On a system of equations arising in meteorology: Well-posedness and data assimilation

Data assimilation plays a crucial role in modern weather prediction, providing a systematic way to incorporate observational data into complex dynamical models. The paper addresses continuous data assimilation for a model arising as a singular limit of the three-dimensional compressible Navier-Stokes-Fourier system with rotation driven by temperature gradient. The limit system preserves the essential physical mechanisms of the original model, while exhibiting a reduced, effectively two-and-a-half-dimensional structure. This simplified framework allows for a rigorous analytical study of the data assimilation process while maintaining a direct physical connection to the full compressible model. We establish well posedness of global-in-time solutions and a compact trajectory attractor, followed by the stability and convergence results for the nudging scheme applied to the limiting system. Finally, we demonstrate how these results can be combined with a relative entropy argument to extend the assimilation framework to the full three-dimensional compressible setting, thereby establishing a rigorous connection between the reduced and physically complete models.

💡 Research Summary

The paper investigates continuous data assimilation for a reduced atmospheric model that emerges as a singular limit of the three‑dimensional compressible Navier‑Stokes‑Fourier system with rotation driven by a temperature gradient. By scaling the Mach and Rossby numbers with a small parameter ε, the authors rigorously derive a “two‑and‑a‑half‑dimensional” rotating Oberbeck‑Boussinesq (rOB) system in which the vertical velocity component disappears and only horizontal velocity and temperature fluctuations remain. This reduced system retains the essential physics of the full model while being mathematically more tractable.

The first major contribution is the proof of global‑in‑time well‑posedness for strong solutions of the rOB system. Assuming sufficiently regular initial data (u₀∈W^{2‑2/p,p}(Ω_h) and Θ₀∈W^{2‑2/p,p}(Ω)∩W^{2,2}(Ω) with p≥2) and smooth boundary temperature, the authors establish existence, uniqueness, and maximal regularity of solutions (Theorem 2.1). The proof relies on a clever variable transformation that converts the non‑local temperature boundary condition into a more manageable form, followed by energy estimates, Sobolev embeddings, and a Galerkin approximation scheme.

Next, the authors analyze the long‑time behavior of the rOB dynamics. They show that the system is dissipative in the sense of Levinson: there exists a bounded absorbing set in the phase space and consequently a compact trajectory attractor. This result confirms that, despite the infinite‑dimensional nature of the PDE, the asymptotic dynamics are governed by a finite number of degrees of freedom, a property that underlies many data‑assimilation strategies.

The core of the paper is the development of a continuous data‑assimilation algorithm based on the Azouani‑Olson‑Titi (AOT) nudging framework. An interpolant operator I_h, representing coarse spatial observations (e.g., low‑frequency Fourier modes or coarse mesh values), is introduced together with a relaxation parameter μ>0. The nudged system adds the feedback term μ(I_h(u)−I_h(v)) to the momentum and temperature equations, where u denotes the unknown reference solution and v the assimilated approximation. Theorem 4.1 proves that, provided μ and the observation resolution h are sufficiently small, the nudged solution converges exponentially fast to the true solution in the L² norm. Remarkably, the analysis shows that observing only the temperature field (or its spatial averages) is enough to recover the full horizontal velocity field, reflecting the strong coupling between temperature and velocity in the rOB equations.

Finally, the authors bridge the reduced model back to the original three‑dimensional compressible system using a relative entropy (or modulated energy) method. For well‑prepared initial data—i.e., data whose ε‑scaled differences are of order O(ε)—the relative entropy between the full solution and the rOB solution decays like C ε² e^{−λt}. Consequently, the error introduced by performing data assimilation on the reduced model remains controlled on any finite time interval, and the assimilation results can be transferred to the full compressible system. This provides a rigorous justification for using the cheaper rOB model in operational weather forecasting while still guaranteeing accuracy for the underlying physical model.

Overall, the paper delivers a comprehensive theoretical framework that connects (i) singular‑limit dimension reduction, (ii) global well‑posedness and attractor theory, (iii) rigorous continuous data assimilation with exponential convergence, and (iv) relative‑entropy based error control linking reduced and full models. The results suggest that substantial computational savings can be achieved in atmospheric data assimilation without sacrificing fidelity, opening avenues for more efficient and reliable weather prediction.