Terrestrial Planets over the Next Decade

By Don Pollaco, Warwick University, UK

There can be no doubt that NASA’s Kepler mission has been a resounding success. In particular, much of what we know about rocky planets has come from this mission. After saying that, ground-based radial velocity surveys had already indicated the existence of super-Earths—a class of planet not found in our solar system (ignoring Planet 9!), and the first exo-rocky planet discovered was found through the French-ESA CoRoT mission (Corot-7b). The list of “firsts” from Kepler is truly amazing:

  1. Planetary size distribution
  2. The commonality of multi-planet systems
  3. The application of transit timing techniques to derive planetary masses and the recovery of unseen components
  4. The detection and modelling of the first circumbinary systems
  5. The diversity of low mass planets
  6. The evaporation and breakup of small planets

and so, while some results are less good, e.g. estimate of eta-Earth (number of habitable zone planets per star) and the masses of low-mass planets, Kepler’s place in history is assured. To me though, Kepler’s greatest result is really the ubiquity of exoplanets; specifically small planets.

Almost as impressive has been Kepler’s contribution to the proving of stellar asteroseismology. Whilst these techniques had been applied to the Sun and individual stars, Kepler has been used to derive stellar parameters for hundreds of stars at a level never before achieved en masse.

Kepler Small Planets

Kepler has given us a tantalizing first glimpse of the small/rocky planet population and some of the results have been absolutely awesome (Figure 1). For example, masses for the fantastic seven planet Kepler-11 system have been derived through modelling the gravitational perturbations giving rise to the transit time variations, and show these planets are much bigger than expected for their masses—maybe they are mini gas planets or have fluffy extended atmospheres.

Figure 1. The known small planet population in the mass-radius plane (x-axis’ units are Earth masses) compared to different compositions and compared to rocky planets in our solar system There is far more diversity than originally expected.

At the other extreme is Kepler-10c. Kepler-10b (mass 3.33ME, radius 1.47RE, Period 0.84d) was well known as Kepler’s first rocky planet, and spectroscopic observations from the ground with HARPS-N on La Palma not only confirmed this, but also detected the stellar reflex motion from the long period Kepler-10c component. Surprisingly, the mass turned out to be 17.2ME, but the Kepler (2.3RE) radius suggested we were still most likely looking at a massive super-Earth. Given that we struggle to understand the internal structure of the Earth, we are quite mystified to explain that of Kepler-10c. These results and others lead us to believe the small planet population is much more diverse than we originally believed.

When Kepler was being designed, it was generally agreed that there would be little variation of compositions in this population, so that from a measurement of planetary radius its mass could be directly inferred. Consequently, it was assumed that there would be little need for follow up observations. Kepler showed us the need to determine the planetary mass directly.

Radial Velocity Surveys: Masses of Kepler Planets

Since the first discovery of a planet around a Sun-like star (Mayor and Queloz 1995), radial velocity surveys—searching for the reflex motion induced in the star—were often the most efficient discovery technique. Compared to the transit method’s strict requirement on the orbital geometry, radial velocity detection is far more lenient. However, without knowledge of the orbital inclination to our line of sight, all we can determine is the planetary minimum mass. Basically, from radial velocity information alone we can learn about the planetary orbit, but essentially nothing about the planet itself.

Radial velocity information is most useful when it is used alongside transit data. With radius and, most importantly, orbital information coming from the light curve modelling, solutions of the equations of motion can give an accurate planetary mass. Thus, we can get an accurate estimation of the bulk density/composition of a planet. What should be emphasized here is that to derive the planetary mass and radii requires better accuracy in the stellar parameters; in fact, for the best transit light curves knowledge of the host star is often the factor limiting that of the planetary component. The study of exoplanets has led to a renaissance in stellar research and especially the proving of asteroseismology.

The low brightness of the Kepler field stars and the prevalence of small planets is a double whammy for our studies of the masses of small planets—the small reflex motion and lack of stellar photons make mass measurements at best somewhat challenging. So while the Kepler photometry has produced highly accurate relative radii, even the brightest Kepler host stars are challenging targets for radial velocity work. It is ironic that the planets with the most accurate accepted masses are massive planets found from ground-based transit surveys such as SuperWASP or HAT.

