Parametric flood trigger sheet on real GloFAS data

Parametric flood trigger sheet · simulated model flow (GloFAS v5.0)

Mengíbar gauge (CEDEX 5004)

Guadalquivir · pin 37.984° N, 3.799° W · model cell 37.975° N, 3.825° W (2.5 km away)

Model cell chosen by: long-term mean flow closest to the gauge's, within 0.1°.

Demo: the "reported flood years" are the years the real gauge measured above its own 1-in-5 flow, standing in for a client's loss years. 27 water years (October to September) in which the gauge and the model overlap.

Attachment · 1-in-5

295 m³/s

payout starts above this

Exhaustion · 1-in-25

534 m³/s

100 % at or above this

Years that would have paid

7 of 27

0 in full · first 1985

Average payout per year

14 %

of the limit, over the record

Highest daily river flow in the flood year (October to September) at the model cell · 1985–2022 (27 years)

Bars: simulated model flow (GloFAS v5.0), highest day in each flood year. Lines: attachment and exhaustion. Labels: payout that year. ▲ marks the reported flood years.

Return levels (GEV, simulated model flow (GloFAS v5.0))

Flood	Level m³/s	90 % band
1-in-2	161	117–221
1-in-5attachment	295	215–371
1-in-10	394	293–467
1-in-25exhaustion	534	416–592
1-in-50	649^†	515–711

† Beyond the 27-year record: an extrapolation. Read it with its band.

Caveats

This is simulated model flow (GloFAS v5.0), not a gauge reading. The model can miss or exaggerate a flood.
This demo uses the 27 water years (1985–2022, with gaps) in which the CEDEX gauge and GloFAS both have a full record, not the full 1980–2024 model record a client run uses.
The global model is not reliable on catchments smaller than about 500 km². This gauge drains 16,166 km².
Return periods longer than the record (27 years) are extrapolations: read them with their band.
Dams upstream on the Guadalquivir can cut real peaks that the model may not see.
Re-test the trigger whenever the model version changes.

The trigger

Each flood year, take the highest daily simulated model flow (GloFAS v5.0) at the model cell. At or below 295 m³/s (1-in-5) nothing pays. At or above 534 m³/s (1-in-25) 100 % of the limit pays. In between, the payout rises in a straight line.

Linear from 0 % at the attachment to 100 % at the exhaustion, on the flood year's highest daily flow.

Basis risk against 6 reported flood years

Hits: 5
Misses: 1
False payouts: 2
Rank score: 0.93

In 93 % of pairs, a flood year had a higher model flow than a year without a flood (50 % is chance; this is strong).

Here the "flood years" are the 6 years in which the real CEDEX gauge measured more than its own 1-in-5 flow (264 m³/s); in a client run they are the client's own loss years.

Provenance

source: GloFAS
dataset: cems-glofas-historical
version: version_5_0
via: Copernicus EWDS, through kai-engine reference guadalquivir
window: 1984-10-01 to 2022-09-30
gauge: CEDEX Anuario de Aforos 2021-2022, station 5004 (used only to pick the cell and for the flood years)
input: kai-website public/assets/kh/triggers_v5.json
sha256: 4b367b94bec5eba973a747664d4f60dc3057c862f525c4131911dea204966e47
generator: hydro-saas scripts/trigger-fixture.mjs

Fuente: CEDEX, Anuario de Aforos 2021-2022 (Ministerio para la Transicion Ecologica y el Reto Demografico). Contains modified Copernicus Emergency Management Service information 2026 (GloFAS v5.0).

KAI Water & Energy · KAI-Hydro parametric flood trigger design. Built on simulated model flow (GloFAS v5.0); it is a design aid, not a flood warning and not a gauge record.