Nepal's debris flow began 5,150 metres up as a 2.2-square-kilometre slab of bedrock

Divide the estimated volume by the area of the scar and the failure gets a depth. Around 110 million cubic metres spread over roughly 2.2 square kilometres averages about 50 metres of rock and ice ^[10]. Both inputs are satellite estimates, and Science News describes the reconstruction as a rough picture of the event's origin ^[27].

Early speculation blamed a collapsing glacier ^[7]. The images instead showed a freshly exposed, uniform slope on Langtang Lirung's north face at around 5,150 metres, which researchers read as bedrock that detached and carried its overlying ice into the valley ^[7][8]. The torrent that followed reached speeds over 160 kilometres per hour and destroyed villages on its way downstream toward India ^[5][4].

The rock at that altitude is fractured. Ancient tectonic uplift of the Himalaya left fractures that ice has sealed for millennia ^[15]. "It's that cryospheric glue that, for millennia, has held together the glaciers and the rocks and the soils and the boulders up at high altitude," Alton Byers of the University of Colorado Boulder said during a September 1 online panel ^[16]. He said human-caused warming has softened much of that permafrost, likening it to "taking a stick of butter out of the refrigerator and putting it on the counter" ^[17].

The days before the collapse were warm at that elevation. On August 24 and 25 the average temperature there was about 5 degrees Celsius, with daily maxima above 10 ^[21]. Historically that time of year averages around 2.7 degrees and rarely exceeds 6 ^[22]. The two-day mean therefore ran roughly 2.3 degrees above the seasonal average, and the daily peak at least 4 degrees above the usual ceiling ^[23]. World Weather Attribution reported on September 17 that warming has raised the annual average temperature on the slope and in the region by about 2 degrees Celsius since preindustrial times ^[18]. Ben Clarke of Imperial College London said at a September 16 briefing that this is "much faster than the average rate of global warming" ^[19].

Why the rock let go is not settled ^[11]. Walter Immerzeel of Utrecht University told the same briefing that the magnitude 7.8 Gorkha earthquake of 2015 may have damaged the bedrock ^[12]. That quake set off a landslide on the same mountain, killing around 200 people in Langtang Valley ^[13]. Immerzeel also said the overlying glacier retreated more than 300 metres between 2010 and 2026, potentially shifting the stresses on the rock wall ^[24]. Nobody has separated those contributions.

The case that this event sat outside the expected range rests on how the people who work these slopes describe it. The magnitude caught experts on the region's geology off guard ^[2]. "I've never seen anything remotely as large as this," said Byers ^[6]. Ranjan Dahal of Tribhuvan University, who has studied Himalayan geohazards for nearly 30 years, said "I don't think even Hollywood could make such a movie" ^[3]. Science News reports no hazard-map revision and no new upper bound for failure size on slopes like these ^[28]. Glaciologist Ulyana Peña said "it's not something that's just limited to Nepal" and that "The mountains are crumbling globally" ^[25].

A recurrence interval takes more than one event with a satellite-estimated volume. The 2015 landslide on the same peak killed about 200 people and this one killed more than 1,200, upward of six times as many ^[13][1][14], though a death toll depends on where people were as much as on how much rock moved.