The California-Kepler Survey. I. High Resolution Spectroscopy of 1305 Stars Hosting Kepler Transiting Planets

TL;DR

The California-Kepler Survey provides high-resolution spectroscopic data for over 1300 Kepler planet host stars, improving stellar and planetary property estimates and enabling detailed statistical studies of exoplanet systems.

Contribution

This work delivers the largest uniform high-resolution spectroscopic catalog of Kepler host stars, enhancing the accuracy of stellar and planetary parameters.

Findings

Achieved 60 K precision in effective temperature

Determined stellar surface gravity to 0.10 dex accuracy

Provided high-quality spectra for community use

Abstract

The California-Kepler Survey (CKS) is an observational program to improve our knowledge of the properties of stars found to host transiting planets by NASA's Kepler Mission. The improvement stems from new high-resolution optical spectra obtained using HIRES at the W. M. Keck Observatory. The CKS stellar sample comprises 1305 stars classified as Kepler Objects of Interest, hosting a total of 2075 transiting planets. The primary sample is magnitude-limited (Kp < 14.2) and contains 960 stars with 1385 planets. The sample was extended to include some fainter stars that host multiple planets, ultra short period planets, or habitable zone planets. The spectroscopic parameters were determined with two different codes, one based on template matching and the other on direct spectral synthesis using radiative transfer. We demonstrate a precision of 60 K in effective temperature, 0.10 dex in surface gravity, 0.04 dex in [Fe/H], and 1.0 km/s in projected rotational velocity. In this paper we describe the CKS project and present a uniform catalog of spectroscopic parameters. Subsequent papers in this series present catalogs of derived stellar properties such as mass, radius and age; revised planet properties; and statistical explorations of the ensemble. CKS is the largest survey to determine the properties of Kepler stars using a uniform set of high-resolution, high signal-to-noise ratio spectra. The HIRES spectra are available to the community for independent analyses.

Tables

Table 1. Table 1 Papers from the California Kepler Survey

Primary CKS Papers

CKS I. High-Resolution Spectroscopy of 1305 Stars Hosting Kepler Transiting Planets (this paper)

CKS II. Precise Physical Properties of 2075 Kepler Planets and Their Host Stars (Johnson et al., submitted)

CKS III. A Gap in the Radius Distribution of Small Planets (Fulton et al., submitted)

CKS IV. Metallicities of Kepler Planet Hosts (Petigura et al., to be submitted)

CKS V. Stellar and Planetary Properties of Kepler Multiplanet Systems (Weiss et al., to be submitted)

Related Papers Using CKS Data

Detection of Stars Within $\sim$0.8″of Kepler Objects of Interest (Kolbl et al. 2015)

Absence of a Metallicity Effect for Ultra-short-period Planets (Winn et al. 2017, submitted)

Identifying Young Kepler Planet Host Stars from Keck-HIRES Spectra of Lithium (Berger et al., in prep)

Table 2. Table 2 CKS Stellar Samples

Sample	$N_{\mathrm{stars}}$	$N_{\mathrm{planets}}$
Magnitude-limited ()	$960$	$1385$
Multi-planet Systems	$484$	$1254$
Habitable Zone Systems	$127$	$127$
Ultra-Short Period Planets	$71$	$71$
Other	$38$	$38$
False PositivesaaThe False Positive sample includes systems for which all planet candidates have been dispositioned as false positives.	$113$	$175$
TotalbbSome stars are in multiple samples.	$1305$	$2075$

Table 3. Table 3 CKS Target Stars

	Stellar Samples
	Magnitude-limited	Multi-planet	Habitable	Ultra-Short		All Planets are
KOI No.	(Kp  14.2)	Systems	Zone	Period Planets	Other	False Positives
1	1	0	0	0	0	0
2	1	0	0	0	0	0
3	1	0	0	0	0	0
6	1	0	0	0	0	1
7	1	0	0	0	0	0

