Intermittent Obukhov-Corrsin Regime

• Intermittent Obukhov-Corrsin regime is a turbulence framework where passive scalars follow classical inertial scaling at second order while exhibiting anomalous higher-order statistics.
• Research approaches combine rigorous PDE analysis with phenomenological methods to uncover non-Gaussian small-scale behavior and codimension-sensitive dissipation.
• Key implications include improved predictions for passive scalar mixing, anomalous dissipation mechanisms, and adaptations to buoyancy-dominated and nonstationary regimes.

The intermittent Obukhov–Corrsin regime is a family of passive-scalar turbulence regimes in which the classical Obukhov–Corrsin inertial-convective picture is retained at second order or in an averaged sense, while intermittency appears through anomalous higher-order scaling, non-Gaussian small-scale statistics, anomalous dissipation, or codimension-sensitive concentration of dissipation. In the classical homogeneous isotropic setting, the baseline prediction is the Obukhov–Corrsin law $E_\theta(k)\sim C\,\chi\,\epsilon^{-1/3}k^{-5/3}$ together with the dimensional estimate $S_p(r)\propto r^{p/3}$. In recent rigorous PDE work, the same phrase is used in a different but related sense: a supercritical regularity regime for transport by rough divergence-free flows, in which uniform regularity and anomalous dissipation coexist. By contrast, some recent mathematical derivations of the Obukhov–Corrsin spectrum explicitly prove annularly averaged second-order spectral scaling without claiming intermittency itself (Divitiis, 2012, Colombo et al., 2022, Hess-Childs et al., 31 Jul 2025, Rowan, 2 Dec 2025).

1. Classical Obukhov–Corrsin framework

In incompressible homogeneous isotropic turbulence with a passive scalar, the basic object is the normalized two-point scalar correlation

$$f_\theta(r,t)=\frac{\langle \vartheta \vartheta'\rangle}{\theta^2},$$

whose evolution is governed by the Corrsin equation

$$\theta^2 \frac{\partial f_\theta}{\partial t} + f_\theta \frac{d \theta^2}{d t} -G -2 \chi \theta^2 \left( \frac{\partial^2 f_\theta}{\partial r^2} + \frac{2}{r} \frac{\partial f_\theta}{\partial r} \right) = 0,$$

with transfer term

$$G = - \frac{\partial}{\partial r_k} \langle \vartheta \vartheta' (u_k'-u_k) \rangle.$$

In the same setting, the scalar spectrum is defined by a sine transform of $f_\theta$, and the inertial-convective regime is identified with

$$\Theta(\kappa)\sim C_\theta \epsilon^{-1/3}\epsilon_\theta\,\kappa^{-5/3}.$$

The same closed theory also recovers the Batchelor viscous-convective regime $\Theta(\kappa)\sim \kappa^{-1}$ at high $Pr$, and a steeper low-$Pr$ range with exponent $S_p(r)\propto r^{p/3}$0 satisfying $S_p(r)\propto r^{p/3}$1 (Divitiis, 2012).

This baseline already implies that the Obukhov–Corrsin law is not a statement about the entire spectrum. It is a finite-window inertial-convective law embedded in a broader regime structure controlled by advection, stretching, viscosity, and diffusivity. A related Lagrangian-Liouville closure program formulates the scalar dynamics in terms of the pair-separation vector $S_p(r)\propto r^{p/3}$2, assumes statistical independence of $S_p(r)\propto r^{p/3}$3 from $S_p(r)\propto r^{p/3}$4 in fully developed turbulence, and derives a non-eddy-viscosity closure

$S_p(r)\propto r^{p/3}$5

within a framework where the temperature correlation reaches a developed form in finite time (Divitiis, 2015).

