Which weather model is most accurate in San Francisco?
San Francisco, California · daily high temperature in °C · trailing 30 days · refreshed at most every 6 hours
Every day, The Spread archives what each major weather model predicted for San Francisco at 1-, 3-, and 7-day leads, then scores those forecasts against the SFOC1 observations once the day has happened. This page covers the trailing 30 days (2026-08-04 to 2026-09-02), for daily high temperature.
One day out, ECMWF IFS has been most accurate (typical miss 0.97°C); at three days, too close to call; at seven days, ECMWF AIFS (AI) (1.33°C).
The largest standing bias belongs to NOAA AIGFS (AI), which has run warm by about 3.38°C at 1-day lead over this window.
The Spread's own bias-corrected blend scored a 0.92°C typical miss at one day. Its corrections come from the same window it is scored on (in-sample), so read it as a ceiling, not a promise.
The scoreboard
Typical miss (MAE) in °C with the signed bias in parentheses — positive bias means the model ran warm. Days = scored days behind the 1-day number.
| Model | 1-day | 3-day | 7-day | Days |
|---|---|---|---|---|
| ECMWF IFS | 1.0° (+0.1°) | 1.2° (-0.0°) | 1.6° (-1.0°) | 30 |
| GEM (Canada) | 1.2° (-0.4°) | 1.4° (-0.7°) | 1.9° (-0.9°) | 30 |
| UKMO (UK) | 1.4° (+1.0°) | 1.5° (+0.6°) | — | 30 |
| ECMWF AIFS (AI) | 1.5° (+1.2°) | 1.6° (+1.2°) | 1.3° (+0.9°) | 30 |
| DWD ICON | 1.8° (-1.8°) | 1.8° (-1.6°) | — | 30 |
| NOAA GFS | 2.5° (+2.5°) | 2.5° (+2.4°) | 2.7° (+2.5°) | 30 |
| NOAA AIGFS (AI) | 3.4° (+3.4°) | 3.0° (+2.8°) | 2.7° (+2.4°) | 30 |
| NBM (US blend) | 1.3° (+0.4°) | 1.5° (+0.3°) | 1.6° (+0.5°) | 30 |
| Blended (bias-corrected) | 0.9° (+0.0°) | 1.1° (+0.0°) | 1.2° (-0.0°) | 30 |
Method, honestly
The Spread's San Francisco point is the San Francisco Downtown station (SFOC1), 0.6 km from downtown, used for both forecasts and verifications. Forecasts come from each model's actual archived runs via Open-Meteo's Previous Runs API — what the model said at the time, not a reconstruction. Truth is SFOC1 observations — the station's own daily highs, a thermometer rather than a grid, and no correction applies: the thermometer is the truth outright. Rankings cover the models themselves; the two consensus products at the bottom of the table — NBM, which is itself a blend of models, and The Spread's own bias-corrected blend — are shown for comparison, never ranked or named "most accurate" here. The blend's corrections come from the same trailing window it is scored on (in-sample), so treat its score as a ceiling. A rolling 30-day window means these numbers move — that is the point.
Receipts — was the forecast right?
One page per settled date: what every model said as the day approached, what happened, and who was closest.
- Tue Sep 1 10 days out the models said 20–24; it hit 21. GFS led (off 0.5°C); ECMWF ran 2°C cold.
- Mon Aug 31 9 days out the models said 19–24; it hit 18. GFS led (off 0.0°C); GEM ran 2°C cold.
- Sun Aug 30 8 days out the models said 15–24; it hit 18. UKMO led (off 0.1°C); ICON ran 2°C cold.
- Sat Aug 29 7 days out the models said 16–25; it hit 19. GFS led (off 0.3°C); ICON ran 2°C cold.
- Fri Aug 28 6 days out the models said 18–24; it hit 20. AIFS led (off 0.2°C); GFS ran 3°C hot.
- Thu Aug 27 5 days out the models said 20–26; it hit 21. ECMWF led (off 0.2°C); AIGFS ran 4°C hot.