Table 4. Table 6 Adopted Parameter Uncertainties

Parameter	1-$\sigma$ Uncertainty
$T_{\mathrm{eff}}$	$\pm$60 K (relative; within this catalog)
	$\pm$100 K (systematic)
$\log g$	$\pm$0.10 dex
$\left[\mathrm{Fe}/H\right]$	$\pm$0.04 dex (relative; within this catalog)
	$\pm$0.04 dex (systematic)
$V\sin i$	$\pm$1 $\text{text}km{s}^{-1}$
	2 $\text{text}km{s}^{-1}$ upper limit for $V\sin i$  1 $\text{text}km{s}^{-1}$

Table 5. Table 7 Comparison with Other Surveys

	Stated Uncertainties				Offset with CKS				RMS with CKS
Catalog	$T_{\mathrm{eff}}$	$\log g$	$\left[\mathrm{Fe}/H\right]$	$N_{\star }$	$T_{\mathrm{eff}}$	$\log g$	$\left[\mathrm{Fe}/H\right]$		$T_{\mathrm{eff}}$	$\log g$	$\left[\mathrm{Fe}/H\right]$
	[K]	[dex]	[dex]	common	[K]	[dex]	[dex]		[K]	[dex]	[dex]
This Paper
CKS aaDispositions: CP = confirmed planet; PC = planet candidate; FP = false positive.	$60$ (rel) 100 (sys)	$0.10$	$0.04$ (rel) 0.06 (sys)	$\cdots$	$\cdots$	$\cdots$	$\cdots$		$\cdots$	$\cdots$	$\cdots$
Validation of CKS with Platinum Stars
Huber et al. (2013) bbMorton et al. (2016)	$\cdots$	0.01	$\cdots$	$71$	$\cdots$	$-0.03$	$\cdots$		$\cdots$	$0.08$	$\cdots$
Bruntt et al. (2012) ccMullally et al. (2015)	60	0.03	0.06	$57$	$-3$	$0.02$	$-0.056$		$70$	$0.11$	$0.056$
Comparison Surveys
KIC (Brown et al. 2011) ddNASA Exoplanet Archive, accessed 2017 February 1; http://exoplanetarchive.ipac.caltech.edu	200	0.40	$\sim$0.30	$1215$	$-52$	$+0.09$	$-0.194$		$161$	$0.29$	$0.254$
Huber et al. (2014) eeErrors for Huber et al. (2014) are specified separately for stars with photometry (ph) or also spectroscopy (sp). $T_{\mathrm{eff}}$ errors are stated as 3.5% (193 K at 5500 K) for photometry and 2% (110 K at 5500 K) for spectroscopy.	110 (sp) 193 (ph)	0.15 (sp) 0.40 (ph)	0.15 (sp) 0.30 (ph)	$1302$	$+128$	$+0.03$	$-0.148$		$193$	$0.26$	$0.227$
LAMOST (De Cat et al. 2015)	100	0.10	0.10	$283$	$-7$	$-0.03$	$-0.053$		$113$	$0.14$	$0.119$
SPC (Buchhave et al. 2014)	50	0.10	0.08	$396$	$-5$	$+0.02$	$+0.031$		$93$	$0.15$	$0.117$
KEA (Endl & Cochran 2016)	100	0.18	0.12	$44$	$+79$	$+0.05$	$-0.053$		$70$	$0.15$	$0.106$
Everett et al. (2013)	75	0.15	0.10	$143$	$-40$	$0.05$	$+0.008$		$102$	$0.18$	$0.076$
Flicker (Bastien et al. 2014) ffFlicker $\log g$ uncertainties are higher that 0.10 dex for some stars.	$\cdots$	0.10	$\cdots$	$232$	$\cdots$	$-0.11$	$\cdots$		$\cdots$	$0.21$	$\cdots$