2. Intermittency as anomalous higher-order scalar statistics

In the phenomenological literature, an intermittent Obukhov–Corrsin regime usually means that second-order statistics remain compatible with Obukhov–Corrsin, while higher-order structure functions depart from the dimensional law

$S_p(r)\propto r^{p/3}$6

Within the finite-scale Lyapunov analysis of temperature fluctuations in homogeneous isotropic turbulence, intermittency does not primarily appear as a modified $S_p(r)\propto r^{p/3}$7 second-order exponent. Instead, it enters through the longitudinal temperature derivative. The normalized derivative is represented as

$S_p(r)\propto r^{p/3}$8

and the resulting PDF is non-Gaussian, has null skewness, and develops broader tails as $S_p(r)\propto r^{p/3}$9 increases. The corresponding flatness and hyperflatness rise with $$f_\theta(r,t)=\frac{\langle \vartheta \vartheta'\rangle}{\theta^2},$$0; in particular, Gaussian values $$f_\theta(r,t)=\frac{\langle \vartheta \vartheta'\rangle}{\theta^2},$$1 and $$f_\theta(r,t)=\frac{\langle \vartheta \vartheta'\rangle}{\theta^2},$$2 are recovered at $$f_\theta(r,t)=\frac{\langle \vartheta \vartheta'\rangle}{\theta^2},$$3, while $$f_\theta(r,t)=\frac{\langle \vartheta \vartheta'\rangle}{\theta^2},$$4 and $$f_\theta(r,t)=\frac{\langle \vartheta \vartheta'\rangle}{\theta^2},$$5 as $$f_\theta(r,t)=\frac{\langle \vartheta \vartheta'\rangle}{\theta^2},$$6 (Divitiis, 2012).

A non-dynamical but explicit statistical analogue appears in the analysis of Ogata Kōrin’s Red and White Plum Blossoms, where the gray-scale luminance field

$$f_\theta(r,t)=\frac{\langle \vartheta \vartheta'\rangle}{\theta^2},$$7

is treated as a passive-scalar surrogate. Using a Gaussian-windowed Fourier transform in two circular river regions, the measured spectrum shows an intermediate-range power law close to $$f_\theta(r,t)=\frac{\langle \vartheta \vartheta'\rangle}{\theta^2},$$8. The second-order structure function is roughly consistent with $$f_\theta(r,t)=\frac{\langle \vartheta \vartheta'\rangle}{\theta^2},$$9, whereas in the scaling range $$\theta^2 \frac{\partial f_\theta}{\partial t} + f_\theta \frac{d \theta^2}{d t} -G -2 \chi \theta^2 \left( \frac{\partial^2 f_\theta}{\partial r^2} + \frac{2}{r} \frac{\partial f_\theta}{\partial r} \right) = 0,$$0 to $$\theta^2 \frac{\partial f_\theta}{\partial t} + f_\theta \frac{d \theta^2}{d t} -G -2 \chi \theta^2 \left( \frac{\partial^2 f_\theta}{\partial r^2} + \frac{2}{r} \frac{\partial f_\theta}{\partial r} \right) = 0,$$1 pixels the fourth- and sixth-order exponents are reported to be approximately $$\theta^2 \frac{\partial f_\theta}{\partial t} + f_\theta \frac{d \theta^2}{d t} -G -2 \chi \theta^2 \left( \frac{\partial^2 f_\theta}{\partial r^2} + \frac{2}{r} \frac{\partial f_\theta}{\partial r} \right) = 0,$$2 and $$\theta^2 \frac{\partial f_\theta}{\partial t} + f_\theta \frac{d \theta^2}{d t} -G -2 \chi \theta^2 \left( \frac{\partial^2 f_\theta}{\partial r^2} + \frac{2}{r} \frac{\partial f_\theta}{\partial r} \right) = 0,$$3, rather than the non-intermittent values $$\theta^2 \frac{\partial f_\theta}{\partial t} + f_\theta \frac{d \theta^2}{d t} -G -2 \chi \theta^2 \left( \frac{\partial^2 f_\theta}{\partial r^2} + \frac{2}{r} \frac{\partial f_\theta}{\partial r} \right) = 0,$$4 and $$\theta^2 \frac{\partial f_\theta}{\partial t} + f_\theta \frac{d \theta^2}{d t} -G -2 \chi \theta^2 \left( \frac{\partial^2 f_\theta}{\partial r^2} + \frac{2}{r} \frac{\partial f_\theta}{\partial r} \right) = 0,$$5. The same work reports scale-dependent flatness and hyperflatness, non-Gaussian fat-tailed increment PDFs with no perfect collapse across scales, and an $$\theta^2 \frac{\partial f_\theta}{\partial t} + f_\theta \frac{d \theta^2}{d t} -G -2 \chi \theta^2 \left( \frac{\partial^2 f_\theta}{\partial r^2} + \frac{2}{r} \frac{\partial f_\theta}{\partial r} \right) = 0,$$6 decomposition indicating that anisotropy is not the main source of the deviations. It also emphasizes that the intermittency claim is sensitive to image quality and less robust in lower-resolution or spikier versions of the image (Matsumoto, 17 Jan 2025).

