Drainage Services Department
Hong Kong suffers from multiple floods each year. Drainage Services Department (DSD) has employed a series of flood prevention strategies, such as setting up a flood protection standard, performing studies on drainage systems, and carrying out preventive maintenance. Despite the continuous efforts, flooding risks still exist because of extreme weather. In this government-initiated project, ASTRI will collaborate with DSD to extend the current hydraulic and hydrological approach for flood monitoring and forecasting by considering multifaceted data and machine learning. Flooding may be caused by heavy rainstorms, storm surge, and overtopping wave spots. Therefore, a flood forecasting system will depend on meteorological data and on-site data. Spatial features from time-series radar images will be extracted by convolutional nets and a rainfall prediction engine will be trained with a recurrent neural network. A LoRaMesh network with intelligent transmission will be leveraged to integrate hydraulic and hydrological data from flooding blackspots. Finally, data from multiple sources will be fed into a deep neural network and a flood forecasting system will be produced for each trial site. This project aligns with The Hong Kong Government’s SmartCity initiative and will provide an essential technical solution for cost-effective flooding monitoring and prediction in Hong Kong. On one hand, it can bring new insights into the flood prevention approaches currently adopted by DSD. On the other hand, it can help the government to make timely strategies to avoid the possible huge economic loss and casualties.