Abstract
For atmospheric turbulence, multiplying an estimate of the convection velocity with the integral time scale is useful for estimating the integral length scale. Velocity scales that have been used to estimate the convection velocity include the local mean velocity, the ratio of e -folding length and time scales, the approximations of the slope of an elongated band of space-time autocorrelation isocontours, and the phase speed of Fourier modes. A knowledge gap is the lack of evaluation of these velocity scales directly against the convection velocity, especially for canopy flows where previous studies have reported somewhat inconsistent results. The objective of this work is to assess the ability of the local mean velocity and each autocorrelation-based velocity scale to estimate the convection velocity of velocity components in canopy flows. Firstly, large-eddy simulation (LES) results of neutral canopy flows are used to compare these velocity scales to directly quantified convection velocity. When the direction of interest roughly aligns with the local mean velocity (specifically, for an angle of $$7.5^\circ $$ 7 . 5 ∘ or smaller), all autocorrelation-based velocity scales generally agree better with the convection velocity than the local mean wind component. When the direction of interest departs from the local mean velocity for more than $$15^\circ $$ 15 ∘ , the ability of each velocity scale to approximate the convection velocity changes substantially. Secondly, data collected during the Canopy Horizontal Array Turbulence Study (CHATS) are used as an example of interpreting estimates of the convection velocity in the field with the guidance from LES findings. The guidance focuses on uncertainties of velocity scale estimates and potential caution needed when using these estimates.