About Follow-up Observations
For all candidate transit cases, complementary follow-up observations are made to confirm that the transits are due to planets and to learn more about the characteristics of the parent stars and planetary systems.
Of Planetary Candidates and Ground-based Follow-up Observations
by Geoff Marcy, August 2010
The Kepler Science Team has been quite busy analyzing all the data Kepler has collected to date. There are many planetary candidates that the team must assess and verify as a true planet or a false signature. The Kepler Mission has a primary goal of measuring the brightnesses of 100,000+ stars with unprecedented precision. If an Earth-sized planet orbits in front of a Sun-like star, the blocking of the starlight causes the star to dim over and over, allowing Kepler to detect the planet. The bigger the planet, the more light it blocks, allowing the Kepler team to determine the diameter of the planet.
Such a discovery is called a "planet candidate" because it has not yet been verified as a true planet. If it isn't a planet, why does the star appear to dim, over and over? One nagging possibility is that behind the star are two additional stars that orbit each other, eclipsing themselves when they cross in front of each other. Such a background "eclipsing binary star" would dim once per orbit, mimicking the dimming signature of a planet. In that case, the "planet candidate" would not be a planet at all. We would be fooled. With hundreds of planet candidates emerging from Kepler, as announced in June 2010, the challenge of weeding out the eclipsing binary stars from the bona fide planets is a daunting task.
The Kepler Mission assesses these false planets with its "Follow-up Observing Program" (FOP) designed to distinguish true planets from the imposters. The FOP consists of 15 team members, each with different expertise in different methods of identifying pesky eclipsing binary stars. The first approach is to obtain high quality pictures of the field of stars around the main stars. The FOP takes images of the field surrounding Kepler stars using a 1-meter telescope at Lick Observatory, 2-meter telescopes operated by the Las Cumbres Observatory, and even the Keck telescope in Hawaii for the highest priority stars. So far the FOP has obtained images of over 400 Kepler stars. To obtain more detailed images, the FOP uses the adaptive optics system on the 5-meter Palomar telescope and the MMT telescope on Mt. Hopkins. Adaptive optics can take pictures capable of detecting any eclipsing binary located exceedingly close to the star. Any star showing no eclipsing binary by adaptive optics is unlikely to have one still hiding, by chance, behind the glare of the star.
Another way the FOP weeds out eclipsing binary star is by taking a "reconnaissance" spectrum of the star. Using telescopes with 3-meter diameter mirrors at Mt. Hopkins, McDonald, Lick and the Canary Islands observatories, the light of a star can be spread out with a spectrometer into the colors (i.e. wavelengths) of which the light is composed. Eclipsing binary stars reveal themselves by the two distinct rainbows of colors they each produce, painted one on top of the other, but displaced from each other by the Doppler Effect. The Doppler effect is what allows a police officer to detect a speeding car on the highway. An eclipsing binary star would exhibit two different speed readings in its spectrum of colors, betraying the existence of two orbiting stars whizzing around each other. The reconnaissance spectrum also permits the FOP to determine how many "spectral lines" the star has and how sharp those lines are. Spectral lines are light at a particular frequency, just as a piano has notes of a particular frequency. The spectral lines come from atoms in the star’s atmosphere, and a large number of lines and their sharpness offers a chance to measure the Doppler effect with extreme precision, measuring the speed of the star to within human walking speed. Indeed, the highest priority planet candidates (those nearly Earth-sized) are then observed with the Keck telescope in Hawaii with its "HIRES" spectrometer, with the goal of measuring the Doppler Effect with extreme precision of one meter per second. A planet will pull gravitationally on its host star, yanking it to and fro, and such motion of the star can be detected by the changing Doppler effect. Thus, the planet candidate can be certified as a bona fide planet by detecting the orderly "wobble" of the star as seen in the continuously oscillating Doppler effect.
Moreover, Kepler FOP measures the amount of Doppler effect of the star. The more massive the planet, and greater the gravitational tug on the host star. So the FOP can use the amount of Doppler effect of the star to measure the mass of the planet. This is a glorious achievement, as the dimming measured by Kepler gives us the planet's diameter, while the Doppler effect gives us the planet's mass. The beauty is that we can directly determine the density of the planet, which is its mass divided by its volume. Planets like Earth have the high density of rock, about 5 grams per cubic centimeter, while gaseous planets like Jupiter have much lower densities of about 1 gram per cubic centimeter. The FOP measurement of the planet's density allows the Kepler team to distinguish true rocky planets, like Earth, from gaseous planets, like Jupiter.