- Wed Aug 26 4 days out the models said 18–24; it hit 21. ICON led (off 0.6°C); UKMO ran 6°C hot.
- Tue Aug 25 3 days out the models said 18–22; it hit 21. GFS led (off 0.6°C); ICON ran 3°C cold.
- Wed Aug 19 13 days out the models said 18–21; it hit 18. ICON led (off 0.8°C); AIGFS ran 6°C hot.
- Tue Aug 18 13 days out the models said 18–25; it hit 18. GEM led (off 0.4°C); AIGFS ran 5°C hot.
- Mon Aug 17 13 days out the models said 18–20; it hit 19. UKMO led (off 0.1°C); AIGFS ran 4°C hot.
- Sun Aug 16 13 days out the models said 17–22; it hit 18. GEM led (off 0.7°C); AIGFS ran 4°C hot.
- Sat Aug 15 13 days out the models said 17–21; it hit 18. ICON led (off 0.3°C); AIGFS ran 2°C hot.
- Fri Aug 14 13 days out the models said 18–24; it hit 18. UKMO led (off 0.0°C).
- Thu Aug 13 13 days out the models said 18–22; it hit 19. ECMWF led (off 0.2°C); AIGFS ran 2°C hot.
- Wed Aug 12 13 days out the models said 17–24; it hit 19. GEM led (off 0.2°C); AIGFS ran 3°C hot.
- Tue Aug 11 13 days out the models said 18–21; it hit 18. GFS led (off 0.3°C); AIGFS ran 4°C hot.
- Mon Aug 10 13 days out the models said 19–23; it hit 19. GFS led (off 0.7°C); AIGFS ran 5°C hot.
- Sun Aug 9 13 days out the models said 17–24; it hit 19. ICON led (off 1.4°C); AIGFS ran 9°C hot.
- Sat Aug 8 12 days out the models said 18–24; it hit 16. ICON led (off 0.4°C); AIGFS ran 9°C hot.
- Fri Aug 7 11 days out the models said 19–25; it hit 17. ICON led (off 1.1°C); AIGFS ran 8°C hot.
- Thu Aug 6 10 days out the models said 21–27; it hit 18. ICON led (off 0.2°C); AIGFS ran 6°C hot.
- Wed Aug 5 9 days out the models said 20–28; it hit 19. ICON led (off 0.1°C); AIGFS ran 7°C hot.
- Tue Aug 4 8 days out the models said 19–30; it hit 20. ICON led (off 1.4°C); AIGFS ran 6°C hot.
- Mon Aug 3 7 days out the models said 21–29; it hit 21. ICON led (off 1.2°C); AIGFS ran 8°C hot.
- Sun Aug 2 6 days out the models said 18–27; it hit 22. GEM led (off 0.3°C); GFS ran 6°C hot.
- Sat Aug 1 5 days out the models said 19–27; it hit 22. ECMWF led (off 0.7°C); AIGFS ran 6°C hot.
- Fri Jul 31 4 days out the models said 18–26; it hit 20. ECMWF led (off 0.9°C); AIGFS ran 3°C hot.
- Thu Jul 30 3 days out the models said 18–25; it hit 18. GEM led (off 0.2°C); AIGFS ran 5°C hot.
Other cities
Atlanta, Georgia Berlin, State of Berlin Boston, Massachusetts Buffalo, New York Charlotte, North Carolina Chicago, Illinois Dallas, Texas Denver, Colorado Detroit, Michigan Houston, Texas London, England Los Angeles, California Miami, Florida Milan, Lombardy Minneapolis, Minnesota New Orleans, Louisiana New York, New York Oklahoma City, Oklahoma Orlando, Florida Paris, Île-de-France Region Philadelphia, Pennsylvania Phoenix, Arizona Salt Lake City, Utah San Antonio, Texas San Diego, California Seattle, Washington St. Louis, Missouri Tampa, Florida Washington, District of Columbia
The Spread — the forecast spread — is in closed beta. These verification pages are public while the rest is gated.