TOWARDS ETA-EARTH: CHARACTERIZING THE DETECTION RATE OF SMALL PLANETS IN THE KEPLER PIPELINE
This proposal lays out a plan for directly measuring the completeness and reliability of the planet sample produced by the Kepler pipeline. This knowledge is essential for determining an accurate picture of the distribution and occurrence rate of small planets. Measuring this distribution is the main goal of the Kepler Extended Mission, in particular the frequency of Earth-size planets in and near the habitable zone of their host stars. As the number of planet candidates detected by Kepler increases, and as their properties become more interesting (smaller planets, further from their host stars), people are turning their attention to characterising the underlying distribution of planets indicated by the planet candidate catalogues.
However, unless we have a rigorous understanding of the completeness of the planet sample being generated by the Kepler pipeline, i.e. what fraction of real, detectable planets are being discovered, and the reliability of that planet sample, i.e. what fraction of candidates in the sample are due to real planets as compared to false positives, we cannot derive an accurate picture of the true underlying planet population. The work proposed here aims to characterise the detection efficiency of the Kepler pipeline. We will inject simulated planet signals into the real Kepler data, which will allow us to empirically measure the types of planets to which the pipeline is sensitive, and more importantly, the types to which it is not sensitive. From this we can derive the completeness of the Kepler planet sample.
We will also separately inject simulated astrophysical false positives in the real Kepler data, which gives us the opportunity to examine the types of false positives that the pipeline is capable of identifying and rejecting. This will directly inform our understanding of the reliability of the Kepler planet sample.
Finally, the set of significant periodic signals identified by the Kepler pipeline is dominated by spurious detections, and therefore needs to be carefully vetted in order to produce the final planet sample. Currently, a team of volunteers is assembled who follow a set of rules in order to identify the truly transit-like signatures. In the future, the Mission will move towards an automated vetting procedure using machine learning. The impact of both vetting procedures on the completeness and reliability of the generated planet sample must be quantified. We will produce a challenge set of light curves, containing both simulated planet signals and simulated astrophysical false positive signals, as well as genuine periodic events found by the pipeline, with which to assess the performance of the vetting procedure.
DEEP AO IMAGING VALIDATION OF KOI'S USING THE LARGE BINOCULAR TELESCOPE
University Of Notre Dame
We propose to significantly enhance and expand Kepler's imaging follow-up program by using observations obtained with the world's premiere diffraction-limited facility: the new ("extreme") adaptive optics (AO) system at the Large Binocular Telescope (LBT). The LBT AO system offers several distinct advantages compared to other comparable large-aperture telescopes: (1) a pyramid wavefront sensor permits closed-loop (on-axis) correction for stars as faint as R=17.2 mag using the primary target as its own natural guide star; (2) use of a deformable secondary mirror results in lower thermal background levels; (3) a red-sensitive (I-band) wavefront sensor provides access to previously inaccessible K-dwarf and M-dwarf stars, increases the number of available off-axis guide stars, and increases the signal-to-noise ratio of starlight at the wavefront sensor, due to larger atmospheric r0 and t0 values. LBT AO measurements are significantly deeper than, yet complementary to, visible "lucky" imaging measurements, and can generate sufficient sensitivity to help reduce false-positive probabilities to essentially zero, particularly as a result of the ~2 mag "red advantage" obtained by
observing in the JHK bands. Our proposed program will augment the overall number of KOIs targeted with AO and improve upon contrast levels achieved at the closest angular separations for high-priority stars of any apparent magnitude. These new capabilities translate directly into more robust validation of small (terrestrial) planets and will improve the overall statistical significance of Kepler's results, by accessing more stars, reducing false-positive probabilities for individual systems, and ultimately decreasing the uncertainty in Keplerâ€TMs determination of the occurrence rate of rocky worlds.
