This paper introduces a new dual-beamwidth (2BW) radar method for inferring the velocity variance σturb2 due to the turbulent kinetic energy density per unit mass. The 2BW method has the advantage of being much less sensitive than the standard single-beamwidth (1BW) method to the necessary instrumental corrections of the observed velocity variance σobs2. We test and compare the 2BW and 1BW methods using σobs2 between 5.0 and 7.5 km MSL measured by the MU VHF radar near Shigaraki, Japan, with beams directed toward 10¿ zenith angle. We find that during periods of light winds the σturb2 profiles from the two methods agree well. But with wind speeds >30 m/s the 1BW values of σturb2 are about 25% larger than the 2BW values. With the 2BW method, typical values of σturb2 between 6.0 and 7.5 km MSL increased from ~0.07 m2/s2 with winds of ~2 m/s to ~0.2 m2/s2 with winds between 30 and 60 m/s.