Equations

$$\Delta T_{\mathrm{eff}}$$

$$\Delta \log g$$

$$\Delta [Fe/H]$$

$$\Delta T_{\mathrm{eff}}$$

$$\Delta \log g$$

$$\Delta [Fe/H]$$

Full text

1. Introduction

The NASA Kepler Mission (Borucki et al. 2010; Koch et al. 2010; Borucki 2016) has ushered in a new era in astronomy, in which extrasolar planets are known to be ubiquitous. The canon of Kepler papers contains descriptions of many remarkable planetary systems. The precision of Kepler photometry enabled the detection of planets as small as Mercury (Barclay et al. 2013), and the long, nearly uninterrupted dataset revealed a plethora of compact systems of multiple transiting planets (e.g. Kepler-11; Lissauer et al. 2011a). These iconic Kepler systems present opportunities to determine planet masses and orbital properties through dynamical effects (Ford et al. 2011; Lissauer et al. 2011b) and have inspired new classes of planet formation models (Hansen & Murray 2012; Chiang & Laughlin 2013). Circumbinary planets were found (Doyle et al. 2011), and searches for moons (Kipping et al. 2012) and rings (Heising et al. 2015) were attempted. Kepler also revealed planets resembling the Earth in size and incident stellar flux (Borucki et al. 2012, 2013; Quintana et al. 2014; Torres et al. 2015).

Doppler measurements of the masses of Kepler-discovered planets provided constraints on the composition of small planets extending down to the size of Earth (e.g., Kepler-78b; Howard et al. 2013; Pepe et al. 2013). Once the sample of such measurements was large enough, patterns began to emerge. Marcy et al. (2014) measured the masses of 49 planets and found evidence for a transition from rock- to gas-dominated compositions with increasing planet size (Weiss & Marcy 2014; Rogers 2015; Wolfgang & Lopez 2015).

The Kepler canon also includes statistical analyses of the properties of thousands of transiting planets and their host stars. Shortly before the launch of Kepler, radial-velocity (RV) surveys found that the occurrence of close-in (< 0.5 AU) planets around FGK stars rises rapidly with decreasing mass, with Neptune-mass planets outnumbering Jovian mass planets (Howard et al. 2010b; Mayor et al. 2011). After just a few months of Kepler photometry, the prevalence of planets smaller than Neptune ($R_{P}$ $<$ 4.0 $R_{\oplus}$) was confirmed and came into sharper focus. Many studies quantified the occurrence of planets as a function of planet radius and orbital period (Howard et al. 2012; Petigura et al. 2013b; Fressin et al. 2013; Dressing & Charbonneau 2013). Further work showed that Earth-size planets are common in and near the habitable zone (Petigura et al. 2013a; Dressing & Charbonneau 2015; Burke et al. 2015).

A important limiting factor in large statistical analyses of Kepler planets is the quality of the host star properties. Using only broadband photometry, the Kepler Input Catalog (KIC; Brown et al. 2011) provided stellar effective temperatures and radii good to about 200 K and 30%. These parameters limit precision planet size and incident stellar flux measurements, obscuring important features. For example, any fine details in the radius distribution of planets are smeared out by the uncertainties associated with photometric stellar radii.

This paper introduces the California-Kepler Survey (CKS), a large observational campaign to measure the properties of Kepler planets and their host stars. CKS is designed in the same spirit as the pioneering spectroscopic surveys of nearby stars targeted in Doppler planet searches (Valenti & Fischer 2005). By providing a large sample of well-characterized stars, those early surveys mapped out the strong correlation between giant-planet occurrence and stellar metallicity (Fischer & Valenti 2005) and planet occurrence as a function of planet mass, stellar mass, and orbital distance (Cumming et al. 2008; Howard et al. 2010b; Johnson et al. 2010).

For the CKS project we measure stellar parameters and conduct statistical analyses of the Kepler planet population. A central motivation for CKS was to reduce the uncertainty in the sizes of Kepler stars and planets from typically 30% in the KIC to 10% using high-resolution spectroscopy. With this improvement, CKS enables more powerful and discriminating statistical studies of the occurrence of planets as a function of the properties of the planet and the host star, including its mass, age, and metallicity.

The CKS project grew out of experience with the Kepler Follow-up Observation Program (KFOP; Gautier et al. 2010), which carried out extensive ground-based observations of hundreds of Kepler Objects of Interest (KOIs) using many facilities operated by dozens of astronomers.111This effort was later enlarged to include any willing observers and renamed the Community Follow-up Observing Program (CFOP). These observations included direct imaging (Adams et al. 2012, 2013; Baranec et al. 2016; Ziegler et al. 2017; Furlan et al. 2017) as well as high-resolution spectroscopy (Gautier et al. 2012; Everett et al. 2013; Buchhave et al. 2012, 2014). The Spitzer Space Telescope was also used for characterization of Kepler-discovered planets (Désert et al. 2015).

