New Astronomy Reviews 83 (2018) 28–36

Contents lists available at ScienceDirect

New Astronomy Reviews

journal homepage: www.elsevier.com/locate/newastrev

Kepler-62f: Kepler's first small planet in the habitable zone, but is it real? T
a,⁎,1 b c,2 d
William Borucki , Susan E. Thompson , Eric Agol , Christina Hedges
a
NASA Ames Research Center, Moffett Field, CA 94035, United States
b
Space Telescope Science Institute, Baltimore, MD 21218, United States
c
Department of Astronomy and Virtual Planetary Laboratory University of Washington, Seattle, WA 98195, United States
d
Bay Area Environmental Research Institute, 625 2nd St Ste. 209, Petaluma, CA 94952, United States

ABSTRACT

Kepler-62f is the first exoplanet small enough to plausibly have a rocky composition orbiting within the habitable zone (HZ) discovered by the Kepler Mission. The
planet is 1.4 times the size of the Earth and has an orbital period of 267 days. At the time of its discovery, it had the longest period of any small planet in the habitable
zone of a multi-planet system. Because of its long period, only four transits were observed during Kepler's interval of observations. It was initially missed by the Kepler
pipeline, but the first three transits were identified by an independent search by Eric Agol, and it was identified as a planet candidate in subsequent Kepler catalogs.
However in the latest catalog of exoplanets (Thompson et al., 2018), it is labeled as a false positive. Recent exoplanet catalogues have evolved from subjective
classification to automatic classifications of planet candidates by algorithms (such as ‘Robovetter’). While exceptionally useful for producing a uniform catalogue,
these algorithms sometimes misclassify planet candidates as a false positive, as is the case of Kepler-62f. In particularly valuable cases, i.e., when a small planet has
been found orbiting in the habitable zone (HZ), it is important to conduct comprehensive analyses of the data and classification protocols to provide the best estimate
of the true status of the detection. In this paper we conduct such analyses and show that Kepler-62f is a true planet and not a false positive. The table of stellar and
planet properties has been updated based on GAIA results.

1. Introduction could reliably detect Earth-size transits in the presence of the noise
expected on orbit. The fifth proposal to the NASA Discovery Program
The Kepler Mission was launched in 2009 with the goals of de- was approved for implementation in 2001 and the Mission launched in
termining the occurrence rate of small (i.e., approximately Earth-size) 2009. Dozens of candidate transiting planets were immediately ob-
planets in the habitable zones (HZ) of solar like stars and to determine vious. The 2010/04/20 (#2) issue of Astrophysical Journal Letters was
the characteristics of exoplanets and the stars they orbit (Koch et al., entirely devoted to the 24 papers describing the Kepler results.
2010). This mission was the culmination of five proposals to NASA, Caldwell et al. (2010) discuss Kepler's on orbit instrument performance.
interspersed with years of tests to demonstrate the technology neces- Jenkins et al. (2010) present an overview of the automated analysis of
sary for high-precision photometry (Borucki, 2016; Borucki, 2017). the Kepler data. Batalha et al. (2010) describe the process of choosing
Quantitative discussions by Rosenblatt (1971), Borucki and Summers the best targets among the millions of stars in the field-of-view.
(1984) generated requirements for the technology needed for the Petigura et al. (2017) and Furlan et al. (2018) discuss the ground-based
photometric detection of Earth-size planets. Borucki and Koch (1994) survey required to characterize the host stars. During its four-year life
compared various methods to observe thousands of stars simulta- span, the Mission discovered more than 4400 planetary candidates with
neously. Robinson et al. (1995) published the results of CCD tests that 2300 confirmed as planets, and of these, 13 were small planets in the
demonstrated CCD arrays had the 10 ppm relative precision needed HZ. Kepler revolutionized our understanding of exoplanet systems with
when systematic errors were measured and corrected. The feasibility of short periods (Lissauer et al., 2014a; Winn et al., 2019), giving insight
doing automated photometry on thousands of stars simultaneously was into their statistical properties, such as occurrence rates, size distribu-
shown by a wide-field-of-view photometric telescope that radioed its tions, orbital periods, multiplicities, as well as turning up examples of
observations to Ames where the results were analyzed for transits by a new systems such as circumbinary transiting planets (Winn and
computer program (Borucki et al., 2001). In the same year, Fabrycky, 2015; Doyle, 2019).
Koch et al. (2000) published results from a laboratory facility that Determining the occurrence rate of planets requires a large, uni-
provided an end-to-end demonstration of prototype photometer that formly-processed sample whereas characterizing particular discoveries

⁎
Corresponding author.
E-mail address: william.j.borucki@nasa.gov (W. Borucki).
1
Ames Associate
2
Guggenheim Fellow

https://doi.org/10.1016/j.newar.2019.03.002
Accepted 19 March 2019
1387-6473/ © 2019 Published by Elsevier B.V.
W. Borucki, et al. New Astronomy Reviews 83 (2018) 28–36

