Sunday, March 27, 2016

The Nearest 20 Parsecs

Figure 1. Familiar bright stars and selected exoplanetary host stars within 20 parsecs (65 light years) of our Sun. This stylized diagram illustrates the perspective of an observer located in the north celestial hemisphere, in the direction of the constellation Draco. Approximate stellar spectral types are coded. Names of host stars appear in turquoise; names of stars without known planets appear in violet.

With a stellar population in excess of 100 billion and a diameter of at least 30,000 parsecs (100,000 light years), our Milky Way is a good-sized spiral galaxy. Astrophotography has unveiled some of its most spectacular features, from the claw-like filaments just north of Sagittarius A* (our resident black hole, about 7900 parsecs away) to the billowing clouds and blazing young stars of the Carina Nebula (2300 parsecs away) and the delicate filigree of supernovae remnants such as Simeis 147 and Vela (920 and 250 parsecs away, respectively). Yet even as these panoramic views of our local universe swim into focus, other details remain fuzzy, especially when we turn to the question of potential planets orbiting all those billions of stars.

Three decades of searching, using several different techniques, have returned data on 2098 confirmed extrasolar planets in more than 1300 different star systems. The most complete and broadly comparable information continues to flow from radial velocity (RV) and transit searches. Apart from two outliers (SWEEPS-04 and -11, both in the Galactic Bulge), the most remote star with planets detected by either method lies at a distance of 3200 parsecs. That’s more than 10% of the diameter of the Galactic Disk.

But if we take a closer look at the combined sample of RV and transiting planets available in early January (n = 1867), stellar distances shrink fast. Searches of the Extrasolar Planets Encyclopaedia (EPE), the Kepler Discoveries Table, and various discovery papers found me distance estimates for just 713 stars in that sample. Although they range from 3 to 3200 parsecs, 50% are closer than 90 parsecs, and 70% are no farther than 200 parsecs. The size of those two fractions exposes a critical deficiency in our picture of extrasolar planets: the subset of robustly characterized systems skews sharply in favor of the nearest stars. Even though more than 1000 transiting planets have been confirmed by the Kepler Mission, fewer than 10% of their stellar hosts have distance estimates. The ones with estimates typically reside between 200 and 2700 parsecs.

Evidently our knowledge of transiting as well as RV planets has gaps of near-cosmic proportions.

If we want to grasp the true diversity of extrasolar planets and system architectures, we could do worse than examine a clearly delimited sample of well-constrained systems. I suggest that the best place to find them is the volume of space within 20 parsecs, where diversity is balanced by precision of observation.
the immediate solar neighborhood
Figure 1 represents a sphere centered on our Sun with a diameter of 40 parsecs. Although this region is devoid of O- and B-type stars – the rarest and most massive species in our Galaxy – all of the more common varieties of main sequence and evolved stars are represented. The brightest nearby objects are young blue-white stars of type A, such as Vega and Sirius, and older red giants, such as Pollux and Aldebaran. Many bright stars in our immediate neighborhood are binary or multiple. For example, both Sirius and Procyon have tiny white dwarf companions that were born as B stars and then rapidly bloomed and withered into their present state. Capella, the third brightest star in northern skies, is actually a quadruple system, consisting of a pair of tightly bound red giants accompanied by a pair of red dwarfs at a separation of 10,000 astronomical units (AU). More than a dozen exoplanetary host stars in our immediate neighborhood are also members of binary or higher-order multiples.

Google couldn’t find me an authoritative estimate of the full stellar population within 20 parsecs, but I did locate some approximate counts of G-type stars in this space (about 130). A rough extrapolation from available sources (Turnbull 2013, RECONS 2012) would suggest a population of 1500-2000 stars inside our 20-parsec radius, with about 75% classed as M dwarfs. (If anyone reading this knows of an authoritative census, please let me know!) Within the full population are 73 host stars accompanied by a total of 129 planets, suggesting that only 3%-5% of stars in this volume can be described as “planetic.” Yet recent studies argue that virtually every star of spectral types M through G (and probably types F through A) harbors at least one planet (Winn & Fabrycky 2015). Evidently a vast population of planetary systems remains to be discovered right on our Galactic doorstep.

