How accurate the fog forecast is
The forecast was checked on 6,115 mornings from October to March since January 2024, at airports in 14 European cities. For each morning, the forecast made the day before was compared with what happened. A morning was counted as foggy when the visibility measured at the airport was under 1 km for at least two hours between 4 and 10 h. Each month was tested with a formula that was fitted without that month.
| Fog chance the forecast gave | Mornings | Fog happened |
|---|---|---|
| about 1% | 4,276 | 1% |
| about 9% | 879 | 10% |
| about 22% | 432 | 20% |
| about 39% | 356 | 40% |
| about 58% | 157 | 55% |
| about 74% | 15 | 67% |
The forecast works for fog that forms overnight in calm air. It does not work for fog that comes in from the sea, so coastal cities such as Los Angeles and San Francisco have no page.
Each forecast is for one weather station, usually the city's airport, and fog is local. The airport can be in fog while the city centre is clear, or the other way round. The station is written on each city's page.
Delhi, Lahore and Amritsar share a formula of their own and were tested separately. In Delhi the forecast gives too high a chance: mornings given about 60% had fog 50% of the time. In Lahore it is a little too high: mornings given about 80% had fog 72% of the time. In Amritsar it gives too low a chance: mornings given about 60% had fog 72% of the time.
The method and all test results are on GitHub.