allows a more comprehensive and detailed examination of all the data habitable-zone candidates, 62e and 62f, motivated further refine-
available for each of the selected individuals. ment of the stellar properties. The stellar parameters are critical for
The Kepler-62 system was one of the earliest confirmed exoplanet obtaining the planet radii, which are only measured as a ratio to the
systems with planets in the HZ. The KOI that became 62e looked like stellar radius from the transit depth. High SNR spectroscopic ob-
our best small HZ planet candidate and was the initial focus of the servations were used to derive an effective temperature, log(g), and
effort but our detailed analysis revealed 62f was an even better metallicity of the star (Furlan et al., 2018). These values were then
candidate in both size and the amount of flux intercepted. matched to stellar evolution models to estimate the stellar size, mass,
Furthermore, Kepler-62f was a nearly Earth-size planet (∼1.4 R⊕) luminosity and age. These values are given in columns 2 and 3 of
and therefore more likely to have a rocky composition. At the time of Table 1. Analyses by Thompson et al. (2018), Berger et al. (2018),
its discovery, Kepler-62f was the smallest exoplanet in the HZ, and Fulton and Petigura (2018) are provided for comparison in
making it a particularly important planet candidate. As such, a columns 4 through 6.
variety of different observation techniques were used to obtain the Based on the data from GAIA DR2 (Bailer-Jones et al., 2018; Luri
data necessary to establish its reality as a planet rather than a false- et al., 2018; GAIA Collaboration, 2018), we have also carried out an
positive (Borucki et al., 2013). Infrared measurements were obtained empirical measurement of the properties of the host star based upon
from ground-based telescopes and the space-based Spitzer telescope. the GAIA DR2 parallax, broad-band photometric measurements from
Telescopes using active optics, spectral masks, and speckle systems SDSS, 2MASS and WISE, the density estimated from the period-
provided information on nearby stars that might cause confusion. duration relation of the five planets, and an empirically-calibrated
Analysis of centroid motion of image centroids and a search for mass-luminosity relation (Eker et al., 2015). The broadband fluxes
transit timing variations (TTVs) (Ragozzine and Holman, 2019; along with the parallax and a small extinction correction give a
Nesvory, 2019) were conducted. High-resolution spectra from sev- precise luminosity of 0.2565 ± 0.0045 L⊙. The mass-luminosity re-
eral ground-based telescopes were used to search for radial velocity lation yields a mass estimate, while the density measurement from
(RV) variations, stellar characteristics, and variability. Upper limits Borucki et al. (2013) then yields a radius estimate. Finally, the lu-
on the masses were placed, based on the lack of TTV and RV varia- minosity and radius together yield a temperature for the star. All of
tions, of < 36 M⊕ for both 62e and 62f at 95% confidence (additional these parameters are discrepant with the measurements based upon
transits from later Quarters and from HST have not yet been included spectroscopy and parallax from Berger et al. (2018) and Fulton and
in this analysis). An extensive analysis of possible false-positive Petigura (2018) at the 2-sigma level. The origin of the discrepancy
sources was conducted. Because of the resources required, this range appears to lie in the density of the star which is inferred from the
of measurements and analyses are not conducted for most exopla- transit models; this should be reexamined in future photometric
nets. All these observations and analyses are consistent with Kepler- models of the light curve. These parameters are also listed in Table 1.
62f being a planet rather than a false-positive. However, in the recent The revised parameters for the star, anchored by the GAIA DR2
exoplanet catalog (Thompson et al., 2018), Kepler-62f has been distance from Bailer-Jones et al. (2018), yield a slightly larger size
listed as a false-positive. We will show that this misclassification is for the star, and hence slightly larger size for both planets. The planet
due to particular circumstances associated with the transits of this 62f remains below the 1.6 to 1.8 R⊕ cutoff for rocky planets (based
planet and with the criteria used in the automatic vetting procedure upon measurements at shorter orbital period, Fulton et al., 2017),
of that catalog. We suggest methods to overcome such mis- while 62e remains slightly above.
classifications. To put Kepler-62 in context, we compare it with other small, tem-
The Kepler-62 system (KIC 9002278) was first designated as perate ("habitable-zone") planet candidates discovered with the Kepler
Kepler Objects of Interest (KOI) 701.01, 701.02, & 701.03 based on Mission. Fig. 1 shows the semi-major axes of these planets with radii less
the first four months of data and all appear in the second catalog than 1.8 R⊕ as a function of host-star temperature. The dimension of 1.6
paper (Borucki et al., 2011 and online data). Further analysis showed R⊕ is thought to be the approximate boundary between rocky and gas-
the presence of two additional KOIs: 701.04 and 701.05. After fur- rich planets at shorter orbital periods (Rogers, 2015), while there ap-
ther data and analyses led to their confirmation as planets, the des- pears to be a marked gap in planetary radius at short periods around 1.8
ignations were changed to Kepler-62 b, c, d, e, & f; consistent with R⊕ (Fulton et al., 2017).
their orbital periods. Overplotted are the runaway and maximum greenhouse limits cor-
To estimate the planetary parameters, a careful characterization responding to the inner and outer orbital distance at which a planet
of the stellar properties was necessary. Preliminary values given in with an Earth-like atmosphere and carbon-silicate cycle might maintain
Borucki et al. (2011) are shown in Table 1. The discovery of the two surface temperatures supporting liquid water (Kopparapu et al.,

Table 1
Comparison of estimates of stellar and exoplanet properties.
Publication Borucki et al. (2011) Borucki et al. (2013) Thompson et al. (2018) Berger et al. (2018) Fulton & Petigura (2018) This paper

M* (M⊙) 0.83 0.69 ± 0.02 0.697–0.014+0.023 0.764 ± 0.011

R* (R⊙) 0.68 0.64 ± 0.02 0.66 ± 0.04 0.711 ± 0.029 0.689 ± 0.008 0.660 ± 0.018
Teff (K) 4869 4925 ± 70 4926 ± 98 4859 ± 97 4859 ± 60 5062 ± 71
log(g [cm/s2]) 4.70 4.68 ± 0.04 4.683 ± 0.023
[Fe/H] −0.37 ± 0.04 −0.34 ± 0.04
⍴ [g/cc] 3.73 3.8 ± 0.3 2.88 ± 0.23 3.8 ± 0.3
L* (L⊙) 0.23 0.21 ± 0.02 0.254 ± 0.029 0.246 ± 0.018 0.2565 ± 0.0045
Age [Gyr] 7±4 10(1.03−.21+0.09)
D [pc] 368 300.87 ± 1.2 300.87 ± 1.2 300.87 ± 1.2
R62e [R⊕] 1.61 ± 0.05 1.72–0.07+0.10 1.84–0.08+0.10 1.827 ± 0.050 1.670 ± 0.051
S62e [S⊕] 1.2 ± 0.2 1.24–0.19+0.27 1.352 ± 0.028 1.3 ± 0.1 1.321 ± 0.098
R62f [R⊕] 1.41 ± 0.07 1.43–0.06+0.08 1.531–0.099+0.084 1.501 ± 0.066 1.461 ± 0.070
S62f [S⊕] 0.41 ± 0.05 0.44–0.07+0.09 0.477 ± 0.010 0.5 ± 0.04 0.466 ± 0.034

(S⊕ is the solar flux at the top of the Earth's atmosphere.).

