Inferring Stellar Parameters from Iodine-Imprinted Keck/HIRES Spectra with Machine Learning

TL;DR

The work tackles stellar parameter inference from iodine-imprinted Keck/HIRES spectra when iodine-free templates are unavailable. It trains The Cannon using the CHIP pipeline to infer $18$ labels, including $T_{ m eff}$, $ ext{log} hinspace g$, $v\,\sin i$, [Fe/H], and 15 abundances, from iodine-contaminated data. Although uncertainties are larger than traditional iodine-free analyses, the method yields meaningful correlations with SPOCS labels across a 372-star test set, enabling rapid characterization of large RV survey samples without separate iodine-free templates. This approach, built on public data (NExScI PRV KOA) and the SPOCS catalog, enhances the practicality of characterizing exoplanet host stars in iodine-cell spectrographs and broadens the applicability of The Cannon to contaminated spectra.

Abstract

The properties of exoplanet host stars are traditionally characterized through a detailed forward-modeling analysis of high-resolution spectra. However, many exoplanet radial velocity surveys employ iodine-cell-calibrated spectrographs, such that the vast majority of spectra obtained include an imprinted forest of iodine absorption lines. For surveys that use iodine cells, iodine-free "template" spectra must be separately obtained for precise stellar characterization. These template spectra often require extensive additional observing time to obtain, and they are not always feasible to obtain for faint stars. In this paper, we demonstrate that machine learning methods can be applied to infer stellar parameters and chemical abundances from iodine-imprinted spectra with high accuracy and precision. The methods presented in this work are broadly applicable to any iodine-cell-calibrated spectrograph. We make publicly available our spectroscopic pipeline, the Cannon HIRES Iodine Pipeline (CHIP), which derives stellar parameters and 15 chemical abundances from iodine-imprinted spectra of FGK stars and which has been set up for ease of use with Keck/HIRES spectra. Our proof-of-concept offers an efficient new avenue to rapidly estimate a large number of stellar parameters even in the absence of an iodine-free template spectrum.

Figures (3)

• Figure 1: A comparison of the stellar properties spanned by the training and test sets. The left diagram depicts the 60% of stars in the training set, while the right diagram showcases the remaining 40% in the test set.
• Figure 2: Left: HD 745's AFS continuum normalization fit (blue) over-plotted on top of the deblazed spectrum (pink). Right: HD 745 spectrum after AFS continuum normalization.
• Figure 3: Each blue point represents one of 372 stars in our test set. All 18 label values from the SPOCS catalogue are plotted against our trained model's predicted values. The pink line through each panel displays the line of equality. Three extreme outlier stars, which performed poorly across all labels, were removed from the sample prior to creating this figure; such outliers would be straightforward to identify in practice, since they correspond to nonphysical stellar properties.