In this paper, we describe the survey (Sec. 2), the spectroscopic pipelines (Sec. 3), the catalog of spectroscopic parameters (Sec. 4), a comparison of results from other surveys (Sec. 5), and a summary of conclusions (Sec. 6). Table 1 outlines the papers in the CKS series. Paper II presents the stellar radii, masses, and approximate ages for stars in the CKS sample, based on the spectroscopic parameters presented here. Papers III, IV, and V are statistical analyses of planet and star properties enabled by this large and precise catalog. A set of related papers make use of the CKS data to conduct complementary analyses.

2. The California-Kepler Survey

2.1. Project Plan

The original goal of the CKS project was to measure the stellar properties of all 997 host stars in the first large Kepler planet catalog (Borucki et al. 2011). As the Kepler planet catalogs grew in size (Batalha et al. 2013; Burke et al. 2014), we decided on a magnitude limit of $Kp<14.2$ (Kepler apparent magnitude) for the primary CKS sample. Most of the spectra were collected during the 2012, 2013, and 2014 observing seasons. During this time the tabulated ‘dispositions’ of some KOIs changed between ‘candidate’, ‘confirmed’, ‘validated’, and ‘false positive’. We discuss the dispositions that we adopted in Sec. 2.5. Planet candidates have low probabilities of being false positives, typically $<$10% (Morton & Johnson 2011). For simplicity, we refer to KOIs as “planets” throughout much of this paper, except when describing known false positives.

The CKS project is independent from the KFOP observations that were in direct support of the Kepler mission. CKS observations of the magnitude-limited sample (see Sec. 2.3) were conducted using Keck time granted for this project by the University of California, the California Institute of Technology, and the University of Hawaii. Observations of the sample of Multi-planet Systems were supported by Keck time from the University of California. The samples of Ultra-Short Period Planets and Habitable Zone Planets were observed using Keck time from NASA and the California Institute of Technology specifically for this project. Most of the CKS results ($\sim$1000 stars) are derived from spectra reported for the first time here. Some of the CKS stars ($\sim$300/$1305$) were observed with Keck-HIRES as part of the NASA Keck time awarded to the KFOP team specifically for mission support and are included in CKS. Those previous observations were for characterization of noteworthy systems or as part of determining precise planet masses. The KFOP observations are described in Kepler Data Release 25 (DR25; Mathur et al. 2016) and include spectroscopic parameters that may vary slightly compared with our results. See Furlan et al., in prep. for a summary of KFOP spectroscopy. All spectra used in this paper are publicly available on Keck Observatory Archive.

2.2. Observations

We observed all $1305$ stars in the CKS sample with the HIRES spectrometer (Vogt et al. 1994) at the W. M. Keck Observatory. We used an exposure meter to stop the exposures after achieving a signal-to-noise ratio (S/N) of 45 per pixel (90 per resolution element) at the peak of the blaze function in the spectral order containing 550 nm. A small subset of targets was observed at higher S/N, usually because a higher S/N was needed to serve as template spectra for precise RV measurements (Marcy et al. 2014). For the faintest targets ($Kp>15.0$) the S/N was limited to $\sim$20 per pixel, given the constraints on the total observing time. The spectral format and HIRES settings were identical to those used by the California Planet Search (Howard et al. 2010a). This includes the use of the B5/C2 decker with dimensions of $0\farcs 86\times 3\farcs 5/0\farcs 86\times 14\farcs 0$, resulting in a spectral resolution of 60,000. For stars with $V>11$ (most of the sample), we used the C2 decker and employed a sky-subtraction routine to reduce the impact of scattered moonlight and telluric emission lines (Batalha et al. 2011). The spectral coverage extended from 3640 to 7990 Å. We aligned the spectral format of HIRES such that the observatory-frame wavelengths were consistent to within one pixel from night to night. This allows for extraction of the spectral orders using the CPS raw reduction pipeline. We used the HIRES guide camera with a green filter (BG38), ensuring that the guiding signal was based on light near the middle of the wavelength range of the spectra. Except for a few stars with nearby companions, we used the HIRES image-rotator in the vertical-angle mode to capture the full spectral bandwidth within the spectrometer entrance slit.

