More layers

How much riverside ground changed, from the source (left) to Betrawati (right)

Rasuwa, Nepal. 26 August 2026

The Rasuwa flood, traced from space by radar

On 26 August 2026 a flood came down this valley in Rasuwa. A NASA radar satellite had passed over a week earlier and came back five days after. I compared the two passes, and the map is coloured wherever the ground looked different the second time.

echo got strongerecho got weakerEither colour means the ground changed.

Scroll to follow the flood downstream, or use the Next button at the top. You can click the map anywhere to see the radar record of that spot.

The source, north of Langtang Lirung

ICIMOD reports that a slab of rock and glacier ice broke off the north face of Langtang Lirung and fell about 1,200 m into the valley below. This is where the radar change begins and where it is strongest.

Sources
  • Radar: NISAR L2 GCOV, track 98 ascending, 19 and 31 Aug 2026 (NASA/JPL and ISRO, via ASF DAAC).
  • Event description: ICIMOD, Kyirong–Rasuwa flood 2026.
  • Peak position and stream lines: OpenStreetMap contributors.

Down the Lende Khola

The debris ran down the Lende Khola, a stream on the Nepal–China border. Some accounts say it blocked the river for a while before bursting through. The radar cannot settle that, but it shows the whole valley floor changed.

Sources

Rasuwagadhi, where two rivers meet

Here the Lende Khola joins the Bhote Koshi at the border crossing. ICIMOD calls Rasuwagadhi and Timure the worst hit.

Sources
  • Radar: NISAR L2 GCOV, track 98 ascending, 19 and 31 Aug 2026.
  • ICIMOD.

The warning minutes

By the reported times, the first SMS alerts went out 38 minutes after the collapse. The flood had already passed Rasuwagadhi.

  1. Seismic signal of the collapse
  2. Reported arrival of the flood at Rasuwagadhi
  3. Hydrologists learn of the flood
  4. SMS alerts sent “to around 679,000 people in Rasuwa, Nuwakot, Dhading and Chitwan”
  5. Reported arrival at Betrawati
  6. Flood reaches Malekhu

Nepal time, as reported by ICIMOD. Some scientists doubt the arrival times, and one analysis puts the first collapse on 25 August.

Sources

Syabrubesi

The Kathmandu Post reported damage at Syabrubesi and at hydropower projects along this stretch.

Sources

Betrawati

Our traced path ends at Betrawati, 66 km from the source. ICIMOD reports the flood carried on to Malekhu.

Sources
  • Radar: NISAR L2 GCOV, track 98 ascending, 19 and 31 Aug 2026.
  • ICIMOD.

How the flood compares with a normal period

The satellite passed ten times from June to October. Each bar is how much of the riverside changed between one pass and the next.

Details

The ice above the valley

Wet snow soaks up radar, so melting ice sends back a weaker echo. On the map, blue is ice whose echo is weaker than it was on 20 June. Drag the slider to move through the summer.

This may be a sign of melting. We could not check it on the ground, so we do not call it a measurement.

Details and sources
  • Fresh snow, bare ice and rock debris also change the echo, and NISAR's long wavelength reacts less to wet snow than shorter ones. Five months cannot show whether a glacier is shrinking.
  • Radar: NISAR L2 GCOV, track 98 ascending, ten dates from 20 Jun to 18 Oct 2026, averaged to about 50 m.
  • Glacial lakes: High Mountain Asia Near-Global Multi-Decadal Glacial Lake Inventory, version 1 (NASA NSIDC DAAC), period 2015 to 2018.
  • Heights: NASADEM, version 1 (NASA LP DAAC). Glacier outlines: OpenStreetMap contributors.

Where would water go?

Click anywhere on the map. A line shows the way downhill from that spot and which villages lie below it. Try one of the ringed glacial lakes.

Shaded strips are low ground beside the rivers: under 10 m above the water, 10 to 20 m.

The map only shows the shape of the land. Please do not read it as a flood forecast or an official hazard map.

Details and sources
  • The route follows the slope only. It does not know how much water there is, how deep it gets or how fast it moves.
  • Elevation: NASADEM Merged DEM Global 1 arc second, version 1 (NASA LP DAAC), measured in 2000, 30 m cells. Too coarse for narrow gorges.
  • Method: flow direction (D8) and height above the nearest drainage, pysheds library. See the data and methods page.

Use the data yourself

  1. Click a place on the map to see its radar record on ten dates.
  2. Download CSV from the card and open it in a spreadsheet.
  3. Get everything else from the downloads list.
  4. Check or repeat the method on the data and methods page.

How we know it was the flood

  1. Seven ordinary 12-day periods on the same track show nothing like it.
  2. Four neighbouring rivers, measured the same way on the same dates, show no jump.
  3. Three satellite tracks saw the valley from different directions, and the change appears on all three.

Each dot is one 12-day period. The red dot is the one with the flood.

What it cannot tell you

  • The radar shows where the ground changed. It cannot show what people lost.
  • Anything else that happened in those 12 days is included.
  • Very steep slopes distort radar.
  • It looks back at an event and cannot be used as a warning system.
The measurement

NISAR is a radar satellite built by NASA and ISRO. It sends its own microwave pulses, so it sees through cloud and at night, and it passes over this valley on the same track every 12 days. For each pass we take the radar brightness of the ground (backscatter, HH and HV), average it over 5 × 5 pixels of 10 m to calm the grain, and divide the later pass by the earlier one. The result is in decibels: 3 dB is a doubling or halving.

The three tracks
TrackBeforeAfterNote
98 ascending19 Aug31 AugSame radar mode on both dates. Used for all numbers here.
48 descending16 Aug28 AugMode differs between dates (10 m and 20 m pixels).
149 descending23 Aug4 SepMode differs between dates; the whole scene is about 0.6 dB brighter after.

Switch between them under “More layers” on the map.

Data and code
  • NISAR L2 GCOV, provisional release (CRID P05023: calibrated, not yet fully validated), NASA/JPL and ISRO, distributed by the Alaska Satellite Facility DAAC; subsets requested through NASA Harmony on 14 Nov 2026.
  • Rivers, places and peaks: © OpenStreetMap contributors (ODbL), retrieved 14 Nov 2026. River lines can sit tens of metres from the real channel.
  • Scripts and provenance: data and methods, credits.

Warnings and preparedness

A satellite that passes every 12 days cannot warn anyone. Warnings come from these agencies.

This site does not show evacuation routes or safe zones. Follow local authorities.