Bed roughness effects on horseshoe vortex dynamics and soil erosion mechanisms in vegetated overland flows

Understanding how bed roughness modulates hydrodynamic processes around vegetation is critical for predicting soil erosion patterns in sloped landscapes. Through flume experiments with high-frequency particle image velocimetry (PIV), this study quantifies the interactions between bed roughness (ks = 0.009, 0.25, 0.75, 1.55) and horseshoe vortex (HV) dynamics under overland flow conditions (ReD = 2627–3815). Time-averaged flow field analysis, based on vorticity and swirl strength methods, revealed that increasing surface roughness disrupted the HV system by reducing the number of vortices, decreasing the vorticity and swirl strength of the primary HV, and shifting its position closer to the bed. Statistical analysis of the instantaneous velocity components showed the emergence of bimodal probability density functions (PDFs) and joint probability density functions (JPDFs) in the near-wall region upstream of the cylinder, representing the backflow and downflow events. As roughness increased, the bimodal region decreased in size and shifted further from the cylinder. Linear stochastic estimation (LSE) was used to characterize the underlying flow modes, indicating that the backflow event was associated with the backflow mode, while the downflow event was linked to the zero-flow mode. Notably, roughness elements enhanced flow stagnation (zero-flow mode dominance > 60 %), suggesting a potential mechanism for erosion mitigation. These findings provide quantitative linkages between micro-scale hydrodynamics and landscape-scale erosion processes, informing the design of vegetation-based erosion control strategies through targeted roughness manipulation.