Author(s): Masahiro Hashiba; Ichiro Fujita; Noriyuki Nishiyama; Yoko Ohta
Keywords: Image velocimetry; Aerial STIV; UAV; GCP; SfM/MVS
Abstract: By using a UAV (Unmanned Aerial Vehicle), it is possible to measure the river flow by image velocimetry wherever the place away. However, to perform image analysis, it is necessary to install the ground control point (GCP) in the image and to set the accurate coordinates to run a geometric correction. In this study, we extract as pseudo-GCP the point cloud of the 3D terrain model, which was made by Structure from Motion (SfM) using a post-processed kinematic system by UAV. After we calculated the geometric correction by the pseudo -GCP used point cloud, we analyzed the surface velocity distribution by Space-Time Image Velocimetry (STIV) method. We conducted the field observation in the Ishikari river (The river width was 150m). Firstly, we took a picture by drone in a zigzag direction from the left bank to the right bank with a lap rate of 70-85% and created the mesh of 4cm square. It took within a 20minute for transverse distance was 250m and longitudinal distance was 300m. Secondly, we took a slant angle video by a drone hovering about 30 seconds near the riverbank. Comparing the surveying-GCP and pseudo-GCP, the plane position error and the height error was within 0.10 m. Moreover, the flow velocity error due to the surveying-GCP and pseudo-GCP was within 10%, except for low velocity and discharge was within 5%, regardless of the divided sectional method and DIEX method. We succeeded to calculate surface flow velocity with the minimum error by the STIV method using pseudo -GCP which is no requirement of surveying-GCP.