29
W. Borucki, et al. New Astronomy Reviews 83 (2018) 28–36

Fig. 1. Stellar effective temperature versus semi-major axis for small planets (Rp < 1.8 R⊕ to within 1-sigma) found with Kepler within and near the conservative
"habitable-zone" (Kane et al., 2016), with modified parameters from Berger et al. (2018). The region shaded in blue/purple represents the “Conservative Habitable-
zone” boundaries as set by the runaway and maximum greenhouse limits (Kopparapu et al., 2014). Green represents the region where tidal synchronization
(Kasting et al., 1993) is expected, while the orange region delineates the region of desiccation (Luger and Barnes, 2015). The red line is the smallest value for the
semi-major axis that would provide for dynamic stability for a 0.01 Earth-mass moon (Barnes and O'Brien, 2002) for Qp=100, Mp=M⊕ and age 5 Gyr. Planets to the
left and right of the blue HZ region are too hot and too cold, respectively, for liquid water to exist on their surface, given the atmospheric compositions assumed in the
model. The portion of the purple region to the right of the colored regions is potentially the most favorable region for habitability; shaded regions are less favorable.
The orbital distance and temperature of the host star are plotted in blue for Kepler-62e and 62f, based upon Fulton & Petigura (2018). The sizes of the planets in Fig. 1
are based on the revised stellar radius utilizing the GAIA DR2 distance measurement based on the paper by Fulton & Petigura (2018). The sizes of the symbols for
planets (relative to the Earth) are proportional to their actual size.(For interpretation of the references to colour in this figure legend, the reader is referred to the web
version of this article.)

2014). In addition to the incident stellar radiation, a star's gravity can In Fig. 1 we plot the approximate boundary (i.e., to the right of the
also impact a planet's habitability (Barnes, 2017). Planets which orbit at orange region) beyond which an Earth-like planet might avoid any
distances closer to the star will experience stronger tidal forcing. This significant water loss due to this process. Similarly, atmospheric or
can lead to tidal synchronization of the planet, slowing down its rota- water loss may occur due to the higher orbital velocity within the ha-
tion, and the inability to host a giant moon. Such results negate two bitable-zone for later type stars (Lissauer, 2007; Zahnle and Catling,
unique qualities of our Earth that have been argued to correlate with 2017); this limit tends to affect only the latest-type stars, and so is not
habitability (Balbus, 2014), although these may not be required for included in the plot. Of course, planets can migrate and water could be
habitability (Joshi et al., 1997; Lissauer et al., 2012a). delivered or outgassed at later times, so this constraint has some lati-
The tidal synchronization timescale and moon instability timescale tude when arguing about potential habitability as defined by the pre-
depend on a complex range of factors that are difficult to estimate for sence of surface liquid water.
any particular system. Nevertheless, a rough estimate can be made for All of these constraints indicate that physical barriers to Earth-like
timescales of 5 Gyr for an Earth-like planet. The red curve in Fig. 1 habitability may occur for planets in the inner regions of "habitable-
shows the orbital distance at which an Earth-like moon would exit the zones" of cooler, late-type stars. These stars have been the focus of
Hill sphere and become unstable in 5 Gyr. The calculation assumes a transit searches because their planets have shorter orbital periods and a
tidal Q factor of 100 and Love number of 0.3 (Barnes and higher geometric probability of transit, and because these stars are
O'Brien, 2002). The green-colored region in Fig. 1 shows the range of especially common. All of these factors lead to an enhanced probability
distances for which a planet is expected to tidally synchronize its ro- for transit detection, in addition to the smaller size of the star which
tation in 5 Gyr based on a constant-time tidal lag model (Kasting et al., leads to a larger depth of transit. In addition, the small stellar size
1993). In order to be habitable in a way similar to that on Earth, the improves the signal-to-noise ratio for the characterization of the pla-
planet should orbit at a distance to the right of the green region. The nets’ atmospheres via transit spectroscopy due to the larger depth of
true effect of tides on these planets, and the resulting impact on ha- transit.
bitability, is likely to be much more complex, however, so these curves Kepler-62f is the first small transiting planet found by the Kepler
should only be viewed as a guide (Barnes, 2017; Shields et al., 2016; Mission which has a sufficiently large orbital distance to experience
Bolmont et al., 2015; Deitrick et al., 2018; Piro, 2018; Sasaki and weak tides and to avoid desiccation at its current location, making it a
Barnes, 2014). unique and particularly valuable planet at the time of its discovery.
Another limit to habitability is the ability of a planet to retain water Since that time, Kepler added additional small planet candidates which
during the period when the star is contracting to the main sequence. also reside in this region (Kane et al., 2016). However, Kepler-62e and f
Late-type stars have extended pre-main sequence phases with elevated are the only pair of temperate, small planets amongst these, which
activity (Lissauer, 2007). This activity is associated with high energy might enable the future characterization of these planets masses with
photons that irradiate the planet and can lead to water dissociation and transit-timing variations (Agol and Fabrycky, 2017).
escape of hydrogen, resulting in desiccation (Luger and Barnes, 2015).

