Climate Acute RisksFlood depth — 10-year return, present climate (GIRI)

Flood depth — 10-year return, present climate (GIRI)

This layer estimates how deep floodwater would stand at a given location during a flood of the kind expected roughly once a decade, under present-day climate.

Source: GIRI (UNEP-GRID / CDRI) Year: 1979-2016 Category: Physical risks · Mitigating services · Climate Acute Risks Coverage: Global Format: Raster grid (~90 m) Used in risk analysis: Yes — gates Mitigating services

What it shows

This layer estimates how deep floodwater would stand at a given location during a flood of the kind expected roughly once a decade, under present-day climate. By giving a site-level water depth rather than a broad category, it can tell apart floodable and non-floodable ground within river corridors, deltas and urban areas, making it well suited to screening flood exposure precisely.

How it is built

The layer is produced by UNEP-GRID for the Global Infrastructure Resilience Index (GIRI), developed with the Coalition for Disaster Resilient Infrastructure. It is based on a distributed hydrological model (Silvestro et al., 2013, 2015) driven by a global meteorological dataset for the present-climate run, coupled with a hydraulic model that converts simulated river discharge into flood depth. The output is a global grid of inundation depth at roughly 90 m (3 arc-second), for a 10-year return period and a present-climate baseline (1979–2016). It represents both river and rainfall-driven flooding; coastal storm surge is not included and is covered by a separate coastal-flood layer. Forward-looking climate-scenario versions are available as separate layers.

How to read it

Higher values mean deeper projected floodwater at that location for a once-in-a-decade flood, indicating greater exposure. Locations with little or no depth are effectively outside the modelled flood footprint at this return period.

Class thresholds

The underlying layer models a physical flux: flood inundation depth, in centimetres (cm) — the modelled depth of standing floodwater at each ~90 m cell for a 10-year return period under the present-climate baseline (1979–2016). The raster stores depth in centimetres; the five risk classes map to the following thresholds (shown in metres, as the legend labels them):

ClassFlood depth (10-year return)
Very low< 0.5 m (< 50 cm)
Low0.5 – 1 m (50 – 100 cm)
Moderate1 – 2 m (100 – 200 cm)
High2 – 4 m (200 – 400 cm)
Very high≥ 4 m (≥ 400 cm)

A site is flagged on the Mitigating services dimension when the layer value exceeds 100 cm (1 m) of flood depth — i.e. it reaches the Moderate class or above — and the entity's matching ENCORE flood-mitigation-services pressure is also material.

How these thresholds were set. This is a Darwin in-house classification: round metre-scale cutoffs (0.5, 1, 2, 4 m) applied to the continuous GIRI flood-depth surface to bin it into five interpretable hazard classes. The breaks are not taken from GIRI or any external published standard — GIRI publishes a continuous depth raster, and the class boundaries are Darwin's own choice of round-number depth thresholds.

Two deliberate threshold choices, noted for transparency.

  • The bands describe submersion severity, not event rarity. This layer and the coastal flood-depth layer use the same 0.5 / 1 / 2 / 4 m bands even though they represent different return periods (GIRI here is a 10-year event; the coastal layer is a 100-year event). The bands answer "how deep would the water be?", which is comparable across return periods; they do not encode how often the flood occurs. A 1 m depth at a 10-year return is a more frequent hazard than 1 m at a 100-year return, so the return period in each layer's title should be read alongside the class.
  • The same layer is binned two ways for two purposes. The site-level map and proximity flag use the absolute physical bands above; the country/region risk profiles re-bin the same GIRI surface by 20/40/60/80 quantiles ([188, 261, 360, 450] cm). This is intentional — absolute depth is what a site needs, while a relative ranking is what comparing whole countries needs (and it mirrors WWF's quantile approach, see below).

Source

GIRI — Global Infrastructure Resilience Index (UNEP-GRID / CDRI, 2023). Licensed under CC BY 3.0 IGO. https://giri.unepgrid.ch/.

Comparison with the WWF Risk Filter Suite

This layer maps to the WWF Water Risk Filter (WRF) "Flooding" risk category (and the Biodiversity Risk Filter S3_7 Flooding theme). In WWF that category is a composite of two basin-level sub-indicators, both aggregated to HydroBASINS Level 7 catchments:

WWF sub-indicatorSourceAggregationClass thresholds
Flood Occurrence (B3_1)Dartmouth Flood Observatory — count of large flood events 1985–2021Sum per basin0 · 1–2 · 3–10 · 10–30 · >30 events
Flood Hazard (B3_2)JRC Global River Flood Hazard Maps — RP100 max depth per pixel (0 outside extent)Median per basin, then even quantiles≤0.0514 · ≤0.0530 · ≤0.0567 · ≤0.0718 · >0.0718 m

Why our thresholds differ from WWF's

At first glance WWF's flood-hazard class breaks (≈5–7 cm) look wildly lower than Darwin's (0.5–4 m). This is not a real disagreement — the two layers measure different things and bin them on different principles:

  • Unit of analysis. WWF takes the median depth over an entire HydroBASINS L7 basin, where most pixels lie outside the 100-year flood extent and are recorded as 0 m. That median is therefore diluted to a few centimetres and only carries meaning relative to other basins. Darwin reads the modelled depth at the individual ~90 m pixel under the site, so the number is the actual water depth at that location.
  • Relative vs. absolute classification. WWF classifies by quantiles (each class is one-fifth of all basins), so its breaks are a ranking, not a physical depth. Darwin uses absolute physical depth bands (knee-deep ≈ 0.5 m, etc.), which is what site-level screening needs.
  • Different hazard and return period. WWF's hazard component is JRC fluvial RP100; Darwin's GIRI layer is fluvial + pluvial at RP10.

In short, the difference is justified by design: Darwin's purpose is to discriminate floodable from non-floodable ground within a basin, which a basin-median quantile cannot do. Reusing WWF's centimetre-scale breaks would be meaningless at pixel level.

Where Darwin does aggregate this layer to a coarser unit — the country/region risk profiles — it switches to a WWF-style 20/40/60/80 quantile classification (thresholds [188, 261, 360, 450] cm over ~450 geographies), i.e. methodologically aligned with WWF's relative approach, differing only in the aggregation unit (country vs. basin) and source (GIRI vs. JRC/Dartmouth).

Risk analysis

A site is flagged on a dimension by combining a proximity trigger (this layer) with an activity trigger (the entity's ENCORE pressure/service). Proximity only → Potentially material; proximity and the matching ENCORE pressure/service is material → Very material; neither → Not material.

DimensionENCORE service / pressureProximity trigger (this layer)Activity trigger (entity)
Mitigating servicesFlood mitigation servicesLayer value above 100“Flood mitigation services” pressure ≥ 4

Legend

Symbolised field: Flood depth (10-year return)

Value (Flood depth (10-year return))LabelColour
1Very low (<0.5 m)#fee08b
50Low (0.5–1 m)#fdae61
100Moderate (1–2 m)#f46d43
200High (2–4 m)#d73027
400Very high (>4 m)#7b3294

Generated from darwin/layers/layer-flood-depth-giri-10yr-present.toml and risk_indicator_pairs.toml (develop).