Global
The whole planet, one to seven months out — and a straight account of where on it a seasonal forecast is worth anything. Being paved — the reference routes below are open now.
What this lane will carry
Route 01 · Route 06Maps
The world, seven months out
Global anomaly fields for temperature, rainfall, sea-surface temperature and pressure, at every lead, against the model's own hindcast climatology — with the regions broken out so a continent is not a smudge on a Robinson projection.
Tropics
Where it actually works
Seasonal forecasting is a tropical science that mid-latitude countries borrow from. The monsoons, the Sahel, the Nordeste, East Africa, the maritime continent — these are the places where a three-month outlook is genuinely actionable.
ENSO
The one index everything hangs on
The Niño-3.4 plume across successive initialisations, and what each system thinks the coming winter's state will be. It is the first thing anyone looks at and it deserves to be shown properly rather than as a single headline number.
The geography of skill
The map behind every other mapIf you rank the planet by how much a seasonal forecast beats climatology, the ordering is stable and it has been stable for thirty years. Tropical sea-surface temperature first, by a long way — the ocean has memory measured in seasons. Then tropical rainfall, which follows the SST. Then the ENSO-teleconnected mid-latitudes in winter: the southern United States, southern Africa, eastern Australia, parts of South America.
Then a long gap. Then European winter, at a level that is real but requires care. Then mid-latitude summer, where the honest answer for precipitation is usually climatology plus the trend.
This ordering is not a limitation of the models so much as a property of the climate system: predictability at a season's range comes almost entirely from the slow components — ocean, land, ice, stratosphere — and it reaches the atmosphere strongly only where the coupling is strong. The tropics are where the coupling is strong.