Surface microtopography measurement is the basis of surface roughness quantification, which is of great significance to the dynamic monitoring of surface morphology and the study of soil erosion process. This research adopted the chain method, cameras and unmanned aerial vehicle (uav) method, selected the conventional artificial rainfall soil bin and field runoff plots, set up the bare land and two processing plant roots respectively, to determine the surface roughness. The results showed that:(1) The average curvature indexes determined by the chain method, cameras and uav method were 0.257, 0.219 and 0.248, respectively, with the average actual bending index deviation was less than 0.242 < 10%. Among them, the bending index measured by the chain method was relatively large as a whole (6.2%), the bending index measured by the camera method was relatively small as a whole (-9.5%), and the bending index measured by the uav method was slightly large, but the error was only 2.5%. (2) The groove of bare soil was deep, and the measured results of chain method and camera method were inconsistent with the reality. The measured results of uav method were relatively consistent with the reality, and the chain method and camera method were slightly smaller than uav method by 50.0% and 40.3% respectively. The results of the chain method with root soil groove were not consistent with the actual conditions, and the results of the camera method and the uav method were relatively consistent with the actual conditions. The chain method was 48.2% and 57.8% respectively smaller than the camera method and the uav method. (3) Limited by the topographic factors, the camera method could not determine the surface roughness of runoff plot. For bare land and rooted plots, the measured results of uav were relatively consistent with the actual situation, while the measured results of chain method were inconsistent with the ups and downs of the surface. The surface bending index measured by chain method was slightly higher than that by uav method by 28.6% and 24.8%, respectively. In general, stereoscopic photogrammetry could effectively reflect surface roughness. However, due to the limitations of shooting height and terrain factors, the measurement results of camera method might be underestimated or limited, and the measurement results of uav method were relatively stable. Due to the influence of human operation and the error of the measurement process, the error of the measurement results of the chain method was relatively large.