We conducted an experiment comparing wind speeds and aerodynamic roughness length (z0) values over three snow surface conditions, including a flat smooth surface, a wavy smooth surface, and a wavy surface with fresh snow added, using the wind simulation tunnel at the Shinjo Cryospheric Laboratory in Shinjo, Japan. The results indicate that the measurement location impacts the computed z0 values up to a certain measurement height. When we created small (4 cm high) snow bedforms as waves with a 50 cm period, the computed z0 values varied by up to 35% based on the horizontal sampling location over the wave (furrow versus trough). These computed z0 values for the smooth snow waves were not significantly different than those for the smooth flat snow surface. Fresh snow was then blown over the snow waves. Here, for three of four horizontal sampling locations, the computed z0 values were significantly different over the fresh snow-covered waves as compared to those over the smooth snow waves. Since meteorological stations are usually established over flat land surfaces, a smooth snow surface texture may seem to be an appropriate assumption when calculating z0, but the snowpack surface can vary substantially in space and time. Therefore, the nature of the snow surface geometry should be considered variable when estimating a z0 value, especially for modeling purposes.