Innovative AI-based Flood Monitoring System: for Enhancing Predictive Insights by Reducing Sensor Dependency

Flood monitoring is necessary in smart cities to ensure the stability and security of cities’ functioning. Physical flood sensors such as water flow sensors are costly, have a high likelihood of being damaged by floods, and difficult to maintain. In this paper, a new flood monitoring system through the integration of Artificial Intelligence (AI) that minimizes the utilization of stationary sensors is presented. While using Convolutional Neural Networks (CNNs), drainage images, road elevation, weather application programming interfaces (APIs), and historical flood data, it provides accurate flood predictions. The degrees of precision, recall, and F1 score achievable by an AI-based system are slightly higher (~0. 90) than the traditional sensor-based system (~0. 85), and therefore, the AI-based system is more accurate and economical for flood prediction. The system increases smart cities’ vulnerability by providing a strong flood monitoring system, minimizing maintenance expenses and optimizing forecast data.