30
W. Borucki, et al. New Astronomy Reviews 83 (2018) 28–36

2. A short history of the discovery of Kepler-62f

After detection of the Kepler-62 system, additional ground-based

and space-based observations and several analyses were used to con-
firm the Kepler-62 system as a real planetary system (Borucki et al.,
2013). For example, Spitzer observations at 4.5 µm of Kepler 62-e by
Désert et al. (2015) showed that the transit depth measured in the IR
(i.e., 570 ± 405 ppm) and visible (700 ± 30 ppm) were similar as is
expected if the transits are associated with the target star.
False positives in photometric searches for transiting planets are
common, often caused by the presence of background eclipsing bin-
aries. The fact that the Kepler-62 system shows transits from multiple
objects greatly reduces the odds that it is a false-positive such as a
background eclipsing binary (Lissauer et al., 2012b; Lissauer et al.,
2014b; Rowe et al., 2014). Mathematical analyses using the BLENDER
algorithm (Torres et al., 2004; 2011; Fressin et al., 2012; Borucki et al.,
2013; Torres et al., 2015; Torres and Fressin, 2019) estimate the odds Fig. 3. Phase-folded light curve for Kepler-62f for Quarters 0 through 17. Data
that any of the five planets is a false positive rather than a planet is less from all 4 transits are averaged to produce the above light curve. One sigma
than 1 × 10−3. errors are shown with vertical bars. The continuous curve shows the model fit
During the writing of the 2013 paper describing the discovery of the from Borucki et al. (2013).
Kepler-62 system of three planets, the draft was sent to the Kepler team
members for their review and additions. Concurrently, one of the team
members (Eric Agol) had been working on an algorithm to detect The results of the analysis showed three strong peaks in Δχ2 plotted
exoplanets with significant transit timing variations. His approach used versus time plot (Fig. 2) with a period of 267.29 days. These peaks were
box-like transit models that were fit to every cadence in the dataset over so strong that the planet was readily recognized by eye when co-author
a grid in transit duration and depth. Each result was compared to a fit Agol first viewed this plot; the QATS periodogram analysis was not
with no transit, and the difference in the chi-square statistic (Δχ2) was needed to detect the planet. As shown in the Figure, the first transit of
computed between the fits with and without a transit. Results were fed the planet is missing. The light curve was checked for an earlier transit
into the QATS algorithm (Carter and Agol, 2013; Agol and and it was found to have occurred during a data gap when the space-
Carter, 2019), that created a periodogram allowing for small fluctua- craft was downlinking data and not observing. Unfortunately the ex-
tions in the times of transit. This approach was developed in the search pected TTVs for Kepler-62f planets are of order 10 s of minutes and
for circumbinary planets that were particularly problematic due to therefore too small for detection when compared to the uncertainty of
noisier light curves. The requirement of a consistent depth for all the center of the transit times as measured with Kepler (i.e., ∼ 18 min).
transits helped to suppress multiple noise fluctuations from masquer- Shortly thereafter, another planet with a period of 12.4 days was
ading as a series of transits to create a false transiting planet. With the found. The discovery of these additional planets caused the draft to be
standard QATS algorithm, a single strong noise feature can change the rewritten to present information on all five planets (Borucki et al.,
baseline level of noise in the periodogram, which makes the signal of 2013). Later, when the Q15 data became available, a fourth transit of
real planets less significant. Requiring the same depth for all transits planet 62f was found at the predicted time, giving further confidence to
helped to clean the periodogram of this spurious noise, making the the veracity of Kepler-62f.
peaks in the periodogram of real planets more significant. This process The initial detection of Kepler-62f occurred at the end of August
was used to analyze the Kepler-62f light curve after the fitting and 2012 just after the Kepler data containing the third transit in Quarter 12
subtraction of the transit signatures from the three previously-known was released (Kepler Data Release 17). During the preparation of the
planetary candidates. paper, the data were made available on the PlanetHunters website.3
After the paper was submitted (in January 2013), but before it was
published (in April 2013), a citizen scientist, Mark Omohundro (@
ajamyajax) identified the three transits by visual inspection, and,
without knowledge of the Kepler paper under review, he reported his
findings on the PlanetHunters website before the Kepler-62 paper was
published.4 Had Agol missed the detection of 62f, the credit for the first
identification of the planet would have fallen to an amateur planet
sleuth.

3. Is Kepler-62f a false positive?

Despite these observations, the last, uniformly-vetted Kepler exo-

planet catalog (Data Release 25, Thompson et al., 2018) does not list
Kepler-62f among planet candidates, although it does list Kepler-62b, c,
d, & e. This is a surprise because the folded transit observations (Fig. 3)
clearly show a well-defined transit light curve. Its measured transit
duration (7.46 ± 0.20 h) is consistent with the measured orbital period
Fig. 2. Three equally-spaced peaks in Δχ2 first demonstrated the presence of and the estimated stellar diameter of 0.64 ± 0.02 Rʘ for an orbital
the Kepler-62f planet prior to the availability of the Q15 data. The lightcurve
inclination of 89.90° ± 0.03°. An analysis of the likelihood of Kepler-
for Kepler-62 was preprocessed to remove the signal of the other 3 planets,
leading to a strong, periodic signal from Kepler-62f. The vertical arrows show
the locations of the three transits observed prior to the publication of the dis- 3
https://www.planethunters.org
covery paper (Borucki et al., 2013). 4
https://oldtalk.planethunters.org/discussions/DPH101owp4

