Drone-based leak detection and real-time monitoring of urban water in Harare, Zimbabwe

Increasing water scarcity resulting from inefficient urban water distribution systems across many African cities threatens the attainment of the Sustainable Development Goals. This is concerning given additional pressures on water resources such as climate change, pollution, increased demand from population growth and agriculture. Thus, there is a need to develop efficient, rapid and repeatable methods for tracking both urban water distribution and water use. This study applied drone-based thermal infra-red imaging for real-time leak detection and monitoring of Harare's aging water network by characterizing water distribution infrastructure with emphasis on spatial variations in pipe material and age. Very high-resolution thermal imagery was acquired using a DJI Mavic 3T drone and then processed before being integrated with water distribution infrastructure data within a GIS to identify temperature anomalies indicative of leaks. A scalable thermally based remote sensing algorithm was developed for detecting water leaks. The model achieved a precision of 82% and a recall of 90%, demonstrating the high accuracy of the model in detecting most leaks. An F1-Score of 0.86 demonstrated a strong balance between detection accuracy and false alarms, suitable for operational use. The integration of drone data within a GIS platform offers a scalable framework for a proactive approach in water-scarce urban contexts.