River floods in Romania

# How high does the water reach where you live?

Type your village or town, drop the pin on your house and see what the European model shows for river floods that recur at 10, 100 and 500 years. Then raise the Danube on the Galați gauge and watch in 3D where the water goes.

## Map of water depth

Where do you live?

under 0.5 m
0.5–1 m
1–2 m
2–4 m
over 4 m
permanent water (river, lake)

Towns are searched in a list held on this page, and nothing is sent elsewhere. The “Find a street” button sends the text you typed to OpenStreetMap, and “My location” does not leave your device. Cells are about 90 m across, so the model cannot tell two neighbouring houses apart.

A “100-year” flood has a 1 in 100 chance of being reached or exceeded in any year. It does not arrive once a century: two can come a few years apart.

## The water on the ground, at Galați and Brăila

The Danube, the Siret and the Prut meet here. The bar below sets the Danube's level on the Galați port gauge, in centimetres. Water spreads from the Danube channel only onto ground it can reach without crossing a rise higher than its own level, and the water surface drops 3.2 cm per kilometre towards the sea. Touch the ground to see, in the model, the gauge reading at which water reaches it.

Touch the ground to see, in the model, the gauge reading at which water reaches it.

Danube level at Galați, in cm on the port gauge 450 cm

under 0.5 m
0.5–1 m
1–2 m
2–4 m
over 4 m
hatched: European map, 100 years

Diked areas such as Insula Mare a Brăilei, Brateșul de Jos and the Brăila–Dunăre–Siret enclosure show as flooded from a certain level, because the model has no dikes.

The 3D terrain cannot be shown on this device. The map above and the table in the check section give the same information.

## We ran the model for April 2006

On 26 April 2006 the Danube reached 661 cm on the Galați gauge and 699 cm at Brăila. We put the same level into the model and compared the result with the water seen from space that month. The source is the Global Flood Database, built from 250 m MODIS images: new water, outside permanent channels and lakes, on 41 km² of the 3,446 km² area.

How much of the water observed in April 2006 each model covers, and how much it shows in total

Model Shows in total (km²) Of the observed water, covers Of what it shows, matches observed water

The model on this page, tree-free terrain, at 661 cm 1,022 72% 3%

The same model, with the raw surface (trees included) 892 34% 2%

European JRC map, 100-year flood 1,158 91% 3%

European JRC map, 10-year flood 1,106 82% 3%

Area shown by each model against the water observed from space in April 2006, in the Galați–Brăila area. The satellite area is written on its bar. Sources are at the end of the page.

### What worked

The model covers 72% of the observed water, and the European map 91%. The correction matters a great deal: with the raw surface, coverage is only 34%, probably because the surface seen from space includes the crowns of the floodplain willows rather than the ground beneath. The correction has two parts, trees and built-up areas, and their effects were not separated. Under trees we put the height of the nearest open ground and subtracted 2 m in built-up areas; according to the documentation of the previous series, the European map uses a similar correction, 8 m under forest and 2 m in towns.

### What did not

At the crest level, the model shows about 25 times more water than the satellite saw over the whole period from 7 April to 24 May 2006. The last column of the table shows the same: only 3% of what the model shows matches the observed water. The difference comes, at least in part, from what the model does not know. Insula Mare a Brăilei, about 720 km², has been diked since 1964–1965 and did not fail in April 2006; according to the Brăila research station, on 15 April the emergency command was willing to decide on a controlled flooding of it, to relieve pressure on the Danube and defend the port of Galați, but the enclosure was spared. The station worked against the flooding, so it is an interested witness. The dikes in the Călărași–Galați sector were designed in 1965 for floods with an annual probability of 10%, not 1%. Before diking, Balta Brăilei flooded once every one to three years. Other diked enclosures in the area are Brateșul de Jos, 136 km², and Brăila–Dunăre–Siret, 53 km². With Insula Mare they add up to about 910 km², but part of Insula Mare lies outside the area and their exact outlines are not to hand, so it is not possible to say how much of the difference they explain.

The satellite has limits too. Images at 250 m may not see water under trees or through cloud, so the 41 km² are probably a minimum.

### How sensitive the result is

If the terrain in the model is 1 m higher relative to the level the gauge measures, coverage falls from 72% to 59%; if it is 0.5 m lower, it rises to 75%. The slope of 3.2 cm per kilometre does not improve this result: with no slope, coverage is 76%. It stays because the drop in level between Brăila and Galați shows in the two crests of 26 April 2006, but this test does not confirm it.

### What this does not show

Among the worst-hit places in 2006 were, upstream of this area, Rast and Bistreț in Dolj, where a dike failed: on the Bistreț polder 11,126 ha of arable land were flooded and 8,322 people were displaced (WWF, 2006). Along the Romanian Danube about 73,000 ha of plain were flooded in all, mostly in polders (Gâștescu and Țuchiu, 2012). Galați was flooded in July 2010, when the Danube overflowed and the sewer network could not cope with heavy rain. According to the Brăila station's paper, the Siret, at 2.5 to 5 times its 2006 flow, then raised the Danube at Brăila and Galați: at the confluence its water stood 1.80 m higher than at Galați. The model applies the Danube's level to the Siret and the Prut as well, so it cannot show such a backwater.