3. Supercritical and codimension-sensitive rigorous formulations

In rigorous PDE theory, the intermittent Obukhov–Corrsin regime is formulated as a supercritical transport regime for rough divergence-free velocities. One precise realization is the condition

$$\theta^2 \frac{\partial f_\theta}{\partial t} + f_\theta \frac{d \theta^2}{d t} -G -2 \chi \theta^2 \left( \frac{\partial^2 f_\theta}{\partial r^2} + \frac{2}{r} \frac{\partial f_\theta}{\partial r} \right) = 0,$$7

with velocity regularity $$\theta^2 \frac{\partial f_\theta}{\partial t} + f_\theta \frac{d \theta^2}{d t} -G -2 \chi \theta^2 \left( \frac{\partial^2 f_\theta}{\partial r^2} + \frac{2}{r} \frac{\partial f_\theta}{\partial r} \right) = 0,$$8 and scalar regularity controlled uniformly in diffusivity in $$\theta^2 \frac{\partial f_\theta}{\partial t} + f_\theta \frac{d \theta^2}{d t} -G -2 \chi \theta^2 \left( \frac{\partial^2 f_\theta}{\partial r^2} + \frac{2}{r} \frac{\partial f_\theta}{\partial r} \right) = 0,$$9. For this supercritical range, there exist bounded initial data such that the unique bounded solutions of

$$G = - \frac{\partial}{\partial r_k} \langle \vartheta \vartheta' (u_k'-u_k) \rangle.$$0

satisfy

$$G = - \frac{\partial}{\partial r_k} \langle \vartheta \vartheta' (u_k'-u_k) \rangle.$$1

while still exhibiting anomalous dissipation,

$$G = - \frac{\partial}{\partial r_k} \langle \vartheta \vartheta' (u_k'-u_k) \rangle.$$2

The proof uses the stochastic flow

$$G = - \frac{\partial}{\partial r_k} \langle \vartheta \vartheta' (u_k'-u_k) \rangle.$$3

and the Feynman–Kac representation

$$G = - \frac{\partial}{\partial r_k} \langle \vartheta \vartheta' (u_k'-u_k) \rangle.$$4

The same work also proves lack of selection by vanishing diffusivity and by convolution mollification: for every $$G = - \frac{\partial}{\partial r_k} \langle \vartheta \vartheta' (u_k'-u_k) \rangle.$$5, there are $$G = - \frac{\partial}{\partial r_k} \langle \vartheta \vartheta' (u_k'-u_k) \rangle.$$6 divergence-free velocities for which distinct inviscid weak limit points coexist, one conserving $$G = - \frac{\partial}{\partial r_k} \langle \vartheta \vartheta' (u_k'-u_k) \rangle.$$7 and another dissipating it (Colombo et al., 2022).