Figure 2 sorts the known exoplanetary systems within 20 parsecs according to their distance from our Sun, with each successive ring in the blue background representing an increment of 5 parsecs in the radius of the expanding sphere.

Figure 2. All host stars at 20 parsecs or less, arranged by distance from the Sun

73 exoplanetary host stars sorted by distance from our Sun. Spectral types are indicated by the color key at lower right. The inner circle has a radius of 5 parsecs, and each successive ring represents an increment of 5 parsecs. Stellar icons are arranged in 2 dimensions by right ascension, which is marked at the edge of the outer circle. (Declination is necessarily ignored.) This diagram shows the relative distance of planetary systems from our Sun, but not from each other.
Inspection of the diagram reveals that the frequency of confirmed host stars per cubic parsec falls rapidly with increasing distance, even within this limited space. So does the ratio of dim stars (spectral type M) to brighter stars. G-type stars, on the other hand, appear to be over-represented. They constitute less than 10% of field stars within 20 parsecs, but almost one-third of the sample of host stars in this space.

About 57% of the planets in Figure 2 are low-mass objects in the same classification as Uranus and Earth (Table 1). These planets have masses smaller than about 0.15 Jupiter masses (0.15 Mjup), equivalent to 48 Earth masses (48 Mea). The remainder are gas giants like Jupiter and Saturn, although very few of them have orbital periods as long as our giants. A similar majority of nearby host stars (55%) harbors a single detected planet. Nonetheless, multiplanet systems are abundant enough that only one-third of all planets in this space occur in singleton systems.

Thus the “average” planet in our neighborhood is a low-mass object with at least one other companion orbiting the same star. Notably, this description applies to six out of eight planets in our Solar System.

Figure 3. Transiting low-mass planets within 20 parsecs

All these planets have been observed in transit from Earth’s orbit. A single transit has also been reported for HD 219134 b. The extrasolar planets in this image have orbital periods shorter than 3 days; the period of 55 Cancri e is shorter than 18 hours.

Our neighborhood also contains several distinctive architectural designs. The most common (11 systems; 15%) features at least two gas giants. More than half of this sample harbors three or more planets in total, and in 9 out of 11 systems, at least one planet orbits outside 2 AU. Again, this description fits our Solar System, but it also applies to quite different systems, including Upsilon Andromedae, 47 Ursae Majoris, and HD 128311.

The next most common architecture (12%) is the compact low-mass configuration, in which at least three low-mass planets orbit inside 1 AU and no gas giants are detected. This architecture is mutually exclusive with the previous type, and it is extremely common in the Kepler sample. Well-studied examples within 20 parsecs include GJ 581 and HD 69830.

The third most common architecture overlaps with the two-giant design. This is the mixed-mass type (10%), in which a minimum of three planets are present: at least one gas giant and at least one low-mass object. 55 Cancri, GJ 876, and Mu Arae are notable nearby examples containing at least two gas giants each.

In the full exoplanetary sample, the most abundant and easily detectable architecture is the Hot Jupiter configuration, defined as a star accompanied by a gas giant planet with a period of 10 days or less. However, this is only the fourth most common architecture in our immediate neighborhood. Just five (7%) of the systems detected to date in this space contain a Hot Jupiter. Three of them (HD 189733, Tau Boötis, and 51 Pegasi) are singletons. Each of the others(Upsilon Andromedae, HD 217107) harbors at least one additional gas giant on a wider orbit, so these two systems can also be assigned to the two-giant architecture.

