A cascade, not a flood: testing the competing accounts of the 26 August 2026 Lhende Khola–Bhote Koshi disaster

Ishfaq Gujree 

On 26 August 2026 a rock–ice mass detached above the Lhende Khola on the China–Nepal border, blocked the channel, and released a flow that travelled about 100 km down the Bhote Koshi and Trishuli, killing hundreds of people and removing about a tenth of Nepal's generating capacity in a single morning. Reporting produced mutually incompatible accounts of the cause, while the agency best placed to judge states the mechanism remains undetermined. Five positions on the trigger are held by credible parties; each makes a prediction, and three can now be tested. Neither the ICIMOD glacial lake inventory nor a seven-year Sentinel-2 water-occurrence record finds a pre-existing surface lake of consequence in the source catchment — two methods agreeing to within 4%, with the detection limit stated. A proposed sequence in which the failure dammed the river for 18–19 hours predict a seismic signal the USGS catalogue does not contain. A stage–area–volume relation from pre-event topography alone, without empirical coefficients, reproduces the reported lake area and identifies which valley is impounded. The catalogue also holds a second failure three hours later, publicly noted but never characterised. Finally, the widely cited figure for capacity lost is only the operational half of a two-part official figure.