As a consequence, researchers have developed our ability to model gravitation perturbations detectable through transit timing variations (as noted earlier) and this is how most Kepler planetary masses have been determined. This has the advantage that they can be derived from the light curve alone and with apparently small errors, but is possible for only a small fraction of the planets. There is still some controversy surrounding the use of masses derived in this way and maybe more importantly in the quoted errors. Maybe this will improve in the future.

However, one of the big lessons from Kepler (and the ongoing K2 surveys of course) is that we need a host star population as bright as possible so we can derive masses, make planetary atmosphere observations, etc. So given this, what does the future hold?

Looking forward—the Transit Roadmap

For exploring the inner parts of solar systems, and in particular the habitable zones, for the next 10–15 years it is likely that transits of bright stars that allow radial velocity observations to be made will dominate (Figure 2). That’s not to say that other techniques and regions of the parameter space will not be important—they will. For example, with SPHERE and GPi we are taking our first steps with dedicated and optimized instruments capable of direct planet detection—at least of luminous, young and massive planets. JWST may also be capable of this. Gaia and various microlensing space missions such as WFIRST (~2025) or EUCLID (2021) will allow us to statistically explore the outer parts of solar systems.

Figure 2. The funded transit roadmap showing facilities that are used for detection and bulk characterisation. Some of the facilities here can also be used for atmospheric characterisation.

In terms of transit experiments, we have a crop of ground-based experiments—including the new NGTS, and the re-tasked Kepler K2 surveys. While still at an early stage, NGTS is proving capable of routinely detecting dips which could be due to Neptune-sized objects. Various experiments have been deployed targeting M dwarf stars, where the low intrinsic brightness and small star size mean that ground based photometry would even be capable of detecting Earth-sized planets in orbits of a few days; corresponding to the habitable zones of the feeblest stars.

In general, finding small planets (smaller than Neptune, say) in habitable zones is a difficult task and is best done from space. This not only avoids limitations in photometric accuracy from the Earth’s atmosphere, but also the interruptions caused by the day/night cycles. Even still, as we push to higher and higher accuracies stellar activity becomes a bigger issue with less stars being suitable for radial velocity work. However, understanding stellar activity is an area of much research and there is hope that small radial velocity signals will be detectable against the activity signal in the future. Nonetheless, we are fortunate that both NASA and ESA have recognized the need for new surveys and we have a series of missions that have transit detection at their heart.

CHEOPS is due for launch in 2017 and is ESA first “Small” satellite. This Swiss led mission is designed to look at objects one at a time. CHEOPS has two science drivers:

  1. The follow up of known planets discovered from radial velocity surveys and especially those targets thought likely to transit, and
  2. High accuracy light curves of transits from other surveys, notably NGTS.

So, while CHEOPS is not a survey instrument it will produce extremely accurate photometry of known planets and hence bulk densities. CHEOPS also has many other potential uses such as monitoring transits for timing variations etc.

NASA’s Transiting exoplanet Survey Satellite, TESS, will be launched around the end of 2017. TESS will be orientated into a highly eccentric and inclined orbit which reaches almost to the lunar orbit. For most of the 27 day orbit the satellite will be far from the Earth, enabling accurate photometry. The clever orbit and observation strategy results in sections of the sky being monitored for 27 days before moving to the next section. These sections overlap at the Ecliptic poles and so a small region is monitored for as long as ~1 year. Given this, it is likely that TESS will find many single transiting systems which would benefit from CHEOPS observations.

TESS is aimed at surveying the nearest and brightest stars (mag) and is therefore preferentially examining M dwarfs. These low luminosity stars have habitable zones close in (periods as short as a week or so for the coolest objects). Furthermore, as these stars are quite small the detection of small planets can be achieved easier. Being extremely red objects they are likely to be ideal targets for the JWST and it is likely that the first observations of the atmosphere of a habitable zone planet will come from TESS.