31
W. Borucki, et al. New Astronomy Reviews 83 (2018) 28–36

62f being a false positive caused by an astrophysical event mimicking a centered at times (BJD 2454833): 589.725, 857.006, 1124.287, and
planetary transit shows only a 1 in 5000 chance (Borucki et al., 2013, 1391.568. It is likely that another transit occurred at 322.434 during
Supporting Online Material). Quarter 3. However there are no observations at that time because the
Catalogs prior to DR25 (Thompson et al., 2018) classified Kepler-62f spacecraft was in a data-transmission mode rather than in a science
as a KOI or planetary candidate. The first mention of what was even- acquisition mode. See Fig. 4.
tually to become the Kepler-62f was based on the Q0 through Q2 data As shown in Fig. 4, Kepler observed four transits of Kepler-62f. Pa-
catalog (Borucki et al., 2011) where KOI 701.03 (later renamed as nels 2 through 5 show clear evidence of transits with depths of ap-
Kepler-62e) was listed (their Table 5) as a super-Earth planet in or near proximately 460 ppm and durations of approximately 7.5 h. However,
the HZ. KOI 701.01, −0.02, and −0.03 are listed in the two of these four transits failed to pass DR25 Robovetter tests (“Rubble”
Batalha et al. (2013) and Burke et al. (2014) catalogs. The and “Skye”), namely the transits during Q9 and Q12. Because at least
Rowe et al. (2015) catalog based on the Q1-Q12 data was the first to list three transits are required for Robovetter to accept a candidate, Kepler-
701.04 (Kepler-62f) and 701.05(Kepler 62c). Kepler-62f is also listed in 62f was labelled as a False Positive. Kepler-62b,c,d, & e continued to be
the Mullally et al. (2015) catalog. All five KOIs were detected and recognized as candidates.
dispositioned as planet candidates in the DR24 catalog (Coughlin et al., The test called “Rubble” checked that each reported transit contains
2016), the first uniform catalog with fully automatic vetting via the at least ¾ of the expected number of data points given the duration of
DR24 Robovetter. It is only in the most recent catalog (DR25) that the transit and Kepler's regular cadence. For the transit in Quarter 9 (i.e.,
Kepler-62f has been classified as a false positive. panel 3 in Fig. 4), the Kepler pipeline removed the first few cadences
The Kepler exoplanet catalogs evolved in order to accomplish (shown as grey “x”s) of the transit because those cadences overlapped
Kepler's goal to determine the occurrence rate of small planets around with the transit of Kepler-62b. Since Kepler-62b was found prior to
other stars. Earlier catalogs (Borucki et al., 2011; Batalha et al., 2013; Kepler-62f and because the Kepler pipeline removes the cadences as-
Burke et al., 2014; Rowe et al., 2015 and to some extent Mullally et al., sociated with transits before searching for another transit, those ca-
2015) manually vetted each system and only deemed signals “false dences were not available for the analysis of Kepler-62f. Thus, the Q9
positives” when there was strong evidence to that effect. The last Kepler transit was deemed to have insufficient data to be counted as a true
catalog, Data Release 25 (DR25) (Thompson et al., 2018), used a ro- transit by the Robovetter. However, a careful examination of the data
botic vetting algorithm called the Robovetter (Thompson et al., 2018). prior to the removal of Kepler-62b shows sufficient data to validate the
It also implemented new criteria (Thompson et al., 2018, Appendix A) transit. Note, the shallow transits of planet c were found by the pipeline
that must be passed to be promoted to “planet candidate” in order to after planet f and this is why the proximity of planet c in Q15 did not
obtain a reliable catalog. impact Rubble's decision of this transit.
Automating the vetting had several advantages, including speed, Another test, called “Skye”, checked for cases where the transit
flexibility and the ability to characterize which transits were in- times are coincident with other Threshold Crossing Events (i.e., po-
correctly called “false positives”. Simulated transits and false posi- tential transit signals found in the data) in the same “skygroup”. (A
tives were processed using the Robovetter in order to find thresholds skygroup includes all the targets observed on the same CCD detector
on each criterion. Each vetting parameter was tuned to maximize the during each Quarter.) When too many transits occur near the same
recognition of simulated transits into “planet candidates” while time, it assumes that such signals are caused by instrumental sys-
minimizing the simulated false positives. Because all of the input tematics; specifically high-frequency oscillations often called
signals to the Robovetter were treated in the same way, the detection “rolling band” artifacts (Caldwell et al., 2010) and illustrated in
system could be understood and its performance could be quantified Fig. 11 of Borucki (2016). However, real transits that occur at such
for use in occurrence rate calculations. These measurements of the times are likely to be inadvertently flagged as “bad”. The Skye test
detection system are especially important near Kepler's detection failed the transit which occurred in Quarter 12; i.e., shown in panel 4
threshold where the vetting algorithms are less reliable. This occurs at of Fig. 4.
orbital periods greater than approximately 200 days and planet radii However, an analysis of the data taken at the time of the Q12 transit
less than approximately 2 R⊕. Ultimately, the candidates near Kepler's shows that the no “rolling band” artifact was present that could have
detection threshold are the ones that will be used to estimate the oc- mimicked a transit. The top panel of Fig. 5 shows the flux time series
currence rate of terrestrial planets in the HZ of GK dwarf stars (e.g. see from the 8 central illuminated pixels of the image of Kepler-62. The
Mulders et al., 2018). right panel shows the flux from the 22 pixels unilluminated (sky
This need for a consistent detection system for accurate occurrence background) pixels that surrounded the 8 central pixels. Also shown is
rates also means that even in obvious cases of misclassification, manual the expected flux reduction in the background pixels that would have
overrides of its decisions were not allowed. While the Robovetter and occurred if a “rolling band” artifact had created the transit observed in
the DR25 catalog are well suited for occurrence rates, it does not always the 8 illuminated pixels. Fig. 5 shows that no dip in flux occurred in
provide the best knowledge of individual systems. Ultimately, failure to surrounding pixels. Therefore no “rolling band” artifact was present at
meet the Robovetter criteria for a “planet candidate” does not in- the time of the Q12 transit. Thus the Q12 transit must be considered a
validate the existence of those planets confirmed by extensive ob- legitimate transit.
servations and analysis. Indeed, measurements of the vetting system The uniform processing of the Kepler threshold crossing events by
(Thompson et al., 2018) indicate that a significant fraction of real Robovetter produces a useful set of events for statistical analyses on
transit events were deemed “false positives” by the Robovetter in this the full Kepler exoplanet catalogue. However, the detailed analysis of
part of the parameter space. The following discussion explains the si- Kepler-62f data shows that there are individual events that are
tuations that caused the Robovetter to disposition Kepler-62f as a false misclassified. While both Rubble and Skye are useful tests to remove
positive in the DR25 catalog. the majority of false positives, in the case of Kepler-62f they have
removed a valuable small exoplanet in the HZ from the Kepler cata-
3.1. Two transit events rejected by Robovetter logue.
These results suggest that the detection of long-period planets
One of the requirements of the DR25 Robovetter protocol is that the would be enhanced if their search were conducted; 1) with a method
light curve show a minimum of three valid transits and that each transit that did not disqualify their transits when they occurred near the time
pass several tests that check that it is consistent with that expected from of the transits of the short-period planets, and 2) by comparing the
a transiting planet. During Kepler Mission science operations, four central image pixels with the surrounding pixels when the transit occurs
transits of Kepler-62f were observed in Quarters 6, 9, 12, and 15 during periods of high-level noise.

32
W. Borucki, et al. New Astronomy Reviews 83 (2018) 28–36

Fig. 4. Observed fluxes (dots) of the five Kepler-62f transits that occurred during the Kepler Mission, centered on the mid-transit time for Kepler-62f and covering the
two days before and after the mid-transit. Top to bottom; Quarters 3, 6, 9, 12, and 15. Note that the transits of Kepler-62f shown in panels 3 and 4 were not
considered valid by the Robovetter. These data use the DR25 data processing and are median-detrended by the Kepler pipeline.5 The red line shows the full transit
model for Kepler-62b, c, d, e, & f as determined by the Kepler pipeline (Twicken et al., 2016). Transits by other members of the Kepler-62 system are indicated by
lower-case letters. Points shown as grey-colored “x” were removed after the detection of other planets in this systems and were not available to the detection pipeline
nor the Robovetter when vetting Kepler-62f.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this
article.)

4. Possibility that Kepler-62f is a false alarm due to random noise scenario is not included in the likelihood function (Mullally et al.,
2018). Although Kepler-62f presents four transits and has a high (∼12)
The reliability of the DR25 candidates (against instrument noise SNR signal, there is always some possibility that Kepler-62f is a false-
sources) has been measured (Thompson et al., 2018). The results have alarm. Only with additional observational evidence of the planet's ex-
implications for validated planets like Kepler-62f. In particular, the istence can this possibility be further reduced. To this end, Kepler-62f
validation of these planets is contingent on the observed change in the was observed with the Hubble Space Telescope on 2017/12/7-8 (Pro-
flux being caused by one of the astrophysical signals considered in the gram 15129, Chris Burke PI) with the express purpose of validating the
validation (like a background eclipsing binary). However, if there is
significant probability that the observed transit signal is caused by
random fluctuations in the noise level rather than by particular types of 5
The light curve shown corresponds to that in the MODEL_INT extensions of
instrument noise, then the statistical estimate is not valid because this the Kepler Data Validation time series files available at the Mikulski Archive for
Space Telescopes (doi:17909/T9CT1P).

