KELT - The Kilodegree Extremely Little Telescope

The KELT project developed out of research I began in 2002, working with Andy Gould at Ohio State. That work turned into the paper that describes the optimal telescope configuration for an all-sky transit survey. In short, we found that a small-aperture, wide-field telescope would be the best instrument to detect planetary transits of bright stars. Based on the calculations from the paper, we decided to build such an instrument. That became the KELT-North telescope, which is based at Winer Observatory in southwestern Arizona.  We published the basic configuration and operation of the telescope in the KELT-North instrumentation paper, and described the results from the KELT-North commissioning survey in a paper describing the variable stars and transit candidates found in that effort.

After taking a postdoctoral position at Vanderbilt University, I began building the KELT-South Telescope - a twin of KELT-North. KELT-South is located at Sutherland, South Africa, and is described in the paper "The KELT-South Telescope".  The two KELT telescopes are separately operated: KELT-North is run by Ohio State, with participation by Vanderbilt and Lehigh.  KELT-South is run by Vanderbilt, with participation by Lehigh, Fisk University, and the South African Astronomical Observatory (SAAO).  Both telescopes are quite similar in construction, and operate coordinated transit surveys. 

KELT has benefited from the dedicated efforts of both Keivan Stassun and Scott Gaudi, who have contributed time, money and effort to keep the survey going, and especially Robert Siverd, who has worked at both Ohio State and Vanderbilt, building the KELT reduction pipeline, managing the KELT-North telescope, and working with me to make the project a success. 

KELT has now discovered 26 transiting companions to bright stars.  Here are some of the highlights:

  • KELT-1b is a 27 MJ, 1.1 RJ transiting brown dwarf in a 1.2-day orbit around a V=10.7, F5 star. It is the shortest period and brightest transiting brown dwarf discovered, and is only the second definitively inflated brown dwarf known. It is described in "KELT-1b: A Strongly Irradiated, Highly Inflated, Short Period, 27 Jupiter-mass Companion Transiting a mid-F Star".
  • KELT-2Ab is a 1.5 MJ, 1.3 RJ mildly inflated hot Jupiter in a 4.1-day orbit transiting a slightly evolved V=8.77, F7 star. At the time it was the ninth brightest transiting planet, and the third-brightest one discovered by a ground-based survey. The evolutionary state of the star means that this exoplanet has one of the best measured ages of any known exoplanet. The host star also has a common proper motion M-dwarf binary companion (KELT-2B) that may be the cause of KELT-2Ab's orbital location. It is described in "KELT-2Ab: A Hot Jupiter Transiting the Bright (V=8.77) Primary Star of a Binary System".

  • KELT-6b is a hot Saturn with 0.43 MJ and 1.2 RJ, on a long (P=7.8-day) eccentric (e=0.22) orbit.  It is transiting a metal-poor ([Fe/H] = -0.28), V=10.4, slightly evolved F9 star.  This planet is especially notable for having the same surface gravity and insolation as the best-studied exoplanet, HD209458b, although its host star (and likely the planet too) has a metallicity lower by 0.3 dex.  Furthermore, radial-velocity observations show that there is an additional, distant orbital companion in the KELT-6 system, that is also likely substellar, described here.  KELT-6b was announced at the June 2013 meeting of the American Astronomical Society, and published by Karen Collins of the University of Louisville, who was one of the early partners in followup observations of KELT transit candidates.  The paper is "KELT-6b: A P ~ 7.9 Day Hot Saturn Transiting a Metal-poor Star with a Long-period Companion".
  • KELT-7b is a hot Jupiter with 1.3 MJ and 1.5 RJ transiting a bright V=8.54 F star on a 2.73-day orbit.  At the time its host star was the fifth most massive (1.53 MSun), fifth hottest (6790 K), and the ninth brightest star known to host a transiting planet.  We have measured the spin-orbit alignment of the system, and the planet is well-aligned with the star.  The planet discovery is described in "KELT-7b: A hot Jupiter transiting a bright V=8.54 rapidly rotating F-star", by Allyson Bieryla of the CfA, who has been another major contributor to KELT followup and confirmation.

  • KELT-9b is the hottest planet ever discovered.  It is a little less than 3 MJ and 2 RJ, and in a 1.5-day orbit.  Its host is a 10,000 K A-star with V=7.6, and the planet's equilibrium temperature is over 4000 K.  That makes the planet hotter than most M stars, and so hot that molecules cannot exist on the dayside of the planet.  The discovery was published in Nature in June 2017 by Scott Gaudi of Oho State as the lead author: "A giant planet undergoing extreme-ultraviolet irradiation by its hot massive-star host".  This planet made the cover!