A more explicitly intermittent formulation refines the regularity threshold by the dimension of the dissipation support. If

$$G = - \frac{\partial}{\partial r_k} \langle \vartheta \vartheta' (u_k'-u_k) \rangle.$$8

and $$G = - \frac{\partial}{\partial r_k} \langle \vartheta \vartheta' (u_k'-u_k) \rangle.$$9 solves

$f_\theta$0

then the dissipation distribution

$f_\theta$1

is assumed to be a Radon measure supported on a set $f_\theta$2 of Hausdorff dimension $f_\theta$3. If $f_\theta$4, then one must have

$f_\theta$5

This is the codimension-sensitive intermittent Obukhov–Corrsin constraint. Its novelty is specific to $f_\theta$6, where dissipation lies on a positive-codimension set. The same paper constructs an explicit divergence-free field $f_\theta$7 whose dissipation measure is supported on countably many timeslices, and proves sharpness at the endpoint $f_\theta$8, $f_\theta$9: whenever

$$\Theta(\kappa)\sim C_\theta \epsilon^{-1/3}\epsilon_\theta\,\kappa^{-5/3}.$$0

there exist $$\Theta(\kappa)\sim C_\theta \epsilon^{-1/3}\epsilon_\theta\,\kappa^{-5/3}.$$1 and $$\Theta(\kappa)\sim C_\theta \epsilon^{-1/3}\epsilon_\theta\,\kappa^{-5/3}.$$2 with non-trivial negative Radon dissipation supported on

$$\Theta(\kappa)\sim C_\theta \epsilon^{-1/3}\epsilon_\theta\,\kappa^{-5/3}.$$3

The same construction is also tied to asymptotic total dissipation, enhanced dissipation, Richardson dispersion, anomalous regularization, and spatial intermittency (Hess-Childs et al., 31 Jul 2025).

4. Anomalous regularization and annular spectral laws

A distinct rigorous direction proves the Obukhov–Corrsin spectrum in rough passive-scalar models without claiming an intermittent regime in the higher-moment sense. For passive scalar advection by a white-in-time Kraichnan-type transport noise with spatial Hölder regularity $$\Theta(\kappa)\sim C_\theta \epsilon^{-1/3}\epsilon_\theta\,\kappa^{-5/3}.$$4, $$\Theta(\kappa)\sim C_\theta \epsilon^{-1/3}\epsilon_\theta\,\kappa^{-5/3}.$$5, the classical rough-flow prediction

$$\Theta(\kappa)\sim C_\theta \epsilon^{-1/3}\epsilon_\theta\,\kappa^{-5/3}.$$6

is shifted by the white-in-time scaling to

$$\Theta(\kappa)\sim C_\theta \epsilon^{-1/3}\epsilon_\theta\,\kappa^{-5/3}.$$7

corresponding to essentially $$\Theta(\kappa)\sim C_\theta \epsilon^{-1/3}\epsilon_\theta\,\kappa^{-5/3}.$$8 regularity in equilibrium. The rigorous theorem does not establish a pointwise lower bound for each Fourier mode. It proves the predicted scaling only after summing over geometric annuli, and only up to logarithmic factors and slightly super-geometric shell widths. In the main Kraichnan examples, $$\Theta(\kappa)\sim C_\theta \epsilon^{-1/3}\epsilon_\theta\,\kappa^{-5/3}.$$9, and the total shell mass is bounded above and below by terms of order $\Theta(\kappa)\sim \kappa^{-1}$0 times logarithmic corrections (Rowan, 2 Dec 2025).

The mechanism is anomalous regularization. For the free-decay problem

$\Theta(\kappa)\sim \kappa^{-1}$1

one has the uniform decay estimate

$\Theta(\kappa)\sim \kappa^{-1}$2

together with the time-integrated gain of roughly $\Theta(\kappa)\sim \kappa^{-1}$3 derivatives,

$\Theta(\kappa)\sim \kappa^{-1}$4

This regularization is transferred to the invariant measure of the forced problem through

$\Theta(\kappa)\sim \kappa^{-1}$5

The analytic core is a Fourier-space $\Theta(\kappa)\sim \kappa^{-1}$6 energy identity and a weighted lattice Poincaré inequality. Writing $\Theta(\kappa)\sim \kappa^{-1}$7, the energy identity takes the form