Among singleton systems within 20 parsecs, giants have a slight edge, accounting for 58% of this subsample. Hot Jupiters are in a minority, however: the median semimajor axis for single giants in this volume of space is 1.3 AU, compared to 0.1 AU for low-mass singletons. Single giants also tend to orbit more massive stars: inside 20 parsecs, the median host star mass in such systems is 0.9 Msol (maximum 1.65 Msol), whereas the median for hosts of low-mass singletons is 0.49 Msol (maximum 0.97 Msol). These numbers suggest a certain antipathy between hot stars and low-mass planets, whether real or the result of observational bias. Low-mass planets might truly be scarce around hot stars, or they might simply be harder to detect in such environments.

One of the least common architectures revealed to date is the Solar System analog. I define this configuration as a Sun-like star (0.7-1.2 Msol) hosting a gas giant on a circular orbit wider than 3 AU, without any gas giants on interior orbits either occupying or perturbing the system habitable zone. This configuration is a subset of the mixed-mass type. Inside a radius of 20 parsecs, only one system besides our own features such an architecture: HD 154345. More than a dozen others have been reported outside this volume, but the most distant is only 57 parsecs away (see How Weird is Our Solar System? and Almost Jupiter). None are known within the Kepler sample.

Their absence is not surprising, since Kepler collected data for less than four years and observed only planetary transits. The cool gas giant in a Solar System analog will have an orbital period longer than five years (and probably at least twice as long) and it might not be co-planar with the inner planets. In such a configuration, an observer with a fixed viewing angle could never observe both the inner and outer planets in transit.

Table 1. Characteristics of three exoplanetary populations

Among objects listed in EPE in early 2016, two substantially larger, non-overlapping samples are available to compare with the discrete population in our immediate neighborhood: 1) 472 confirmed Kepler systems hosting 1038 planets, and 2) 606 confirmed transiting and RV systems outside 20 parsecs hosting 707 planets, excluding Kepler discoveries. Table 1 compares both samples to the population of 73 host stars and 129 planets within 20 parsecs.

The descriptive statistics summarized in the table suggest at least six salient points:

1. Although all three groups are strongly biased against M dwarfs (defined as stars < 0.7 Msol) and in favor of hot stars (those ≥ 1.2 Msol), our neighborhood population appears to have the most representative selection of stellar hosts. For example, if we want a “typical” M-type sample, the 26 systems orbiting red dwarfs within 20 parsecs are a good place to start.

2. Only the Kepler sample has a minority of single-planet systems, alerting us that even the host stars in our local sample must have many additional planets. These objects have eluded detection so far because they are either very low in mass (and thus undetectable by ground-based RV monitoring) or traveling on long-period orbits (which require a substantial investment of resources to ensure regular observations over a period of decades).

3. The abundance of multiplanet systems in the Kepler sample (78%) is evidence of the opposite bias. It is much easier to confirm the detection of a transiting planet with one or more candidate companions than it is to rule out all sources of false positives for a single planet with similar transit data. Hence Kepler systems with a single transiting planet (“tranet”) are less likely to be confirmed than those with multiple tranets. Nevertheless, a system with a confirmed tranet might still harbor invisible companions if their orbits are widely spaced (Xie et al. 2014) or non-coplanar (Johansen et al. 2012).

4. The relative fractions of low-mass planets in these three samples indicate a more straightforward bias: the RV technique is simply less sensitive to small planets than transit monitoring is, and sensitivity declines rapidly outside 10 parsecs. That limitation probably indicates that hundreds of terrestrial planets are awaiting discovery within our immediate neighborhood.

5. The diversity of system architectures is greatest within 20 parsecs, as demonstrated by the low frequencies of two-giant, compact low-mass, and mixed-mass systems in the other two populations. This feature makes our neighborhood sample especially valuable for comparison studies.