  • KELT-11b is one of the least dense planets ever discovered.  It is less massive than Saturn at 0.2 MJ, but almost twice the size of Jupiter, with a bulk density similar to Styrofoam.  At the time of discovery it had the brightest transit host star in the southern sky at V=8.0.  I published the discovery in "KELT-11b: A Highly Inflated Sub-Saturn Exoplanet Transiting the V = 8 Subgiant HD 93396".

All of these discoveries are extremely exciting, and are exactly what KELT was built to do.  So far we have published fourteen planets from KELT-North, four planets from KELT-South, and two jointly discovered by both telescopes.  One more planet discovery is in press.

Lehigh University Joshua Pepper - KELT-SouthHere is a photo of the KELT-South building at Sutherland. The telescope is fully automated - no human interaction is involved in regular operations. Each night, the control computer checks the weather, opens the roll-off roof, and starts the telescope scanning a series of fields across the sky.





Predicting the Yields of Transit Surveys

Before designing and constructing a survey for planetary transits, it is important to understand the factors that will increase or decrease the number of transits one expects to find. Simple calculations that consider the total numbers of stars observed, the frequency of planets, and the transit probability dramatically overestimate the number of expected transits by ignoring significant issues. A full, consistent model for a survey must account for many more effects, and is performed either as forward modeling, using assumed distributions of the stellar and planetary properties to statistically predict the ensemble properties of the target stars and the detection efficiency of the survey as a whole; or reverse modeling, in which one uses the known properties of an observed sample of stars to determine the survey yield. For an introduction to these topics, see the conference proceeding, "Statistics and Simulations of Transit Surveys for Extrasolar Planets" by Scott Gaudi. A full example in great detail can be found in "Predicting the Yields of Photometric Surveys for Transiting Extrasolar Planets", which is probably the best reference paper for this topic.

While performing an all-sky survey is one strategy for a transit search, another option is to focus on a single population of stars that share some fundamental property, to learn something about how planet frequency varies within a somewhat homogeneous population. That is the strategy behind transit surveys of star clusters. The basic statistics of such surveys are fundamentally different than all-sky surveys, and in "Searching for Transiting Planets in Stellar Systems" Scott Gaudi and I outline the formalism needed to conduct such modeling. In "Toward the Detection of Transiting Hot Earths and Hot Neptunes in Open Clusters" we apply the methods from the first paper to a number of specific star clusters, with special attention to the potential for the discovery of low-mass planets. The analysis in that paper was updated in "On the potential of transit surveys in star clusters: Impact of correlated noise and radial velocity follow-up" by Suzanne Aigrain and Frederic Pont to include consideration of red noise and follow-up issues.


The Transiting Exoplanet Survey Satellite (TESS) has been selected by NASA as a SMEX mission to be launched in 2018.  It is intended in many ways as a successor to the Kepler mission.  While Kepler has been fantastically successful, it observes a small fraction of the sky in the constellations Cygnus and Lyra, and so most of the stars it observes are too faint for detailed characterization.  TESS plans to observe all the bright (4 < V < 12) dwarf stars across the entire sky, discovering all the transiting planets in the solar neighborhood, which will be the focus of decades of careful study for years into the future.  Predictions of the TESS yield can be found in "The Transiting Exoplanet Survey Satellite: Simulations of Planet Detections and Astrophysical False Positives".  I like to think of TESS as "KELT in space".  I am a collaborator of the TESS mission, and co-chair of the TESS Target Selection Working Group.


The Large Synoptic Survey Telescope (LSST) is the future of astronomy.  It is a planned 8.4-meter telescope to be located in Chile with a 3.2 Gigapixel camera that will observe the entire visible sky in multiple filters once every three nights for 10 years.  It will generate roughly 200 Pbytes of data (A Pbyte is a million Gbytes) and will impact every area of astronomy.  I am a member of two LSST Science working groups: the Stars, Milky Way and Local Volume group, and also the Transients and Variable Stars working group, in which I am the co-chair of the Transiting Exoplanets subgroup.  My work on eclipsing and variable stars overlaps with the focus of both groups.

