Testing 2D temperature models in Bayesian retrievals of atmospheric properties from hot Jupiter phase curves

Spectroscopic phase curves of transiting hot Jupiters are spectral measurements at multiple orbital phases, giving a set of disc-averaged spectra that probe multiple hemispheres. By fitting model phase curves to observations, we can constrain the atmospheric properties of hot Jupiters such as molecular abundance, aerosol distribution and thermal structure, which offer insights into their dynamics, chemistry, and formation. In this work, we propose a novel 2D temperature scheme consisting of a dayside and a nightside to retrieve information from near-infrared phase curves, and apply the scheme to phase curves of WASP-43b observed by HST/WFC3 and Spitzer/IRAC. In our scheme, temperature is constant on isobars on the nightside and varies with cos$^n$(longitude/$\epsilon$) on isobars on the dayside, where $n$ and $\epsilon$ are free parameters. We fit all orbital phases simultaneously using the radiative transfer package NEMESISPY coupled to a Bayesian inference code. We first validate the performance of our retrieval scheme with synthetic phase curves generated from a GCM, and find our 2D scheme can accurately retrieve the latitudinally-averaged thermal structure and constrain the abundance of H$_2$O and CH$_4$. We then apply our 2D scheme to the observed phase curves of WASP-43b and find: (1) the dayside temperature-pressure profiles do not vary strongly with longitude and are non-inverted; (2) the retrieved nightside temperatures are extremely low, suggesting significant nightside cloud coverage; (3) the H$_2$O volume mixing ratio is constrained to $5.6\times10^{-5}$--$4.0\times10^{-4}$, and we retrieve an upper bound for CH$_4$ at $\sim$10$^{-6}$.