CHARACTERIZATION OF THE STELLAR POPULATION IN THE KEPLER FIELD: PUTTING KEPLER'S SMALL PLANETS INTO A GALACTIC CONTEXT
University Of Texas, Austin
In its extended mission, Kepler focuses on the occurrence rate (eta_Earth) and physical properties of small planets (R < 2.5 R_Earth). To make this historic measurement of eta_Earth, Kepler observes nearly continuously a rich star field in the Cygnus/Lyra region for photometric transits of planetary companions. The Kepler search field is located 10 degrees off the galactic plane and thus represents a different stellar population than the immediate solar neighborhood. In order to set the population of small Kepler obtained by planets into a Galactic context, we need to characterize the stellar population in the Kepler field.
The PI proposes to perform a large-scale spectroscopic survey of the stars in the Kepler field using the wide field Visible Integral-field Replicable Unit Spectrograph (VIRUS) at the Hobby-Eberly Telescope. With VIRUS we have the unique capability to obtain tens of thousands of spectra of Kepler field stars in a very efficient way. VIRUS is an array of Integral-Field-Unit spectrographs with a total of >30,000 fibers and covers a area of 67 square arcminutes in a single 20-minute exposure. The goal of this investigation is to use VIRUS to obtain spectra for more than 50,000 stars, down to 16th magnitude, in the Kepler field. The PI will use the existing tools, that were developed to analyze VIRUS data on a large scale, to determine stellar parameters like effective temperature, metallicity (including alpha-element abundances), surface gravity and radial velocity.
With these data, we can determine the underlying stellar context of the planet population that Kepler is finding. We can distinguish between members of the thick- and thin-disk of the Galaxy and even find possible halo stars. As the ultimate goal we can compute eta_Earth for different stellar populations and use this to extrapolate eta_Earth to different regions of the Galaxy and the Milky Way as a whole.
THE ARCHITECTURES OF EXTRASOLAR TERRESTRIAL SYSTEMS
University Of Chicago
The field of extrasolar planets is now hitting close to home. The Doppler technique, having firmly established the statistical properties of gas giants, has made inroads to the sub-Neptune and super-Earth systems. The Kepler spacecraft has brought these latter populations into sharp focus. However, because many of these planets are thought to have a gas layer (with Rp > 2.5 RE) and most are at very short orbital periods, it remains unclear whether this population is analogous to the Solar System's terrestrial planets. We propose to determine whether their formation environment was similar, using the *architecture* of these systems. This term means the orbital distributions: absolute and relative spacings, eccentricities, and mutual inclinations. We propose to continue and advance our work with Kepler, using the architectural tools of transit timing, duration ratios and drifts, and stability analyses. Using transit timing, particularly the relative phases of transit timing variation (TTV) signals in the ~100 pairs of planets near resonance with each other, we will determine the degree of excitation of eccentricity, which indicates how much gas was present during the planet accretion epoch. Using duration ratios and drifts, we will determine both spatial components of the mutual inclinations of pairs of planets, which indicates if the planets formed in a disk which was flattened by gaseous dissipation processes. Using stability analyses, we will determine whether planets of small size are packed next to their neighbors due to gaseous dissipation, or whether planets grew by giant impacts in which case they would be more spread out. The new focus is to answer the vital question of whether the small planets from Kepler are "true" terrestrial planets, having the same dynamical properties as the four terrestrial planets of the Solar System. If they do not, then the Kepler sample is the statistical ensemble on which planet formation theory needs to be rebuilt from scratch. If they do, the theories of terrestrial planet formation built for the Solar System - in particular growth by giant impacts in the absence of gas -- can still serve as the paradigm of planet formation. This determination also impacts the interpretation of Kepler's central measurement, the fraction of Earth-sized planets, ηRe. To determine if this number represents ηterrestrial, i.e. the fraction of planets similar to Earth in history, chemistry, and even biology, we must first compare the dynamical record of Kepler-discovered planets to our terrestrial planets.
