reshaping our planet.
On August 26, 2026, a massive ice-rock avalanche near the Nepal-Tibet border triggered a catastrophic flood, killing hundreds and leaving 1,900 missing. The event created a violent \"pulse\" of water and debris that destroyed critical infrastructure and monitoring systems, highlighting the extreme danger of glacial failures and the urgent need for cross-border hazard data sharing.
On the morning of August 26, 2026, a piece of glacier roughly 610 meters wide broke off a
slope near Langtang Lerung on the Nepal-Tibet border and fell about 1,200 meters. It landed
on a river. What happened next traveled almost 100 kilometers and killed hundreds of people in
a matter of minutes. As of the Disaster Authority's situation update at 10 a.m. on August 29,
626 bodies had been recovered in Nepal. Around 1,900 people are still missing.
Here on DiscoverTube, we've been following the official numbers as they move. And they
have moved every single day. The morning the mountain let go.
Satellite analysis reported by Al Jazeera on August 28 put the detachment point at roughly
That's the lower end of the glacier, the part sitting on the steepest,
warmest ground. It sheared away and dropped about 1,200 meters onto the valley floor.
CNN's analysis of Landsat imagery published on August 28 and 29 gave the width. Around 610
meters of glacier, with a similar-sized section of rock beside it coming down at the same time.
The scar left behind on the mountainside measures roughly 900 meters across.
One number is still missing, and it matters. Nobody has calculated the volume of ice and
rock that fell. Width and drop height are straightforward to measure from orbit.
Depth is not. Until researchers get better data on how thick that slab was,
the total mass remains an open figure. The mechanism here is important.
An ice avalanche on its own is destructive but relatively light. This one entrained rock and
sediment on the way down. So what reached the valley floor was a dense, abrasive slurry carrying
car-sized boulders. That's an ice rock avalanche, and it hits with far more force per cubic meter
than ice alone. Then geography did the rest. The mass came down into the Lende Kola,
a river in Tibet about 20 kilometers northeast of the border crossing. It didn't land on empty
ground. It landed in flowing water in a narrow gorge with settlements downstream. That is the
moment a mountain failure became a flood. And the impact was violent enough that the first
official reading of the event was simply wrong. Correcting it took the better part of a day.
90 Seconds of Warning That Never Came
The initial reports described a magnitude 4.4 earthquake. That's what the early seismic
bulletins suggested, and that's what a lot of first response decisions were briefly built on.
The United States Geological Survey later revised the interpretation. There was no earthquake.
The seismic signal had been generated by the landslide itself, and once reanalyzed,
the mass movement registered as a magnitude 5.2 event. This distinction is not academic.
Seismometers can separate a tectonic rupture from a collapsing mountainside because the two
produce different signatures. A fault slipping releases energy in a sharp, high-frequency burst.
A vast mass sliding down a slope generates a longer, lower-frequency rumble as it grinds
and accelerates. Reading that correctly tells responders whether to expect building damage
across a region or a wall of water in one specific valley. For the first hours,
the wrong answer was on the record. Meanwhile, the flood was already gone.
Jeffrey Cargill of the Planetary Science Institute, whose velocity analysis was reported
on August 29, calculated that the wave descended the first 22 kilometers at an average speed of
193 kilometers per hour. That is faster than any vehicle on the road that morning.
Anyone who saw it coming and tried to drive out of the valley was already too late.
Further downstream, the gauges caught what arrived. At Galchi on the Trishuli,
monitoring data from the International Center for Integrated Mountain Development
recorded the water level rising by as much as 9 meters in roughly 30 minutes.
That rate is diagnostic. Monsoon rainfall produces a rising curve. Water accumulates
across a catchment, feeds into tributaries, and the main river swells over hours.
Nine meters in half an hour is not a curve. It's a pulse, and a pulse means something
upstream released all at once. The height explains the death toll.
Floodwaters reached settlements as high as 80 meters above the normal river level.
In a Himalayan valley, 80 meters above the water is where you build.
It's where the road goes, where the shops are, where people assume they're safe from the river.
