US 1-in-1000-year rainstorms become much more common
Today’s thousand-year three-day downpours over the United States could strike several times more often at 2–4 °C of global warming.
Source
Informing Future Risks of Record-Level Rainfall in the United States
Study at a glance
- Design
- Other — Pooled GPCC extremes mapped onto CMIP5 RCP8.5 warming levels for CONUS regions
- N
- N=834 · 834 GPCC 1° cells clustered into 15 precipitation-regime regions (1988–2013)
- Population
- CONUS extreme 3-day rainfall regimes under historical and warmed climates
- Outcome
- Frequency increases of present-day 1000-year 3-day events at 2–4 °C warming
Structured fields used in claim comparison tables when every cited study has a complete layer.
What they did
The authors clustered 834 GPCC 1° cells into 15 precipitation-regime regions, pooled 1988–2013 annual-max 3-day rain to estimate 1000-year events, then mapped those events onto CMIP5 RCP8.5 climates at 0.67, 1.5, 2, 3, and 4 °C above 1851–1880.
What they found
Historical 1000-year 3-day events occur 2–5 times more often at 2 °C and 5–10 times more often at 4 °C, depending on region. All clusters intensify 500-year events; East Coast intensity more than doubles while the western-states cluster changes less. Only four models reach 4 °C in RCP8.5.
The limits
What it doesn't show
1° grids hide local orographic extremes. Pooling assumes independence that is only tested, not guaranteed. High-warming results use a four-model subset. The 26-year GPCC record is short even after spatial pooling.
Key terms
- Return period
- Average wait between events of a given magnitude; a 1000-year event has a 0.1% chance each year in a stationary climate.
- GEV
- Generalized extreme value distribution fitted to block maxima of 3-day precipitation.
- Spatial clustering
- Grouping grid cells with similar seasonal rain statistics to enlarge the extreme-value sample.
Flashcards
Research intelligence for this paper
See its role on concept claims, tensions it is part of, placement history, and related discoveries.
Quiz yourself
The paper’s target rarity is a:
Common questions
Why cluster stations instead of fitting each grid cell?
A 26-year record cannot constrain a 1000-year event; pooling ≥40 cells yields at least 1040 grid-years per cluster.
Why 3-day totals?
They proxy flood-producing storms better than 1-day totals for regional risk.
Is 4 °C sampled by the whole CMIP5 ensemble?
No; only a starred subset of four models reaches 4 °C under RCP8.5.
More on Extreme events