## What this page cannot know

Dikes and polders. The model on this page has no dikes, and the documentation of the previous series of the European map says the same of it. Inside a diked enclosure, the colour shows where the water would go if the dike did not exist.

Streams and flash floods. The European map rests on the flows of large rivers; the documentation of the previous series excludes basins under 500 km². A flood on a small stream does not appear here, though it can be the most dangerous.

Rain and sewers. Water that gathers from a heavy downpour or comes out of the sewers is not in the model. Galați was flooded that way too in July 2010.

Buildings and trees. The terrain comes from a surface model, Copernicus GLO-30, which includes roofs and tree crowns. We removed the trees, but buildings stay in the model: in town, the real street level may be lower than the model's.

Elevation. The error of the terrain could not be measured here, and the terrain heights and the gauge have no transformation between them: they were treated as identical, with heights from the EGM2008 model of the Earth and the gauge tied to the Black Sea at Constanța. On a floodplain, an error of one metre in height can move the edge of the flooded ground by kilometres.

The Galați gauge. Its zero, 0.42 m above the Black Sea at Constanța, comes from a statement by an engineer at the Lower Danube River Administration, quoted by local press in August 2026; the same source says works on the embankment may call for a new check.

Water slope. The 3.2 cm per kilometre is calculated from two crests, at Brăila and at Galați, on 26 April 2006, and from the distance between the gauges, which was not verified. The real slope depends on the flow.

The Siret and the Prut. Here they receive only the Danube's level. Their own flow is not in the model.

This page is not an official risk determination and does not replace the ANAR maps, site permits or the advice of the Inspectorate for Emergency Situations. For a decision about a house, a purchase or insurance, use the official map and ask for a permit.

## Frequently asked questions

How do I find out whether I live in a flood zone in Romania?
The official maps are the flood hazard and risk maps of Apele Române (ANAR), at harticiclul2.inundatii.ro. This page shows a European model next to them, at 90 m, which ignores dikes and is not an official map.
What does a 100-year flood mean?
A 100-year flood has a 1 in 100 chance of being reached or exceeded in any year. It does not come once a century: two can arrive a few years apart.
How high did the Danube rise at Galați in April 2006?
On 26 April 2006 the Danube reached 661 cm on the Galați port gauge, 61 cm above the flood level of 600 cm, and 699 cm at Brăila (Gâștescu and Țuchiu, 2012).
Where is the zero of the Galați gauge?
The zero of the Galați port gauge is 0.42 m above Black Sea level at Constanța, according to an engineer at the Lower Danube River Administration quoted in local press in August 2026; the same source says the value may need a new check.

## Sources and licences

Water depth at 10, 100 and 500 years: Copernicus Emergency Management Service / European Commission, JRC, “River flood hazard maps for Europe and the Mediterranean Basin region”, version 3.1.1, 3 arc-second cells (about 90 m), licence [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/). We clipped the map to Romania and coloured the depths. [Dataset](https://jeodpp.jrc.ec.europa.eu/ftp/jrc-opendata/CEMS-EFAS/flood_hazard/). Documentation of the previous series: Dottori et al., 2022, [Earth System Science Data 14](https://essd.copernicus.org/articles/14/1549/2022/), which states that the maps do not include dikes.

Terrain: Copernicus DEM GLO-30 (ESA / Airbus), 30 m. Land cover: ESA WorldCover 2021, 10 m, [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/).

The official map: Administrația Națională “Apele Române”, flood hazard and risk maps, cycle II, [harticiclul2.inundatii.ro](https://harticiclul2.inundatii.ro/). Link only, no reproduction.

April 2006, levels and damage: Gâștescu, P., Țuchiu, E., 2012, conference paper, Tulcea, [limnology.ro](https://www.limnology.ro/water2012/Proceedings/001.pdf) (661 cm at Galați and 699 cm at Brăila, flood level 600 cm, enclosures that failed); WWF, 2006, [2006 Floods in the Danube River Basin](https://wwfeu.awsassets.panda.org/downloads/2006_floods_in_the_danube_river_basin_wwf_working_paper.pdf).

Diked enclosures: SCDA Brăila, “Diked enclosures on the Danube: current problems”, [asas.ro](https://www.asas.ro/noutati/interes-public/Incintele%20indiguite%20la%20Dunare%20-%20probleme%20actuale%20si%20de%20perspectiva.pdf); flood risk management plan, [inundatii.ro](https://inundatii.ro/).

Zero of the Galați gauge: Adrian Maizel, RIS Hydrographic Bureau, AFDJ Galați, quoted by [Viața Liberă](https://www.viata-libera.ro/economie/263525-cota-dunarii-la-galati,-la-minimul-ultimilor-30-de-ani-cat-este-diferenta-fata-de-nivelul-marii), 18 August 2026.

Water seen from space, April 2006: Global Flood Database (Tellman et al., Nature, 2021), Dartmouth Flood Observatory event 2839, used only to compute the numbers in the check section; it is not published here.

Places: GeoNames, [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/). Outlines: Natural Earth, public domain.

Translated by Marius Comper.

Share