A Comparison of 1D and 3D Exoplanet Atmosphere Model Grids: ScCHIMERA and the SPARC/MITgcm

TL;DR

This study quantifies how 1D radiative-convective equilibrium (RCE) models compare to 3D general circulation models (GCMs) for exoplanet dayside emission, focusing on the biases these differences can introduce in JWST-era analyses. By constructing matched grids of 3D SPARC/MITgcm and 1D ScCHIMERA models across equilibrium temperatures, gravities, metallicities, and rotation, and by equating their outgoing bolometric flux, the authors dissect the dayside pressure–temperature structures into four regions and examine the resulting spectra, including cases with TiO/VO-driven inversions. Benchmarking against identical substellar PT profiles and scaling TiO/VO opacities shows that opacity differences and vertical gradients can yield residuals up to tens of percent at certain wavelengths, though bolometric fluxes remain closely matched ($\leq1.5\%$) across the grid. JWST-simulated observations reveal that, for some instrument configurations, the 1D vs 3D differences could be detectable, implying potential biases in inferred planetary properties unless 3D effects are accounted for or effectively approximated in faster codes; this motivates developing pressure-dependent heat redistribution or pseudo-2D approaches to bridge the gap between 1D and 3D treatments.

Abstract

Inferring the properties of transiting exoplanet atmospheres relies on comparing models to spectroscopic observations. Atmosphere models, however, make a range of assumptions, from one-dimensional (1D, varying with altitude) radiative-convective equilibrium (RCE) to three-dimensional (3D) general circulation models (GCMs). The goal of this investigation is to determine the causes of differences in dayside thermal emission spectra resulting from 3D-GCMs (using SPARC/MITgcm) and 1D-RCE models (using ScCHIMERA). We conduct a one-to-one comparison of 1D-RCE models and 3D-GCMs with the same outgoing bolometric thermal flux over a grid of equilibrium temperatures, gravities, metallicities, and rotation periods. Each 1D-RCE model assumes heat redistribution in the planet's atmosphere consistent with that in the corresponding 3D-GCM's photosphere. Comparing corresponding models, the dayside average pressure-temperature (PT) structures can be broken into four vertical regions, each influencing wavelength-dependent differences in their spectra. Furthermore, the dayside average 3D-GCM PTs for planets with Teq=1400 K exhibit a temperature inversion, whereas corresponding 1D-RCE models do not. We find that spectral differences between 1D-RCE models and 3D-GCMs with the same parameters decrease for hotter planets because the spectral shapes more closely resemble blackbodies. To a lesser extent, spectral differences increase for planets with longer rotation periods because of smaller day-night temperature contrasts in the photosphere. Finally, we compare spectral differences to realistic observational uncertainties from JWST with the NIRISS SOSS, NIRSpec G395H, and MIRI LRS instrument modes. We find that 1D-RCE models and 3D-GCMs with the same parameters can produce dayside spectral differences larger than JWST's uncertainty, potentially biasing data-model inferences.

Figures (12)

• Figure 1: Left Panel: Pressure-temperature profiles for 1D-RCE and 3D-GCM models with T$_{\textrm{eq}}$=1800 K, [M/H]=0.0, log$_{\textrm{10}}$(g)=1.3, M$_*$=0.8 M$_{\textrm{Sun}}$, and TiO and VO removed from the atmosphere. The initial profile used in the 3D-GCM before allowing the atmosphere to evolve is shown with a dashed line, and grey lines are all PT profiles from the 3D-GCM dayside. Key regions of the atmosphere are shaded in different colors. Right Panel: Secondary eclipse planet flux spectra for 1D-RCE and 3D-GCM models with the same parameters as the left panel.
• Figure 2: Left Panel: Pressure-temperature profiles for 1D-RCE and 3D-GCM models with T$_{\textrm{eq}}$=1800 K, [M/H]=0.0, log$_{\textrm{10}}$(g)=1.3, M$_*$=0.8 M$_{\textrm{Sun}}$, and TiO and VO included in the atmosphere. The initial profile used in the 3D-GCM before allowing the atmosphere to evolve is shown with a dashed line, and grey lines are all PT profiles from the 3D-GCM dayside. Key regions of the atmosphere are shaded in different colors. Right Panel: Secondary eclipse planet flux spectra for 1D-RCE and 3D-GCM models with the same parameters as the left panel.
• Figure 3: Relative planet flux residual (%) versus wavelength, averaged at each temperature across the other parameters within our grid. The mean residual is plotted as a solid line, and the 1$\sigma$ distribution is plotted as a shaded region. Left: Models with TiO and VO removed. Right: Models with TiO and VO included.
• Figure 4: The wavelength-averaged root mean square (RMS) of the fractional planet flux residuals for all models without TiO and VO, i.e., $\sqrt{([F_{\textrm{p,3D}}-F_{\textrm{p,1D}}]/F_{\textrm{p,1D}})_{\textrm{avg}}^{2}}$. White squares indicate a missing 3D-GCM and/or 1D-RCE model from the grid due to numerical instabilities in the respective model.
• Figure 5: The wavelength-averaged root mean square (RMS) of the fractional planet flux residuals for all models with TiO and VO, i.e., $\sqrt{([F_{\textrm{p,3D}}-F_{\textrm{p,1D}}]/F_{\textrm{p,1D}})_{\textrm{avg}}^{2}}$. White squares indicate a missing 3D-GCM and/or 1D-RCE model from the grid due to numerical instabilities in the respective model.