The Spread Maps Request an invite

Which weather model is most accurate in Paris?

Forecast Highs Lows Rain Model accuracy

Paris, Île-de-France Region · daily high temperature in °C · trailing 30 days, through Fri Oct 16

Every day, The Spread archives what each major weather model predicted for Paris at 1-, 3-, and 7-day leads, then scores those forecasts against the LFPB observations once the day has happened. This page covers the trailing 30 days (2026-09-17 to 2026-10-16), for daily high temperature.

One day out, NOAA GFS has been most accurate (typical miss 0.89°C); at three days, too close to call; at seven days, GEM (Canada) (1.8°C).

The largest standing bias belongs to NOAA AIGFS (AI), which has run cool by about 2.93°C at 3-day lead over this window.

Paris's full forecast, every model day by day, with the record on this page beside it. Request an invite; it's free during the beta.

The scoreboard

Typical miss (MAE) in °C with the bias in parentheses. Bias is the mean signed error of the predicted daily high — positive means the model ran warm. A model's high is the warmest of its hourly values and the station's is the warmest it got at any moment, so a model's high reads a little cool, and its low a little warm, even on a day it gets every hour right. ECMWF IFS gives a value only every 3 hours (every 6 past day 6), and ECMWF AIFS (AI) and NOAA AIGFS (AI) only every 6, with the hours between filled in, so their highs read cooler still and their lows warmer. In the forecasts scored here GFS also steps to every 3 hours from day 5, ICON and GEM from day 4, and UKMO from day 3, every 6 by day 6. Days = scored days behind the 1-day number.

Model 1-day 3-day 7-day Days
NOAA GFS 0.9° (-0.1°) 1.4° (-0.8°) 2.4° (-1.0°) 29
GEM (Canada) 1.2° (-0.2°) 1.4° (-0.6°) 1.8° (-0.1°) 29
UKMO (UK) 1.3° (-1.0°) 2.3° (-2.1°) — 29
DWD ICON 1.3° (-1.0°) 1.6° (-1.2°) — 29
ECMWF IFS 1.5° (-1.4°) 1.9° (-1.8°) 2.6° (-1.9°) 29
ECMWF AIFS (AI) 1.9° (-1.9°) 2.0° (-2.0°) 2.3° (-2.2°) 29
NOAA AIGFS (AI) 2.3° (-2.3°) 2.9° (-2.9°) 3.9° (-3.7°) 29

Method, honestly

The Spread's Paris point is the Le Bourget station (LFPB), 14 km from downtown, used for both forecasts and verifications. Each forecast is the last one The Spread archived on the day it was made, the forecast the model was showing at the time, not a reconstruction. A day the station did not report is left out. Truth is the thermometer at Le Bourget (LFPB) — 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; consensus products such as the NBM, itself a blend of models, are not scored here. 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.