ACCURATE FUNDAMENTAL PROPERTIES OF KEPLER TARGET STARS
We propose to perform a homogeneous and accurate re-determination of fundamental properties of the full Kepler target sample, with an emphasis on precise measurements for planet-candidate host stars. The primary goals of the extended Kepler mission - the determination of the frequency of habitable planets and the discovery of Earth-sized planets orbiting in the habitable zone of Sun-like stars - are fundamentally connected to the properties of the observed stars. Systematic errors in the stellar properties of the parent sample can significantly bias planet occurrence studies, while uncertainties in the properties of planet-candidate host stars can prevent firm conclusions about the characteristics of small planets found in the habitable zone. Nearly all studies on planet occurrence have so far relied on stellar properties listed in the Kepler Input Catalog, which was designed for target selection and is known to suffer from significant biases. Since the creation of the Kepler Input Catalog, a large amount of new data has become available, ranging from new photometric surveys to spectroscopic follow-up observations, and the Kepler light curves themselves which can be used to infer stellar properties using asteroseismology. The aim of this proposal is to combine these new datasets to determine accurate fundamental properties for all Kepler target stars, and to optimize precise asteroseismic studies of Kepler planet-candidate hosts. The results are expected to 1) enable an accurate determination of the frequency of habitable planets around sun-like stars, 2) establish robust correlations between host-star and planet properties and 3) identify and characterize new planets detected in or close to the habitable zone.
COMPLETING THE KEPLER CENSUS OF EARTH-LIKE PLANETS
California Institute of Technology
The central goal of the Kepler Mission is to characterize the statistical distribution of small planets in the Galaxy. Of particular interest are Earth-like planets---rocky planets in the habitable zones of their stars. The main product of the Kepler detection pipeline, however, is a large population of transit candidates, not secure planets, and any of a number of astrophysical eclipse scenarios can mimic the signal of a transiting planet. While there have been several a priori studies--the pioneering one by the PI--suggesting that the rate of such false positives among well-vetted candidates should be low, a large, detailed, and observationally informed study of this rate has yet to be undertaken. Such a study is critical to properly understanding the demographics of small planets. The PI proposes to apply two existing unrivaled Kepler follow-up datasets - a Keck/HIRES spectroscopic survey of nearly 1000 Kepler Objects of Interest (KOIs) and a Robo-AO adaptive optics survey of over 1100 KOIs - to this effort, in order to calculate observationally informed false positive probabilities for every KOI. This will likely lead to the validation of nearly 1000 Kepler planet candidates. Equally critical to a proper understanding of the true demographics of small planets is the development of a statistical framework to distill the true radius and period distributions of exoplanets from the results of the Kepler survey, properly incorporating all the subtleties of reliability and completeness. The PI has done a pilot study along these lines with a small sample of cool-star KOIs, for which he has demonstrated the use of some novel statistical techniques. This study has identified some hitherto-unknown features of the radius distribution of planets around cool stars, including that planets around the size of Earth appear to be the most common type of planet around those stars. As a Participating Scientist, in addition to characterizing the false positive probabilities of all the planet candidates, the PI will bring his substantial statistical expertise to the Kepler Project to continue to develop these statistical techniques and apply them to the whole survey, in order to characterize in detail the demographics of small planets in the Galaxy and to calculate the frequency of Earth-like planets.
A DEEP SEARCH FOR SMALL PLANETS AND MULTIPLE SYSTEMS USING THE SARS PIPELINE
Weizmann Institute of Science
The Kepler extended mission is tightly focused on measuring Eta-Earth - the frequency of terrestrial planets in the habitable zone of solar-like stars - as accurately as possible. Measuring Eta-Earth using Kepler requires being able to detect shallow and long-period transit signals. We propose to apply the SARS (Simultaneous Additive and Relative Sysrem) pipeline to accomplishing this goal. The SARS pipeline was already used to find >80 planet candidates in Kepler's data, all of which in multiple systems. Experience shows that this pipeline can allow detecting new planetary candidates which are typically shallower than the ones detected by the nominal Kepler pipeline, and with no period bias, making it well suited to improving the measurement of Eta-Earth. Moreover, the SARS pipeline has a well-defined evolution program that will improve it significantly: it is now already dramatically faster to compute and is expected to be about twice as sensitive to long-period signals in the near future. We propose to use these sensitive tools not just to look for Earth-like planets, but we also investigate multiplicity in planetary systems - cases where there are more than one planet and/or more than one star in the system. Such studies allow placing powerful constraints on theoretical models of star- and planet- formation. We present the array of innovative techniques developed to attack this problem, improving the detection limit for both single- and multiple- planet systems and spanning every stage of the pipeline: from data detrending, through data decorrelation, to signal detection and global modeling. We demonstrate that the techniques above compare well with the state of the art, and we also present the multiple achievements so far of using these techniques on Kepler data.
