Research on Flood Forecasting Model Based on Artificial Intelligence Technology
Xue-jun Yu, Jian Meng, Lei Sha, Lin-hai Liu · 2025
In recent years, extreme rainfall events leading to large-scale catastrophic flooding have posed a severe threat to public safety and property. Consequently, achieving rapid and high-precision flood forecasting became a critical issue that needed to be addressed. With the rapid development of artificial intelligence (AI) technologies, applying AI to river level forecasting significantly improved the accuracy and precision of flood predictions. This paper integrated AI with flood forecasting by processing basin water level data, flow, and rainfall information to construct corresponding feature values. The Xin'anjiang model was used to calculate the runoff generation and concentration results for the Yongxing Bridge watershed, which were then used as inputs for the LightGBM model. Together with the developed ensemble model, these inputs were employed to predict the water levels at Yongxing Bridge. The results indicated that this model effectively captured the fluctuations of the station's water level over time, demonstrating high prediction accuracy. It met most flood forecasting needs and provided a novel solution for basin water level flood forecasting.