FORECASTSEASON.COMRoute 07 · openSEAS5 ExplorerProbabilityAnalogsSecond OpinionCitiesModelsDriversAbout
SEASON90Forecast SeasonSeasonal Outlooks · U.S. · Europe · Global
Forecast Season/The Season Ahead
Route 07 · City briefings

The Season Ahead

A seasonal outlook for one place, stated the only honest way — as a distribution rather than a number, and with the skill of that particular gridcell printed next to it. City outlooks are the most widely published product in seasonal forecasting and the least often qualified; the column that matters most on this page is the one that says no skill.

SEASON90

Cities

40
every continent, 51 members each

Temperature

27 of 40
where the terciles beat climatology

Rainfall

14 of 40
the same test, and it is a harder one

Leaning warm

23 of 40
P(warm) at or above the 33 % null

One city at a time

40 cities · 51 members · 6 months

The band is the ensemble: the outer edge spans the 10th to 90th percentile of the 51 members, the inner one the 25th to 75th, and the line through it is the median. The dashed line is the model's own climate for that month, moved to the present. Where the two are far apart and the band is narrow, the model is saying something.

City
Field

All forty

SON 2026 · anomaly, probability, verdict

The first three whole months ahead, summarised. °C and mm are the median member against the model's own climate; P(warm) and P(wet) are the share of members in the top third, where 33 % is the null. The verdict columns are the reforecast: verified means the tercile forecast for that cell beat climatology over 1993–2016, no skill means it did not, and the anomaly beside it should be read as a curiosity rather than a forecast.

Africa

City°CP(warm)TemperaturemmP(wet)Rainfall
Egypt+0.656%verifiedarid
South Africa+0.449%verified+748%no skill
DR Congo+0.892%verified+1150%verified
Nigeria+0.799%verified+1549%verified
Morocco-0.426%verified+748%verified
Kenya+0.358%verified+3057%verified

Asia

City°CP(warm)TemperaturemmP(wet)Rainfall
Thailand+1.094%verifiedarid
India+0.134%verifiedarid
India+1.684%no skillarid
United Arab Emirates+0.774%verifiedarid
Indonesia+1.399%verified-930%verified
Philippines+0.482%verified-7710%verified
India+1.595%verifiedarid
China+0.342%verified+1554%no skill
Singapore+0.697%verified-3322%verified
Japan+0.558%verified+1040%no skill

Europe

City°CP(warm)TemperaturemmP(wet)Rainfall
Greece+0.342%no skillarid
Norway-0.136%no skill-3621%verified
Germany+0.344%no skill-135%no skill
United Kingdom+0.548%no skill+743%no skill
Spain-0.035%verifiedarid
Russia-0.321%no skill+242%no skill
France+0.752%no skill+1043%no skill
Iceland-0.128%no skill-427%no skill
Italy+0.652%no skill+2144%no skill

North America

City°CP(warm)TemperaturemmP(wet)Rainfall
United States-0.620%no skill+640%no skill
Mexico+0.990%verified+1244%no skill
United States+0.150%verified+234%verified
United States-0.324%verified+840%no skill
United States+0.853%no skill-1025%verified
Canada+0.858%no skill-1527%verified

Oceania

City°CP(warm)TemperaturemmP(wet)Rainfall
Australia+0.235%verifiedarid
New Zealand-0.128%verified-522%verified
Australia+0.142%no skill-819%verified
Australia+0.858%verified-1324%no skill

South America

City°CP(warm)TemperaturemmP(wet)Rainfall
Colombia+1.5100%verified-3118%no skill
Argentina-0.422%verified-329%no skill
Peru+2.4100%verifiedarid
Brazil+2.299%verified-785%no skill
Brazil+0.756%verified+4864%verified

How to read this

Four things it is not

Resolution

A gridcell, not a city

Every briefing names the 1° cell it actually used and how far that cell centre sits from the city — usually 30–80 km. That is the right resolution for a six-month outlook, because there is no seasonal skill at scales below it. A page implying street-level precision half a year out would be claiming something the science does not support. Coastal cities are pinned to a land cell deliberately: Lima's nearest cell is mostly Pacific, and built from it the outlook read +6 °C — true of that water, nonsense for the city.

Spread

A narrow band is not confidence

The 51 members sample the uncertainty SEAS5 knows how to represent. They cannot sample the errors the model shares with itself, so a systematic bias appears as a tight band rather than a wide one. Ensemble spread is a lower bound on uncertainty, never the whole of it — which is what the skill column is for.

Baseline

Normal means the model's normal

Anomalies here are departures from SEAS5's own 1993–2016 reforecast climatology, shifted to the present — not from an observed 1991–2020 normal. A coupled model drifts, and measuring it against itself cancels the drift. It also means these numbers are not interchangeable with the ones your national service publishes.

Rainfall

Usually the honest answer is no

Far fewer cities carry a verified rainfall forecast than a verified temperature one, and in the driest places the tercile is not even a category — below 5 mm a month the gap between "below" and "above normal" is under two millimetres, so those cities are marked arid and no rainfall number is shown.