I have been working with Vanderbilt graduate student Michael Lund to determine whether and how LSST will be able to discover transiting exoplanets.  In our first paper, "Transiting Planets With LSST. I. Potential for LSST Exoplanet Detection", we showed that LSST will definitely be able to detect transits, especially for Hot Jupiters.  However, the detection will be difficult, as there will be very large numbers of false positives.  Furthermore, nearly all detections will not be possible to dynamically confirm via RV observations, so we expect that analysis will have to be done statistically. 

Eclipsing Disks

Starting in 2012, Joey Rodriguez, a graduate student at Vanderbilt, started working with me and Keivan Stassun to explore other phenomena that could be detected with data from KELT.  Joey started looking for large transiting objects like circumprotoplanetary ring systems, inspired by the discovery of J1407.  While we did not find any planetary rings, we did find several cases of stellar disks eclipsing their host stars.  Joey published several discoveries, describing disk systems in RW Auriga, V409 Tau, and DM Ori.

Furthermore, Joey discovered, in conjunction with astronomer Sumin Tang, the longest period eclipsing binary system known, called TYC 2505-672-1, with an orbital period of over 69 years.   That discovery is now listed in the Guiness Book of World Records here.


I became involved with the NASA Kepler mission shortly after the satellite launched in 2009.  I was a member of the Kepler Users Panel for a year, advising the Kepler team on how to best coordinate and communicate with the larger astronomical community.  I worked with members of the Kepler Guest Observer office - including Karen KinemuchiMartin StillTom Barclay, and Mike Finelli - to write a paper for those unfamiliar with Kepler data on how to use the data products supplied by Kepler.  That paper is "Demystifying Kepler Data: A Primer for Systematic Artifact Mitigation".  I am also a member of the Kepler Eclipsing Binary Working group.

I have also been working with Fabienne Bastien, Keivan Stassun, and Gibor Basri to analyze the fundamental variability of stars.  Before the Kepler mission, no instrument was able to measure the intrinsic variability of middle and lower main sequence stars, except for the sun.  The Kepler mission changed that, and our team has determined that by monitoring the intrinsic variability of sunlike stars on short (8-hour) timescales, we can determine the surface gravity of these stars to exquisite precision.  This discovery has been published by Nature as "An observational correlation between stellar brightness variations and surface gravity", and promises to be a powerful new tool for understand stellar astrophysics.

Non-transiting Planets

I am a member of the Sloan Digital Sky Survey-III, which consists of four projects, including the MARVELS and APOGEE surveys.  MARVELS is a multifiber spectrograph designed to detect exoplanets using the radial-velocity method.  I became involved with MARVELS early on, taking the lead on the target selection process at Vanderbilt. 

I am also involved in several projects associated with the APOGEE survey, a multifiber infrared spectrograph designed to map out populations of stars throughout the Milky Way, determining their kinematics and chemical compositions.  The APOGEE spectrograph has proven to be so sensitive that it is able to detect exoplanets via the radial-velocity method, as well as to discover many binary star systems.

Eclipsing Binary Stars

Any comprehensive time-domain photometric survey will discover large numbers of eclipsing binary stars.  These systems are of special interest because eclipsing binaries (EBs) allow us to use the mutual eclipses and gravitational interactions to determine the fundamental properties of the stars in those systems.  That is especially interesting when the binaries contain stellar components whose nature we do not fully understand, including low-mass stars, certain types of pulsating stars, or young stars.  I am working with a number of collaborators to examine particular EBs from KELT.  I also worked with Andrej Prsa and Keivan Stassun to determine what information the LSST project (see below) will be able to derive from the millions of EBs that it will discover: "Expected Large Synoptic Survey Telescope (LSST) Yield of Eclipsing Binary Stars"


My time at Vanderbilt made me aware of how crucial the science of astroinformatics is to the more data-driven discipline that astronomy is becoming.  One aspect of that transformation is that we need more tools that are nimble, flexible, and simple to use, in order to view and manipulate large data sets.  I am proud to have participated in the development of Filtergraph, a tool built by Daniel Burger.  Filtergraph is a remarkably useful online tool for uploading data and graphing it in multiple dimensions with fully customizable filtering and display options.  I highly recommend that anyone who frequently finds themselves replotting the same data set in many different ways to make use of this free resource.


Joshua Pepper  |  413 Deming Lewis Lab |   16 Memorial Drive East
Bethlehem, PA 18015  |  |  (610) 758-3649