Two events, no link
On 26 August 2026 roughly 0.2 km² of ice and rock detached from Langtang Lirung in Nepal and fell over a kilometre; the resulting debris flow ran 72 km of valley at around 50 m/s. More than 1,300 people are confirmed dead and thousands remain missing. Twelve days later, Hurricane Marie — which never touched land — delivered Southern California's best south swell of the year and 177 lifeguard rescues at Newport Beach in a single day. The two events share no mechanism and no basin; any narrative that joins them is rhetoric. What they share is a failure mode.
What attribution science can honestly say
The warming trend in the Langtang catchment is measured — roughly five times the global average above 4,000 m since the 1960s — and human influence is very likely the main driver of global glacier retreat. What no study yet supports is that climate change caused this particular collapse: researchers identify six candidate mechanisms and cannot narrow them to one.
Attribution also cut both ways in this window. One rapid study found West African flooding made about five times more likely by warming, with models underestimating the trend; another found no detectable climate trend behind the Assam floods at all. Both point at the same dominant impact driver: exposure and urbanisation. An honest research practice publishes both results at the same volume.
The part humans own
Across the window's death toll, the decisive layer was rarely rainfall. Of 159 dead in one Chinese flood region, 107 died because six dams breached. Of 261 dead in one Indian state, 154 died on roads. Nepal's early-warning network was built for glacial lake outbursts — and in this disaster there was no lake; monitoring had concentrated for years on a short list of famous lakes while the disasters came from ice that was not on the list. The distance between a natural event and a mass-casualty disaster is made mostly of human decisions about exposure and warning.
What is loading now
The U.S. Climate Prediction Center's August discussion put a greater-than-90% chance on a very strong El Niño for the 2026–27 fall and winter, and a 69% chance that October–December exceeds every event on record since 1950. The fingerprint is already visible on the ground: a drenched East and South Asia, a dry maritime continent, East Africa's flood risk displaced into October–December. The agency's own caveat travels with every figure we quote: impacts become more likely, not guaranteed.
For the coasts this program studies, that outlook is not a prediction of any specific event. It is a research question with a closing window — and the program's response is to freeze its expectations before the season, publish them while they are still falsifiable, and score them in public afterwards, in both directions.
This is research. Estimates are models, not guarantees; a model is never an observation, and a machine estimate is never “confirmed” — only a human reviewing it in the field makes it verified. Nothing on this page is an operational-capability claim, a safety promise, or a prediction of a specific event at a specific place.
References & data
- 1.NOAA Climate Prediction Center — ENSO Diagnostic Discussion, 13 August 2026 — El Niño Advisory; all probability figures quoted.
- 2.USGS / public event record — 2026 Nepal–Tibet floods: mechanism, seismic reading, toll.
- 3.Hausfather, Z. — "What we can and cannot say about climate change and the Nepal disaster", The Climate Brink, 2026.
- 4.World Weather Attribution — Gulf of Guinea floods rapid study (16 Jul 2026); Assam floods rapid study (14 Aug 2026).
- 5.NASA Earth Observatory — "A trio of tropical cyclones in the Pacific", 4 Sep 2026 — season activity figures.
- 6.ABC News (wire) — Newport Beach rescue counts and lifeguard statements, 6 Sep 2026.
- 7.China Daily / Asahi Shimbun / press archive — Dam-breach toll revision (21 Aug 2026); Chiba rainfall and first-issue landslide warning (Aug 2026).