Was the forecast right for San Francisco on Wed Oct 28?
San Francisco, California · daily high temperature in °F · forecasts from each model's archived runs, verified against the SFOC1 observations
14 days out, 4 models put San Francisco's Wed Oct 28 high anywhere from 59° to 77°; by the final call, the day before, the field had settled to a median of 66°. The day verified at 67° against the SFOC1 observations.
AIFS landed closest, off by 0.7°. GEM ran 5° cool.
What every model said, at every lead
Forecast highs for Wed Oct 28 as each model's run approached it, recorded from the last archived model run per day. NBM and The Spread's own blend are omitted because they average the other models.
| Lead | AIFS | AIGFS | ECMWF | GEM | GFS | ICON | UKMO | Range |
|---|---|---|---|---|---|---|---|---|
| 15d · 2026-10-13 | — | 61° | — | — | 67° | — | — | 61–67° |
| 14d · 2026-10-14 | 72° | 59° | 70° | — | 77° | — | — | 59–77° |
| 13d · 2026-10-15 | 65° | 60° | 75° | — | 67° | — | — | 60–75° |
| 12d · 2026-10-16 | 56° | 63° | 61° | — | 62° | — | — | 56–63° |
| 11d · 2026-10-17 | 66° | 59° | 67° | — | 59° | — | — | 59–67° |
| 10d · 2026-10-18 | 65° | 64° | 67° | — | 59° | — | — | 59–67° |
| 9d · 2026-10-19 | 62° | 68° | 70° | 63° | 70° | — | — | 62–70° |
| 8d · 2026-10-20 | 63° | 66° | 80° | 63° | 72° | — | — | 63–80° |
| 7d · 2026-10-21 | 63° | 61° | 70° | 62° | 69° | — | — | 61–70° |
| 6d · 2026-10-22 | 64° | 61° | 64° | 60° | 62° | 62° | 64° | 60–64° |
| 5d · 2026-10-23 | 62° | 64° | 62° | 61° | 61° | 66° | 63° | 61–66° |
| 4d · 2026-10-24 | 60° | 60° | 61° | 60° | 65° | 62° | 65° | 60–65° |
| 3d · 2026-10-25 | 65° | 64° | 65° | 62° | 69° | 64° | 65° | 62–69° |
| 2d · 2026-10-26 | 65° | 64° | 66° | 62° | 65° | 66° | 67° | 62–67° |
| 1d · 2026-10-27 | 66° | 64° | 66° | 62° | 65° | 66° | 68° | 62–68° |
| actual | 67° |
What one day can show
A score for one day isn't very meaningful; a useful ranking needs many days of data. See which model has been most accurate in San Francisco, or in every city The Spread scores.
Forecasts are what each model's archived runs said at the time they were recorded by The Spread; nothing is reconstructed after the fact. Truth is the SFOC1 observations. The actual values are the ones in hand when this day was verified, usually 2 days after the date; a correction published after that is not carried into it. Archive depth varies by city; verified days exist only for dates the archive watched arrive.