The FOP has its work cut out for it. With hundreds of candidate planets from Kepler, there are thousands of imaging and spectroscopic observations that must be made. The FOP scientists are incredibly hard-working, spending hundreds of long nights at telescopes around the world. The two goals of the FOP are crucial to the Kepler mission, namely to weed out the eclipsing binary stars that mimic planets and to measure the masses of the credentialed planets. So far, many hundreds of images and over 700 spectra have already been taken of the Kepler planet candidates. A dozen eclipsing binaries have indeed been found, cleansing them from the planet candidates that Kepler continues to pursue. In the end, Kepler plus the FOP will provide the hard data that secure the discovery of Earth-sized planets around other stars.
[end of section written by Geoff Marcy]
Elimination of non-Planetary Candidates
White dwarfs have radii similar to that of the Earth and can produce a light curve that mimics a transit by a terrestrial planet. Also, a grazing eclipse by a stellar companion can mimic a transit. In both cases, the stellar companions have masses that are usually >>100 MJ. The radial-velocity variations induced by such companions are typically >>1 km/s and can easily be detected, such as with the SAO Digital Speedometer available to Latham (1992). Thus, transit candidates where the companion is stellar can be eliminated. Similarly, brown dwarfs (10 MJ < M < 100 MJ, R~1 RJ) can be distinguished from giant planets.
Detection of Giant Planet Companions
Radial velocity data also serve to explore or delineate the structure of the planetary systems by detecting giant planets not seen in transit or by reflected light. Normally, such data only yield the quantity, M sini, where M is the planetary mass and i the orbital inclination. The presence of transits implies that i ~90 º, which establishes the mass of the planet. By sampling the Doppler shift a few times during an orbit, the orbital parameters of those planets showing transits are completely determined and the planet mass is established to within 3%. The mass (from spectroscopy) and planet radius (from photometry) yield the density of the planet.
Kepler team member Marcy has developed the technique to measure velocities to 3 m/s on Keck (Marcy, et. al. 2000). Kepler team members also have institutional access to the proven Hamilton Echelle (Basri) and to new spectrometers on the Hobby-Eberly Telescope (Cochran) and the MMTO (Latham) all with <10 m/s capability.
Stellar Mass, Size and Metallicity
For those stars found to have planets, high-resolution, high signal to noise ground-based spectroscopy is performed to clearly establish the spectral type and luminosity class. Stellar evolution models are used to estimate the mass, radius and metalicity of the parent star (e.g., Mazeh et al. 2000). The stellar mass is required to calculate the semi-major axis of the orbit and the stellar radius to calculate the planet's size. The frequency of planets with respect to spectral type and other stellar characteristics can then be established. There is now some preliminary evidence that planetary systems may be found more often around stars whose spectra show high metalicity (Gonzalez 1997, 1998) or depleted lithium (Cochran et al. 1997). Firmly establishing the existence of any such trends provides extremely valuable constraints on models of planetary system formation. Since the Kepler Mission FOV is along a galactic arm at the same galacto-centric distance as the Sun, the stellar population sampled is indistinguishable from the immediate solar neighborhood. Thus, these results can guide target selection for future planet searches by SIM and TPF for planetary systems in the very near solar neighborhood.
Additional constraints on the parent star's radius and other properties are obtained from p-mode oscillations (e.g., Brown and Gilliland 1994) using Kepler's 1-min sampling mode. In the Sun a series of modes with periods of about 3 min and equal spacing in the frequency domain are excited to a level of about 3 ppm in white light. This level of precision requires the detection of at least 1012 photons. Kepler provides the necessary photon levels in one month for the 3,600 dwarf stars brighter than mv=11.4 in the FOV. Two members of the Kepler team have considerable experience and scientific interest in this type of stellar seismology that can be done with Kepler (Gilliland and Brown)
Where Kepler has found a planet, SIM can be used to find the masses (if they are jovian or larger) or set upper limits to the masses of the detected planets. SIM data complement the Doppler data since Doppler spectroscopy can not be used for stars hotter than F5. SIM can also be used to search for additional giant planets in wide orbits. For example, SIM can detect a 0.4 MJ planet in a 4-year orbit around a solar-like star at a distance of 500 pc.
Using parallax, the geometric distance to a system at 500 pc can be determined to 0.2%, placing all inter-comparisons of discovered planetary systems on a very firm foundation. Two of the Kepler team members (Latham and Boss) are also members of the SIM key projects proposal team for planetary systems studies. The single-pointing field-of-regard of SIM of 15 deg. is nearly identical to that of Kepler and is a highly efficient use of SIM observing time.
From infrared measurements with SOFIA or NGST any IR excess or lack thereof will indicate the fraction of systems having terrestrial planets that are or are not embedded in large amounts of extrasolar zodiacal dust.