The GAPS Programme at TNG. LXIX.The Dayside of WASP-76b revealed by GIANO-B, HARPS-N and ESPRESSO: Evidence for Three-Dimensional Atmospheric Effects

TL;DR

This work targets the dayside emission of the ultra-hot Jupiter WASP-76 b by combining high-resolution spectra from GIARPS (GIANO-B and HARPS-N) and ESPRESSO to detect CO and Fe I and to probe three-dimensional atmospheric structure. The authors implement a homogeneous data-analysis pipeline, including telluric/stars removal and cross-correlation with both 1D emission templates and 3D GCM-informed templates, to interpret phase- and Doppler-related signatures. They detect CO in GIANO-B (S/N ≈ 10.4) and Fe I in HARPS-N (S/N ≈ 3.5) and ESPRESSO (S/N ≈ 6.2), with a marginal Fe I signal in one GIANO-B night, and find that 3D, dynamic Global Circulation Models best explain the ESPRESSO data, aligning the peak with the expected $K_p$ and $V_{ ext{rest}}$ within a couple of sigma. The results support strong three-dimensional atmospheric effects and phase-dependent Doppler shifts in UHJs and demonstrate the value of integrating multi-band, high-resolution spectroscopy with GCM-based interpretation under the GAPS program.

Abstract

The study of the atmosphere of ultra-hot Jupiters (UHJs) with equilibrium temperature $\geq$2000 K provides valuable insights into atmospheric physics under such extreme conditions. We aim to characterise the dayside thermal spectrum of the UHJ WASP-76b and investigate its properties. We analysed data gathered with three high-resolution spectrographs, specifically two nights with simultaneous observations of HARPS-N and GIANO-B, and four nights of publicly available ESPRESSO optical spectra. We observed the planet's dayside covering orbital phases between quadratures (0.25 < $φ$ < 0.75). We performed a homogeneous analysis of the GIANO-B, HARPS-N and ESPRESSO data and co-added the signal of thousands of planetary lines through cross-correlation with simulated spectra of the planetary atmosphere. We report the detection of CO in the dayside atmosphere of WASP-76b with a signal-to-noise ratio (S/N) of 10.4 in the GIANO-B spectra. In addition, we detect Fe I in both the HARPS-N and ESPRESSO datasets, with S/N of 3.5 and 6.2, respectively. A signal from Fe I is also identified in one of the two GIANO-B observations, with a S/N of 4.0. Interestingly, a qualitatively similar pattern - with a weaker detection in one epoch compared to the other - is also observed in the two HARPS-N nights. The GIANO-B results are therefore consistent with those obtained with HARPS-N. Finally, we compared our strongest detections of CO (GIANO-B) and Fe I (ESPRESSO), with predictions from Global Circulation Models (GCMs). Both cross-correlation and likelihood analyses favour the GCM that includes atmospheric dynamics over a static (no-dynamics) model when applied to the ESPRESSO data. This study adds to the growing body of literature employing GCMs to interpret high-resolution spectroscopic measurements of exoplanet atmospheres.

Figures (11)

• Figure 1: Top panel: Schematic representation of the orbit of WASP-76 b, with the orbital phases covered in this study with GIARPS, colour-coded according to observing night. Bottom Panel: S/N averaged over orders as a function of the orbital phase for nights in emission, the markers' colour is proportional to the airmass. The exposures at a higher S/N ratio are the HARPS-N ones, while those at a lower S/N are from GIANO-B.
• Figure 2: Left Panel: Temperature-Pressure (TP) profile used for our cross-correlation analysis, from Yan2023. Right panel: VMRs adapted in the single-species models used in this paper. The string "mol" represents each of the investigated molecules.
• Figure 3: Detection S/N maps as a function of $V_{\mathrm{rest}}$ and $K_{\mathrm{p}}$ for CO in GIANO-B (left panel), Fe I in HARPS-N (central panels), and Fe I in ESPRESSO (right panel) data. Red (white) dotted lines mark the expected (obtained) planetary position. For each map, the top and right sub-panels show the 1D CC functions (in terms of S/N) at the peak position, with the best-fit Gaussian overplotted in orange
• Figure 4: 3D templates with dynamics (left panels) vs 3D templates without dynamics (right panels) comparison for Fe I in ESPRESSO data. Upper panels: detection S/N maps as a function of $K_{\mathrm{p}}$ and $V_{\mathrm{rest}}$ with the 1D CC functions (see Fig \ref{['detections']} for details). Bottom panels: Likelihood confidence intervals as a function of $K_{\mathrm{p}}$ and $V_{\mathrm{rest}}$. Red (black) dotted lines mark the expected (obtained) planetary position. The 3D templates were computed based on the drag-free GCM of WASP-76 b from Wardenier2025.
• Figure 5: Cross-correlation results for CO (left panels) and Fe I (right panels) in GIANO-B data. For each species, we show the results for individual nights (top two rows) and their combined signal (bottom row). The greyscale maps represent the cross-correlation functions as a function of velocity in the telluric rest frame ($V_{\mathrm{tell}}$) and orbital phase. Masked regions are affected by stellar residuals (see Sect. \ref{['data_analysis']}). The black and white dashed lines indicate the stellar and planetary trails, respectively. The colour maps display the 2D cross-correlation function in the ($K_{\mathrm{p}}$, $V_{\mathrm{rest}}$) plane, expressed in terms of S/N. A detailed description of these plots is provided in the caption of Fig. \ref{['detections']}. Since we adopted the CC approach from Gibson2020, Cont2022, and Nortmann2025, the noisier lines have larger uncertainties, and a bigger relative scatter among the data.