Title: Real-Time Flash Flood Prediction System Using Artificial Intelligence (AI) Based on Sensor and Terrain Data
Authors: Nguyen Minh Tri and Trinh Thanh Do
Volume: 9
Issue: 6
Pages: 225-232
Publication Date: 2025/06/28
Abstract:
Flash floods can be extremely destructive and are hard to forecast precisely especially in mountainous areas. This paper proposes an artificial intelligence-based early warning system that uses real-time data from Internet of Things Sensor data alongside digital elevation models (DEM) for risk reduction. Essential environmental factors - rainfall, soil moisture, water and stream height, flow rate, and the slope on the terrain - are obtained via high-risk zones and processed using LSTM and Random Forest models. The system provides approximate predictions of up to 85% accuracy from 30 minutes up to 2 hours prior to a flash flood event.What's special is this system can "self-learn" from continuously integrating new data and the ability to notify alerts via mobile apps and dashboards to local communities. Therefore, this paper will present the suggested approach to using AI and IoT for disaster prevention; especially to developing countries with complex terrains such as Vietnam.