Same Temperature, Different Run: 684 City-Months of Running Weather Data
Two runs at 28°C can score 22 and 66. We scored every hour of the year in 57 cities to find out why temperature alone never explains a hard run, and how much the hour you pick is worth.
Short answer
Across 57 cities and 684 city-months of climate normals, temperature explains only about a quarter of the variation in running conditions. Hong Kong in August and Phoenix in May sit at the same 28°C and score 22 and 66. In peak summer the typical city swings 24 Run Score points between its best and worst hour, and in 52% of the months that average out as poor, at least one hour of the day is still good.
Two runners set out at 28°C. One has a decent run. The other falls apart at 5k and spends the evening wondering what went wrong with their fitness.
Nothing went wrong with their fitness. We scored every hour of the year in 57 cities, 684 city-months in total, and the single clearest result is this: temperature explains only about a quarter of how good conditions actually are for running. The rest is moisture, timing, and wind, and the gap it opens is much larger than most runners expect.
The method, briefly
Every number below comes from the same dataset that powers the Run Score calculator: three full years of hourly Open-Meteo ERA5 reanalysis per city, 2022 through 2024, reduced to typical conditions for each hour of each month, then scored 0 to 100 by the same engine the RunWeather app runs. The scoring thresholds are published in full at how the Run Score works, so every figure here can be checked rather than taken on trust.
A Run Score of 65 or above is good running weather, 80 is great, and below 45 conditions are actively working against you.
Finding 1: the same temperature, a 44-point difference
The clearest pair in the whole dataset:
| City and month | Air temp | Dew point | Run Score |
|---|---|---|---|
| Hong Kong, August | 28.0°C (82°F) | 25.7°C (78°F) | 22 |
| Phoenix, May | 28.1°C (82°F) | -1.7°C (29°F) | 66 |
One tenth of a degree apart in air temperature. Forty-four points apart in running conditions.
The difference is the dew point. At a dew point near 26°C the air is already close to saturated, so the sweat on your skin has nowhere to evaporate to, and evaporation is how a running body sheds heat. Phoenix in May is genuinely hot, but the air is bone dry, so cooling still works. Hong Kong in August is a fundamentally different run wearing the same number on the thermometer.
Across all 684 city-months, air temperature correlates with the daytime score at r = -0.52 and dew point at r = -0.49. Worth being precise about what that means: neither one is the master variable. Temperature is very slightly the stronger single predictor, and it still leaves roughly three quarters of the variation unexplained. Anyone who tells you one number is enough is selling something. What actually predicts a hard run is the combination, which is why we score it rather than display it.
Finding 2: the hour you pick is worth 24 points
Take each city's peak summer month, July in the northern hemisphere and January in the southern, and compare its best hour against its worst:
| City | Month | Best hour | Worst hour | Swing |
|---|---|---|---|---|
| Nashville | July | 69 | 29 | 40 |
| Dallas | July | 62 | 24 | 38 |
| Atlanta | July | 69 | 33 | 36 |
| Los Angeles | July | 84 | 49 | 35 |
| Phoenix | July | 62 | 27 | 35 |
| Rome | July | 72 | 38 | 34 |
| Tokyo | July | 58 | 25 | 33 |
The median city swings 24 points, and 43 of the 57 swing at least 20. For scale, 24 points is roughly the distance between "good running weather" and "conditions are working against you". It is a bigger effect than most training variables you agonise over, and it asks nothing of you but a change of alarm.
Finding 3: the best hour is earlier than you think
We expected a split between morning cities and evening cities. There isn't one.
- 42 of 57 cities peak at 6am in their hottest month.
- 55 of 57 peak at or before 8am.
- Only two cities score best in the evening.
Evening running feels reasonable because the air temperature has dropped. But the ground has been absorbing sun all day and is still radiating it back, humidity often climbs after sunset, and the dew point barely falls. The thermometer improves while the run does not. Dawn is when the air has actually finished cooling, and the data is close to unanimous about it.
Finding 4: a bad month still contains good runs
126 of the 684 city-months average below 65 during daylight hours. It would be reasonable to write those months off.
But of those 126, 66 of them, 52%, still contain at least one hour that reaches 65 or better. Half of all "bad" months are not bad months at all. They are good months with a narrow window, and the runners who find that window train through the summer while everyone else moves to the treadmill.
What to actually do with this
- Stop reading the thermometer first. Read the dew point next to it. Our dew point threshold chart has the bands.
- Move the alarm, not the workout. In most cities, 6am is worth 20 to 40 points over an afternoon run in the same day's weather.
- Judge the hour, not the month. Half the months that look unrunnable on average still have a window.
- Stop blaming your fitness for the weather's work. If your pace fell apart at the same effort as last week, check what the conditions were doing before you rewrite your training plan. That is the whole reason we put a number on it.
You can check any city and any hour yourself in the Run Score calculator. It is free, needs no account, and runs the same scoring engine as everything above.
