Unmanned aerial vehicle system integrating high-resolution sensors for detecting leaks in irrigation canals-Proof of concept*

A system for the remote detection of water leaks in irrigation canals is presented which integrates visual, thermal, and multispectral imagers into a specially designed platform mounted on small unmanned aerial vehicles (UAV), along with an image processing and analysis procedure employing three different software packages and customized scripts. The end results are processed images of different colour bands and indexes which are shown to be effective at highlighting the leaks' signatures. A custom payload platform weighing 1 kg was designed and built, consisting of high-resolution thermal, multispectral, and visual cameras, GPS units, a flight control camera, and a data acquisition system. A total of 27 flights in 4 irrigation districts were conducted to test and evaluate the payload platform and image processing system. The portable platform was successfully flown on three different UAV. A total of 46,387 images were captured and processed. The image processing and analysis system worked well and identified all known leaks, as well as leaks that the district's staff were unaware of. Thermal image indexes were effective at showing cold spots which were caused by canal leaks. The visual and the blue and green near-infrared indexes were also found to be effective at identifying leaks. The Normalized Difference Vegetation Index (NDVI) did help differentiate between vegetation and wet soils, but we did not obtain consistent results with the NDVI.