6. Hot Jupiters are heavily represented in all three populations, with the most obvious excess in the RV sample outside 20 parsecs. Nevertheless, the Kepler fraction is quite similar to the fraction in the local population. If only about 1% of Sun-like stars harbor a Hot Jupiter, as recent studies conclude (Bayliss & Sackett 2011, Wright & al. 2012, Wang & al. 2015), both of these samples are unbalanced.

local bubble

If we expand our perspective by an order of magnitude to a radius of 200 parsecs, we see a more varied starscape (Figure 3). This larger volume of space encompasses most of the region known as the Local Bubble, an irregular, gas-free cavity surrounded by molecular hydrogen clouds that trace the Orion Spur of our local spiral arm. Whereas stars in the immediate vicinity of our Sun appear to be a random sample of the Galactic population, our enlarged perspective reveals localized structures.
Figure 3. Selected star clusters and exoplanetary systems within 200 parsecs

Selected exoplanetary systems and star clusters within 200 parsecs (652 light years) of our Sun. Each successive ring in the colored circle represents an additional 50 parsecs. In this flattened, two-dimensional schema, stellar icons are distributed by right ascension according to their distance from our Sun, not by their distance from each other. Right ascension is marked by the numbers ascending counterclockwise around the outer ring. 

Visible are 10 star clusters, some young and bright, others older and dimmer. A star cluster is a compact association in which the space density of stars is much higher than the surrounding environment, and member stars share a similar motion through space. Each cluster population is far from random, since cluster members were born together in the same molecular cloud, and thus have very similar ages and chemical compositions. All stars of the same mass in the same cluster are at the same evolutionary stage.

Star clusters can be described as either embedded or open, depending on the presence (embedded) or absence (open) of remnant hydrogen clouds. Several nearby clusters are still embedded: Rho Ophiuchi, Taurus-Auriga, Chamaeleon I, and Coronet. Their estimated ages range from 1 to 10 million years. The open clusters are substantially older, with the Pleiades about 125 million years old, Praesepe 600 million, and the Hyades 625 million. Notably, planets have been discovered in both Praesepe and the Hyades.

Apart from these astronomically interesting and aesthetically captivating structures, we can also see several field stars that harbor unusual planetary systems. Kepler-16, located only 65 parsecs (212 light years) away, was the first circumbinary system ever detected, and the nearest by a margin of several hundred parsecs. This system includes a K-dwarf and an M-dwarf sharing a binary orbit of 41 days, with an undersized gas giant (Kepler-16b) orbiting their common center of mass in a period of 229 days. Remarkably, even though gas giant environments are inimical to life as we know it, this giant happens to reside in the system’s habitable zone. While we can be pretty sure that circumbinary planets are rare, they might not be so rare that Kepler-16 is alone in our Local Bubble.

Venturing into deeper space, we find not one but two examples of another rare species: conjoined planetary systems, defined as stellar binaries in which each star hosts its own system of planets. XO-2NS comprises two G-type stars with a binary semimajor axis of 4600 AU. Star N hosts a Hot Jupiter; star S hosts two warm gas giants. WASP-94AB is a pair of F-type stars with a binary semimajor axis of 2700 AU. Both stars host Hot Jupiters, with no other planets in evidence.

Then we find the rarest of the rare, the only one of its kind: WASP-47, which includes a Hot Jupiter flanked by two low-mass planets inside 0.1 AU. No other Hot Jupiter has any nearby companions. WASP-47b also belongs to another small club: Hot Jupiters with outer giant companions, of which we identify two inside 20 parsecs, but few others in deeper space.

An unmistakable feature of Figure 3 is the Kepler treasure trail extending outward along right ascension 19 into infinity. Joining Kepler-16 in this procession of wonders is Kepler-444 and its Hot Martian quintuplets (five planets less massive than Earth orbiting a K-type star in periods shorter than 10 days) as well as two M dwarf systems (Kepler-186, 438) hosting small planets that have been widely promoted as potentially habitable Super Earths. Also represented is one of the few mixed-mass systems in the Kepler sample: Kepler-68, which hosts two small transiting planets inside 0.1 AU as well as a non-transiting gas giant detected by radial velocity measurements orbiting just outside 1 AU. All these Kepler systems enhance the diversity of the region between 50 and 200 parsecs, which otherwise presents a bland continuum of relatively bright stars hosting lonely giants.

