The Spread

ECMWF AIFS temperature

Valid tomorrow evening.

Valid 18Z 1 Sep · ECMWF AIFS run 12Z 30 Aug+54 h

15.015.015.015.015.015.020.020.020.020.020.020.020.020.020.020.020.020.020.020.020.020.020.020.025.025.025.025.025.025.025.025.025.025.025.025.030.030.030.030.030.030.030.030.030.030.030.030.030.035.035.035.035.035.035.0 38.6° 5.9° −1.4 Oklahoma City: −1.4 °C at 3-day lead −1.9 Salt Lake City: −1.9 °C at 3-day lead −2.9 Seattle: −2.9 °C at 3-day lead −2 Orlando: −2 °C at 3-day lead −0.2 New York: −0.2 °C at 3-day lead −2.3 Los Angeles: −2.3 °C at 3-day lead −1.7 Chicago: −1.7 °C at 3-day lead −0.5 Houston: −0.5 °C at 3-day lead −1.2 Phoenix: −1.2 °C at 3-day lead −0.6 Dallas: −0.6 °C at 3-day lead −2.6 Atlanta: −2.6 °C at 3-day lead −1.4 Boston: −1.4 °C at 3-day lead −0.5 Washington: −0.5 °C at 3-day lead −1.1 Denver: −1.1 °C at 3-day lead −1.1 Minneapolis: −1.1 °C at 3-day lead −2.8 Miami: −2.8 °C at 3-day lead +1.6 San Francisco: +1.6 °C at 3-day lead −1.2 New Orleans: −1.2 °C at 3-day lead −1.4 Buffalo: −1.4 °C at 3-day lead −1.3 Charlotte: −1.3 °C at 3-day lead −0.7 Philadelphia: −0.7 °C at 3-day lead −1.3 Detroit: −1.3 °C at 3-day lead −1.2 Tampa: −1.2 °C at 3-day lead +2.2 San Diego: +2.2 °C at 3-day lead −1.5 St. Louis: −1.5 °C at 3-day lead −1 San Antonio: −1 °C at 3-day lead

Warmest 38.6 °C, 50 km WNW of Reynosa, Mexico · coolest 5.9 °C, 130 km WNW of Lillooet, British Columbia

Methodology

ECMWF AIFS's own 2 m temperature, on the model's own 0.25° grid — which is why this map is finer or coarser than its neighbours rather than all four being drawn alike. Every model here shows the same moment from the same run time, so switching between them is a comparison and not four different questions. Raw model output: nothing on this map is bias-corrected, unlike the city forecasts elsewhere on this site — which is what the numbers on it are for.

The scale is the same every day — -35 to 50 °C — so a colour means a temperature rather than "warm for today", and the changeover from cool to warm sits at 60 °C rather than at freezing, because that is nearer where a person stops calling it cool. Values are printed on the lines where the bands change. The 0 °C line is drawn wherever the field crosses it, which in high summer is nowhere.

The warmest and coolest cells are ringed on the map, and both are read from the unsmoothed grid — the fill is lightly smoothed for legibility, and smoothing fills extremes, so the two numbers would be quietly wrong if they came off the picture rather than off the model's own field.

The numbers on the map are each city's measured bias over the last 30 days at 3-day lead: how far ECMWF AIFS's daily high has run from what the thermometer recorded, in °C, warm positive. Switch models above and they change — that gap is the thing this site is for. Three ways to hold them lightly: the bias is measured on daily highs where the map is one instant, it is a 30-day window rather than a constant, and each number is one thermometer, not the region around it.

Every map on this page is drawn from model output we store ourselves, not from anyone's picture of it. The per-city read of the same day — which cities the models are fighting about, and what the discussion makes of it — is on the national page.