Flood Radar: Floods Through the Clouds
Sentinel-1 radar sees through cloud and at night: map the land under new water after any flood since 2014, with its area in km².
Floods come with clouds, so optical satellites rarely see them. Radar does: it sends its own microwave signal, which passes through cloud and works at night, and calm water reflects it away so it looks dark. Choose a place anywhere on Earth and the day of the flood; the tool takes the first Sentinel-1 pass after it and a dry reference made of up to three earlier passes from the same orbit, finds the water in both with an automatic threshold, confirms new water by the drop in backscatter, leaves out steep slopes with an elevation model, and maps the flooded land, the permanent water and the water that went back with their true areas. It can follow the flood through every radar pass from a month before to six weeks after, and exports the maps, the flood polygons and the numbers.
Overview
In Iraq the spring floods of the Tigris and the marshes, and the flash floods of the north and the western desert, come with low cloud and rain that hide them from optical images for days. Radar gives the extent of the water on the day it matters: for rescue, for damage to crops and roads, and for planning.
The same method is used by emergency mapping services worldwide; here it runs in the browser, on any place, for any flood since October 2014.
What you enter · What you get
What you enter
- A place from the list, found by name or drawn on the map (up to about 120 × 120 km)
- Date of the flood and search window after it
- Dry reference period before it
- Water threshold (automatic or set), change test, slope mask, speckle filter
What you get
- Flood map: flooded land, permanent water and water that went back
- Areas in km² and share of the area
- Radar before, after and change composite
- Histogram of backscatter with the threshold
- Water area through every pass around the flood
- Flood polygons (GeoJSON) and table (CSV)
How it works in the field
Method and equations
Radar images
Sentinel-1 C-band (5.4 GHz) interferometric wide-swath images, radiometrically terrain-corrected to gamma-nought at 10 m, VV polarisation, converted to decibels. Passes are grouped by relative orbit, direction and day, and joined where one frame does not cover the area.
Water and flood
Open water is found below a threshold chosen by the minimum-error method of Kittler and Illingworth on the histogram of the after image (kept between −24 and −14 dB, −18 dB when the water is too little to form a peak). Flooded land is water after but not before, and by default must also be at least 3 dB darker than the dry reference, the mean of up to three passes of the same orbit. A 3 × 3 median filter reduces speckle.
Terrain and area
Slopes steeper than 5°, computed from the elevation model, are left out because radar shadow looks like water. Each pixel is given its true ground area from its latitude.
Accuracy and error limits
- Open floodwater on flat land is mapped with 85–95 % agreement with reference maps in published tests; its edge is located to one or two pixels (10–20 m). Areas are therefore within about 5–10 % for large floods and less certain for small ones.
- What is not covered: water between buildings, under forest or below standing crops (often not dark, so underestimated); flash floods that drain before the next pass (passes are 6–12 days apart); wind-roughened water (missed) and smooth dry surfaces such as sand, tarmac and airports (false water); and steep terrain, which is left out.
- To reduce the error: use a reference from the same orbit, keep the 3 dB change test, set the threshold by hand when the histogram shows two clear peaks, check the change composite, and compare with optical images or field reports when the sky clears.
Who uses it
Standards and references
- Torres, R. et al. (2012). GMES Sentinel-1 mission. Remote Sensing of Environment, 120, 9–24.
- Small, D. (2011). Flattening gamma: radiometric terrain correction for SAR imagery. IEEE Transactions on Geoscience and Remote Sensing, 49(8), 3081–3093.
- Kittler, J. & Illingworth, J. (1986). Minimum error thresholding. Pattern Recognition, 19(1), 41–47.
- Martinis, S., Twele, A. & Voigt, S. (2009). Towards operational near real-time flood detection using a split-based automatic thresholding procedure on high resolution TerraSAR-X data. Natural Hazards and Earth System Sciences, 9, 303–314.
- Twele, A., Cao, W., Plank, S. & Martinis, S. (2016). Sentinel-1-based flood mapping: a fully automated processing chain. International Journal of Remote Sensing, 37(13), 2990–3004.
- UN-SPIDER (2019). Recommended Practice: Flood Mapping and Damage Assessment Using Sentinel-1 SAR Data. United Nations Office for Outer Space Affairs.
At a glance
- Category
- Natural hazards
- Runs on
- Phone, tablet and desktop browser
- Export
- Flood map and radar views as PNG, JPG and PDF; flood polygons as GeoJSON; areas as CSV; charts as PNG, JPG and PDF; full PDF report
- Access
- Free
Run this survey on your next site
All 64 tools, report export and the AI assistant come with one subscription.