Rain that was not forecast
On Sunday 18 October in Tegueste, the models forecast 2.5 mm and 31 mm fell
It was very concentrated rain: 10.2 mm in the heaviest hour, while the 6 stations within 25 km collected a median of 6.7 mm. That is the typical signature of a thunderstorm.
Las Mercedes-Llano los Loros station (AEMET), 6 km from Tegueste.
What happened
- No model saw it coming: the closest, the ECMWF IFS, gave 3.9 mm.
- It was very concentrated rain: 10.2 mm in the heaviest hour, while the 6 stations within 25 km collected a median of 6.7 mm. That is the typical signature of a thunderstorm.
- The Atlas Forecast's chance of rain did not get it right either (Brier score 0.5, the models' 0.13; lower is better).
What each model gave
| Model | For that day (0–24 h) | A day earlier (24–48 h) |
|---|---|---|
| ECMWF IFS | 3.9 mm | 8.1 mm |
| ECMWF AIFS | 3 mm | 4.8 mm |
| GFS | 2 mm | 2.7 mm |
| ICON | 0 mm | 0 mm |
| ARPEGE | 0.8 mm | 0.4 mm |
| UKMO | 3.3 mm | 9.5 mm |
| Median | 2.5 mm | 3.8 mm |
| Measured | 31 mm | |
Hour by hour
Rain accumulated since midnight UTC at the Las Mercedes-Llano los Loros station, with each model's forecast for that same day (0–24 h). Spanish mainland time.
- Measured
- Median of the models
- Each of the other models
Show the data in a table
| Time | Measured | ECMWF IFS | ECMWF AIFS | GFS | ICON | ARPEGE | UKMO |
|---|---|---|---|---|---|---|---|
| 03:00 | 0 | 0.1 | 0.1 | 0.1 | 0 | 0 | 0 |
| 04:00 | 0 | 0.1 | 0.1 | 0.1 | 0 | 0 | 0.1 |
| 05:00 | 0.2 | 0.1 | 0.1 | 0.1 | 0 | 0 | 0 |
| 06:00 | 0 | 0.1 | 0.1 | 0 | 0 | 0 | 0 |
| 07:00 | 0 | 0.1 | 0.1 | 0 | 0 | 0 | 0 |
| 08:00 | 0 | 0.1 | 0.1 | 0 | 0 | 0 | 0.1 |
| 09:00 | 0 | 0.1 | 0.1 | 0 | 0 | 0 | 0 |
| 10:00 | 0 | 0.1 | 0.1 | 0 | 0 | 0 | 0 |
| 11:00 | 0 | 0.1 | 0.1 | 0.1 | 0 | 0 | 0 |
| 12:00 | 0.2 | 0.1 | 0.1 | 0.2 | 0 | 0 | 0.1 |
| 13:00 | 0.2 | 0.1 | 0.1 | 0 | 0 | 0 | 0.6 |
| 14:00 | 0 | 0.1 | 0.1 | 0 | 0 | 0 | 0.5 |
| 15:00 | 1.4 | 0.4 | 0.2 | 0.1 | 0 | 0.1 | 0.2 |
| 16:00 | 4.8 | 0.4 | 0.2 | 0.3 | 0 | 0 | 0.2 |
| 17:00 | 3.2 | 0.4 | 0.2 | 0.1 | 0 | 0 | 0.1 |
| 18:00 | 0.8 | 0.2 | 0.2 | 0.3 | 0 | 0 | 0.2 |
| 19:00 | 1.8 | 0.2 | 0.2 | 0.2 | 0 | 0 | 0.1 |
| 20:00 | 2.8 | 0.2 | 0.2 | 0.1 | 0 | 0 | 0 |
| 21:00 | 0 | 0.1 | 0.1 | 0.2 | 0 | 0 | 0 |
| 22:00 | 0.2 | 0.1 | 0.1 | 0.1 | 0 | 0 | 0 |
| 23:00 | 0 | 0.1 | 0.1 | 0 | 0 | 0 | 0 |
| 00:00 | 0 | 0.2 | 0.1 | 0 | 0 | 0 | 0.2 |
| 01:00 | 10.2 | 0.2 | 0.1 | 0 | 0 | 0 | 0.4 |
| 02:00 | 5.2 | 0.2 | 0.1 | 0 | 0 | 0.7 | 0.5 |
Weather in Tegueste · Model accuracy · The model league · Other misses
How we spot them
Every day we compare what the Meteocat (in Catalonia) and AEMET weather stations measured the day before with what each model had forecast for that same day (0–24 h ahead). It counts as a notable miss if 15 mm or more fell and the median of the models gave less than 3; if it gave 15 mm or more and less than 1 fell; if the high was off by 4 °C or more (and by at least 4 °C more than that station usually is); or if a gust went over 60 km/h and was 25 km/h or more above the forecast. Stations within 40 km with the same miss count as one episode, and we show at most 5 per day.
The explanations come from the same data: which model saw it, whether the rain came earlier or later, whether it fell (or was forecast) a few kilometres away, whether it was very concentrated… When none of that fits, we say so. Days run from midnight to midnight UTC (2 am to 2 am in summer and 1 am to 1 am in winter, Spanish mainland time).
Missing now and then is unavoidable: the weather is chaotic and every model has its days. How often each one gets it right, on average, is in Accuracy and in the model league.