PUB, Singapore’s National Water Agency
Rainfall in general is a complex phenomenon which is very difficult to monitor and forecast. Rainfall in tropics is admittedly more complex due to high spatiotemporal variability, especially true in case of convective precipitation events. It is challenging to accurately quantify the amount of rainfall on the ground at any given location after a rainfall event and even more challenging to forecast rainfall with significant lead time and accuracy.
The use of rain gauges is the de facto standard for gathering rainfall data. Ideally, these devices are installed in open spaces to minimise environmental interference. However, due to space constraints, the spatiotemporal resolution of data collected from these traditional gauges is typically not very dense. It is thus of interest to explore solutions that would increase the spatial density of monitored rainfall information where use of rain gauges is impractical.
Technological advances in image processing and computer vision techniques enable extraction of diverse features and knowledge discovery from images and videos. Optical sensors such as CCTV’s and other Internet of Things (IoT) cameras are already widely available in cities. These cameras also record rainfall events and street water levels, and this can be considered as an “opportunity” for rainfall monitoring. By processing the camera feed via advanced data analytics methods, and distilling the data for knowledge using machine learning approaches, this optical sensor network can be used for early flood warning, and to supplement the real-time information from water level sensors.
H2i is collaborating with PUB, Singapore’s National Water Agency on the development of a “camera gauge”. The streets in Singapore are already equipped with several optical sensors, including CCTV cameras, that are installed for various purposes. Thus, this project will explore the idea of re-purposing such sensors into a new source of rainfall measurement. We envision that the project will allow for the conversion of readily available camera feeds to an instant associated measurement of the rainfall rate. Together with the existing network of traditional rain gauges, this will help to capture street level rainfall information in real time!