A swirl-coherent Q-criterion for vortex identification via statistical thresholding

Accurate identification of vortical structures is central to the analysis of complex fluid flows. While the Q-criterion remains one of the most widely used vortex identification methods, it is known to suffer from sensitivity in shear-dominated regions and a lack of robustness in threshold selection. In this study, a novel vortex identification technique is proposed that enhances the classical Q-criterion by incorporating a local measure of rotational coherence and applying a data-driven thresholding approach. Specifically, the method modulates Q by a swirl coherence metric derived from spatial variations in the vorticity field, emphasizing rotationally consistent regions. A Gaussian mixture model is then employed to determine an adaptive threshold that separates vortical regions from background flow structures. The resulting framework enables automatic and physically interpretable vortex segmentation. The method is tested on numerical simulations of flow over a cambered plate, on particle image velocimetry data of colliding vortex rings, and in geophysical flow. Comparisons with conventional criteria demonstrate that the proposed method yields more accurate results and avoids false identifications in high-shear zones. This hybrid physical–statistical approach offers a robust and generalizable tool for vortex identification across a wide range of flow conditions.