2017/02/25 by James Glimm, Glimm, James, Vinay Mahadeo +1
Earth and Planetary Sciences · Engineering · Environmental Science · #Climate variability and models #FOS: Physical sciences #Fluid Dynamics (physics.flu-dyn) #Fluid Dynamics and Turbulent Flows #Mathematical Physics (math-ph) #Meteorological Phenomena and Simulations
paper · pdf · doi:10.48550/arxiv.1702.08439
openalex publication_date 2017/02/25 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
We study the spatial statistics of velocity gradient volatility (i,e., the energy dissipation rate) in turbulent flow. We extend the Kolmogorov-Obukhov theory but also narrow its scope. The models are log normal, with verification from finely resolved large eddy and direct numerical simulations. They are parameterized by a mean and a covariance operator. Addressing applications to large eddy simulations, the mean and the covariance depend on the resolved scale solution. Removing this resolved scale dependence by a locally defined rescaling of the turbulent statistics yields a universal theory for the subgrid statistics, specifically for the mean and variance of turbulent fluctuations, i.e., the log of the energy dissipation rate ε. The variance of the velocity gradient statistics is found to be log normal, in accordance with Kolmogorov 1962. Rescaling by the resolved scale mean and variance removes the influence of the resolved scales on the subgrid scales, justifying conceptually the universality we observe. The universality is the basis of new power scaling laws. The new power laws, in turn, allow a simple parameterization of the mean and covariance of the subgrid rescaled dissipation rate, i.e., velocity gradient volatility. We treat the coarse grid resolved space- time location as a random variable. A restriction of the theory to a small range of unresolved but stochastically modelled scales is proposed, with renormalization group ideas offered to overcome this restriction.