River velocity measurements using optical flow algorithm and unoccupied aerial vehicles: A case study

Abstract

Accurate and reliable measurements of river flow are critical for a multitude of hydrologic engineering applications. However, flow rate measurements using in-situ sensors are uncertain in many applications and physical measurements of velocity may not be practical due to inaccessible sites or flood conditions. Recent advances in remote sensing using unoccupied aerial vehicles have overcome these limitations through non-contact measurements of river velocities; however, existing approaches have several shortcomings, including the need for artificial tracers in the absence of debris and prior knowledge of tracer size, shape, and flow direction. This case study seeks to overcome these shortcomings through the development of a system that utilizes drones, video imaging, and state-of-the-art optical flow algorithms to measure velocity in rivers. This system was applied along Menomonee River in Wauwatosa, WI. To remotely sense river flow, a DJI Matrice 210 RTK drone equipped with a Zenmuse X5S camera was used to capture video. The video data from the drone was analyzed using optical flow algorithms to generate velocity estimations. River velocity was measured directly at point locations using a hand-held velocimeter. Results indicate that the optical flow algorithms estimate the magnitude of surface velocity to within 13–27% of hand-held measurements without the use of artificial seeding. These outcomes suggest that this system could be used as a possible method to measure velocities in rivers.