Two-layer Quasi-geostrophic Flow Research Articles

Long-term stability and generalization of observationally-constrained stochastic data-driven models for geophysical turbulence

Abstract Recent years have seen a surge in interest in building deep learning-based fully data-driven models for weather prediction. Such deep learning models, if trained on observations can mitigate certain biases in current state-of-the-art weather models, some of which stem from inaccurate representation of subgrid-scale processes. However, these data-driven models, being over-parameterized, require a lot of training data which may not be available from reanalysis (observational data) products. Moreover, an accurate, noise-free, initial condition to start forecasting with a data-driven weather model is not available in realistic scenarios. Finally, deterministic data-driven forecasting models suffer from issues with long-term stability and unphysical climate drift, which makes these data-driven models unsuitable for computing climate statistics. Given these challenges, previous studies have tried to pre-train deep learning-based weather forecasting models on a large amount of imperfect long-term climate model simulations and then re-train them on available observational data. In this article, we propose a convolutional variational autoencoder (VAE)-based stochastic data-driven model that is pre-trained on an imperfect climate model simulation from a two-layer quasi-geostrophic flow and re-trained, using transfer learning, on a small number of noisy observations from a perfect simulation. This re-trained model then performs stochastic forecasting with a noisy initial condition sampled from the perfect simulation. We show that our ensemble-based stochastic data-driven model outperforms a baseline deterministic encoder–decoder-based convolutional model in terms of short-term skills, while remaining stable for long-term climate simulations yielding accurate climatology.

Solitary waves of two-layer quasi-geostrophic flow and analytical solutions with scalar nonlinearity

In this paper we investigate the instability of two-layer quasi-geostrophic model with scalar nonlinearity, which is a geophysical fluid dynamics model, and give a nonlinear stability criterion. Furthermore, the analytical solutions of the two-layer quasi-geostrophic model are obtained via the Jacobi elliptic periodic function transform method and symbolic computation. It is efficient in solving a range classes of nonlinear partial differential equations for this method. The result shows that the structure of Rossby waves is influenced by many factors such as base flow and scalar nonlinearity.

Hetonic quartets in a two-layer quasi-geostrophic flow: V-states and stability

We investigate families of finite core vortex quartets in mutual equilibrium in a two-layer quasi-geostrophic flow. The finite core solutions stem from known solutions for discrete (singular) vortex quartets. Two vortices lie in the top layer and two vortices lie in the bottom layer. Two vortices have a positive potential vorticity anomaly, while the two others have negative potential vorticity anomaly. The vortex configurations are therefore related to the baroclinic dipoles known in the literature as hetons. Two main branches of solutions exist depending on the arrangement of the vortices: the translating zigzag-shaped hetonic quartets and the rotating zigzag-shaped hetonic quartets. By addressing their linear stability, we show that while the rotating quartets can be unstable over a large range of the parameter space, most translating quartets are stable. This has implications on the longevity of such vortex equilibria in the oceans.

Geostrophic tripolar vortices in a two-layer fluid: Linear stability and nonlinear evolution of equilibria

We investigate equilibrium solutions for tripolar vortices in a two-layer quasi-geostrophic flow. Two of the vortices are like-signed and lie in one layer. An opposite-signed vortex lies in the other layer. The families of equilibria can be spanned by the distance (called separation) between the two like-signed vortices. Two equilibrium configurations are possible when the opposite-signed vortex lies between the two other vortices. In the first configuration (called ordinary roundabout), the opposite signed vortex is equidistant to the two other vortices. In the second configuration (eccentric roundabouts), the distances are unequal. We determine the equilibria numerically and describe their characteristics for various internal deformation radii. The two branches of equilibria can co-exist and intersect for small deformation radii. Then, the eccentric roundabouts are stable while unstable ordinary roundabouts can be found. Indeed, ordinary roundabouts exist at smaller separations than eccentric roundabouts do, thus inducing stronger vortex interactions. However, for larger deformation radii, eccentric roundabouts can also be unstable. Then, the two branches of equilibria do not cross. The branch of eccentric roundabouts only exists for large separations. Near the end of the branch of eccentric roundabouts (at the smallest separation), one of the like-signed vortices exhibits a sharp inner corner where instabilities can be triggered. Finally, we investigate the nonlinear evolution of a few selected cases of tripoles.

