The town of Grazalema, Spain is no stranger to rain. Its geography — positioned at the western end of the Betic Cordillera — makes it a natural funnel for humid Atlantic winds, and it holds Spain’s record for highest annual rainfall. But what unfolded from January 1, 2026 is statistically staggering for any practising hydrologist.
As of February 9, 2026, the cumulative daily rainfall total reached 2,449.90 mm.
The Numbers in Context
Put that figure on a scale an engineer can feel:
- In just 40 days, this region received nearly three times the total annual rainfall of London
- On February 4, Storm Leonardo delivered 512.5 mm in 24 hours — smashing a 78-year-old record
- That single day’s rainfall exceeded Madrid’s entire annual average
This isn’t just a big storm. Meteorologists tracking the event pointed to a convergence of a negative North Atlantic Oscillation (NAO) phase and a persistent atmospheric river carrying tropical moisture — a configuration standard return period analysis wasn’t calibrated to capture at this intensity and duration.
The Hydrological Consequence
When soil saturation reaches 100%, the runoff coefficient effectively becomes 1.0. At that point, the distinction between rainfall and runoff disappears — we are no longer managing stormwater, we are managing a moving river.
New streams appeared where none had existed. Infrastructure designed for known flood envelopes was pushed past its limits.
The Design Question This Raises
From a design and risk management standpoint, this is a direct challenge to the tools we use every day. We routinely talk about “1-in-100-year events” as if those return periods are stable. But when the cumulative rainfall trace goes near-vertical like this, we have to ask honestly:
Are our current IFD (Intensity-Duration-Frequency) curves prepared for a climate that no longer follows historical averages?
The 2019 Australian Rainfall and Runoff revision acknowledges non-stationarity as a concern, but the bulk of our design flood estimation practice still relies on frequency analysis fitted to pre-2000 records. Events like Grazalema are not outliers to be discarded — they’re data points about what the tail of the distribution actually looks like. Whether configurations like this one are becoming more frequent is precisely the kind of question non-stationarity analysis is for, not something a single event can settle.
If the historical baseline has genuinely shifted, the question for practitioners is whether our standards have kept pace.
Data visualisation: Nahel Belgherze (@WxNB_) via r/meteorology. Data: AEMET (Spanish Meteorological Agency). Originally shared on LinkedIn where it reached 13,236 impressions.