Hydro-Meteorological Flood Data Sensing, Prediction and Classification using Internet of Things

A flood is a natural and seasonal calamity whose real time information is critical for engineers, researchers and public sector agencies. High speed communication technologies and Internet of things (IoT) systems can help in predicting the occurrence of the floods. To be effective, a flood event prediction system should be able to constantly monitor hydrometeorological factors. In this paper, we have developed an IoT system to sense, monitor, and detect the occurrence of flood events in real-time. Our system uses a machine learning (ML)-based predictor capable of correctly detecting and classifying flood events into various classes. To improve the system's classification efficiency, a novel approach to estimating water discharge based on cross sectional area and water flow is also proposed. Our system uses K-Nearest Neighbor (KNN) algorithm, and performance metrics like F1-score has been used to assess the system's effectiveness.