  • Fri Oct 16 14 days out the models said 11–23; it hit 22. GEM led (off 0.1°C); AIFS ran 4°C cool.
  • Thu Oct 15 14 days out the models said 13–17; it hit 24. GFS led (off 0.6°C); AIGFS ran 4°C cool.
  • Wed Oct 14 14 days out the models said 15–18; it hit 21. GFS led (off 1.0°C); AIGFS ran 4°C cool.
  • Tue Oct 13 14 days out the models said 15–18; it hit 24. GEM led (off 0.1°C); AIGFS ran 4°C cool.
  • Mon Oct 12 14 days out the models said 14–19; it hit 20. GEM led (off 1.8°C); ICON ran 5°C cool.
  • Sun Oct 11 14 days out the models said 13–19; it hit 17. UKMO led (off 0.3°C); AIGFS ran 2°C cool.
  • Sat Oct 10 14 days out the models said 16–29; it hit 18. AIFS led (off 1.4°C); UKMO ran 3°C cool.
  • Fri Oct 9 14 days out the models said 12–24; it hit 17. GFS led (off 0.1°C); AIGFS ran 2°C cool.
  • Thu Oct 8 14 days out the models said 16–24; it hit 17. GFS led (off 0.3°C); GEM ran 2°C cool.
  • Wed Oct 7 14 days out the models said 15–21; it hit 19. ECMWF led (off 0.2°C); GEM ran 3°C warm.
  • Tue Oct 6 13 days out the models said 16–19; it hit 25. GFS led (off 0.3°C); AIGFS ran 2°C cool.
  • Mon Oct 5 14 days out the models said 18–28; it hit 25. GEM led (off 0.0°C); AIGFS ran 2°C cool.
  • Sun Oct 4 14 days out the models said 17–22; it hit 25. GEM led (off 0.2°C); AIGFS ran 3°C cool.
  • Sat Oct 3 13 days out the models said 18–25; it hit 23. GFS led (off 0.1°C); AIGFS ran 2°C cool.
  • Fri Oct 2 14 days out the models said 19–25; it hit 22. GFS led (off 0.1°C); AIGFS ran 2°C cool.
  • Thu Oct 1 14 days out the models said 17–27; it hit 22. GFS led (off 0.2°C); GEM ran 4°C cool.
  • Wed Sep 30 14 days out the models said 15–24; it hit 24. GFS led (off 0.3°C); GEM ran 3°C warm.
  • Tue Sep 29 14 days out the models said 13–28; it hit 29. UKMO led (off 0.1°C); AIGFS ran 4°C cool.
  • Mon Sep 28 14 days out the models said 20–26; it hit 25. ICON led (off 0.7°C); AIFS ran 2°C cool.
  • Sun Sep 27 14 days out the models said 17–27; it hit 28. GEM led (off 0.1°C); AIGFS ran 5°C cool.
  • Sat Sep 26 14 days out the models said 18–27; it hit 23. AIGFS led (off 0.1°C); GFS ran 2°C warm.
  • Fri Sep 25 14 days out the models said 15–23; it hit 29. GEM led (off 0.9°C); AIGFS ran 4°C cool.
  • Thu Sep 24 14 days out the models said 20–29; it hit 24. GFS led (off 0.2°C); AIFS ran 2°C cool.
  • Wed Sep 23 13 days out the models said 20–26; it hit 27. GFS led (off 1.9°C); AIGFS ran 4°C cool.
  • Tue Sep 22 14 days out the models said 23–25; it hit 24. ICON led (off 0.1°C); AIGFS ran 2°C cool.
  • Mon Sep 21 14 days out the models said 20–23; it hit 23. ICON led (off 0.5°C); AIGFS ran 2°C cool.
  • Sun Sep 20 14 days out the models said 17–29; it hit 24. ECMWF led (off 0.3°C); AIFS ran 2°C cool.
  • Sat Sep 19 14 days out the models said 19–28; it hit 25. GFS led (off 0.3°C); AIFS ran 2°C cool.
  • Fri Sep 18 14 days out the models said 15–25; it hit 22. GFS led (off 0.1°C); AIGFS ran 2°C cool.
  • Thu Sep 17 14 days out the models said 14–34; it hit 22. ICON led (off 0.6°C); GFS ran 1°C warm.
  • Wed Sep 16 14 days out the models said 14–25; it hit 23. ICON led (off 1.2°C); UKMO ran 3°C cool.
  • Tue Sep 15 14 days out the models said 17–32; it hit 33. GFS led (off 0.8°C); AIGFS ran 4°C cool.
  • Mon Sep 14 14 days out the models said 14–32; it hit 29. GFS led (off 0.9°C); ECMWF ran 4°C cool.
  • Sun Sep 13 14 days out the models said 20–28; it hit 27. GFS led (off 0.1°C); AIGFS ran 4°C cool.
  • Sat Sep 12 13 days out the models said 19–25; it hit 26. GFS led (off 0.8°C); ECMWF ran 3°C cool.
  • Fri Sep 11 12 days out the models said 20–25; it hit 25. UKMO led (off 0.7°C); ICON ran 2°C cool.
  • Thu Sep 10 11 days out the models said 20–25; it hit 21. ECMWF led (off 0.2°C); GFS ran 2°C warm.
  • Wed Sep 9 10 days out the models said 20–24; it hit 22. GFS led (off 0.6°C); AIGFS ran 2°C cool.
  • Tue Sep 8 9 days out the models said 20–29; it hit 27. GFS led (off 0.9°C); UKMO ran 4°C cool.
  • Mon Sep 7 8 days out the models said 22–32; it hit 29. GEM led (off 0.1°C); AIFS ran 3°C cool.
  • Sun Sep 6 7 days out the models said 24–30; it hit 28. GEM led (off 0.1°C); GFS ran 3°C warm.
  • Sat Sep 5 6 days out the models said 23–30; it hit 25. GFS led (off 0.9°C); AIFS ran 2°C cool.
  • Fri Sep 4 5 days out the models said 25–31; it hit 28. ECMWF led (off 1.1°C); GFS ran 3°C warm.
  • Thu Sep 3 4 days out the models said 26–30; it hit 27. UKMO led (off 0.8°C); ECMWF ran 3°C cool.
  • Wed Sep 2 3 days out the models said 22–27; it hit 26. GEM led (off 0.8°C); ECMWF ran 2°C cool.

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 Philadelphia, Pennsylvania Phoenix, Arizona Salt Lake City, Utah San Antonio, Texas San Diego, California San Francisco, 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.