GWX — Flooding

Flooding

Floods account for about 30% of global weather/climate economic losses and about 16% of deaths, 1980-2025 (see Loss Normalization), the second-largest share of both. River discharge here comes from GloFAS (ERA5-forced LISFLOOD reanalysis), sampled at 1,074 pour-point cells: a continuous high-flow percentile metric and GloFAS's Alert/Warning/Severe return-period tiers, both starting in 1982 after excluding the record's documented model spin-up years.

Summary Judgment

Consistent with IPCC AR6

AR6 itself declines to state a clean global river-flood direction, citing regional inconsistency. The data here shows the same pattern: no detected change globally once corrected for the number of variables tested at once, masking sharp regional divergence underneath (decreasing in Africa/South America, no change in Asia/Europe/North America/Oceania).

IPCC finding (AR6 WG1 Ch.11 Executive Summary): "Significant trends in peak streamflow have been observed in some regions over the past decades (high confidence)... Regional changes in river floods are more uncertain than changes in pluvial floods because complex hydrological processes and forcings, including land cover change and human water management, are involved." — see the full Detection & Attribution comparison for every phenomenon this site tracks.

Flooding — River Discharge High-Flow Extent

Monthly

% of 1,074 GloFAS 'pour-point' cells (the highest-upstream-area grid cell in each of the same 5° blocks drought.py uses) classified two ways -- above the 90th percentile of that same cell's same-calendar-month history ('high flow'), or exceeding GloFAS's published 2/5/20-year return-period discharge threshold ('Alert'/'Warning'/'Severe') -- averaged to one monthly figure, selectable below by region and metric. Unit: % of pour-point cells.

Series

Source: GloFAS (Global Flood Awareness System), Copernicus Early Warning Data Store

Computing trend…

Select a region and metric above -- the invented "high flow" percentile series, or GloFAS's Alert/Warning/Severe return-period tiers. See Methodology for the pour-point construction and the 1979-1981 exclusion.

Both series -- the percentile-based "high flow" metric and the GloFAS return-period-based Alert/Warning/Severe tiers -- match documented flood events independently: once the record's early-year model spin-up bias is excluded (see the panel's note above), each region's highest-flood months line up with well-known historical floods -- 1998 (the El Niño-driven East African floods and the catastrophic Yangtze River floods in China), 1982-83 (one of the 20th century's strongest El Niño events, with severe flooding across Peru and Ecuador), 1988 (the historic Khartoum/Sudan flood, one of the worst in that country's history), 2010 (the catastrophic 2010 Pakistan floods) and 2010-11 (the La Niña-driven Queensland floods), 2019 (the historic US Midwest/Missouri-Mississippi basin flooding), and 2022-2024 (the 2022 Lismore/eastern Australia floods, Storm Daniel's Greece/Balkans flooding, and the historic 2024 West/Central African floods).

Regional literature context

Independently verified literature findings for each continent, researched before the live GloFAS pipeline above existed -- kept here as additional context the live percentile index alone doesn't capture (e.g. within-continent north/south divergence, gauge-record-length caveats). Full citations in the Library.

Oceania (Australia). Multiple peer-reviewed studies (Ho, Nathan, Wasko, Vogel & Sharma 2022; Wasko & Sharma 2017; Wasko & Nathan 2019) document a north/south divergence in Australian flood behavior under warming: rainfall extremes have intensified in the country's north while runoff efficiency and flood response have weakened in parts of the south, meaning rainfall trends and flood trends don't move together -- a distinction the single continent-wide series above can't show on its own.
Africa. Tramblay et al. 2021 (399 gauges, 1981-2018) found African flood magnitude is more strongly driven by antecedent soil moisture than by rainfall extremes.
Europe. Blöschl et al. 2019 found no single continent-wide trend, but clear opposing regional patterns -- increasing floods in the northwest, decreasing in the south and east, each with a distinct physical driver.
North America. Hodgkins et al. 2017 found no widespread significant trend in major-flood occurrence (1961-2010) across 1,200+ minimally-altered gauges -- dominated by multidecadal variability.
South America. Barichivich, Espinoza et al. 2018 found severe Amazon flooding at Manaus has increased roughly fivefold since the early 2000s, linked to a strengthened Walker circulation.
Asia. Do, Westra & Leonard 2017 found Asia's average streamflow gauge record length (~17 years) runs short relative to other continents -- a data-coverage caveat for gauge-based studies, distinct from the live GloFAS series above (which is model-derived, not gauge-length-limited, but carries separate limitations -- see the methodology page).