Turbulence modeling by time-series methods

TL;DR

This paper introduces a novel time-series based model for stationary turbulent velocity, aligning with Kolmogorov's hypothesis, and proposes a non-parametric method to estimate turbulence intermittency from data.

Contribution

It presents a new turbulence modeling framework that separates second-order statistics from higher order ones, inspired by recent modeling ideas and classical hypotheses.

Findings

Model aligns with K41 hypothesis of local isotropy

Provides a non-parametric method for intermittency estimation

Separates deterministic and stochastic components in turbulence

Abstract

A general model for stationary, time-wise turbulent velocity is presented and discussed. This approach, inspired by modeling ideas of Barndorff-Nielsen and Schimgel, is coherent with the K41 hypothesis of local isotropy, and it allows us to separate second-order statistics from higher order ones. The model can be motivated by Taylor's hypothesis and a relation between time and spatial spectra. Second order statistics are used to separate the deterministic kernel function and the weakly stationary driving noise. A non-parametric estimation method for the turbulence intermittency is suggested.