Bayliss DD, Sackett PD. (2011) The frequency of Hot Jupiters in the Galaxy: Results from the SuperLupus survey. Astrophysical Journal 743, 103.
Kraus AL, Hillenbrand LA. (2007) The stellar populations of Praesepe and Coma Berenices. Astronomical Journal 134, 2340-2352.
Converse JM, Stahler SE. (2008) The distribution of stellar mass in the Pleiades. Astrophysical Journal 678, 431-445.
Johansen A, Davies MB, Church RP, Holmelin V. (2012) Can planetary instability explain the Kepler dichotomy? Astrophysical Journal 758, 39.
Research Consortium on Nearby Stars. (2012) RECONS Census of Objects Nearer than 10 Parsecs. Available at:
Turnbull M. (2013) Nearby Stars: Distance, Type, Luminosity. Global Science Institute. (Excel spreadsheet found online)
Urquhart JS, Figura CC, Moore TJT, Hoare MG, Lumsden SL, Mottram JC, Thompson MA, Oudmaijer RD. (2014) The RMS survey: galactic distribution of massive star formation. Monthly Notice of the Royal Astronomical Society 437, 1791-1807.
Wang J, Fischer DA, Horch EP, Huang X. (2015) On the occurrence rate of Hot Jupiters in different stellar environments. Astrophysical Journal 799, 229.
Winn JN, Fabrycky DC. (2015) The occurrence and architecture of exoplanetary systems. Annual Review of Astronomy and Astrophysics 53, 409-447.
Wright JT, Marcy GW, Howard AW, Johnson JA, Morton TD, Fischer DA. (2012) The frequency of Hot Jupiters orbiting nearby Solar-type stars. Astrophysical Journal 753, 160.
Xie JW, Wu Y, Lithwick Y. (2014) Frequency of close companions among Kepler planets – A TTV study. Astrophysical Journal 789, 165.
Zucker C, Battersby C, Goodman A. (2015) The skeleton of the Milky Way. Astrophysical Journal 815, :23.

Sunday, March 20, 2016

Almost Jupiter

Figure 1. Saturn has the rings, but Jupiter has the Great Red Spot: a cyclone large enough to swallow Earth. Telescopic observations attest that this storm system has been raging for centuries, although it has shrunk by half in the past hundred years. Similar vortices circulate elsewhere in the planet’s deep atmosphere. Objects like Jupiter – i.e., gas giants following circular orbits with periods of several Earth years – are apparently rare in our region of the Galaxy. Image credit: NASA/Voyager 1

David Kipping and colleagues recently announced the discovery of a transiting gas giant with an orbital period of 2.9 years, the longest ever confirmed for a transiting planet. Their analysis opens new horizons in our understanding of planetology and system architecture. Kipping’s group found the object by searching archival Kepler data, which reveal a mixed-mass system of four planets orbiting Kepler-167 (alias KIC-3239945). The star was already known as the host of two short-period Super Earths, to which Kipping’s group has added a third. All three have semimajor axes smaller than 0.15 astronomical units (0.15 AU) and radii smaller than twice Earth’s (2 Rea).

The new giant is the outermost of the four planets, orbiting at a semimajor axis of almost 2 AU (which would fall between Mars and the Asteroid Belt in our Solar System). According to the standard naming protocol, it is designated Kepler-167e. Only two transits could be detected during the four-year span of Kepler data collection, but that’s just enough to validate the object’s reality. 

Table 1. Characteristics of the Kepler-167 planetary system

Column 1 shows the planet name; column 2, the radius in Earth units (Rea); column 3, the semimajor axis (a) in astronomical units; column 4, the eccentricity (e); and column 5, the orbital period in days.

The host star has a likely spectral type of K3 or K4, given its estimated mass of 0.77 Solar masses (Msol) and effective temperature of 4890 K. Surface gravity measurements confirm its evolutionary status on the main sequence. The discovery team estimated the star’s age as 3.3 billion years and its distance as 330 parsecs (1075 light years). 

