Integrating Opportunistic Data Streams for Real-Time Flood Forecasting

<p dir="ltr">Traditional hydrologic sensing relies on established monitoring networks, including rain gauges, weather radars, and hydrometric stations. While effective, updating/expanding these systems for real-time flood forecasting and environmental monitoring often presents significant financial and logistical challenges. In contrast, unconventional and opportunistic data sources such as commercial microwave link attenuation measurements, CCTV footage, audio surveillance, road traffic behaviour, and smart technologies like car auto windscreen wipers offer a promising alternative for enhancing environmental sensing at a fraction of the cost. The utility of such data depends on effective data fusion techniques, which enable the integration of multiple heterogeneous sources. Some data streams provide broad spatial coverage, while others offer high accuracy at specific locations. Combining these complementary sources allows for a more robust flood forecasting system, reducing uncertainty and enhancing resilience in flood-prone areas. This study explores the potential of these opportunistic data streams and proposes a structure to integrate them into an operational framework.</p><p dir="ltr">This paper was presented at the 21st Computing and Control in the Water Industry Conference (CCWI 2025) at the University of Sheffield (1st - 3rd September 2025).</p>