Preventing Irrigation Canal Breaches Using Small Unmanned Aerial System with Multispectral Payload

摘要

This article proposes a solution for detecting and localizing water leaks in irrigation canals using a small unmanned aerial system (UAS) with a multispectral payload to enable timely prevention of irrigation canal breaches. The current state of the art is to use multispectral imagery from satellites orhigh-altitude manned aircraft, but this lacks sufficient resolution to find seepage and erosion before a catastrophic failure occurs. The other current practice is manual ground inspection, which is time-consuming and expensive plus inspectors may not be able to getthrough the brush to fully inspect the canal. This project is adapting and expanding multispectral payloads developed for manned aircraft and applying them to an AirRobot AR180 and AR200 quadcopter UAS. A custom payload weighing 1 kg was designed and built, fusing two nadir-mounted thermal FLIR Vue Pro R and multispectral MicaSense RedEdge cameras and incorporating a forward-facing CMOS Camera Module 728 px × 488 px visual pilot camera. A total of 27 flights were flown at 13 sites in 4 irrigationdistricts in Garwood, TX, and the Lower Rio Grande Valley region to establish ramifications for UAS, payload, and concept of operations as well as to evaluate the proposed solution. The fused imagery identified all known surface leaks but was unable to detect any underground leaks. This work is expected to reduce leaks in irrigation canals which account for 10 % to 40 % of all water loss. A low-cost platform would enable the local water districts to fly their canals each month rather than annually and catch leaks before they cause major problems. The work is also expected to aid in detecting seepage in levees before they fail, thus preventing major damage from flooding.