A Physically Driven Parameterisation of Multidimensional Atmospheres: Application to the JWST Phase Curve of WASP-121b
Understanding the multidimensional structure of strongly irradiated exoplanets is essential for interpreting their atmospheric dynamics, chemistry and energy transport, yet current analyses remain limited by the difficulty of extracting reliable phase-resolved spectra and by the lack of physically interpretable parameterisations for retrievals. We combine a data-driven eclipse-normalisation method with an analytical three-dimensional temperature parameterisation derived from radiative, advective and diffusive energy balance and controlled by a few characteristic timescales. Applied to JWST/NIRSpec G395H observations of WASP-121b, the method yields spectra consistent with conventional phase-curve fitting, while the parameterisation reproduces the large-scale thermal structures predicted by general circulation models. The preferred retrieval reveals a pronounced day--night contrast, a dayside thermal inversion extending to both limbs, an inversion over part of the nightside, and limb temperatures differing by several hundred kelvin. Dynamical transport strengthens with pressure, and the hotspot offset increases from $\sim4^\circ$ to $\sim9^\circ$ across the pressures probed by G395H. The confined dayside hot region and the small, pressure-dependent offsets lie closer to the $\sim$3~G GCM than to its non-magnetic counterpart, although Rayleigh drag cannot be excluded. The spectra also favour distinct dayside and nightside chemical states, with more nightside CH$_4$ than the cooler temperatures alone can explain, pointing to disequilibrium chemistry. The retrieved thermal structure further implies an inhomogeneous cloud distribution, with condensation favoured on the nightside and cooler morning limb. The framework provides a computationally efficient, physically interpretable path from spectroscopic phase curves to multidimensional atmospheric structure.