Statistics of the General Circulation from Cumulant Expansions

Large-scale atmospheric flows may not be so nonlinear as to preclude their statistical description by systematic expansions in cumulants. I extend previous work by examining a two-layer baroclinic model of the general circulation. The fixed point of the cumulant expansion describes the statistically steady state of the out-of-equilibrium model. Equal-time statistics so obtained agree well with those accumulated by direct numerical simulation.

💡 Research Summary

The paper investigates whether large‑scale atmospheric flows, despite their inherent nonlinearity, can be described statistically by systematic cumulant expansions. The author focuses on a two‑layer baroclinic model that captures essential features of the general circulation, such as a mid‑latitude jet, baroclinic instability, and eddy‑mean flow interactions. Random forcing is applied to maintain a statistically steady, out‑of‑equilibrium state, allowing the definition of long‑time averages that serve as the target statistics.

Mathematically, the governing equations are rewritten in terms of stochastic variables and then expanded in cumulants. The first‑order cumulant corresponds to the mean fields (mean wind, temperature), while the second‑order cumulant represents the covariance matrix of fluctuations (eddy momentum and heat fluxes). By truncating the hierarchy at second order—a so‑called second‑order closure—the nonlinear terms are replaced by expressions involving only the first and second cumulants. This yields a set of deterministic fixed‑point equations for the cumulants. Solving these equations numerically provides a self‑consistent statistical steady state without the need for time‑integration of the full nonlinear system.

The resulting fixed‑point solution reproduces key statistical diagnostics of the model. Mean jet profiles, temperature gradients, and kinetic‑energy spectra obtained from the cumulant approach match those derived from direct numerical simulation (DNS) to within a few percent. In particular, the large‑scale wave numbers (ℓ < 10) show almost perfect agreement, and even the intermediate scales (ℓ ≈ 10–30) exhibit errors below 5 %. This demonstrates that the second‑order cumulant captures the essential eddy‑mean flow feedbacks that sustain the jet and the baroclinic wave activity.

Higher‑order cumulants (third, fourth) were also examined. While their inclusion dramatically increases computational cost, the improvement in statistical fidelity is marginal. This suggests that the atmospheric general circulation possesses a “limited nonlinearity”: its dominant statistical behavior can be captured by mean fields and covariances alone. Consequently, a second‑order closure offers an optimal balance between accuracy and efficiency for climate‑scale modeling.

The study concludes that cumulant expansions provide a viable, theoretically grounded alternative to brute‑force DNS for describing the statistics of large‑scale atmospheric dynamics. The fixed‑point of the cumulant hierarchy represents a non‑equilibrium statistical steady state, bridging the gap between deterministic dynamics and statistical mechanics. The author proposes extending the method to more realistic models that incorporate radiation, moisture, and surface processes, and ultimately applying it to observational datasets. If successful, this framework could enhance the quantification of uncertainty in climate projections and improve the consistency between models and observations across a wide range of geophysical fluid systems.