Ribbon turbulence

We investigate the non-linear equilibration of a two-layer quasi-geostrophic flow in a channel with an initial eastward baroclinically unstable jet in the upper layer, paying particular attention to the role of bottom friction. In the limit of low bottom friction, classical theory of geostrophic turbulence predicts an inverse cascade of kinetic energy in the horizontal with condensation at the domain scale and barotropization in the vertical. By contrast, in the limit of large bottom friction, the flow is dominated by ribbons of high kinetic energy in the upper layer. These ribbons correspond to meandering jets separating regions of homogenized potential vorticity. We interpret these results by taking advantage of the peculiar conservation laws satisfied by this system: the dynamics can be recast in such a way that the initial eastward jet in the upper layer appears as an initial source of potential vorticity levels in the upper layer. The initial baroclinic instability leads to a turbulent flow that stirs this potential vorticity field while conserving the global distribution of potential vorticity levels. Statistical mechanical theory of the 112 layer quasi-geostrophic model predicts the formation of two regions of homogenized potential vorticity separated by a minimal interface. We explain that cascade phenomenology leads to the same result. We then show that the dynamics of the ribbons results from a competition between a tendency to reach the equilibrium state and baroclinic instability that induces meanders of the interface. These meanders intermittently break and induce potential vorticity mixing, but the interface remains sharp throughout the flow evolution. We show that for some parameter regimes, the ribbons act as a mixing barrier which prevents relaxation toward equilibrium, favouring the emergence of multiple zonal (eastward) jets.

PROBABILISTIC DYNAMICS OF TWO-LAYER GEOPHYSICAL FLOWS

The two-layer quasigeostrophic flow model is an intermediate system between the single-layer two-dimensional barotropic flow model and the continuously stratified three-dimensional baroclinic flow model. This model is widely used to investigate basic mechanisms in geophysical flows, such as baroclinic effects, the Gulf stream and subtropical gyres. We consider the two-layer quasigeostrophic flow model under stochastic wind forcing on the top layer. The fluctuating part of the wind forcing is modeled as the generalized time derivative of a Wiener process. We first transform this stochastic two-layer fluid system into a coupled system of random partial differential equations. Then we prove that the stochastic two-layer fluid system has finite sets of asymptotically determining functionals (such as determining modes and determining nodes) in probability. Furthermore, we show that the asymptotic probabilistic dynamics of this system depends only on the top fluid layer. Namely, in the probability sense and asymptotically, the dynamics of the two-layer quasigeostrophic fluid system is determined by the top layer. In other words, the bottom layer is slaved by the top layer. This conclusion is true provided that the Wiener process and the fluid parameters satisfy a certain condition. In particular, this latter condition is satisfied when the trace of the covariance operator of the Wiener process is small enough and the Ekman constant r is sufficiently large.

Point vortex dynamics for coupled surface/interior QG and propagating heton clusters in models for ocean convection

The dynamic behavior of baroclinic point vortices in two-layer quasi-geostrophic flow provides a compact model for studying the transport of heat in a variety of geophysical flows including recent heton models for open ocean convection as a response to spatially localized intense surface cooling. In such heton models, the exchange of heat with the region external to the compact cooling region reaches a statistical equilibrium through the propagation of tilted heton clusters. Such tilted heton clusters are aggregates of cyclonic vortices in the upper layer and anti-cyclonic vortices in the lower layer which collectively propagate almost as an elementary tilted heton pair even though the individual vortices undergo shifts in their relative locations. One main result in this paper is a mathematical theorem demonstrating the existence of large families of long-lived propagating heton clusters for the two-layer model in a fashion compatible to a remarkable degree with the earlier numerical simulations. Two-layer quasi-geostrophic flow is an idealization of coupled surface/interior quasi-geostrophic flow. The second family of results in this paper involves the systematic development of Hamiltonian point vortex dynamics for coupled surface/interior QG with an emphasis on propagating solutions that transport heat. These are novel vortex systems of mixed species where surface heat particles interact with quasi-geostrophic point vortices. The variety of elementary two-vortex exact solutions that transport heat include two surface heat particles of opposite strength, tilted pairs of a surface heat particle coupled to an interior vortex of opposite strength and two interior tilted vortices of opposite strength at different depths. The propagation speeds of the tilted elementary hetons in the coupled surface/interior QG model are compared and contrasted with those in the simpler two-layer heton models. Finally, mathematical theorems are presented for the existence of large families of propagating long-lived tilted heton clusters for point vortex solutions in coupled surface/interior QG flow.

