On 29 October 2024, the Valencia region was flooded after extremely heavy downpours. Nathalie Rombeek , a PhD candidate in flood forecasting at TU Delft, reconstructed the cause of these devastating floods. Using data from 225 personal weather stations, installed by residents at home, she was able to trace the dynamics of the dramatic event in detail. By making smart use of these weather stations, flood warnings can in the future be issued with greater accuracy. The results have been published in the scientific journal Hydrology and Earth System Sciences.
The flooding began with torrential rainfall in the morning, upstream in the Magro river basin. Rainfall of more than 150 mm in 24 hours had been forecast. -But by the morning, personal rain gauges had already recorded 130 mm,- Rombeek observed in the data. By early afternoon, the Magro river had burst its banks. -At 13:00 the situation upstream was already so severe that the first fatalities occurred.-
With this flood wave still in the channel system, another peak of rainfall occurred later in the day further downstream, adding more water to the system. The coincidence of these discharge peaks caused disastrous flooding.
If false alarms are raised too often, people stop feeling the urgency.
General warning
Six days before the floods, the Spanish Meteorological Institute (AEMET) had already identified an increased risk of flooding. The first warnings covered a large area of the province of Valencia (20,000 km˛), since the storm was difficult to predict. On the day of the storm, the warning was quickly escalated to red, indicating extraordinary danger, but still for a broad area. For residents of downstream areas, such as the city of Valencia, the urgency probably only became clear when an emergency message was sent at 20:00 by the regional civil protection agency. By then, personal rain gauges upstream of the Magro river had already measured 200 to 300 mm of rainfall. The flood wave had caused enormous damage and a large number of casualties.
-It is very difficult to warn the right people at the right time,- Rombeek reasons. -Warnings must be reliable. If false alarms are raised too often, both citizens and government staff stop feeling the urgency. It is therefore crucial not only to warn earlier, but above all more accurately. That increases credibility in the long term.-
Left: One of the personal weather stations on a roof in the Valencia region. Source: Avamet
Right: Thanks to these personal rain gauges, Rombeek was able to reconstruct in greater detail how much rain fell at different locations in the catchments around Valencia. Source: Nathalie Rombeek.
Real-time rainfall data
Personal weather stations are usually (depending on brand and type) connected via Wi-Fi and send data to an online platform every few minutes. This information is of great added value, says one of Rombeek-s supervisors Remko Uijlenhoet , Professor of Hydrology & Water Management: -The network of personal weather stations is much denser than that of national meteorological services. The information is often available in real-time.- This makes them a welcome addition for producing accurate flood forecasts in both space and time.
Based on comparisons with rain gauges from the Spanish Meteorological Institute and statistical analyses, Rombeek concluded that personal stations provide a reliable picture. Not all’home stations are well installed, and in extreme weather some fail or are washed away. -A single station is therefore not very reliable. But this is about the wisdom of the crowd. There are so many that we can afford to lose or filter out a few,- Uijlenhoet explains. The combined information from all these gauges provides a clear picture.
The network of personal weather stations is much denser than that of national meteorological services.
Improved warning system
Combining measurements from personal weather stations with those from government agencies improves the detail and accuracy of forecasts. By taking all’available data into account, warnings in Valencia could have been earlier, more accurate and precise, the researchers conclude.
Personal weather stations in Limburg
Personal weather stations exist all’over the world. Nathalie Rombeek is currently working on historical forecasts of more floods, like the devastating floods of the Guadalupe river in Texas during the summer of 2025 and the floods in Limburg, The Netherlands, in 2021. Again, she is using data from rain gauges in the gardens and on the roofs of residents. By cleverly exploiting these data, Rombeek and Uijlenhoet hope to better predict and model river discharges, enabling authorities to issue accurate and timely flood warnings.
Scientific article published on 26 November 2025 in Hydrology and Earth System Sciences: Rombeek, N., Hrachowitz, M., and Uijlenhoet, R.: Torrential rainfall in Valencia, Spain, recorded by personal weather stations preceding and during the 29 October 2024 floods, Hydrol. Earth Syst. Sci., 29, 6715-6733, https://doi.org/10.5194/hess-29-6715-2025, 2025.
Calling for rain
Measuring weather phenomena such as rainfall can be far more accurate than when relying solely on traditional meteorological instruments. Cellular communications networks also provide a valuable source of information for weather forecasting, Remko Uijlenhoet discovered by chance. The major challenge is persuading telecom companies to make their data available. If that succeeds, this opportunistic sensing technique could be of great value worldwide, according to Uijlenhoet. Read his Story of Science.
