Flood detection and forecast by IoT technology

J. H. Jang, T. W. Li

研究成果: Paper同行評審


Under the influence of climate change, the scale and impact of flood disasters have become more and more severe in Taiwan due to the increase in rainfall intensity and urbanization. To strengthen the technology of flood detection and forecast in urban areas, an IoT (Internet of Things) based flood sensor (named flood box) and a flood prediction model based on machine learning (ML) technology were developed in this study. For flood detection, the flood boxes are installed at several low-lying locations in Tainan, Taiwan, for inundation depth measurement. Pressure tests show that the flood boxes functioned normally under outdoor rainy weather conditions. For flood forecast, the observed data by flood sensors are processed by a SVR (Support Vector Regression) ML model to predict the inundation depth at the locations where flood sensors are absent or malfunctioned in a flood event on 13 August in 2019. Satisfactory agreements between prediction and observation are found with the overall RMSE (Root-Mean-Square Error) equivalent to 5.73 cm.


Conference22nd Congress of the International Association for Hydro-Environment Engineering and Research-Asia Pacific Division: Creating Resilience to Water-Related Challenges, IAHR-APD 2020
城市Sapporo, Virtual

All Science Journal Classification (ASJC) codes

  • 生態學
  • 環境工程


深入研究「Flood detection and forecast by IoT technology」主題。共同形成了獨特的指紋。