For the first time, artificial intelligence has been applied to hydraulic risk mitigation

Francesco Vincenzi, president of Anbi:“Floods in streams can also be predicted up to 6 hours in advance”

ROME - Artificial intelligence can help predict the flow rates of even torrential waterways, the most difficult to manage due to the speed with which floods arrive:it is the result of the experimentation born from the collaboration of the Reclamation Consortium 1 Toscana Nord with the Department of Earth Sciences of the University of Pisa;the two entities have worked together to improve flood forecasting times by applying innovative techniques based on "machine learning".This was announced by theNational Association of Land and Irrigation Water Management and Protection Consortia.

“The importance of innovation is that, thanks to artificial intelligence, Big data from large territorial information banks can be processed in real time, allowing flood times to be predicted up to 6 hours in advance. – indicates Francesco Vincenzi, president of ANBI – This is a big step forward in soil protection, because it is also effective on smaller riverbeds, characterized by sudden changes in flow and which are more affected by the extremes of atmospheric events.”

“Thanks to the use of artificial intelligence we can calculate probable flow scenarios based on rainfall data, detected by the entire system of rain gauges of a large basin and not only by those located near the bed of a watercourse", explains Ismaele Ridolfi, president of the Reclamation Consortium 1 Northern Tuscany.

“It is well known that the new rainfall patterns, more violent and concentrated in time and space, are a harbinger of sudden floods with often disastrous consequences.Artificial intelligence can help alert the area, activating the necessary precautionary measures – he underlines Massimo Gargano, general director of ANBI – This, however, draws attention to the question of water culture:too often, in fact, human tragedies derive from a lack of knowledge of the safety rules in flood events.It is necessary to invest in a structured civil prevention program:from works to widespread information.”

The agreement between Reclamation Consortium 1 Toscana Nord and the Department of Earth Sciences of the University of Pisa is active and tested on three waterways (Freddana, Versilia and Carrione) and on Lake Massaciuccoli.

“The artificial intelligence system also works in cases of intense and sudden events, which are the most difficult to predict, but which are increasingly frequent due to global warming,” he explains Monica Bini, scientific director of the university Department of Earth Sciences.

He adds Marco Luppichini, university professor, who analyzed the system:“We verified that local physical detections can lead to an incorrect estimate of the trend.This is largely overcome by applying machine learning models, which analyze a wide range of data”.“The management of big data is the frontier on which the research promoted by ANBI is working and these results push the Consorzio di Bonifica 1 Toscana Nord and the Department of Earth Sciences to strengthen collaboration for the future, also involving the Consorzio di Bonifica 4 Lower Valdarno".