$\Theta(\kappa)\sim \kappa^{-1}$8

and the weighted lattice estimate gives control of weighted Fourier mass by directional lattice differences. The admissible Kraichnan-type models include an isotropic choice

$\Theta(\kappa)\sim \kappa^{-1}$9

and an anisotropic shear model supported on coordinate axes,

$Pr$0

subject to structural conditions on

$Pr$1

that exclude purely one-dimensional shears and ensure sufficient directional mixing. The paper explicitly states that intermittency corrections are expected only for moments $Pr$2 and are not its focus; its contribution is a rigorous verification of the Obukhov–Corrsin regime with annular averaging and logarithmic losses, not a derivation of an intermittent spectrum (Rowan, 2 Dec 2025).

5. Nonstationary and buoyancy-dominated generalizations

The Obukhov–Corrsin framework has also been generalized to nonstationary settings in which the scalar is globally active but effectively passive at inertial scales. In a turbulent puff evolving under the Oberbeck–Boussinesq equations, the strong-buoyancy regime occurs for

$Pr$3

and the bulk scales obey

$Pr$4

$Pr$5

The underlying assumption is an adiabaticity hypothesis: small-scale fluctuations relax rapidly to the slowly evolving large-scale background. Under this hypothesis, the inertial range remains K41/OC-like, with time-dependent fluxes inherited from the puff dynamics (Mazzino et al., 2021).

The generalized Obukhov–Corrsin prediction for temperature increments in the buoyancy-dominated puff is

$Pr$6

while the viscous-range scalar increments satisfy

$Pr$7

Intermittency is then incorporated through anomalous exponents imported from stationary turbulence, specifically

$Pr$8

The DNS evidence reported in that work includes collapse of second-, fourth-, and sixth-order structure functions after applying the predicted temporal prefactors, ESS slopes consistent with the intermittency-corrected theory, and clear disagreement between the data and the no-intermittency lines. The same paper explicitly contrasts this with a Bolgiano-type scenario and states that such a scale-by-scale buoyancy balance is not observed in the simulations. The resulting interpretation is that buoyancy modifies the large-scale amplitudes and time dependence, but the inertial-range scalar cascade remains in a generalized intermittent Obukhov–Corrsin class (Mazzino et al., 2021).

6. Scope, ambiguities, and recurrent misconceptions

Across the literature, the phrase “intermittent Obukhov–Corrsin regime” is used in several related but non-identical senses. In homogeneous isotropic turbulence and in statistical analogues, it usually means that the second-order Obukhov–Corrsin law remains approximately valid while higher-order structure functions, flatness, hyperflatness, or increment PDFs display anomalous scaling and non-Gaussianity (Divitiis, 2012, Matsumoto, 17 Jan 2025). In rigorous transport theory, it refers instead to supercritical regularity regimes, codimension-localized dissipation measures, and integrability-dependent constraints on the admissible pair $Pr$9 (Colombo et al., 2022, Hess-Childs et al., 31 Jul 2025).

A recurrent misconception is to identify every rigorous derivation of an Obukhov–Corrsin spectrum with an intermittency result. The Kraichnan-type annular spectrum theorem does not define or prove an intermittent Obukhov–Corrsin regime in the turbulence-physics sense; it derives the expected second-order spectrum on geometric annuli, modulo logarithmic losses, from anomalous regularization. Its own discussion states that intermittency corrections are expected only for moments $Pr$0 and are not being addressed (Rowan, 2 Dec 2025).

Another ambiguity concerns dynamical versus statistical evidence. Image-based analyses may show spectra, structure functions, and anisotropy tests that are statistically consistent with passive-scalar intermittency, but such evidence is not a dynamical derivation from transport equations. Conversely, closure theories based on Lagrangian separation or finite-scale Lyapunov analysis can supply dynamical mechanisms for cascade and finite-time development without, by themselves, fixing anomalous multiscaling exponents. The intermittent Obukhov–Corrsin regime is therefore best understood as a technically structured extension of the Obukhov–Corrsin framework, not as a single universally standardized model.