Pore-resolved direct numerical simulations of turbulent flow over a randomly packed sediment bed reveal a quantity that most hydraulic models ignore: the depth to which turbulence penetrates the bed. At three permeability Reynolds numbers — 2.56, 5.17, and 8.94, spanning conditions representative of natural streams — the mean flow and shear penetration depths increase nonlinearly with permeability. The peaks of Reynolds stresses, form-induced stresses, and pressure fluctuations concentrate in the top layer of sediment spheres, not uniformly through the bed.
Classical hydraulics treats the streambed as an impermeable boundary — a wall. The no-slip condition applies at the bed surface, and the turbulent flow above is modeled independently of what lies beneath. But real streambeds are porous. Water moves through the gaps between grains, creating a hyporheic zone where surface water and groundwater exchange. The DNS results show that this exchange is not a passive seepage but a turbulence-driven process: eddies in the stream imprint their pressure fluctuations into the bed, and those fluctuations drive flow through the pore spaces below.
The probability distributions of normalized bed stress collapse across all three Reynolds numbers, suggesting a universal statistical structure underlying the turbulence-bed interaction. This collapse means the statistics can potentially be parametrized for larger-scale transport models without requiring pore-resolved simulation at every scale.
The through-claim is about boundaries that participate. A wall that absorbs momentum and transmits pressure fluctuations is not a boundary in the classical sense — it is a mediating layer. The bed does not merely resist the flow; it filters it, admits parts of it, and returns modified fluid to the stream. Transport models that treat boundaries as passive surfaces miss the physics of exchange — and the exchange is where much of the geomorphic work happens.