Experimental Study on the Development of Scour Holes

Abstract

Scour holes are localized erosion features that develop in a range of river morphologies, including transitions between non-erodible and erodible riverbeds, river narrowings, heterogeneous bed compositions, and at river bifurcations and confluences. They are of practical relevance in managed river systems, where they affect navigability and can threaten bank and infrastructure stability. Field observations from the Dutch Rhine Delta show that scour holes vary considerably in their behaviour: some respond dynamically to discharge fluctuations, while others remain stable over time. Although previous laboratory studies have investigated the formation of scour holes from a flat bed, the response of already-existing scour holes to changing discharge conditions has not been systematically studied in a controlled experimental setting. Such knowledge is critical for predicting whether a scour hole will deepen or stabilize under flood events, with direct implications for navigation safety and the protection of riverbank infrastructure.Five physical model experiments are conducted in a recirculating glass flume at the hydraulic engineering laboratory of the TU Wien, using a Froude-scaled model with a non-erodible to erodible bed transition. Bed profiles are recorded using underwater laser scanning, and flow velocities are measured using acoustic Doppler velocimetry (ADV and ADVP). The study is guided by three research questions addressing the reproducible pre-formation of scour holes, their morphological evolution under varying discharge, and the internal flow structure within the scour hole.Regarding pre-formation, a wooden stencil based on a parametrized scour profile is used to reproducibly carve initial scour geometries into the sand box before each experiment. The standard deviation of the deepest point elevation across repeated pre-forming attempts is 1.6 mm, comparable to the maximum sediment grain size. Experiments conducted under the same flow conditions show closely similar morphological development, confirming that the method produces consistent and reliable initial conditions. The resulting scour geometry is described using a two-segment analytical parametrization, which serves as a reference for assessing morphological changes in subsequent experiments.Regarding morphological evolution, the response of the scour hole is found to be strongly dependent on discharge. Under base flow conditions, deepening is subtle and confined to the vicinity of the deepest point, with the overall profile shape remaining largely stable. Under increased flow, incision is more pronounced across the full profile. Under peak flow, the scour hole undergoes the most significant morphological change: erosion extends over the full sandbox length from early in the experiment, the rate of deepening is more than six times the deepening observed under base flow, and the profile transitions from the initial asymmetric shape towards a bowl-like geometry as the deepest point migrates progressively downstream. The two-segment analytical parametrization introduced to describe the initial scour geometry remains applicable under base flow and during the early stages of increased flow but fails to capture the evolved geometry under peak flow conditions.Regarding flow velocities, the presence of the scour hole causes significant deceleration of the flow relative to the bulk flume velocity. The strongest deceleration is observed in the vicinity of the deepest point, with minimum depth-averaged streamwise velocity ratios of approximately 56-68% of the bulk velocity under peak flow conditions, decreasing progressively as the hole deepens. Locally, within the deeper parts of the scour hole, velocities fall to 10-20% of the bulk flow, consistent with findings from previous studies. A distinct shear layer develops just below the original bed elevation, where the slower flow inside the scour hole meets the faster overlying flow. Turbulent kinetic energy and Reynolds shear stress are elevated in this region and increase progressively as the hole deepens. An analysis of coherent turbulence structures using cross-correlation of vertical velocity and Reynolds shear stress time series does not yield conclusive results, likely due to the device spacing exceeding the coherence length of individual eddies.The results of this study are broadly consistent with findings from comparable flume experiments by Stenfert (2017) and Van Zuylen (2015), as well as with field observations of dynamic scour hole behaviour reported by Oldenhof et al. (2026). The study contributes to a better understanding of how existing scour holes respond to varying discharge conditions. Future work could extend the findings by investigating three-dimensional scour hole evolution, live-bed conditions with upstream sediment supply, and longer experiment durations to capture the full migration of the deepest point under peak flow.