Table 1 and Figure 2 describe a virtual twin of HD 219134 (for two different perspectives on that system, see Motalebi & al. 2015 and Wright & al. 2015). Both Kepler-167 and HD 219134 are early K dwarfs with identical masses, and both host a cluster of small planets inside 0.25 AU as well as a presumed gas giant at an approximate semimajor axis of 2 to 3 AU. 

Figure 2. Kepler-167 system architecture

The four planets of Kepler-167 are represented at their relative sizes. All orbits must be co-planar, since all planets are observed in transit. As an early K dwarf, the parent star is dimmer and cooler than our Sun. Thus planet e orbits outside the system ice line.

This architecture has conspicuous structural similarities with the other multiplanet systems summarized in Figure 3. All 10 systems include a Sun-like star (0.7 to 1.2 Msol) accompanied by at least two low-mass planets on short-period orbits and at least one gas giant on a wider orbit. In 9 out of 10 systems, at least one low-mass planet orbits inside 0.1 AU, and in 7 out of 10, the gas giant orbits outside 1 AU. In two systems, an additional low-mass planet orbits outside the gas giant. One system (WASP-47) hosts two gas giants, with the second one dominating the inner cluster of planets. Another (Kepler-90) might also host two gas giants, since planets g and h both have radii larger than 8 Rea. In Kepler-90, however, both these planets are part of the outer cluster of planets.

A striking feature shared by most systems is the pronounced gap between the inner aggregation of low-mass planets and the outer giant. This gap is absent from the two most compact systems (Kepler-89 and -289), where all known planets orbit inside 0.6 AU, and it has a different configuration in Kepler-90 and HD 10180, which host seven planets each. Altogether, these 10 systems look like a suite of variations on a single theme. 

Figure 3. Selected mixed-mass planetary systems

Tags: ME = Earth masses; RE = Earth radii; AU = astronomical units (Earth-Sun separation = 1). Selection criteria: at least one gas giant and at least two low-mass planets on interior orbits. In each system with transiting planets, all transiting orbits are co-planar. Note that gas giants Kepler-89d and WASP-47b are also observed in transit, while a single transit is reported for the innermost planet of HD 219134.

Despite the proximity of the gas giant to the low-mass planets in each one, many of these systems have an extremely “flat” configuration, with all planets traveling in the same orbital plane (see Mass Matters). In half of them (Kepler-87, 89, 90, 167, and 289), all planets are observed in transit, implying that all orbits are approximately co-planar. In WASP-47, all three inner planets are seen in transit, with the same implication. Sensitive, long-term monitoring will be needed to determine whether the outer giant also transits.

However, in two other systems (Kepler-48 and -68), no transits of the outer giant were detected, suggesting that the inner and outer systems are misaligned. Furthermore, we have no way to estimate the orbital alignment of the last two systems. None of the planets around HD 10180 have been observed in transit, and in HD 219134, a single transit has been reported for the innermost planet only. Even in the case of a perfectly co-planar system, our viewing angle might limit transit detection to the planet with the shortest period, with all the rest orbiting just out of sight.

Co-planarity is commonly interpreted as evidence of a calm dynamical history. Notably, the eight planets in the Solar System are approximately co-planar (depending on viewing geometry), and in more than 100 Kepler systems containing only low-mass planets, at least three planets per system are co-planar. Data showing that at least half of the compact mixed-mass exoplanetary systems currently known are also co-planar provide a useful constraint on their formation history. 

planetology of Kepler-167 

We have little information on the composition of the planets orbiting Kepler-167, since we know only their radii. No transit timing variations are available to constrain their masses. Although the discovery team offered hope that future radial velocity programs could characterize the outer planet, that prospect remains hypothetical in view of the system’s distance.

Kepler-167e has a radius of 10.15 Earth units (Rea), or about 90% of Jupiter’s, which is consistent with either a gas giant planet or a brown dwarf star. Thus Kipping’s group describes this object as a “degenerate world,” given the uncertain implications of its radius. The minimum goal of a radial velocity study would be to break this degeneracy by establishing whether the object’s mass exceeds 13 Jupiter masses (Mjup), the nominal threshold for brown dwarfs.

