The harrowing stories of families and businesses affected by recent floods highlighted the urgent need for more accurate warning systems for such events. This makes the work of the University of Galway-based StopFloods research team all the more timely.
The team is developing a cutting-edge, artificial intelligence (AI) powered flood forecasting and decision-support system which integrates meteorological, tidal and river-flow data. By transforming fragmented data into clear, actionable insights, the team aims to equip emergency managers and communities with the means to anticipate, prepare for and respond to flood threats more effectively.
The team, led by Indiana Olbert and Tom McDermott, with societal impact champion Ciaran Broderick of Met Éireann, has won the Digital for Resilience Challenge under the National Challenge Fund. The National Challenge Fund was established under the Government’s National Recovery and Resilience Plan (NRRP), funded by the EU’s Recovery and Resilience Facility.
Co-ordinated and administered by Research Ireland, the fund calls on researchers to identify problems related to Ireland’s green transition and digital transformation, and works directly with those most affected to solve them.
RM Block
“StopFloods is a collaborative project strongly supported by the Flood Forecasting Centre at Met Éireann, Cork City Council, and local authorities nationwide, and we are grateful for their contributions,” says Olbert.
The project has its origins in the major Cork city floods of 2009. “Academic researchers were asked by the OPW to help to understand the mechanism that caused the flood,” Olbert explains. “Cork is quite complex, with the river Lee draining into the city, and there is also a large impact from the sea.”
Having developed a high-resolution hydrodynamic model to mimic the flood and the different factors that led to it, it was a natural next step to investigate how it could be used to forecast future events.
“We spent a lot of time talking to decision makers in local authorities, homeowners and businesses to understand the problem,” says McDermott. “We found a significant gap. People don’t have flood warnings at a local scale. They have weather forecasts and county by county flood warnings but there is a real need for local-level forecasts which are always on.”
He explains the complexity of the problem facing local decision makers. “Even with a yellow or orange warning, it is very difficult to predict floods. “Other factors include hydrogeology, land use in the area, the nature of the soil, saturation levels and so on. It’s not just about rainfall. You have to model soil moisture content and a number of different hydrological variables. That’s a real challenge for local decision makers. They have to look at all those different variables and figure out if conditions on the ground point to a flood being likely in the next 24 to 48 hours. That’s complex and difficult enough without trying to do it on a Zoom call at midnight.”
That’s the issue StopFloods aims to address, he continues. “We will provide all of that information on a simple-to-read dashboard. Decision makers will have a map of the local area showing the flood risk for the next 36 hours for different locations.”
The role played by AI is critically important. “Our team is expert at modelling flood risk but the models we have developed are slow and very computer intensive,” says Olbert. “This project uses AI to imitate what the models do and uses that for real-time forecasting at a local level. We are developing tools on the platform that work for decision makers and stakeholders in a form they can use. Our aim is to get the forecasts to the right people at the right time in the right place. There is nothing like this available at such high resolution.”
The €1 million in funding received from the National Challenge Fund will finance further development of the platform over the next two years. “The more data we have to train the AI model to predict the better,” says McDermott. “We have a working prototype for the city of Cork which has been validated against past events. It’s now about scaling up and putting it into the hands of decision makers.
“We are moving from development to deployment,” he continues. “The pilot site is Cork city, and we now need to move to forecasting for the entire country and from looking at historic data to creating forecasts in real time. By the end of the project, we hope to have the platform fully deployed in Cork city. We are starting work on Galway, and we hope to begin work on other locations as well.”
The ultimate aim is to have a real-time flood forecasting system available for the entire country. But that won’t be the end of the story. “Other countries are struggling with this as well,” McDermott says. “This will have global application.”
