Using Stellar Spectral Energy Distributions to Measure Exoplanet Parameters

TL;DR

This study presents a photometric SED fitting framework to derive precise stellar radii $R_\star$ and SED temperatures $T_{\rm SED}$ for exoplanet host stars using only archival photometry, Gaia parallaxes, and Gaia-based extinction maps. By fixing extinction with 3D Gaia maps, the method reduces $T_{\rm eff}$–extinction degeneracy and delivers typical radius precisions of about 2% without relying on stellar interior models or age priors. Applied to the PanCET sample, exoplanet radii reach ~2% precision and masses ~10% precision, improving literature uncertainties by roughly a factor of two and enabling self-consistent planetary parameters across the sample. The approach is all-sky and scalable, allowing broad, model-independent exoplanet parameter determinations for a large fraction of known planets.

Abstract

The ability to make accurate determinations of planetary parameters is inextricably linked to measuring physical parameters of the host star, in particular the stellar radius. In this paper we fit the stellar spectral energy distributions of exoplanet hosts to measure their radii, making use of only archival photometry, the $Gaia$ parallaxes and $Gaia$ extinction maps. Using the extinction maps frees us of the degeneracy between temperature and extinction which has plagued this method in the past. The resulting radii have typical random uncertainties of about 2 per cent. We perform a quantitative study of systematic uncertainties affecting the methodology and find they are similar to, or smaller than, the random ones. We discuss how the stellar parameters can be used to derive the properties of both transiting exoplanets, and those where only a radial-velocity curve is available. We then explore in detail the improvements the method makes possible for the parameters of the PanCET sample of transiting planets. For this sample we find the best literature measurements of the planetary radii have mean uncertainties about 40 per cent larger than those presented here, with the new measurements achieving precisions of 2 per cent in radius and 10 per cent in mass. In contrast to much recent work, these transiting exoplanets parameters are derived without using theoretical models of stellar interiors, freeing them of the assumptions those models contain, and any priors for stellar age. As the data used are available for the whole sky, the method can be used for self-consistent measurements of the planetary parameters of a very large fraction of known exoplanets.

Figures (21)

• Figure 1: A schematic showing the process of the SED fitting method, starting with generating grids of synthetic photometry (top), how we measure stellar parameters (middle), and finally how we determine their uncertainties (bottom).
• Figure 2: The residual between the synthetic and observed $G$ photometry of our chosen exoplanet hosts as a function of temperature. The synthetic photometry was generated using stellar atmospheres from the $T_{\rm sp}$ and $R_\star$ presented in the literature and placed at the distance from $Gaia$.
• Figure 3: The coverage of all of the system responses used in the fitting process. The chosen systems sample the UV (GALEX, dotted), optical ($Gaia$ DR2, dashed), near-IR (2MASS, solid) and mid-IR (WISE, dot-dashed). The model spectra correspond to the best fitting models for GJ 3470 (red) and KELT-7 (blue), which lie at the $T_{\rm SED}$ extrema of the sample. They illustrate that the bands used in the fit are more than adequate to sample the SED of the stars in our input catalogue.
• Figure 4: The host star properties in $T_{\rm SED} - R_\star$ space, along with their associated 68% confidence contours. For comparison we show a series of isochrones from 2008ApJS..178...89D and the M-dwarf temperature radius relationship (red dashed line) derived in Morrell:2019aa and its bounds (blue shaded region). We emphasise that we do not expect all the stars to lie on these relationships, for example HD 97658 which is known to be metal poor 2011ApJ...730...10H and so should lie below the Solar metallicity isochrones.
• Figure 5: As Fig. \ref{['fig:lit-lum-comp-names-litonly']}, but with the added comparison between the synthetic $G$ band photometry generated from our parameters (red) shown alongside those generated from the literature values (blue). We have also estimated and plotted uncertainty bounds for each target. Note that these uncertainty bounds are likely over-estimates, as we did not have access to the correlation between the bounds in $T_{\rm eff}$ and $R_\star$.