During rainy seasons, Hong Kong will, more often suffer the impact of the inclement weather such as prolonged heavy rains and typhoons. The risk of flooding due to the prolonged heavy rains, the risks of flooding in both urban and rural areas, especially in low-lying areas, cannot be excluded. Therefore, relevant government departments have been taking multi-pronged measures to strengthen flood management with a view to reducing the risk of flooding.
To strengthen the monitoring of real time water level of watercourses in rural area, this department plans to deploy ultrasonic water level sensors and surveillance cameras to monitor the flooding conditions round-the-clock at about 20 locations of flood prone villages and low-lying areas. The water level sensors will connect to the Government Wide IoT Network (GWIN) through the long range (LoRa) gateway and real time video will be transmitted to department’s monitoring system directly through surveillance cameras. The images captured by the surveillance cameras together with the data transmitted by the water level sensors at regular intervals (i.e. every 10 minutes) will be kept at the department’s database automatically. When the reading of a water level sensor exceeds its pre-defined threshold, alert emails will be sent by the monitoring system to the staff concerned automatically, and the staff will access the surveillance camera remotely through the monitoring system to check the actual flooding situation of the concerned location. Staff gauges will also be installed at each location to provide reference of flooding situation which can be viewed through the surveillance cameras.
Workflow Automation
The project aims to automate the workflow in order to expedite the handling of flood reports. When the reading of water level sensor exceeds its pre-defined threshold and an alert email from the monitoring system is received, corresponding image(s) captured by the surveillance camera system should be retrieved automatically from the department’s monitoring system / database and sent through emails to the staff concerned for follow up actions immediately. Furthermore, it is proposed to develop a new mobile app for showing the monitoring results and displaying the relevant information to the mobile app users through their mobile devices.
Flood Detection based on AI System
Apart from the alerts triggered by water level sensors, the surveillance camera system should also monitor the water levels continuously. In the long run, AI model should be developed to provide 24x7 real-time flood detection at selected locations through the surveillance camera system by using cost effective, sustainable and low privacy impact technologies. Alert emails as well as notifications from the mobile app for necessary information such as affected location, date and time recorded as well as the captured image(s) when flooding is detected by the AI system should be sent to the staff concerned for immediate follow up. As the surveillance camera system is newly installed in each of the selected location, no previous images and videos of flooding could be provided at this stage.