33
W. Borucki, et al. New Astronomy Reviews 83 (2018) 28–36

decreased signal. Comparing the transit depth of the central pixels used
to measure the star brightness with those of the surrounding pixels
demonstrated that the artifact could not have been the source of the
transit. Therefore the four expected transits of Kepler-62f were present
and valid.
This is presented as an example of how uniform processing using
data only from Kepler, while necessary for statistical studies, is not well
suited for making determinations of individual objects. It is a mistake to
consider the DR25 catalog disposition reason to cast doubt on the
confirmation of Kepler-62f, which used data beyond that directly
available from Kepler system. The analysis presented here are consistent
with the conclusions in Borucki et al. (2013); i.e., that Kepler-62f
should be considered a confirmed planet.
Given the likely veracity of this planet, it remains one of the most
interesting habitable-zone planets found to date. The longer period of
this planet may make it possible for it to avoid desiccation, tidal syn-
chronization and moon loss, all characteristics which may relate to
habitability. Several bars of CO2 should be sufficient to allow for surface
liquid water, if the planet hosts an Earth-like atmosphere (Shields et al.,
2016). Unfortunately, due to the faintness of the host star, atmospheric
transmission spectroscopy will not be feasible with JWST. However,
given these intriguing prospects for habitability, Kepler-62f and its
companion 62e will continue to be interesting targets for character-
ization, perhaps with transit-timing variations.
A table of updated stellar and exoplanet properties that includes
corrections based on recent GAIA results is also included.

Acknowledgements
Fig. 5. Light curve of Kepler-62 during and near the time of the Q12 transit.
The upper panel displays the results for the sum of the count rate for the 8 pixels Funding for the Kepler Mission was provided by NASA's Science
encompassing the star image and shows a prominent dip in flux at the expected Mission Directorate. Eric Agol acknowledges NSF grants AST-0645416
time. The lower panel shows the sum for the surrounding, unilluminated 22 and AST-1615315, and NASA grant NNX13AF62G. Support for this
pixels at the same time. If a “rolling band” artifact had been present that caused work was provided by NASA through an award issued by JPL/Caltech.
the dip in the center pixels, the artifact would necessarily have reduced the flux
The authors would like to acknowledge the helpful suggestions and
in the 22 surrounding pixels to the level shown by the dotted line in the lower
valuable discussions from Jeffrey L. Coughlin and Jack Lissauer and the
panel. The data show no such signal.
anonymous reviewer.

References
planet transits. Preliminary results indicate a transit was observed at
the expected time. Agol, E., Fabrycky, D.C., 2017. Transit-timing and duration variations for the discovery
and characterization of exoplanets. In: Hans, J. Deeg, Juan, Antonio Belmonte (Eds.),
Handbook of Exoplanets. Springer Living Reference Work ISBN: 978-3-319-30648-3,
5. Summary 2017, id.7.
Agol, E., Carter, J.A., 2019. Discovery and characterization of Kepler-36b. New Astron.
This short history of the discovery of Kepler-62f points out the Rev Special Issue on Kepler Exoplanet Firsts (in this issue).
Bailer-Jones, C.A.L., Rybizki, J., Fouesneau, M., Mantelet, G., Andrae, R., 2018.
usefulness of detailed examinations of exoplanet false positives by a Estimating distance from parallaxes. IV. Distances to 1.33 billion stars in Gaia data
variety of approaches. This is particularly important for the validation release 2. AJ 156, 58.
of small, long-period planets when few transits are available for ana- Balbus, S.A., 2014. Dynamical, biological and anthropic consequences of equal lunar and
solar angular radii. Proc. R. Soc. A 470, 20140263.
lysis. Automated classification systems (such as Robovetter) provide Barnes, J.W., O'Brien, D.P., 2002. Stability of satellites around close-in extrasolar giant
uniformly vetted catalogs of exoplanet systems which is crucially planets. Astrophys. J. 575, 1087–1093.
needed for robust statistical analysis. However such systems can miss Barnes, R., 2017. Tidal locking of habitable exoplanets. Celest. Mech. Dynamic. Astron.
129, 509–536.
events when the data do not conform to simplified criteria. In parti-
Batalha, N.M., Borucki, W.J., Koch, D.G., et al., 2010. Selection, prioritization, and
cular, the classification system (Robovetter) used to produce the DR25 characteristics of Kepler target stars. ApJL 713, L109–L114.
catalog ignored two of the four transits of Kepler-62f and subsequently Batalha, N.M., Rowe, J.F., Bryson, S.T., et al., 2013. Planetary candidates observed by
classified it as a false-positive because it required a minimum of three Kepler. III. Analysis of the first 16 months of data. ApJS 204, 24.
Berger, T.A., Huber, D., Gaidos, E., van Saders, J.L., 2018. Revised radii of Kepler stars
acceptable transits. This paper presented a comprehensive examination and planets using Gaia data release 2. arXiv 1805, 00231.
of the time-series data for two transits that were not recognized by Bolmont, E., Raymond, S.N., Leconte, J., Hersant, F., Correia, A.C.M., 2015. Mercury-T: a
Robovetter. new code to study tidally evolving multi-planet systems. Applications to Kepler-62. A
&A 583, 116.
The transit occurring during the Quarter 9 observations took place Borucki, W.J., Summers, A.L., 1984. The photometric method of detecting other planetary
near the transit time of another planet in the system. That propinquity systems. Icarus 58, 121.
caused some of its data points to be ignored. The transit occurring Borucki, W.J., Koch, D.G., 1994. Multiplex approach to the photometric detection of
planets. Ap&SS 212, 293–298.
during the Q12 period was ignored by Robovetter because of the pos- Borucki, W.J., Caldwell, D., Koch, D.G., et al., 2001. The Vulcan Photometer: A dedicated
sibility that a “rolling band” artifact might be responsible for the