Tearing of an aligned vortex by a current difference in two-layer quasi-geostrophic flow

A study of two-layer quasi-geostrophic vortex flow is performed to determine the effect of a current difference between the layers on a vortex initially extending through both layers. In particular, the conditions under which the vortex can resist being torn by the current difference are examined. The vortex evolution is determined using versions of the contour dynamics and discrete vortex methods which are modified for two-layer quasi-geostrophic flows. The vortex response is found to depend upon the way in which the current difference between the layers is maintained. In the first set of flows studied, the current difference is generated by a (stronger) third vortex in the upper layer located at a large distance from the (weaker) vortex under study.

Two-Layer Quasigeostrophic Flow over Finite Isolated Topography

Abstract A quasigeostrophic two-layer model of flow over finite topography is developed. The topography is a right circular cylinder that extends through the lower layer and an order Rossby number amount into the upper layer (finite topography model). Thus, each layer depth remains constant to first order, and the quasigeostrophic approximation can be applied consistently. The model solutions axe compared to those found when the total topographic height is order Rossby number (small topography model). The steady solution for the finite topography model consists of two parts: one similar to the small topography solution and forced by the anticyclonic potential vorticity anomaly over the topography and the other similar to the solution of potential flow around a cylinder and forced by the matching conditions on the edge of the topography. The finite topography model breaks down when the interface goes above the topography. This occurs most easily when the stratification is weak. Closed streamlines occur mor...

Multiple states for quasi-geostrophic channel flows

Abstract We consider nonlinear baroclinic instabilities of two-layer quasi-geostrophic flow in a rectilinear channel. The full potential vorticity equations are shown to possess a countable infinity of invariant wavenumber sets. Each set is composed of a particular pattern in wavenumber space in which many Fourier modes have zero energy. Solutions with initial conditions confined to a particular wavenumber pattern will remain forever in that pattern. There is also a general asymmetric state with non-zero energy in all wavenumbers. The final state of a long-time evolution calculation depends on initial conditions and internal stability.

Dynamical Criterion for a Marginally Unstable, Quasi-linear Behavior in a Two-Layer Model

A two-layer quasi-geostrophic flow forced by meridional variations in heating can be in regimes ranging from radiative equilibrium to forced geostrophic turbulence. Between these extremes is a regime where the time-mean (zonal) flow is marginally unstable. Using scaling arguments, it is concluded that such a marginally unstable state should occur when a certain parameter, measuring the strength of wave-wave interactions relative to the beta effect and advection by the thermal wind, is small. Numerical simulations support this proposal. A transition from the marginally unstable regime to a more nonlinear regime is then examined through numerical simulations with different radiative forcings. It is found that transition is not caused by secondary instability of waves in the marginally unstable regime. Instead, the time-mean flow can support a number of marginally unstable normal modes. These normal modes interact with each other, and if they are of sufficient amplitude, the flow enters a more nonlinear regime.

On the Predictability of Quasi-Geostrophic Flow: The Effects of Beta and Baroclinicity

Abstract The equilibrium statistics and predictability properties of one- and two-layer quasi-geostrophic flow are examined with the aid of a numerical model. The effect of beta in one-layer flow is to slow the transfer of energy into larger scales and to increase the predictability. In two-layer flow, when beta is zero, energy caters the system via baroclinic instability of the mean Row at very large scales and most energy transfer is confined to low wavenumbers. When beta is non-zero, energy enters at higher wavenumber (in baroclinic modes mainly) before cascading preferentially to lower wavenumber zonal barotropic modes. The predictability of two-layer flow is not significantly altered by beta, because beta increases the range of wavenumber over which significant nonlinear energy transfer occurs. The predictability times of the long waves are found to be always larger than those of the short waves, even when the initial error is spread evenly acres wavenumbers. Reducing the mean baroclinicity increases...

The Effect of a Bottom Boundary Layer on the Stability of a Baroclinic Zonal Current

Abstract The effect of a depth-limited bottom boundary layer on the stability of a baroclinic zonal current is investigated. The model is of a two-layer quasi-geostrophic flow. Limiting the height ...