Validation of High-resolution and Simple River Monitoring Technique using UAV-SFM Method

Abstract

This study tested the accuracy and precision of the UAV-SfM method, an automated photogrammetry technique called SfM (Structure from Motion) using multiple pictures taken by UAV (Unmanned Aerial Vehicle), in a section of Saba river, Yamaguchi, Japan. The method was applied in the submerged area as well as in the exposed area, taking into account the refraction at the water surface, for the first time in domestic rivers. When the resultant DEM (Digital Elevation Model) is considered as the map of riverbed elevation, the RMS (Root Mean Square) error and R2 (coefficient of determination) of UAV-SfM were 0.165 m and 0.93, respectively. In pixels with thick algae cover, large apparent overestimations reaching 0.351 m at maximum were observed because UAV-SfM actually measures the algae surface elevation, not the riverbed elevation. Error analyses also showed that the refraction correction method adopted in this study is working well in spite of its simplicity.