On August 26, that assumption failed.
There is also an independent estimate of the flood's raw power.
The engineer Anup Kanal, whose figures were published on August 28,
calculated an average travel speed between Rasuwagadi and Betrabati of 15 to 20 meters per
second. Working from an estimated flood height of around 20 meters and a width near 40 meters,
he put peak discharge somewhere between 12,000 and 15,000 cubic meters per second.
That is an estimate, not a measurement, and it should be treated as one.
The instruments that would normally have measured it directly
no longer existed by the time the peak passed, which leaves a gap in the record.
The seismic data tells you the mountain moved and roughly how hard.
The gauges downstream tell you what eventually arrived.
Neither tells you what happened to the river in the minutes in between,
up in the gorge where nobody was left to see it.
For that, investigators had to go to orbit.
What the satellites actually saw
The reconstruction came from Copernicus Sentinel-2 imagery,
analyzed by researchers and published in The Conversation on August 28.
It covers the window from 24th to 26th August,
and it fills in the part of the story no instrument on the ground survived to record.
Here is what the imagery shows.
A glacier in the Langtang region collapsed.
The resulting ice and rock avalanche appears to have entered the Lende Kola,
a tributary of the Botei Koshi.
There, it either caused or contributed to a temporary blockage of the river.
When that blockage failed, the flood ran almost 100 kilometers downstream.
That phrasing matters.
Either caused or contributed to is not hedging for its own sake.
It reflects a genuine gap in the evidence.
Nobody has confirmed whether the avalanche debris alone dammed the river,
or whether it destabilized material that was already sitting in the gorge.
The distinction between a flood and a dammed release
is the whole story of why the death toll is what it is.
A river in flood rises.
A river that has been blocked stores energy.
Water backs up behind loose debris, gains head,
and when the barrier fails, it does not drain gradually.
It leaves as a single pulse,
carrying the accumulated volume of however long the blockage held.
That is why the wave arriving downstream behaved nothing like monsoon water.
There is one more observation from the days before, and it needs careful handling.
The British Antarctic Survey noted unusually high ground temperatures
in the area in the two days preceding the collapse.
That is an observation, not an established cause.
It has not been shown to have triggered anything.
Twyla Moon, a scientist at the National Snow and Ice Data Center at CU Boulder,
spoke on the record to NPR on August 28th about melting glaciers
and what that means for communities living below them.
Mohamed Farouk Azam at Isamad pointed to monsoon saturation as an amplifier,
waterlogged slopes being less stable slopes.
Prashant Baral, a cryosphere analyst at the same institution,
described the sheer size of what detached.
None of them said climate change caused this collapse.
Researchers are actively investigating whether melt contributed to the instability.
A direct link has not been established,
and saying otherwise would be running ahead of the evidence.
There is also no record of anyone forecasting
an ice rock avalanche at this specific location.
Warnings about hazardous glacial lakes across the wider Himalaya exist.
A prediction for this mountain, on this slope, does not.
If you want the follow-ups as the analysis firms up,
subscribe to DiscoverTube, which brings up the limitation buried in all of this.
The satellites reconstructed the event beautifully, but only afterwards.
The instruments that were supposed to see it coming were already gone.
Why the sensors died first
Online Khabar reported the losses on August 28th, and the list is stark.
Flood monitoring equipment at Langtang, Rasuagadi,
Betrabadi, Siafrubesi, and Malakou was destroyed.
Three stations remain operational, Devgat, Borle, and Kalikola near Muglin.
Look at where those survivors sit. They are all far downstream, near Muglin.
That leaves effectively the entire upper Bote-Koshi corridor,
the exact stretch of river the flood crossed at speed, without a single functioning sensor.
Parajuli, an official at the Department of Hydrology and Meteorology,
explained why.
Most of the equipment had been installed on bridges,
because bridges were considered the strong, stable structures in the valley.
The bridges themselves were swept away, and the instruments went with them.
That is a design assumption failing in exactly the scenario it was meant to cover.
Bridges are engineered against river forces.
They are not engineered against a debris flow
moving at 15 to 20 meters per second carrying boulders.