34
W. Borucki, et al. New Astronomy Reviews 83 (2018) 28–36

photometer for extrasolar planet searches. PASP 113, 439–451. Petigura, E.A., Howard, A.W., Marcy, G.W., et al., 2017. The California-Kepler survey. I. High-
Borucki, W., Koch, D.G., Basri, G., et al., 2011. Characteristics of planetary candidates resolution spectroscopy of 1305 stars hosting Kepler transiting planets. AJ 154, 107.
observed by Kepler. II. Analysis of the first four months of data. Astrophys. J. 736, 19. Piro, A.L., 2018. Exoplanets torqued by the combined tides of a moon and parent star.
Borucki, W., Agol, E., Fressin, F., et al., 2013. Kepler-62: a five-planet system with planets arXiv 1803, 01971.
of 1.4 and 1.6 Earth radii in the habitable zone. Science 340, 587–590. Ragozzine, D., Holman, M., 2019. Kepler-9: the first multi-transiting system and the first
Borucki, W.J., 2016. Kepler mission: development and overview. Rep. Prog. Phys. 79, transit timing variations. New Astron. Rev Special Issue on Kepler Exoplanet Firsts (in
036901. this issue).
Borucki, W.J., 2017. Kepler: a brief discussion of the mission and exoplanet results. Proc. Robinson, L.B., Wei, M.Z., Borucki, W.J., et al., 1995. Test of CCD precision limits for
Am. Philos. Soc. 161, 38. differential photometry. Publ. Astron. Soc. Pac. 107, 1094.
Gaia Collaboration,Brown, A.G.A., Vallenari, A., Prusti, T., de Bruijne, J.H.J., Babusiaux, Rogers, L.A., 2015. Most 1.6 earth-radius planets are not rocky. Astrophys. J. 801, 41.
C., Bailer-Jones, C.A.L., 2018. Gaia data release 2. Summary of the contents and Rosenblatt, F., 1971. A two-color photometric method for detection of extra solar pla-
survey properties. arXiv 1804, 09365. netary systems. Icarus 14, 71R.
Burke, C.J., Bryson, S.T., Mullally, F., et al., 2014. Planetary candidates observed by Rowe, J.F., Bryson, S.T., Marcy, G.W., et al., 2014. Validation of Kepler's multiple planet
Kepler IV: planet sample from Q1-Q8 (22 months). ApJS 210, 19. candidates. III. Light curve analysis and announcement of hundreds of new multi-
Caldwell, D.A., Kolodziejczak, J.J., Van Cleve, J.E., et al., 2010. Instrument performance planet systems. Astrophys. J. 784, 45.
in Kepler's first months. ApJL 713, L92–L96. Rowe, J.F., Coughlin, J.L., Antoci, V., et al., 2015. Planetary candidates observed by
Carter, J.A., Agol, E., 2013. The quasiperiodic automated transit search algorithm. Kepler. V. Planet sample from Q1-Q12 (36 months). ApJS 217, 16.
Astrophys. J. 765, 1322. Sasaki, T., Barnes, J.W., 2014. Longevity of moons around habitable planets. Int. J.
Coughlin, J.L., Mullally, F., Thompson, S.E., et al., 2016. Planetary candidates observed Astrobiol. 13, 324–336.
by Kepler. VII. The first fully uniform catalog based on the entire 48-month data set Shields, A.L., Barnes, R., Agol, E., Charnay, B., Bitz, C., Meadows, V.S., 2016. The effect of
(Q1-Q17 DR24). ApJS 224, 12. orbital configuration on the possible climates and habitability of Kepler-62f.
Deitrick, R., Barnes, R., Quinn, T.R., Armstrong, J., Charnay, B., Wilhelm, C., 2018. Exo- Astrobiology 16, 443–464.
Milankovitch cycles. I. Orbits and rotation states. AJ 155, 60. Thompson, S.E., Coughlin, J.L., Hoffman, K., et al., 2018. Planetary candidates observed
Désert, J.M., Charbonneau, C., Torres, G., et al., 2015. Low false positive rate of Kepler by Kepler. VIII. A fully automated catalog with measured completeness and reliability
candidates estimated from a combination of Spitzer and follow-up observations. based on data release 25. ApJS 235, 38.
Astrophys. J. 804, 59. Torres, G., Konacki, M., Sasselov, D.D., Jha, S., 2004. New data and improved parameters
Doyle, L., 2019. The discovery of “Tatooine”: Kepler-16b new astronomy reviews; Special for the extrasolar transiting planet OGLE-TR-56b. Astrophys. J. 614, 979.
Issue on Kepler Exoplanet Firsts (in this issue). Torres, G., Fressin, F., Batalha, N.M., et al., 2011. Modeling Kepler transit light curves as
Eker, Z., Soydugan, F., Soydugan, E., et al., 2015. Main sequence effective temperature false positives: rejection of blend scenarios for Kepler-9, and validation of Kepler-9d,
from a revised mass-luminosity relation based on accurate properties. AJ 149, 131. a super-earth-size planet in a multiple system. Astrophys. J. 727, 24.
Fressin, F., Torres, G., Rowe, J.F., et al., 2012. Two earth-sized planets orbiting Kepler-20. Torres, G., Kipping, D.M., Fressin, F., et al., 2015. Validation of 12 small Kepler transiting
Nature 482, 195. planets in the habitable zone. Astrophys. J. 800, 99.
Fulton, B.J., Petigura, E.A., Howard, A.W., Isaacson, H., Marcy, G.W., Cargile, P.A., Hebb, Torres, G., Fressin, F., 2019. Discovery of the first Earth-sized planets in the Kepler-20
L., Weiss, L.M., Johnson, J.A., Morton, T.D., Sinukoff, E., Crossfield, I.J.M., Hirsch, system. New Astron. Rev Special Issue on Kepler Exoplanet Firsts (in this issue).
L.A., 2017. AJ 154, 109. Twicken, J.D., Jenkins, J.M., Seader, S.E., et al., 2016. Detection of potential transit
Fulton, B.J., Petigura, E.A., 2018. The California Kepler survey VII. Precise planet radii signals in 17 Quarters of Kepler data: result of the final Kepler mission transiting
leveraging Gaia DR2 reveal the stellar mass dependence of the planet radius gap. planet search (DR25). AJ 152, 158.
arXiv 1805, 01453. Winn, J.N., Fabrycky, D.C., 2015. The occurrence and architecture of exoplanetary sys-
Furlan, E., Ciardi, D.R., Cochran, W.D., et al., 2018. The Kepler follow-up observation tems. Annual Rev. Astron. Astrophys. 53, 409.
program; II. Stellar parameters from medium- and high-resolution spectroscopy. Winn, J.N., Sanchis-Ojeda, R., Rappaport, S., 2019. Kepler-78 and the ultra-short-period
Astrophys. J. 861, 149. planets, New Astron. Rev.; Special Issue on Kepler Exoplanet Firsts (in this issue).
Jenkins, J.M., Caldwell, D.A., Chandrasekaran, H., et al., 2010. Overview of the Kepler Zahnle, K.J., Catling, D.C., 2017. The cosmic shoreline: the evidence that escape de-
science processing pipeline. ApJL 713, L87–L91. termines which planets have atmospheres, and what this may mean for Proxima
Joshi, M.M., Haberle, R.M., Reynolds, R.T., 1997. Simulations of the atmospheres of centauri B. AJ 843, 122.
synchronously rotating terrestrial planets orbiting M dwarfs: conditions for atmo-
