Uncertainty Growth in Stably Stratified Turbulence
Mrinal Jyoti Powdel, Samriddhi Sankar Ray
TL;DR
This study quantifies uncertainty growth in statistically steady stably stratified turbulence by employing twin DNS runs and decorrelator diagnostics within the Boussinesq framework. It reveals a monotonic decrease of the largest Lyapunov exponent $\lambda$ with increasing stratification (via $N$ and ${\rm Fr}$), while preserving a universal sequence of decorrelation: initial decay, exponential growth with self-similar spectra, and saturation. The growth is highly anisotropic, with vertical spread strongly suppressed by buoyancy, and horizontal spread dominating as stratification strengthens. The analysis shows that chaos suppression arises primarily from strain-mediated alignment with the compressive eigen-direction of the rate-of-strain tensor, rather than direct buoyancy effects, highlighting the utility of decorrelator-based methods for anisotropic geophysical flows.
Abstract
We investigate uncertainty growth and chaotic dynamics in statistically steady, stably stratified three-dimensional turbulence. Using direct numerical simulations of the Boussinesq equations, we quantify the divergence of initially infinitesimal perturbations via twin simulations and decorrelator diagnostics. At short times, perturbations exhibit exponential growth, allowing us to define a (largest) Lyapunov exponent. We systematically examine how this exponent depends on stratification strength, quantified by the Brunt--Väisälä frequency and the Froude number, in a parameter regime relevant to oceanic flows. We find that increasing stratification leads to a monotonic reduction of the Lyapunov exponent, indicating suppressed chaoticity. Despite this reduction, uncertainty growth retains the universal temporal sequence observed in homogeneous isotropic turbulence -- initial decay, exponential growth, and saturation. The growth phase is characterized by self-similar decorrelator spectra, but exhibits strong anisotropy: uncertainty spreads much more slowly along the stratification direction than horizontally, with the disparity increasing with stratification strength. An analysis of the decorrelator evolution equation reveals that the suppression of chaos arises primarily from strain-mediated alignment dynamics rather than direct buoyancy coupling. Our results provide a quantitative characterization of predictability and uncertainty growth in stratified turbulence and highlight the utility of decorrelator-based methods for anisotropic geophysical flows.
