Orthonormal wavelet transformations are used to decompose velocity signals of grid turbulence into both space and scale. The transforms exhibit small-scale enhancements of (i) the spatial fluctuation, (ii) the correlation in space between the adjacent scales, and (iii) the correlation in space between the longitudinal and transverse components. The spatial fluctuation and the scale–scale correlation at small scales are more significant in the transverse component than in the longitudinal component. These features are the same for different families of wavelets.

Turbulence contains tube-like structures of vorticity. We demonstrate that wavelet transforms of velocities are enhanced at the positions of the tubes, by using a direct numerical simulation. Thus our wavelet analyses have captured the effects of those coherent structures on velocities measured in the experiment, which would be difficult for traditional analysis techniques such as those with velocity increments.