A New Wall-Dependent Identification Method for Coherent Streamwise Streaks in Turbulent Channel Flows

A wall-dependent identification method for low-speed streamwise streaks in turbulent channel flows is presented to overcome a key limitation of uniform thresholds, which tend to produce either an excess of detected structures near the wall or an apparent lack of streaks farther from it. The method is constructed following wall-dependent threshold requirements established in the literature and is applied to DNS data of Couette–Poiseuille flows at Reτ≈250 using percolation analysis and clustering to extract three-dimensional objects. In addition, a percolation study at a significantly higher Reynolds number is included to assess the effect and robustness of the filter beyond the reference case. Although some global trends remain consistent with previous results obtained using uniform thresholds, the present work provides a direct comparison between both approaches and shows more clearly how the threshold definition affects the identification of streaks away from the wall. The detected structures are classified into wall-attached and wall-detached families, separated by a transition region at ymin+=20, and both exhibit self-similar geometric trends. Their spatial distribution is also analysed, showing differences mainly associated with the alignment of attached streaks in pure Poiseuille and pure Couette flows.