The Reynolds number for flow in a street canyon, Re = UrefH/ν (where Uref is a reference velocity, H the street canyon height, and ν the kinematic viscosity), cannot be matched between reduced-scale experiments and full-scale field measurements. This mismatch is often circumvented by satisfying the Re independence criterion, which states that above a critical Re (Rec), the flow field remains invariant with Re. Rec = 11,000 is often adopted in reduced-scale experiments. In deep street canyons with height-to-width aspect ratio ≥ 1.5, reduced-scale experiments have shown two recirculation vortices induced by the mean flows, but full-scale field measurements have observed only one vortex. We investigated this discrepancy by conducting water channel experiments with Re between 104 and 105 at three aspect ratios. The canyons with aspect ratio 1.0 have Rec = 11,000, the canyons with aspect ratio 1.5 have Rec between 31,000 and 58,000, while the canyons with aspect ratio 2.0 have Rec between 57,000 and 87,000. Therefore, the widely adopted Rec = 11,000 is not applicable for canyons with aspect ratio greater than 1.5. Our results also confirm that there is only one vortex in deep canyons at high Re. This single-vortex flow regime could change our fundamental understanding of deep canyons, which are often assumed to exhibit multiple-vortex flow regimes. Applications such as numerical model validation based on the multiple-vortex regime should be revisited. Our experimental data with Re up to 105 could be used to validate numerical models at high Re.