The Future: ESA’s M3 PLATO Mission

Over the years, there has been a succession of transit survey concepts studied by ESA, but in 2014, PLATO was finally selected as the “Medium 3” mission with launch date in 2024. PLATO was designed from the outset to characterize habitable zone rocky planets with Sun-like host stars that are bright enough for observation with the new generation of radial velocity spectrograph’s such as ESPRESSO at ESO’s Very Large Telescope (VLT) in Chile. PLATO is a multi-telescope system which provides a huge field of view (>2,200 square degrees—about 20 times that of Kepler) with excellent sensitivity and it will be stationed in a thermally stable environment at the L2 point, several million kilometers from Earth. While the dynamic range is from 4–13 magnitude, most of the interesting science will be for stars with magnitudes allowing asteroseismic characterization of accurate stellar parameters including their age. Figure 3 shows the predicted rocky planet catch for stars that can be fully characterized through asteroseismology compared to those from Kepler and TESS.

Figure 3. Simulations of PLATO transit signal detection performance (in green) for super-Earth planets (less than 2RE) for stars brighter than 11 mag, hence with RV follow-up and host star asteroseismology possible. For comparison, Kepler results are shown (in blue, Fressin et al. 2013) and expected yields for TESS (in red) assuming 27 day observing coverage per field and 2% of the sky observed for 1 year

The time requirement for the ground-based follow up will be almost entirely driven by the smallest, longest period, planets and will represent a significant investment by the astronomical community. For some of the multi-planet systems, masses will also be available from models of any transit timing variations. While PLATO will certainly produce lots of interesting and no doubt unique systems and maybe even moons, rings, etc., the real PLATO reward will be the database of uniformly characterized planetary systems that can be used for future theoretical and observational experiments.

By the end of the next decade, we will have fully characterized hundreds of systems containing rocky planets. Many of these will be bright enough to have their atmospheres examined with the instruments of the day. The database of PLATO systems with known ages will allow us to take the first steps in comparing the observed planet population with theoretical studies, hence throwing light on the important processes that are sculpting the architectures of these systems. In many ways PLATO can be considered the descendant of Kepler and indeed one of the options for PLATO is to revisit the Kepler field to examine the variations in transit timing variations accumulated after a delay of some 15 years.

We live at a very fortunate time. Kepler has opened the window and shown us some of the landscape. The new missions will enable us to make great gains in comparative planetology so that we can understand our place in the Universe.

Dr Don Pollacco, astro-physicist and planet hunter from Queens University, Belfast.

About the author. Don Pollaco is a Professor of Astronomy at Warwick University, UK. He was awarded his PhD in 1990 from St Andrews University. From 1990–1995 he worked as a PDRA and then lecturer at St Andrews and Liverpool John Moores Universities. Between 1995-2000 he was based at the Isaac Newton Group of Telescopes (ING) La Palma. His PhD and postdoctoral studies were concerned with evolved stars and binary systems. From 2000–2012 he worked at Queens University Belfast developing the exoplanet group there. During this period he led the development as PI of the SuperWASP Project and the initial development of NGTS. Since arriving at Warwick in 2012 he has been the Science Coordinator for ESA’s PLATO mission.



5 thoughts on “Terrestrial Planets over the Next Decade”

  1. To play an armchair critic, the recent high precision – as it were – objective classification of planets predicts the Kepler-11 planets fluffiness well. [“PROBABILISTIC FORECASTING OF THE MASSES AND RADII OF OTHER WORLDS”, Jingjing Chen and David Kipping; http://arxiv.org/pdf/1603.08614v1.pdf ] (But instead not Kepler-10 and other high mass, low radius planets.)

    On the other side it shrinks eta-Earth considerably, since the mass cutoff is now 2 Earth masses (or in radius ~ 1.2 Earth radii).

    On the flip side it makes our system perfectly ordinary. Earth is now the super-Earth we have been looking for. And since Saturn is more likely a giant Neptune than a small Jupiter we have a majority of them – with or without Planet 9 – as we should.