Flood Forecasting with Bayesian Approach

Abstract

A wireless sensor network is made up of low-cost, energy-autonomous devices capable of monitoring physical or environmental conditions (temperature, humidity, noise, vibration, pressure, movement, pollution, etc.), performing specific calculations, and collaborate to transmit their data over wireless links to a recipient. With climate change, the occurrence of heavy rains becomes a danger that often leads to flooding. Heavy rains are characterized by their magnitude, duration, severity, and extent, controlled by sensor networks. This work will present a distributed decision support system that can be deployed to assist decision-makers in flood mitigation operations, namely a real-time flood prediction warning system featuring IBM web services. This system is based on the Bayesian approach, allowing decision-makers to take the necessary actions before the disaster occurs. A regression model is used to infer whether an alert is true or false.