Global Extreme Weather and Climate Change Dashboard

Drought

Drought accounts for about 7% of global weather/climate economic losses but the single largest share of deaths, about 33%, 1980-2025 (see Loss Normalization) -- famine-driving droughts in Africa dominate that toll. "Drought" names three different deficits, and this page measures each one separately rather than folding them into a single number. Meteorological drought is a rainfall deficit. Agricultural drought is a soil-moisture deficit -- the water a crop can actually reach, which lags rainfall because the soil store has to draw down first. Hydrological drought is a deficit in river flow and the groundwater and snowpack feeding it -- the water people actually withdraw. Every index below ranks each location against its calendar-month history and classifies the result into the same USDM-equivalent D0-D4 categories, so the three panels read the same way and support direct comparison.

Which drought types show a detected increase, and where

Each of the three deficits below at its D2+ (severe or worse) default, against this site's combined detection standard: IPCC's likelihood criterion at each variable's multiple-comparisons-adjusted bar, plus a magnitude-vs-variability check. Meteorological and agricultural drought come from ERA5, hydrological from GloFAS — so where two or three agree, they agree across unrelated data.

no* marks a borderline case that clears the flat 10% bar and misses its adjusted one: Africa agricultural at p=0.066 against 0.051. Reported as what it is rather than rounded into agreement. Every verdict comes from the same computation the panels below render, so this table cannot disagree with them. See the Methodology page for the construction of each index.

Variables on this page

This site evaluates every variable below over a full reliable-trend window against the combined "detected change" standard (IPCC's likelihood criterion plus a magnitude-vs-variability check -- see the homepage or Methodology for the full definitions). Shown as alternative ways of characterizing this same phenomenon -- some from independent sources, others a different construction on the same underlying data (each variable's note says which).

The precipitation-only (SPI-style) series puts its five worst years at 2023, 2024, 2021, 2025 and 2022 — the Amazon and Panama Canal droughts, the Horn of Africa's worst drought in 40 years, and the Yangtze and European drought of 2022, each independently documented. Those are also the five most recent complete years, which a strengthening trend produces, and it matches how the river low-flow index already behaves. Before the 15 September 2026 weighting correction the ranking put 2021 first and 2011 second; area weighting moved the emphasis toward the tropics, where the recent droughts sit, and 2011 now ranks 16th.

All three drought types now show a detected global increase, and a 15 September 2026 audit explains why. The aggregation behind every index on this grid counted 5° cells equally, which handed Antarctica 3.08 times its share of the land area and handed Africa, Asia, South America and Oceania between 0.55 and 0.66 of theirs -- so the count discounted the continents where the signal sits. Weighting each cell by its spherical area lifted the precipitation-only global series from no detected change into a detected increase at every severity level, and the soil-moisture series with it. The audit also restricted "global" to 60°S–90°N, so the global figures and the six continental ones now describe the same land, and floored the Hargreaves evaporative demand at zero, which the equation had been returning negative for 21.9% of land cell-months. Full measurements sit in the repository underaudit/drought/.

The three types agree on where, and disagree on how much. Taking each at its D2+ (severe or worse) default, South America shows a detected increase on all three independent measures -- rainfall, soil moisture and river flow, built from two unrelated data sources. Africa shows one on rainfall and river flow, and its soil-moisture result clears the flat 10% bar while missing its multiple-comparisons-adjusted one. Asia and North America show one in soil moisture and nowhere else. Europe andOceania show none on any of the three. The continents showing no detected change count as much toward that result as the ones showing one.

Agricultural drought reached a detected global increase on 15 September 2026, having shown none before the weighting correction. The surface (0-7cm) series climbs more steeply than the root zone at every level, which is what a faster-responding shallow layer should do. Africa's soil-moisture result stays a genuine borderline case: it clears the flat 10% significance bar but not the multiple-comparisons-adjusted one, and themethodology page lists it among the results that flip.

Hydrological drought separates from rainfall. The river low-flow index shows a detected increase globally at every severity level (D2+: +0.76 percentage points per decade, p=0.013), and since the weighting correction the precipitation-only series at the same severity joins it. Before that correction it cleared statistical significance while staying inside this site's variability bar, which read as no detected change and as evidence that the drying sat in evaporative demand rather than in rainfall. Regionally the low-flow index detects an increase in Africa (+2.06 points/decade) andSouth America (+3.01 points/decade) and detects nothing in Asia, Europe, North America or Oceania -- the same two continents the precipitation-based index singles out, reached through a different measurement.

The index recovers documented events without being told about them: its worst years are 2023, 2024, 2022 and 2025 (the Amazon and Panama Canal droughts, the Yangtze and European drought of 2022), then 2015 and 1983, the two strongest El Ninos of the later 20th century. The record starts in 1982 rather than 1979 because GloFAS's model groundwater store is still draining over the reanalysis's first years; leaving those years in would suppress early drought and manufacture a drying trend out of a model artifact. That cutoff is not what produces the trend -- recomputing from scratch at start years from 1979 to 1995 moves Sen's slope only between +0.55 and +0.87 points per decade.

The precipitation-minus-AED (SPEI-style) series replicates and extends through the present the central finding of Vicente-Serrano et al. 2022 ("Global drought trends and future projections," Phil. Trans. R. Soc. A 380: 20210285): including atmospheric evaporative demand drives a much steeper drying trend than the precipitation-only series. Comparing 1980-2002 to 2003-2025, the precipitation-only series' global drought trend rises +3.01 percentage points, while the AED-inclusive series rises +6.06 points over the identical window -- about twice as far. 2023, the hottest year on record globally at the time and a documented case where extreme heat drove severe drought beyond what rainfall deficit alone would give, shows the AED-inclusive series at 23.15% against the precipitation-only series' 18.73% for that year. See the methodology page for the Hargreaves-method simplification this uses in place of the source paper's full Penman-Monteith calculation.

This page's meteorological drought series comes from a reanalysis, and an observational record does not reproduce its trend. ERA5 is a physical model constrained by whatever observations the assimilation could reach; where a rain-gauge network is dense that constraint is tight, and where it is sparse the model supplies what the observations do not. Select D2+ GPCP in the chart above to see the identical index -- same grid cells, same land mask, same three-month window, same percentile ladder, same area weights -- computed from GPCP satellite-and-gauge precipitation instead. Over the full record the ERA5 series gives +1.36 percentage points per decade (p < 0.001), a detected increase; the GPCP series gives −0.26 per decade (p = 0.195), no detected change.

An audit of that disagreement (September 2026) found it concentrated in the regions with the thinnest gauge coverage. Across Europe, North America and Oceania the two records agree and neither detects a rise. Africa and South America supply 85% of the ERA5 increase, and only South America's appears in the observations. A gauge-only record, GPCC, covers 1979–2019 and returns a detected decrease over that window where ERA5 returns a detected increase. Which record is right is not settled here, and this site publishes both rather than choosing. Treat the meteorological drought trend as a property of ERA5 until an observational record corroborates it; the soil-moisture and streamflow series on this page descend from the same ERA5 precipitation and are not independent of it.