2016 Volume 19 Issue 1 Pages 91-100
Rapid development and diffusion of unmanned aerial vehicles （UAVs） have been changing the natural environmental measurements in field survey and research. In this study, we attempted to validate the accuracy of natural-environmental measurements based on aerial photos from UAVs, and discussed the potential and problems in the use of UAVs. As a specific example, we took photographs of 1/70 “Gokase River” replication （width: approx. 10 m, length: approx. 95 m）, which was set up under open sky conditions, through a UAV at heights of 10 and 50 m, and conducted three-dimensional measurements based on the aerial photos using the structure from motion （SfM） technology. Digital surface models （DSMs） and ortho photos based on the UAV photos （number of pictures– 10-m shots: 126, 50-m shots: 24） were analyzed in the SfM using “Photoscan pro.” for each height. The accuracy was validated by comparing actual cross-sectional contours with predicted contours from the DSMs at four sites in river replication. The following results were obtained: 1） the predicted contours from DSM well captured the complex shape of the river replication, and 2） two predicted contours among four contours were especially accurate within 5 cm whereas other predicted contours near the margins of DSMs were susceptible to error （i.e., over 10-20 cm）. Thus, these results suggested that the combination of UAV and SfM facilitates efficient natural-environmental measurements.