The engineer Anup Kanal noticed something else in the data before the Rasuagadi station died.
There was no recorded drop in water level beforehand.
That absence is interesting,
because a river being dammed upstream usually starves briefly downstream.
The record does not clearly show it,
and Kanal raises the possibility that this was a landslide dam outburst flood,
an LDOF, rather than a straightforward avalanche-driven surge.
That remains preliminary.
The Chinese side did not catch it either.
According to information received by ICIMOD,
Chinese monitoring systems had not detected any natural event upstream
capable of producing a flood of this magnitude.
But here is the nuance that gets lost.
Nepal's warning system did not fail everywhere.
The Department of Hydrology and Meteorology
sent 679,295 warning messages through Nepal Telecom and NCEL.
The dissemination machinery worked.
The problem was detection.
There was nothing left upstream to pull the trigger in time.
Rebuilding a sensor network takes months.
The hazard did not offer months.
Within a day, both governments were describing something new taking shape upstream.
Two New Lakes Above the Wreckage
Al Jazeera reported on August 28th that the collapse had created not one new lake, but two.
One sits high, near the glacier, holding trapped meltwater.
The other is lower, larger, and far more dangerous,
formed where landslide debris blocked a river valley outright.
China's Ministry of Water Resources issued the warning on August 27th.
A barrier lake of roughly 2 million cubic meters had formed just north of the border,
where the Chochenkola meets the Parepu Tsangpo.
It was already overflowing.
Chinese hydrologists projected another 3 million cubic meters would drain into it by August 30th.
Nepal's disaster authority, the NDRRMA, said the lake level was rising, which meant it could burst.
It put the blockage roughly 18 kilometers upstream of Rasuwagadi,
with a pond behind it covering about 0.11 square kilometers.
These are not abstract predictions.
They are operational hazard warnings about a specific, physically located body of water,
issued by named agencies watching it.
A debris dam is unstable for a simple reason.
It is loose, unsorted material with no engineered spillway.
Water finds the weakest path over the crest and starts cutting downward.
The failure usually begins as erosion, not a sudden structural break.
On August 28th, that lower lake stopped being a forecast.
The breach that wasn't the second disaster
On Friday, August 28th, Rasuwa's chief district officer, Narendra Parriar,
told reporters the Nepali army had just informed him the barrier lake had breached.
That single sentence stopped the rescue effort cold.
Helicopter operations over Rasuwa were suspended.
Residents were filmed running uphill,
away from the same riverbank they had been searching two days earlier.
On the Chinese side, state broadcaster CCTV reported the newly formed lake had begun overflowing,
with a risk of flash flooding or landslides,
and teams working to reach the Girong land port paused as well.
Both sides resumed later the same day.
CCTV described the risk level as having returned to what it called manageable.
Then came the measurement, and it is the most important number in this section.
After the breach, the river rose by about 0.6 meters,
followed by a reported drop of roughly 10 meters in the lake level itself,
with the hazard gradually decreasing.
Compare that to August 26th,
when the gauge at Galchi recorded up to 9 meters of rise in half an hour.
The same corridor, the same river, two days apart,
and the second event produced roughly 1 15th of the water level response of the first.
The reason is mechanical, and it is the difference between a scare and a catastrophe.
When a debris dam fails by sudden structural collapse,
the whole stored volume leaves at once.
When it fails by overtopping erosion,
water spills over the crest and slowly cuts a channel downward.
The lake drains over hours,
and the outflow never greatly exceeds what the river was already carrying.
Every indication so far is that the lower barrier lake drained the second way.
The threat has not been formally closed.
On August 28th, the Chinese embassy in Kathmandu
said the lake formed by the ice rock avalanche remained at a high level
and was still overflowing,
with officials on both sides continuing to monitor for a possible breach.
A Chinese team was dispatched to watch it directly.
The lake sits roughly 10 kilometers upstream of Jirong Port,
close enough that any sudden change would reach the ruined border complex in minutes,
which raises something worth your input.
Nepal first heard about an upstream event on its own river
through a research institution passing along information from across the border.
