UPV unveils digital twin that forecasts Poyo ravine floods 48 hours ahead

A shared forecasting platform

The Universitat Politècnica de València (UPV) presented PGTEC on 2 November 2026, a platform that predicts river overflow and flood spread from real-time rainfall data and artificial intelligence. PGTEC stands for the Plataforma para la Gestión de Prevención Temprana de Emergencias Climáticas. It was developed by the Instituto Universitario Valenciano de Investigación en Inteligencia Artificial (VRAIN), the UPV's Instituto de Ingeniería del Agua y Medio Ambiente (IIAMA) and the Valencian Graduate School and Research Network of Artificial Intelligence (ValgrAI). Researchers described the system as a linked set of hydrological, hydraulic and meteorological digital twins, which they said can be carried to any region of the world. In only a few hours, the models can determine the flooding that a given rainfall will cause, giving administrations and emergency services time to act before the effects arrive. The presentation took place alongside the AIIA Innovation Day of ValgrAI.

Two models working together

The system combines two complementary tools. The TETIS hydrological model uses recorded rainfall and forecasts from several weather models to estimate how much water could flow through each reach of a channel and when overflows could occur. It colours the most exposed reaches, ranks ravines by danger level and flags which ones are already on alert. The second tool, HIDRIS, reproduces how water would spread over the terrain if a channel overflowed, including its depth, speed and which buildings or critical infrastructure could be affected. Emergency managers can read both outputs in one visual environment instead of comparing separate maps and forecasts. Each simulation takes about one to two minutes and follows the expected evolution over the following 48 hours.

Test on the Poyo ravine

The team applied the system to the Barranco del Poyo, using the rainfall recorded during the dana of 29 October 2024. In the demonstration, one simulation showed how the river flow rises over time, while another showed how the water footprint spreads across the area as it floods. Vicent Botti, director of VRAIN at the UPV and director general of ValgrAI, explained that the same approach also applies to wildfires. Satellite imagery is used to detect combustible material on hillsides, which shows the danger of a fire when temperatures are high. Users can move a storm, raise its intensity or replay historical episodes to see how the consequences would change.

Coverage and data governance

The project already includes a digital twin of the L'Horta Sud basins and the Carraixet ravine. The presentation also showed the effects the same October 2024 dana would have had on the Carraixet ravine. Researchers said the approach can be applied to any region of the world to anticipate floods, wildfires and droughts. Sergio Martí, a VRAIN researcher at the UPV, stressed that the data space uses a common semantics so that a rainfall figure means the same thing wherever it comes from. He added that each organization keeps control over its data, deciding who may access it and under which usage contract, agreed before any exchange. The platform brings hydrometeorological data from several official bodies into one interoperable space, giving emergency services a shared view of the situation.