UNIFORM MODELING OF KEPLER OBJECTS OF INTEREST
The central objective of this proposal is perform uniform state-of-the-art lightcurve modeling with Kepler's list of planetary candidates (Kepler Objects of Interest or KOIs) using tested and validated algorithms. This process includes modeling of planetary transits, phase curves and providing orbital solutions. We will use Kepler-photometry and other groundbased observables to determine key planetary parameters such as the radius and mass. More importantly, we will also determine posterior probability distributions for the fitted parameters by employing state-of-the-art Markov chain Monte Carlo algorithms.
Our work will aid Kepler to determine the frequency of terrestrial and larger planets in or near the habitable zone of a wide variety of spectral types of stars by providing accurate stellar mass and radii and accurate planetary orbits and radii. As such, our work as contributes towards the determination of the distributions of sizes and orbital semi-major axes of these planets and provides an estimate the frequency of planets and orbital distribution of planets in multiple-stellar systems. This project also determines the distributions of semi-major axis, albedo, size, mass and density of short-period giant planets.
CONFIRMING SMALL PLANETS AND MEASURING THEIR MASSES WITH TRANSIT TIMING VARIATIONS
Northwestern University, Evanston
I propose to analyze the transit times of Kepler planets to confirm the planet nature of small planets in systems with multiple planet candidates, to constrain or identify the presence of additional nontransiting planets in those systems, and to eliminate astrophysical false positive systems that can produce transit timing variations (TTVs). I propose augment my existing, fast TTV software to incorporate additional bodies and apply it to all suitable Kepler data to give mass measurements for the planets in TTV systems. With the data from the extended mission, these TTV studies, which have already confirmed most of the smallest Kepler planets, will be able to confirm yet smaller planets
and those with longer orbital periods---especially planets within and around the habitable zone of the host stars. I propose to study possible long-term correlations between the age of a planetary system and its orbital architecture and planet sizes. The proposed research specifically addresses one of the scientific goals of the Kepler extended mission: 'Characterize the size and orbital distributions of small planets (R<2.5REarth)' and contributes to the other two: 'Identify correlations between the characteristics of planetary systems and their host stellar systems'; and 'Provide a statistically significant determination of the frequency of Earth-size planets in and near the habitable zone of their host stars.'
VALIDATION OF SMALL KEPLER CANDIDATE TRANSITING PLANETS WITH BLENDER
Smithsonian Institution/Smithsonian Astrophysical Observatory
NASA'S Kepler Mission was launched with the goal of detecting extrasolar planets by the transit method, characterizing them, and determining the frequency of Earth-size planets in the habitable zone of their parent stars. A known challenge of transit searches is the fact that other phenomena unrelated to planets can produce photometric signals indistinguishable from those of a true planet. One example of such a false positive is a chance alignment with a background eclipsing binary (a "blend"), whose typically deep eclipses would then be diluted and reduced to planetary proportions. Confirming the true planetary nature of a candidate usually requires the measurement of the transiting object's mass. This can be done either by detecting the star's acceleration due to the planet, or by detecting departures from strict periodicity in the signals, caused by mutual interactions in systems with multiple planets. Confirmation by these means is not always possible or practical, either because the planetary masses are too small and/or the orbital periods too long to yield a detectable dynamical signal, or because the star is too faint to allow measurements with the requisite precision, or for other reasons. The only alternative is a probabilistic "validation" to demonstrate that the candidate is much more likely to be a transiting planet than a false positive. Without such validations we cannot be sure of the nature of the signals. This is the objective of this project, and the BLENDER procedure we propose to use is designed for that purpose.
