Skip to content

WRI Aqueduct riverine flood risk — introduction#

The WRI Aqueduct Global Flood Analyzer measures river flood risk — the expected exposure of GDP, population, and urban area to riverine flooding — aggregated by administrative unit across the globe. earthlens ships a single aqueduct backend that reads the 2015 Analyzer data directly from WRI's public file host (files.wri.org, CC-BY-4.0, no credentials) and returns the admin polygons carrying the risk attributes.

This page orients the backend. For the hands-on download walkthrough see Usage; the rendered API is the Reference page.

What it is (and is not)#

Aqueduct ships two distinct products, and this backend covers only one:

  • The flood impact / exposure tables — "how much GDP / population / urban area sits in the flood zone" summed per admin unit. This is what the aqueduct backend fetches.
  • The flood hazard depth rasters — the gridded inundation-depth maps. Those are a separate product, already reachable through the gee backend as the WRI/Aqueduct_Flood_Hazard_Maps/V2 asset. The aqueduct backend does not duplicate them.

Why it matters here#

Like the GDACS and risk-indicators backends, Aqueduct departs from the gridded backends (CHC rainfall, ERA5, GEE imagery) in two ways:

  • The output is a vector table, not a grid. A query returns the admin polygons — country, state, or river basin — each carrying one exposure column per flood return period. So Aqueduct is a vector backend (Aqueduct.OUTPUT_KIND == "vector"), and download() returns a pyramids FeatureCollection (a geopandas.GeoDataFrame subclass) rather than writing a raster. A geometry=False request returns a geometry-dropped pandas.DataFrame instead. Because there is no meaningful gridded reduction of an admin-aggregated exposure table, the facade rejects an aggregate= argument.
  • It is a static snapshot, not a time series. The 2015 Analyzer has no time axis, so start / end are optional; the "time" dimension is instead the choice of a 2010 baseline or a 2030 projection.

What is available#

Dimension Choices
admin_level country, state, basin
hazard riverine only (coastal is the paywalled 2020 product)
metric gdp_affected, population_affected, urban_damage
year 2010 (baseline) or 2030 (projection)
scenario baseline (2010); or 2030 climate × socio-economic combinations
return_period 2, 5, 10, 25, 50, 100, 250, 500, 1000 yr

The 2030 scenarios combine a socio-economic pathway (SSP2 / SSP3 / baseline) with a climate pathway (RCP4.5 / RCP8.5 / baseline): ssp2-rcp4p5, ssp2-rcp8p5, ssp3-rcp8p5, base-rcp4p5, base-rcp8p5, ssp2-basehydro, ssp3-basehydro.

The values are per-return-period exposure, not a single pre-integrated expected-annual-damage figure.

Scope note — the 2020 product#

The richer 2020 Aqueduct Floods product (coastal flooding, 2050 / 2080 horizons, city level, pre-computed expected-annual damage) is not freely downloadable: its aggregated tables sit in a WRI S3 bucket that denies anonymous reads. Only the 2015 riverine Analyzer used here is public, so coastal / 2050 / 2080 / city selections are out of scope.

Licence#

The data is licensed CC-BY-4.0. Attribution: World Resources Institute 2015; Winsemius et al. 2013; Ward et al. 2013.