Soil surface roughness (SSR) plays an important role in the physical and hydrological processes of soil surfaces. In order to achieve nondestructive, fast, and large area measurement of SSR, unmanned aerial vehicle (UAV) photogrammetry method was used to take digital images at the altitude of 10 m on three plots and generated the digital elevation model for calculating SSR. From the results of UAV-based SSR, the following conclusions were obtained. First, the domain of soil surface height was consistent with the designed height of the three plots: smooth (all pixels: -5.5-6.5 cm and 80% pixels: -2.3-2.3 cm) < medium (all pixels: -8.5-8.5 cm and 80% pixels: -3.4-3.4 cm) < rough (all pixels: -16.0-13.0 cm and 80% pixels: -6.8-6.8 cm). Second, UAV-based SSR can represent their differences among the three plots, indicated by a consistent root-mean-square height (rmsh) and correlation length (cl) with the pin-profiler results. Third, UAV-based SSR results can reveal the anisotropy of SSR, and for the medium plot, the maximum variation of rmsh and cl with observed azimuth angle is 0.77 cm and 14.35 cm, respectively. The UAV-based SSR method has the advantages of low cost, high efficiency, and all-directional measurement, and can be used in remote sensing model and hydrological simulation.