Should real-time hazard data from shared river basins
be legally required to cross borders instantly?
Tell DiscoverTube what you think.
Nepal's own response was aerial.
The Rescue Coordination Center at Tribhuvan International Airport
prepared reconnaissance flights over the Botakoshi and surrounding valleys,
using helicopters already stationed at Dunche.
That is currently the only practical way to assess a lake nobody can drive to.
And the second lake, high near the glacier holding trapped meltwater,
still has no published volume.
It has been described. It has not been measured.
That is an open hazard above a valley already stripped of its monitoring stations.
As of August 29th, search teams were still working under a fresh warning
after another rise in the Botakoshi.
They have evacuated uphill, waited, and gone back down more than once,
all while looking for bodies in the debris of the first flood.
With the immediate water threat easing, the accounting began.
Nobody has been able to total it yet.
A country counting losses it can't yet price.
The Disaster Authority's situation update at 10 a.m. on August 29th
put the confirmed dead in Nepal at 626, with 4,451 people rescued.
The geography of those recoveries tells you how far the water carried people.
As of 6 a.m. that morning, police had recovered 233 bodies in Chitwan,
158 in Nawalparasi East, 48 in Gorkha, 47 in Nawalparasi West, 45 in Dading,
41 in Newakot, 31 in Tanahun, and just 13 in Rasuwa.
Rasuwa is where the flood hit.
Chitwan is far downstream on the lowland plains.
Most of the victims were found nowhere near where they died.
The breakdown of the missing, released on August 28th, explains who was in that valley.
Of roughly 1,924 unaccounted for on the Nepali side,
933 were linked to hydroelectric project sites.
517 were foreign nationals.
127 were Nepalis living abroad who had come home to visit.
Another 119 were state employees, including 45 army personnel, 28 police,
15 customs officers, and 15 immigration staff.
The foreign nationals are there for a specific reason.
The Rasuagadi-Kirang crossing is the main overland gateway for the Kailash-Mansarovar Yatra,
and late August is peak season.
Every overland pilgrim entering from Nepal funnels through that one narrow gorge.
The Nepal Tourism Board said on August 29th that 187 tourists had been rescued or accounted for.
420 remain out of contact.
The physical damage is easier to count and still enormous.
Relief Agency Care reported on August 28th that 32 bridges
and nearly 40 kilometers of road were swept away,
including the entire route from Betrawadi in Nuwakot up to the Rasuagadi crossing.
Timor and Siafrabesi were among the worst-hit settlements.
14 hydropower and solar projects were affected, totaling around 748 megawatts of capacity.
The World Health Organization said six health facilities were damaged
and eight health workers are missing.
On energy, two official figures are circulating and they do not match.
The Nepal Electricity Authority lists six damaged facilities,
including Rasuagadi, Chalime, Trishuli 3A and Devigat,
removing an estimated 431 megawatts from the grid, plus another 470 megawatts under construction.
Nepal's Energy Ministry, quoted via Bloomberg, cited close to 360 megawatts.
Both are on the record. Neither has been reconciled.
The government moved on August 28th.
Cabinet directed the finance ministry to build a relief and rehabilitation package,
and the prime minister and ministers agreed to donate a month's salary.
For context on what the economy can absorb,
the World Bank's Nepal Development Update in April 2026
recorded 3.4 percent growth in the first half of the fiscal year
and projected 2.3 percent for the full year,
recovering to around 4.4 percent afterwards, partly on hydropower expansion.
That expansion just lost hundreds of megawatts.
There is still no official total economic loss estimate.
The next hard checkpoint is a date.
Chinese hydrologists projected a further 3 million cubic meters of water
would drain into the barrier lake system by August 30th.
The upper glacial lake remains unmeasured.
The volume of ice and rock that fell has never been calculated.
The finance ministry package has no figure attached,
and with roughly 1,900 people still missing, the death toll will move again.
If you want this tracked as those numbers land,
subscribe and leave a comment telling DiscoverTube
which unresolved figure you most want followed.
The upper lake's volume or Nepal's final energy loss.