Even without mass estimates, we can make informed guesses about the composition of the three inner planets of Kepler-167. The radius of planet d corresponds to a rock/metal world of about 2 Mea, with a structure similar to Earth’s. If only its orbit were wider – 0.35 AU instead of 0.14 AU – it would be a top candidate for habitability. The other two planets might also be rocky, if they’re relatively massive (5-9 Mea), but the discovery team favors a mix of rock/metal and volatiles for each one. Regarding the volatile contribution, radii smaller than 2 Rea suggest water instead of hydrogen/helium envelopes. For Kepler-167b and -167c, a substantial water fraction might take the form of a steam atmosphere surrounding a high-pressure ice layer. Unfortunately, available models of planet structure suffer from ice aversion, so I haven’t found much theoretical guidance for this range of radii. 

a Jupiter analog? 

Now we come to the most remarkable claim in the discovery paper, evident in its title. The authors describe Kepler-167e as a Jupiter analog. This rare species has attracted growing interest in the past decade, along with a few competing definitions (see How Weird Is Our Solar System?). According to Kipping & colleagues, Kepler-167e fulfills three essential requirements: its radius is consistent with a gas giant planet, its semimajor axis places it outside the system ice line, and its orbital eccentricity is low.

However, these criteria are not universally regarded as sufficient. In two successive studies, Robert Wittenmyer & colleagues (2011, 2016; hereafter W11 and W16) proposed that Jupiter’s most important characteristic is its dynamical role in the Solar System, both historically and at the present epoch. Their first study (W11) defined a Jupiter analog as a gas giant with an orbital period of at least 8 years and an eccentricity smaller than 0.2. In their view, these parameters implied in situ formation and a relatively calm dynamical history. Their more recent publication (W16) revised this formulation to a minimum mass of 0.3 Mjup, a semimajor axis outside the system ice line and wider than 3 AU, and an orbital eccentricity no more than 0.3. W16 note that planets as lightweight as 0.15 Mjup could also fulfill Jupiter’s role, but they set their cut-off at twice that mass because so few gas giants under 0.3 Mjup are known. To calculate the frequency of these “Jupiter analogs” (see below) they also limited their analytic sample to stars with at least 30 observations spanning 8 years.

All these criteria make sense, but they omit a critical aspect of Jupiter’s dynamic role: our system has no gas giants inside Jupiter’s orbit. However that architecture came into being, the underlying mechanism is very likely to involve Jupiter’s orbital dynamics. Thus, when Wittenmyer’s group recently applied their criteria to the sample of planets detected by the Anglo Australian Search Program (W16), the results were odd. They identified 8 systems altogether, among which only 4 host what I would regard as a plausible Jupiter analog (though only one member of this quartet has a semimajor axis of 5 AU or more). In the other 4, the so-called Jupiter analog is accompanied by a gas giant on an interior orbit with a semimajor axis near 1 AU. In one of those systems (Mu Arae), a third gas giant and a Uranus-mass planet also occupy the inner system. As these system architectures imply dynamical histories very different from the Solar System, it seems unlikely that such candidate Jupiters could have played a role truly analogous to the original Jupiter. As in the past, therefore, I argue that a bona fide Jupiter analog must have no interior giant companions. 

Figure 4. Jupiter and Io

Extrasolar analogs of Jupiter are likely to be accompanied by extensive retinues of satellites, since satellite formation appears to be the final stage in the evolution of a gas giant planet. This photograph shows Jupiter with Io, the innermost of the giant’s four Galilean moons. Apart from our own Moon, Io is the only spherical satellite in the Solar System with a purely rocky composition. It supports extensive volcanism, as the image reveals: the red glow and bluish plume signal active eruptions. Image credit: NASA/JPL-Caltech.

