ETRI launches AI-based urban flood prediction pilot with Busan

South Korea is stepping up efforts to develop technology that uses AI and digital twin systems to predict urban flooding from extreme rainfall before it strikes and reduce the resulting damage.

The Electronics and Telecommunications Research Institute (ETRI) announced Wednesday that it is launching a field demonstration of an AI-based urban flood response platform in partnership with the Busan Metropolitan Government, the Korea Institute of Civil Engineering and Building Technology, the Korea Institute of Geoscience and Mineral Resources, and the Korea Railroad Research Institute.

The collaboration aims to validate and commercialize technology developed under a National Research Council of Science and Technology (NST) convergence program titled "Development of Digital Urban Flood Control Technology for Realizing a Flood-Safe City (No WET City)."

Climate change has brought increasingly frequent localized downpours exceeding 100 millimeters per hour and repeated urban flooding, exposing the limits of disaster response systems that rely primarily on conventional meteorological observation.

Over the next four years, the participating institutions plan to link their specialized technologies and data to build an integrated digital disaster response system capable of predicting, analyzing and managing urban floods — and disseminating alerts in real time.

ETRI will lead the development of AI model-based hazard detection and urban flood prediction technology.

Researchers will develop core AI platform technology that integrates and analyzes a wide range of urban data — including rainfall information, CCTV footage, drainage networks, retention facilities, topographic and infrastructure data, sensor readings, and historical flood records. Going beyond simple statistical analysis, the team will build a "hybrid urban flood prediction model" combining physical simulation with AI. The system will digitally replicate actual drainage flows and flooding patterns in a city, allowing AI to learn from that data and detect hazardous conditions at an early stage.

The technologies ETRI is developing include AI-based urban flood hazard detection, flood prediction, physics- and numerical analysis-based flood simulation, flood risk assessment, compound disaster scenario generation, and a digital standard operating procedure (SOP) service.

The ultimate goal is to link these capabilities with real-time field data and geospatial information to deliver integrated disaster early warnings and situational alerts.

Busan will supply data on the city's drainage networks, topography, structures, urban facilities and flood history, and will carry out phased field demonstrations focused on areas prone to repeated flooding — verifying the technology's applicability and effectiveness in a real urban environment.

The Korea Institute of Civil Engineering and Building Technology will develop urban flood prediction and control technology linked to city infrastructure. The Korea Institute of Geoscience and Mineral Resources will build a 4D urban geological environment-based platform for responding to geological disasters. The Korea Railroad Research Institute will develop technology to strengthen railway infrastructure against extreme rainfall, extending the collaboration's scope to urban transportation safety.

"Through this collaboration, we expect AI and digital technology to become a core means of protecting citizens' lives and property," said Jang Yun-seop, a principal researcher at ETRI's Disaster Safety ADX Research Laboratory. "We will focus ETRI's ICT capabilities on making Busan a leading model of a smart, climate-resilient safe city."