Physics-based universal outlier detector for flow statistics

Outlier detection for PIV velocity fields is still nowadays an active field of research. In the last decades, several image pre-processing and processing algorithms have been developed aiming at increasing dynamic range measurements reducing measurement uncertainty. Nevertheless, are often characterised by presence outliers, which potentially hamper correct interpretation flow physics negatively affect evaluation statistics. The outlier strategies presented in literature mainly based on statistical analysis vector with its immediate neighbour. Most these demonstrated to be effective instantaneous fields, where errors associated outliers order magnitude larger than expected However, approaches not as statistics, yield same To overcome this limitation, paper proposed approach agreement statistics constitutive equations, more specifically turbulent kinetic energy (TKE) transport equation. focus posed ratio between local advection terms TKE a robust estimation production along streamline. It that, principle yields clear separation erroneous vectors. assess performance principle, three different test cases considered. For all them, results compared reference methodology, namely universal method Westerweel Scarano (2005). turbulence transport-based exhibits higher percentage correctly identified