spheric collapse and the implications for habitability. Icarus 129, 450.
Kane, S.R., et al., 2016. A catalog of Kepler habitable zone exoplanet candidates. William Borucki is a space scientist at the NASA Ames
Astrophys. J. 830, 1. Research Center in Mountain View, California. He received
Kasting, J.F., Whitmire, D.P., Reynolds, R.T., 1993. Habitable zones around main se- a MSc in physics from the University of Wisconsin in 1962
quence stars. Icarus 101, 108–128. and then moved to NASA Ames where he first worked on
Koch, D.G., et al., 2000. CCD Photometry Tests for a Mission to Detect Earth-size Planets the development of the heat shield for the Apollo Mission in
in the Extended Solar Neighborhood 4013. SPIE, pp. 508K. the Hypersonic Free Flight Branch. After the successful
Koch, D.G., Borucki, W.J., Basri, G., Batalha, N.M., Brown, T.M., et al., 2010. Kepler Moon landings, he transferred to the Theoretical Studies
mission design, realized photometric performance, and early science. ApJL 713, Branch where he investigated lightning activity in plane-
L79–L86. tary atmospheres and developed mathematical models to
Kopparapu, R.K., Ramirez, R.-M., Schottel-Kotte, J., Kasting, J.F., Domagal-Goldman, S., predict the effects of nitric oxides and chloro-
Eymet, V., 2014. Habitable zones around main-sequence stars: dependence on pla- fluoromethanes on the Earth's ozone layer. In 1983, he
netary mass. ApJL 787, L29. began advocating the development of a space mission that
Lissauer, J.J., 2007. Planets formed in habitable zones of M dwarf stars probably are could detect Earth-size planets and determine the frequency
deficient in volatiles. Astrophys. J. 660, L149. of Earth-size planets in the habitable zone of Sun-like stars. In the succeeding years he
Lissauer, J.J., Barnes, J.W., Chambers, J.E., 2012a. Obliquity variations of a moonless developed the techniques required to find small planets and showed that the technology
earth. Icarus 217, 77. and analysis techniques were sufficiently mature to proceed to flight status. He served as
Lissauer, J.J., co-authors, 23, 2012b. Almost all of Kepler's candidate multiple planet the Science Principal Investigator for the Kepler Mission. The Mission was designed to
candidates are planets. Astrophys. J. 750, 112 (15pp). determine the frequency of terrestrial planets orbiting in and near the habitable zones of
Lissauer, J.J., Dawson, R.I., Tremaine, S., 2014a. Advances in exoplanet science from other stars. It used high-precision photometry to monitor the brightness of over 170,000
Kepler. Nature 513, 336. stars to search for transits caused by orbiting planets, was launched on March 6, 2009,
Lissauer, J.J., Marcy, G.W., Bryson, S.T., et al., 2014b. Validation of Kepler's multiple completed its data acquisition phase in 2013, and is now in the data analysis phase. Based
planet candidates. II. Refined statistical framework and descriptions of systems of on the first four years of observations, over 2300 planets have been confirmed and an
special interest. Astrophys. J. 784, 44. additional 2200 planetary candidates await confirmation.
Luger, R., Barnes, R., 2015. Extreme water loss and abiotic oxygen buildup on planets
throughout the habitable zones of M dwarfs. Astrobiology 15, 119–143.
Luri, X., Brown, A.G.A., Sarro, L.M., Arenou, F., Bailer-Jones, C.A.L., Castro-Ginard, A., de Susan Thompson (also known as Susan Mullally) obtained her
Bruijne, J., Prusti, T., Babusiaux, C., Delgado, H.E., 2018. Gaia Data Release 2: using PhD at the University of North Carolina, Chapel Hill where
Gaia parallaxes. arXiv:1804.09376. she explored new methods to determine the mode identity of
Mulders, G.D., Pascucci, I., Apai, D., et al., 2018. The exoplanet population observation pulsating white dwarf stars. She is currently Senior Archive
simulator. I. The inner edges of planetary systems. AJ 156, 24. Scientist at the Space Telescope Science Institute and is
Mullally, F., Coughlin, J.L., Thompson, S.E., et al., 2015. Planetary candidates observed working to archive the data from Kepler, TESS and JWST. She
by Kepler. VI. Planet sample from Q1—Q16 (47 months). ApJS 217, 31. previously worked in the Science Office for the Kepler Mission
Mullally, Fergal, Thompson, Susan E., Coughlin, Jeffery L., et al., 2018. Kepler's earth-like and as Associate Director for the Whole Earth Telescope. She
planets should not be confirmed without independent detection: the case of Kepler- is known for leading the efforts to create and characterize the
452b, arXiv:1803.11307M. final catalog of exoplanets based on the entire Kepler data set;
Nesvorny, D., 2019. How to find a planet from transit variations. New Astron. Rev Special she identified the class of dynamic, tidally distorted, eccentric
Issue on Kepler Exoplanet Firsts (in this issue). binary stars known as Heartbeat stars; and she mapped the
location of calcium on the surface of a white dwarf star.

35
W. Borucki, et al. New Astronomy Reviews 83 (2018) 28–36

Eric Agol obtained his PhD at the University of California, Christina Hedges is an astrophysicist at NASA's Ames
Santa Barbara, followed by a postdoctoral fellowship at Research Center and is a support scientist for the Kepler and
Johns Hopkins University and a Chandra Fellowship at K2 missions. She supports the exoplanet and aster-
Caltech, where he worked on black holes and gravitational oseismology community in getting the best out of Kepler
lensing, carrying out the computations which inspired the and K2 data. She earned her Masters degree at the
Event Horizon Telescope. He is currently a Guggenheim University of Birmingham working on asteroseismology in
Fellow and Professor of Astronomy at the University of red giant stars. She earned her PhD at the University of
Washington. He developed a model for light curves of Cambridge working on detecting signals exoplanet atmo-
transiting exoplanets which has been used to find and spheres and using machine learning to classify unusual
characterize thousands of exoplanets; he coined the term young stellar objects with warped circumstellar disks.
“transit timing variations;” he first proposed that planets
could be “mapped” from phase-variations; and he dis-
covered the small, temperate exoplanet Kepler-62f.

36