2.3. Stellar Samples

The CKS sample comprises several overlapping sub-samples listed below. Table 2 provides a summary of the number of stars and planets belonging to each subsample while Table 2.3 provides the star-by-star designations. Figure 1 shows the distribution of stellar brightness and of the number of planets per star, for the entire CKS sample.

Magnitude-limited. This sample is defined as all stars with $Kp<14.2$. We set out to observe a magnitude-limited sample of KOIs chosen independent of the number of detected planets or previously measured stellar properties. As the project progressed, we added additional samples of fainter stars, as described below.

Multi-planet Systems. This sample is defined as KOIs stars orbited by two or more transiting planets (excluding false positives). We also observed nearly all of the multi-transiting systems appearing in the Rowe et al. (2014) catalog, with priority given to the highest multiplicity systems and the brightest stars. CKS Paper V (Weiss et al. in prep.) performs a detailed analysis of the multi-planet systems.

Habitable-Zone Systems. We observed $127$ host stars of Kepler planets residing in or near the habitable zone defined by (Kopparapu et al. 2013). Some of the individual habitable-zone planets have been studied extensively and validated (Borucki et al. 2013; Torres et al. 2015; Jenkins et al. 2015). It is not clear what to adopt as the boundaries of the liquid-water habitable zone, because of the many uncertainties in exoplanet atmospheric properties and other factors that impact planet habitability (Seager 2013). The NASA Kepler Team constructed a list of habitable-zone targets using the best available stellar parameters at the time. They selected stars for which the flux received by the planet fell (within 1$\sigma$) between the Venus and “early-Mars” habitable-zone boundaries (Kopparapu et al. 2013). After the revision to the stellar parameters based on our CKS spectra, we now know that some of these planets are well outside of the habitable zone. CKS Paper II (Johnson et al., submitted) gives the newly determined values for stellar flux and planetary equilibrium temperature for all the CKS stars.

Ultra-Short Period Planets. Ultra-short period (USP) planets (Sanchis-Ojeda et al. 2014) have orbital periods shorter than one day. Winn et al. (2017, submitted) have performed an investigation of this sample, in particular on the metallicity distribution.

Other. We observed 38 additional Kepler planet host stars for reasons that do not fall into any of the preceding categories. Often these ad hoc observations were for studies of unusual or noteworthy planetary systems (e.g. Dawson et al. 2015; Désert et al. 2015; Holczer et al. 2015; Kruse & Agol 2014).

False Positives. The planetary candidate status (“disposition”) of some KOIs has changed over time. Inevitably we observed KOIs that are now recognized as false positives. For completeness we report on the parameters for these false positives. Importantly, though, the false positives were not used for the cross-calibration between our two spectroscopic analysis pipelines (see Sec. 4.2). More details on this sample are given in Sec. 2.5.

It is important to recognize that the samples in the CKS survey are built upon the foundation of the Kepler mission. Assembling the Kepler planet catalogs required the extraordinary effort and devotion of the Kepler team members (Borucki et al. 2011; Batalha et al. 2013; Burke et al. 2014; Rowe et al. 2015). Also essential was the painstaking engineering behind the photometer (Caldwell et al. 2010; Gilliland et al. 2011; Bryson et al. 2010; Haas et al. 2010), as well as the software engineering that transformed CCD pixel values into planet candidates (Jenkins et al. 2010; Gilliland et al. 2010; Stumpe et al. 2012; Smith et al. 2012, 2016; Batalha et al. 2010a, b; Torres et al. 2011; Bryson et al. 2013; Christiansen et al. 2012, 2013, 2015, 2016; Thompson et al. 2015; McCauliff et al. 2015; Tenenbaum et al. 2013, 2014; Twicken et al. 2016; Kinemuchi et al. 2012).