Skip to content

EuroFlood

Query Europe's satellite flood-depth maps as GeoDataFrames, in a few lines.

The CEMS-EFAS archive is open but has no index: just a coordinate in each filename, which can sit 2,870 km from the flooding. EuroFlood discovers where and when floods happened across Europe from a published index (zero config), then downloads only the depth rasters you want, no ~19 GB bulk archive, no setup.

PyPI version License: MIT CI Python 3.13

Flood recurrence over Zutphen on the IJssel Flood recurrence over Zutphen on the IJssel

Get started Tutorials Case studies API reference

Install

pip install "euroflood[viz]"

uv add "euroflood[viz]"

Three lines

import euroflood as ef

cat = ef.floods("Zutphen, Netherlands")  # discover: a GeoDataFrame, one row per flood
cat.plot()                                # visualize: recurrence heatmap, no download
cat.download().stats()                    # measure: fetch depth rasters + per-event stats

Zero configuration: the published index is read remotely and cached on first use.

See it across Europe

A decade of observed flood recurrence across Europe (2015–2024), computed entirely from the 138 MB index.

Every place the archive has seen flooded, 2015–2024: 376,857 km² in all, computed entirely from the 138 MB index without downloading a single depth raster. Then zoom into any of it: the case studies work through real events, from Storm Boris to the Valencia DANA.

Why EuroFlood

  • One API, all of Europe

    Continental coverage from ~3,280 Sentinel-1 flood-depth maps (2015–2024), indexed so a query streams a few MB, never the ~19 GB archive.

    How it works

  • GeoDataFrames, not tiles

    Every query returns a GeoPandas GeoDataFrame: filter, clip, join, and export with the tools you already use.

    Discover & filter

  • Visualize without downloading

    Recurrence heatmaps, per-event footprints, and interactive Leaflet maps straight from the index, no rasters fetched.

    Visualize

  • Global hazard maps

    The same API queries modelled CEMS-GLOFAS hazard depth for seven return periods (10–500 yr): ef.hazard("Zutphen", return_period=100).

    Hazard

  • CLI & offline

    Every query is also a euroflood command; mirror the index once for fully offline / HPC use.

    CLI & configuration

  • Cited & reproducible

    A versioned, checksummed index bundle with a manifest: pin a version, cite the source, reproduce a result.

    Citing EuroFlood

Learn by doing

The tutorials take you from a first query to hazard maps and quantitative analysis: runnable notebooks that follow one place (Zutphen, on the IJssel) throughout. Then see the case studies for real-world analyses of genuine events across Europe.

EuroFlood code is MIT-licensed. The index and flood-depth maps derive from the JRC / Copernicus CEMS-EFAS dataset (CC-BY-4.0). See Citing EuroFlood for how to cite the software, the index, and the source data. Developed at Princeton University (Complex Infrastructure Systems Group).