What a seasonal forecast is, and how it differs from a weather forecast
Seasonal forecasts predict the odds of a warmer, cooler, wetter or drier season months ahead. Why they are probabilistic, what they can and cannot tell you.
Seasonal forecasts describe the statistics of weather
A weather forecast can estimate whether rain will arrive on Thursday afternoon, often with probabilities too. At seasonal range the timing of individual weather events cannot be predicted reliably, because the atmosphere is chaotic: tiny errors in the starting state grow until the exact sequence of weather is unpredictable. A seasonal forecast does not try. It asks a different question: over the coming month or three-month season, is the average more likely to be warmer or cooler, wetter or drier, than usual?
Where the predictability comes from
Seasonal signal comes from parts of the climate system that change slowly. The ocean stores heat for months, so sea-surface temperature anomalies such as El Niño persist and steer the atmosphere above them. Soil moisture, snow cover, sea ice and the stratosphere can also contribute, with importance varying by region and season. A seasonal model couples an atmosphere to an ocean and land surface so that these slow components can shape the statistics of the weather it simulates.
Ensembles, not single runs
Because the weather itself is unpredictable at this range, a seasonal centre runs the model many times, typically 50 or so, to sample uncertainty in initial conditions and, where represented, model processes. Each member produces a possible season. The mean minus the hindcast climate gives the ensemble-mean anomaly; the share of members in a category gives a raw ensemble probability. See ensemble-mean anomaly and tercile and quintile probabilities.
What a seasonal forecast can tell you
- Whether the season leans warm or cold, wet or dry, relative to the model’s own climate.
- How strong that lean is, from the size of the anomaly and the probability of the outer terciles.
- Whether different models and successive runs agree. This provides context but does not establish skill.
What these maps cannot tell you
- When within the season the weather happens, or how many cold snaps or storms there will be.
- Conditions for a specific day, week or event.
- Anything with certainty. A raw 60% ensemble probability leaves 40% of members in other outcomes; observed frequencies depend on reliability.
Seasonal forecasts are issued monthly by centres such as ECMWF, the UK Met Office, Météo-France, DWD and CMCC, and collected by the Copernicus Climate Change Service. This site shows those five models on the same maps so that their agreement and disagreement is visible at a glance.
Sources and method
These references explain the underlying science and provider conventions. Descriptions of this site’s calculations, colours and coverage were checked against its processing code and imported data.