Recovering subgrid-scale features in turbulent flows through compressive sensing

Compressive sensing is a powerful technique in image processing that can overcome the classical Nyquist criterion in resolving details of the flow. Yet, this has found little applications in thermal-fluid imaging, to our knowledge at this time. We demonstrate that compressive sensing can be used to recover fine-scale features of turbulence, at imaging resolutions far below those thought possible. Several different turbulence geometries and processes are used as examples, and at different sampling geometries and scales. The results show that the pyramidal sampling configuration is by far the most efficient, and also that compressive sensing in general has important applications in sensing of turbulence. In addition, further applications are suggested on resolving subgrid features using compressive sensing.