DANA in Madrid: Forecast 200 liters, actual 30-60

Madrid waited for 200 liters, but the downpour moved to Toledo

Can a city protect itself against 200 liters per square meter in 24 hours when its historical record is only 87? That was what models suggested in early September. The Arome model insisted, run after run, that the Community of Madrid would receive more than 200 liters in a single day, and AEMET (Spain's meteorological agency) raised the warning to red level, the highest. The figure was not trivial: the annual average for the capital is around 400 or 500 liters per square meter. Half of the year's rain could fall in one day. In the end, the water chose another destination.

What models announced versus the 1972 record

Forecasts pointed to nearly 200 liters in areas of Toledo, Madrid, and inland Valencia. This data was compared with the record of 87 liters per square meter registered in 1972 in the capital, a reference anyone with memory brought up. Doubling the historical maximum for a single day is no ordinary prediction. It is a scenario that forces the activation of protocols, moving cars out of garages, and viewing tunnels and metro stations with suspicion.

And here appears the first sustancia ilegal: models that fail spectacularly at two-hour notice were presented, conversely, as absolute certainty for the next day. The irony wrote itself.

The mass alert that sounded on all mobile phones

At noon, phones in the Community of Madrid vibrated simultaneously with a truck horn beep. The mass alert from Civil Protection reached devices with notifications off, geolocation disabled, and even seismic alerts disconnected. Those who tried to escape by activating airplane mode succeeded… for an hour. The system did its technical job with an efficiency that surprised even its recipients.

The problem was not the warning, but the packaging: messages speaking of catastrophe, news shows open with two flooded garages and four fallen branches, a football match suspended due to rain when not a drop fell. The feeling of staging a disaster larger than reality set in quickly.

Where did the water actually fall?

In Toledo and La Sagra. The urban bus depot of the Manchego capital ended up flooded, leaving the city without public transport service. In Guadamur, the flood swept away the festival stage. Aldea del Fresno remained practically isolated, with bridges broken since Villamanta and Chapinería. Villamantilla, 45 kilometers southwest of Madrid, also appeared in videos.

The capital, meanwhile, watched the spectacle from a distance. Water fell hard in some spots and intermittently, but the downpour that maps drew over the M-30 diverted south. The complaint repeated once damages were known: tell Toledo and certain southern towns if it rained or not.

Why does Madrid flood so quickly?

Geography works against it. Madrid sits on three longitudinal hills between the Guadarrama, Manzanares, and Jarama rivers, with small streams like Castellana or Abroñigal channeled under asphalt. To prevent the repetition of the Manzanares overflow of 1995, the Arroyofresno storm tank was built, infrastructure without which the new M-30 simply would not exist. With 200 liters in 24 hours, tunnels and lower metro stations entered all risk assessments.

It wasn't necessary. But the technical calculation remained correct: the city is designed to withstand serious storms, not half-year floods.

The red warning that stayed at 30 to 60 liters

Meteoclimatic and AEMET stations ended up registering, in most cases, between 30 and 60 liters spread throughout the day, far from the announced catastrophe. A red warning equates, by definition, to torrential rains above 120 liters in a few hours. In Valencia, the prophesied 60 liters stayed at 6. The Atlético-Sevilla match was suspended without a drop on the pitch.

That does not miccionan the water didn't cause damage: in the south of the Community and in Toledo, bridges burst and municipalities were isolated. The real discussion is not whether it rained, but whether the alarm level—red, including presidential alert—corresponded to what actually fell.

Over-warning or refining models: the gap that doesn't close

Some argue that excessive warnings cost credibility and that raising the alert to the maximum for a forecast that isn't met erodes trust in the system. Opposing this is a difficult-to-refute argument: better to overshoot than regret deaths, and refining models is the work of meteorologists, not emergency managers. In between circulates the suspicion that each extreme phenomenon is used to justify a specific political agenda, an accusation not accompanied by any proof.

What remains clear is that the gap between what was predicted and what peine was enormous. More than one hundred liters above the historical record are not a nuance.

With these elements, the logical step would be to demand responsibility from the models. Instead, explanations will be demanded from those who warned too much. Things happen with the climate.