Forecast of extremes
Where the forecasts for the coming days depart most from normal: percentiles of temperature, 850 hPa temperature and 500 hPa geopotential height against 1991–2020
Forecast vs climatology
Percentiles of the ECMWF forecast against the ERA5 1991–2020 climatology.
Regional summary
- Forecast cycle
- —
- Generated
- —
Percentile = position of the forecast value within the distribution of the same day of the year in 1991–2020 (ERA5 reanalysis, ±7-day window, 1° grid), with temperature and geopotential adjusted for the 1991–2020 trend to the 2026 climate. Temperature, T850 and Z500 use the daily mean; precipitation, gust, CAPE, shear and “severe” show only the high percentiles and use the day total or maximum. “Record” means outside the range observed in that period. Data: ECMWF IFS (open data, CC BY 4.0) · ERA5, Copernicus C3S · WeatherBench 2 · ARCO-ERA5 · Visualisation: Luso Meteo
📖 About the data
Each map compares the ECMWF IFS forecast with what was observed on the same day of the year in 1991–2020 in the ERA5 reanalysis (±7-day window, 1° grid), and shows the percentile of that forecast. Contour lines show the forecast absolute values; tap or click the map to see the forecast value, the normal and the percentile at that point. For the “extremes” variables (precipitation, gust, CAPE, shear, severe) only percentiles above normal are shown, and only where the value is relevant (for example precipitation ≥ 1 mm).
Days are complete UTC days; the first partial day is not shown, and gusts are only available up to about 5 days. CAPE and shear (850–500 hPa) use the 00, 06, 12 and 18 UTC instants; “severe” is √(2·CAPE) × shear. The high-resolution IFS tends to forecast more intense local rain and gusts than ERA5, so precipitation and gust percentiles may be somewhat inflated. Forecasts update when ECMWF publishes a new cycle (00 and 12 UTC). An extreme forecast is not a measurement: uncertainty grows with lead time. Indicative information, not intended for safety decisions.