Figure 1

Left: vorticity field from DNS $(\mathrm{Re}=5000)$. Right: photos near the first horizontal grid and further downstream in the soap film channel. Third photo: a zoom on the side of the channel to show the intense activity created by the vertical cylinders in the soap film.Reuse & Permissions

Figure 2

Power spectra from experiments (a) and simulations (b) for the velocity (longitudinal $U$ and transverse $V$). The velocity spectra and the energy density are shown in arbitrary units. The frequency axis and the wave-number axis in (b) and the inset to (b) are in reduced units in such a way that a frequency $f$ of 1 and a value of 1 for $k∕2\pi$ correspond to the channel width. Inset of (b): Energy density spectrum obtained from the spatial velocity field for the DNS. The frequency axis in the experimental data can be converted to length scales using the frozen turbulence assumption (in units of channel width): 160 Hz corresponds to $1∕5$ (or 1 cm in physical units). For the numerics the same thing can be achieved: $f=10$ corresponds to $1∕10$ (in units of channel width).Reuse & Permissions

Figure 3

Energy transfer function $T\left(k\right)$ from the experiments; $W$ is the channel width of 5 cm. Inset: the energy flux $\Pi \left(k\right)$.Reuse & Permissions

Figure 4

Power spectrum of thickness fluctuations. $U_{\text{mean}}=160\mathrm{cm}∕\mathrm{s}$ so 100 Hz corresponds to 1.6 cm.Reuse & Permissions

Figure 5

${⟨\omega \left(0\right)\omega \left(r\right)⟩}^{3∕2}$ versus $r$ in semilog plot: (a) experiments; (b) numerical simulations. The insets show the vorticity structure functions.Reuse & Permissions