Hierarchical Clustering of Modes in Numerical Turbulence Fields

Abstract

We propose an automated method for decomposing turbulence, where the modes are derived by applying singular value decomposition to the electrostatic potential field in turbulence simulations. After classifying the modes into zonal flows and turbulence based on azimuthal integration, the turbulence is further decomposed into finer components through hierarchical clustering. This approach systematically breaks down the turbulence into hierarchical levels, providing flexible control over the degrees of freedom in the mode decomposition process.