Data-driven prediction of reversal of large-scale circulation in turbulent convection

Quasi-stable large-scale circulation in turbulent thermal convection intermittently reverses its rotational direction. A fusion of physical insight into turbulent convection and a data-driven method known as reservoir computing enables accurate prediction of these chaotic reversals using only non-intrusive sensing via measurements of shear stresses and temperatures on the sidewalls. The successful prediction using such non-intrusive and sparse sensing opens up possibilities for closed-loop control of turbulent flows and feasibility in industrial applications.