Explore · Stages 2 and 3 · Real data
What actually happened?
Cyclone Ditwah flooded the lower Kelani in late 2025. These are published flood layers, not predictions. There is no animation, because nothing was mapped between these dates and this project won't invent it.
The mapped flood
Average rainfall over time
Area average for the rectangle over the lower Kelani and the hills upstream, in mm/h. The band is the range the published colours allow.
Rainfall by day, as a table
| Day (Sri Lanka time) | Average rain, low (mm) | Average rain, high (mm) |
|---|---|---|
| 2025-11-25 | 21.0 | 22.8 |
| 2025-11-26 | 62.9 | 66.8 |
| 2025-11-27 | 120.2 | 127.5 |
| 2025-11-28 | 26.9 | 29.4 |
| 2025-11-29 | 3.7 | 6.1 |
| 2025-11-30 | 0.1 | 2.5 |
| 2025-12-01 | 0.0 | 2.4 |
| 2025-12-02 | 0.1 | 2.5 |
| 2025-12-03 | 1.5 | 3.9 |
| 2025-12-04 | 14.1 | 16.9 |
| 2025-12-05 | 15.7 | 18.4 |
| 2025-12-06 | 0.1 | 0.6 |
Shown: Copernicus Emergency Management Service (© 2025 European Union), [EMSR851] Kelani Ganga River (AOI02): Delineation Map.
Coarse rain, fine terrain
These two layers work at very different scales. One rain cell is about 11 km across and covers about 129,600 terrain cells of about 30 m. In the rain cell at 80.1° E, 6.9° N, the surface runs from 5 m to 253.9 m, yet IMERG gives that whole cell one rain value.
Stretching rain onto the terrain grid would look sharper but add no information, so the map keeps it as coarse squares. It can't say which street got more rain. Terrain says where water can go, and rainfall says roughly how much arrived across the area.
What the rainfall shows
- Most rain came before both flood maps. Area-average rainfall was about 235 to 254 mm before the 30 November image, and only 0.4 to 7.7 mm between the two images, across the rectangle shown.
- This is not the gauge-corrected record. These rates come from NASA's public IMERG imagery, whose latest frames arrive within hours, so it is probably a near-real-time run. The service doesn't state which. The gauge-adjusted Final Run wasn't listed for these dates when checked.
- It agrees in shape with a separate source. Daily rain in one cell correlates 0.90 with NASA POWER's MERRA-2 reanalysis, over 11 days. That is not a check against rain gauges, and amounts differ on some days.
- Time zones: half-hourly times are in UTC, and the daily table uses Sri Lanka time, which is 5 hours 30 minutes ahead.
- Terrain is a surface model. Heights include trees and buildings, and they are metres above the EGM2008 geoid. Observed flood lies at a median height of 7.9 m, against 22.9 m for land mapped dry.
When each layer was mapped
| Copernicus image | Observed (ha) | Flood trace (ha) | Extrapolated (ha) | Total (ha) | Not analysed (ha) |
|---|---|---|---|---|---|
| 30 Nov 2025, 05:39 UTC, PlanetScope | 3,377.0 | None | 3,382.3 | 6,759.3 | 636.2 |
| 3 Dec 2025, 07:06 UTC, Legion | 688.3 | 122.9 | 2,108.1 | 2,796.4 | 7,122.3 |
The UNESCO layer is one merged extent from several sensors. Its file name gives 26 November to 2 December 2025, and its catalogue entry gives 28 November to 5 December. It has no date for each area, so it shows where water was mapped during the window, not when.
The fair comparison
Far more ground was not analysed on 3 December, so the headline fall overstates the change. Counting only the 16,071.0 ha analysed on both dates, observed flood went from 1,969.6 ha to 685.4 ha. The source doesn't say why areas were not analysed, and the sensors differ, so some of this change may still not be real.
How far to trust these layers
- UNESCO: its publisher describes it as a preliminary analysis that has not been validated in the field.
- Copernicus observed: extracted from satellite images, excluding permanent water. Stated position error is 6.25 m and 4 m.
- Copernicus extrapolated: observed flood extended with a terrain model into areas images can't see well, such as towns and forest.
- Depth: every depth polygon is marked as modelling. It is not a measurement.
- Alignment: the best match between UNESCO and Copernicus needs a shift of 0 pixels, at about 11 m per pixel. That shows no systematic offset, not that the layers agree: they cover different times.
Sources
- Cyclone Ditwah and Floods in Sri Lanka (November 2025), UNESCO IHP-WINS, doi:10.63253/hvu5r3bf. Licensed CC BY-SA. The derived map layers here are shared under the same licence.
- Copernicus Emergency Management Service (© 2025 European Union), [EMSR851] Kelani Ganga River (AOI02): Delineation Map, and Monitoring 1.
- Rainfall: imagery from NASA's Global Imagery Browse Services (GIBS), part of NASA's Earth Science Data and Information System (ESDIS). Layer: IMERG Precipitation Rate, 30-minute average. Cross-check: NASA POWER.
- Produced using Copernicus WorldDEM-30 © DLR e.V. 2010-2014 and © Airbus Defence and Space GmbH 2014-2018 provided under COPERNICUS by the European Union and ESA; all rights reserved. The organisations in charge of the Copernicus programme by law or by delegation do not incur any liability for any use of the Copernicus WorldDEM-30.
- Basemap: OpenFreeMap, with data from OpenStreetMap contributors.