How we track drought in the world’s breadbaskets
What the dashboard shows, where the numbers come from, how current they are, and what they can’t tell you.
The short answer
Most of the world’s grain grows on farmland watered only by rain. When a dry spell hits at sowing, flowering or grain filling, harvests fall. This dashboard rates 26 of the main rain-fed crop regions on every inhabited continent on two measures that share no data: rain over the last three months from rain gauges, and soil moisture from a model. It counts a region as in drought when both agree, shows how many regions are in drought at once compared with a typical month since 1992, adds the official US Drought Monitor for the US, and lists the countries FAO says need outside help to feed their people.
How we know
Rain. The Global Precipitation Climatology Centre (GPCC), run by Germany’s weather service DWD, collects rain-gauge reports that weather services exchange worldwide and turns them into monthly rainfall on a one-degree grid. Three to five days after a month ends it publishes a first estimate from the gauges that report quickly (the First Guess Product). About two months later it publishes a quality-checked analysis with more gauges (the Monitoring Product, which goes back to 1982); we then replace the first estimate with it. For each crop region we average two or three grid points in its main growing area, add up the last three months, and rank the total against the same three months in every year from 1991 to 2020. In a dry season, when those three months normally bring less than 30 mm, we use the last twelve months instead.
Soil moisture. NOAA’s Climate Prediction Center (CPC) runs a simple model of the soil as a bucket about 1.6 metres deep, filled by rain and emptied by evaporation and runoff (Fan and van den Dool, 2004). It publishes the result every month for the whole world on a half-degree grid, back to 1948. We read the same points and rank the latest month against the same month in 1991–2020.
From rank to drought. On either measure, a value among the driest 20% of the 30 comparison years counts as drought. That is the threshold at which the US Drought Monitor’s “moderate drought” class begins. A region is “in drought” on this page when both measures are that dry, and “dry on one measure” when only one is.
Why two measures, and why both must agree. Neither is reliable alone, and we can show it where an official drought map exists. For May 2021 to October 2026 we compared both with the US Drought Monitor, calling a region officially in drought when at least half the area of its states was in drought. In the Corn Belt (Iowa, Illinois, Indiana) the official map had 8 such months; NOAA’s soil model called 45, DWD’s gauges 17. On the southern Great Plains (Kansas, Oklahoma) the official map had 31; the soil model called 25 and was right in 23, while the gauges called 15 and were right in 12; no gauge has reported inside our two grid cells there since 2020, so GPCC estimates their rain from gauges further away. Counting only months when both agreed gave 15 in the Corn Belt (5 of them right) and 10 on the Plains (all right). So requiring both gives fewer false alarms and misses real droughts; the “dry on one measure” group is where the missed ones are.
A known problem with the soil model’s rain. NOAA’s model runs on NOAA’s own rain analysis, and since about 2020 that input has drifted away from the gauges in several regions. For northern Nigeria it shows three to four times the 1991–2020 average in each year from 2020 to 2023, where GPCC shows 75% to 132%. For Mato Grosso it shows 54% to 79% of average in every year from 2020 to 2025, where GPCC shows 71% to 116%. For the Paris Basin it shows 36% for 2025, where GPCC shows 96%. Each night we compare the two over the last two years, and where they have moved apart by more than they did in almost any two-year period before 2020, the table marks the region’s soil moisture as doubtful.
What we used before. Until November 2026 the page ranked soil moisture alone and showed DWD’s GPCC Drought Index as a second opinion. Checked against past droughts at our points, that index stayed at normal levels through Germany’s drought of 2018 and the Canadian Prairies’ of 2021, while the plain rain totals from the same institute were among the driest of the record. We now rank the rain totals ourselves.
Why not satellite or daily data? We first tried NASA’s POWER service, which offers daily soil wetness. Its most recent weeks come from a different model than its long record, with a jump on the changeover day, so the latest values cannot be compared with history.
The United States. The US Drought Monitor is a weekly map drawn by experts from the National Drought Mitigation Center, the US Department of Agriculture and NOAA, combining soil moisture, rainfall, streamflow and reports from the ground. Where such a map exists it is the better guide, so the two US regions carry its figures for their states. We also show the share of each main farm state in each drought class, and the share of the lower 48 states in severe drought or worse over five years.
Food crises. FAO’s Global Information and Early Warning System lists the countries expected to need outside help with food, updated three times a year, with the main reasons for each. Bad weather is among the reasons for most of them, usually together with conflict, displacement or high food prices.
How current each part is
- Rain worldwide: monthly. DWD’s first estimate appears three to five days after the month ends; the quality-checked analysis replaces it about two months later, which can shift a region’s rank.
- Soil moisture worldwide: monthly; NOAA adds each month in the first week of the next one. Until both have arrived, the headline says which month each measure refers to.
- US Drought Monitor: weekly, released on Thursdays for the week to Tuesday.
- FAO list: three times a year.
- El Niño state: from our El Niño dashboard, monthly.
What the dashboard can’t tell you
Two or three grid points stand for a large region, and some grid cells have no rain gauge of their own, so their rain is estimated from gauges nearby; the table says where. Thirty years of history give only a coarse ranking at the extremes.
Outside the US we have no official drought map to check against, so a region shown as dry on one measure may be in a real drought or may reflect an error in one data set. National services such as Météo-France, Brazil’s Monitor de Secas or Australia’s Bureau of Meteorology are the better guide for their countries.
The soil model has no irrigation, so heavily irrigated regions (Punjab, the Nile delta, California’s Central Valley) are left out. The page cannot say whether a dry spell falls in a critical crop stage; a dry month after harvest matters little.
It is not a harvest forecast. For that, see the GEOGLAM Crop Monitor and national agricultural services.
Further reading
- GPCC Monitoring Product (Schneider et al., 2022), Deutscher Wetterdienst: monthly land-surface precipitation from rain gauges. doi.org/10.5676/DWD_GPCC/MP_M_V2022_100
- GPCC First Guess Product (Ziese et al., 2011), Deutscher Wetterdienst: the first monthly estimate. doi.org/10.5676/DWD_GPCC/FG_M_100
- Fan & van den Dool (2004), Journal of Geophysical Research: the CPC global monthly soil moisture data set. doi.org/10.1029/2003JD004345
- Iizumi et al. (2014), Nature Communications: impacts of El Niño and La Niña on the global yields of major crops. doi.org/10.1038/ncomms4712
- US Drought Monitor: the weekly map and how its classes are defined. droughtmonitor.unl.edu
- FAO GIEWS: countries requiring external assistance for food. fao.org/giews
- GEOGLAM Crop Monitor: monthly expert assessment of crop conditions worldwide. cropmonitor.org
Credits
Rainfall from the Global Precipitation Climatology Centre at Deutscher Wetterdienst (CC BY 4.0); soil moisture from NOAA’s Climate Prediction Center; the US Drought Monitor from the National Drought Mitigation Center at the University of Nebraska–Lincoln, USDA and NOAA; the list of countries from FAO’s GIEWS. The ranks, the rating of each region and the counts are our calculations from their public data.
← Back to the dashboard