Searching Sentinel-2 scenes
Floods. Earthquakes. Live signals. One clear view.
Safety firstRescue 1234 · Police 100 · Ambulance 102 · NEOC 1144
Sentinel-2 before and after the 2026-08-26 Bhote Koshi–Trishuli flood. The blue overlay is the satellite’s own classification of surface water; the pale grey is cloud, drawn deliberately so you can tell “no water was seen here” from “nothing could be seen here”.
Satellite water condition, not a water-quality test. An optical satellite sees where water is and how far it reaches; it cannot see bacteria, arsenic, metals or anything dissolved. Cloud is drawn faintly because those are places the satellite could not see the ground at all.
This is a satellite water condition layer. An optical satellite records reflected light, so it can show where water is and how far it reached. It cannot see bacteria, arsenic, metals or anything dissolved, and nothing here is a water-quality test or a basis for deciding whether water is safe to use.
The water mask is the Sentinel-2 Level-2A scene classification produced by ESA’s own processor, not a threshold invented here. That is deliberate: a cut-off chosen by this project would be an unvalidated number carrying the authority of a satellite. The overlay is drawn on top of the true-colour image rather than replacing it, so the evidence stays visible underneath.
Scenes are picked for clarity rather than for being nearest the event date. During the monsoon the closest pass is usually solid cloud, and comparing two white rectangles shows nothing — so the gap between the two captures is stated, because a wide gap may be showing seasonal change rather than the flood.
A limitation worth knowing before you read the overlay. Checked against the corridor at Betrawati, the 11 August scene classifies no water there at all, while the 27 August scene classifies water plus a great deal of cloud. Narrow, sediment-heavy Himalayan rivers often fall into the classifier’s bare-ground class rather than its water class, so an absence of blue is not evidence of an absence of river. Read the overlay as “water the classifier was confident about”, and the true-colour image underneath as the record.
Treat derived indicators as experimental. They have not been validated against ground observations for this region.