BLENDER is a sophisticated light-curve analysis technique developed by the PI that systematically explores the wide range of possible blend scenarios and places strict constraints on the kinds of configurations that might mimic the transit signal for a given candidate. It does this by making optimal use of the shape information contained in the Kepler light curve, comparing it with the detailed shapes of blend light curves. The procedure incorporates additional, complementary constraints from follow-up observations by the Kepler Project and the community such as high-resolution imaging, high-resolution spectroscopy, color information, and the analysis of the motion of the flux centroids from the Kepler images themselves. In this way BLENDER is able to rule out a majority of blends. The ones that remain viable will be quantified statistically in this project through a suite of Monte Carlo procedures and informed estimates of the number density of stars in the direction of the target, the known frequencies of background eclipsing binaries, and other stellar and planetary properties. We will then proceed to estimate the false positive probability, and validate the planet if this probability is small enough.
Having concluded its 3.5-year nominal Mission, Kepler has entered its Extended phase in which the objectives are tightly focused on the smallest and most interesting planet candidates, those that are under 2.5 Earth radii in size. Because of their correspondingly small masses and the fact that most Kepler targets are faint, dynamical confirmation is currently not possible for the vast majority of these candidates. We propose to use BLENDER for their validation, with particular emphasis on the smallest ones most similar in size to the Earth that are in the habitable zone of their parent stars. These are among the most valuable of all of Kepler's candidates, but are also the most challenging to validate, and BLENDER is the only way to achieve this. The procedure has been used previously by the PI with great success to validate some of Kepler's most iconic results, including the first transiting Earth-size planets, and the first super-Earth-size planet in the habitable zone of a Sun-like star. This project relates directly to NASA's strategic goals and the science objectives of the Extended Kepler Mission to discover and study planets smaller than 2.5 times the size of the Earth.
CIRCUMBINARY HABITABLE ZONE PLANETS
San Diego State University
The P.I proposes to continue to discover and measure the properties of circumbinary planets (those orbiting a pair of stars) with emphasis
on the detection of smaller (R<2.5 Re) planets.
The P.I. has an outstanding record as a Kepler Participating
Scientist, having generated a range of products that have been used in key discovery papers (e.g. Kepler-9 and Kepler-11), and most importantly, leading much of the effort to detect and characterize circumbinary planets. The P.I. was the lead author of three significant Kepler discovery papers (ellipsoidal variations in HAT-P-7; the remarkable tidally pulsating binary KOI-54; and the discovery of the circumbinary planets Kepler-34 and Kepler-35). The latter paper appeared in the prestigious journal Nature,
generating significant positive international press attention for the Kepler Mission.
The proposed work is more tightly focused than the P.I.'s previous Kepler research, dealing exclusively with the detection and subsequent characterization of circumbinary planets. In particular, the search
for the larger planets will continue, using tools already in place
that rely heavily on visual inspection and
pattern-recognition. Because circumbinary planet transits are neither periodic nor of the same duration, simple automatic detection algorithms (such as the ones used in the Kepler pipeline) will not suffice for finding these planets. Visual scrutiny, however, has been very successful thus far, with all 7 known circumbinary planets having been found in this way. The proposed refinement of the tools that assist our visual search method will allow easier and more complete detection of small planets (~3 Re). For even smaller planets (< 2.5 Re),
visual inspection will generally not work. Thus we are developing automated tools that take the binary system's geometry and kinematics into account and allow for detection at thresholds far below what the eye can detect as a single event. The powerful eclipse and transit modeling software ``ELC'' will be modified to carry out this
search. ELC will be used to measure properties of both the binary system and the planets in it. As part of this project, ELC will be improved and enhanced and will be made available to members of the Kepler Working Groups.
The P.I. of this proposal is experienced and well versed in the skills relevant to this research, as evidenced by the Kepler papers he has
led and contributed to, and the Working Groups he actively
participates in. As the proposed investigation is directly related to
the objectives of the Kepler Mission and to NASA's interests as a whole, the work will help Kepler achieve its primary goals of the
Extended Mission. In particular, this proposed research is relevant due to the tendancy of circumbinary planets to be placed near the habitable zones of their parent stars (two out eight or 25% of the cases so
far). This is because of the combination of observational bias and
planet migration. The detection and characterization of circumbinary planets is therefore not only important for a complete census of exoplanets, but also for the determination of Eta-Earth itself, which is the fraction of planets that are sufficiently Earth-like to
potentially harbor life.
Questions concerning Kepler's science opportunities and open programs, public archive or community tools? Contact us via the