Guri City implements AI-driven flood prediction for Wangsukcheon Stream

Guri City has developed an advanced flood prediction system for the Wangsukcheon Stream by integrating real-time public data with generative AI coding capabilities. The project, titled ‘Establishment of a Flood Prediction System through Analysis of Upper Wangsukcheon Stream Rainfall Forecasts,’ received an excellence award at the ‘2026 Gyeonggi Province AI and Data Policy Best Practice Presentation Contest.’

The system utilizes real-time data from the Han River Flood Control Office and the Korea Meteorological Administration API hub. By employing the coding functions of the generative AI ‘Gemini,’ the city built its own water level prediction model and real-time monitoring system without relying on external contractors.

The model calculates water levels by analyzing various hydrological factors, including rainfall amounts, runoff coefficients, arrival times, and the impact of drainage delays caused by water discharges from the Paldang Dam. By combining forecast-based and measurement-based prediction curves, the system has reduced the average error margin to approximately 5cm. This precision allows the city to predict water level changes 1 to 3 hours before heavy rainfall occurs, providing a golden time to preemptively control high-risk areas such as riverside roads and parking lots.