Another recent study by Dominick Rowan & colleagues (2016; hereafter R16) adopts a definition based on W11, and thus very similar to W16: a minimum mass of 0.3 Mjup, a semimajor axis 3 AU or more around a G-type star (scaling as a period of at least 5 years around stars of other spectral types), and an orbital eccentricity under 0.3. Their study reports the discovery of HD 32963 b, which they describe as a Jupiter analog. To demonstrate the application of their criteria, they offer a list that includes the new candidate along with 20 others already known. This list is longer than the one in W16 because R16 drew from all published planets, whereas W16 limited their selection to planets discovered by their own group. (For whatever reason, R16 omit four of the eight planets presented by W16, even though all four were published before R16 submitted their manuscript for peer review.)

Both lists are similar in one important way: R16, like W11 and W16, extend the designation of Jupiter analog even to planets with gas giant companions on interior orbits. This architecture characterizes 6 of the 21 systems presented by R16, with 3 of the 6 shared with W16. Notably, R16 also include HD 219134 (omitted by W16), but only as analyzed by Vogt & al. 2015. Vogt’s group described a six-planet system in which the outermost object is a gas giant named HD 219134 g with a minimum mass of 0.34 Mjup, a semimajor axis of 3.11 AU, and a period of 6.16 years. A competing analysis by Motalebi & al. 2015 found a distinctly different line-up, in which a gas giant named HD 219134 e, the outermost of four planets, has a minimum mass of 0.19 Mjup, a semimajor axis of 2.14 AU, and a period of 3.74 years. The former description meets all the criteria adopted by R16; the latter meets none of them.

Given the close resemblance between the architecture of HD 219134, as presented by  Motalebi and colleagues, and that of Kepler-167 system, as presented by Kipping and colleagues, the criteria of W16 and R16 also exclude Kepler-167e.

Another perspective on this issue is available from Sean Raymond, who in 2006 published an article titled “The search for other Earths: limits on the giant planet orbits that allow habitable terrestrial planets to form.” It’s significant that the term “Jupiter analog” occurs neither in the title nor the text. Instead, Raymond’s concern was to define the orbital architectures that would permit the accretion of a terrestrial planet with a minimum mass 0.3 times Earth (0.3 Mea) in the habitable zone of a Sun-like star. Anticipating W11, W16, and R16, Raymond noted the critical importance of orbital circularity. He found that semimajor axes smaller than 2.5 AU were inconsistent with habitable terrestrial planets, while wider separations became increasingly friendly, such that a sufficiently large semimajor axis could mitigate eccentricities of 0.2 to 0.4. Even though his analysis is 10 years old, its emphasis on the possibility of terrestrial planet formation in the habitable zone is still relevant to current investigations.

Yet despite the differences among these approaches, none of them would characterize Kepler-167e as a Jupiter analog. Whether their exclusion implies that systems like Kepler-167 are unlikely to host habitable planets is another question. 

birth versus survival 

Let’s look again at the 10 systems summarized in Figure 3. In 9 out of 10, the (outer) gas giant orbits inside 2.5 AU, disqualifying all 9 of them as Jupiter analogs according to the criteria of W16, R16, and Raymond 2006. However, the giant in the tenth system (HD 10180 h) has a semimajor axis of 3.4 AU and an eccentricity of 0.08, well within their limits. Why did R16 omit this system from their list? I’d guess the deal-breaker was the planet’s low minimum mass: 0.203 Mjup (64 Mea). Nevertheless, as W16 noted, the actual mass implied by this value is sufficient to play a toned-down version of Jupiter’s dynamic role, since it exceeds 0.15 Mjup.

The HD 10180 system had not yet been announced when Raymond presented his analysis, so he did not discuss the compatibility of its architecture with rocky planet formation. Nor did he discuss the possibility that the inner companions of a Jupiter-like planet might include both rocky planets and gas dwarfs analogous to Neptune. Finally, none of his simulations produced “hybrid” architectures like those featured in Figure 3. By hybrid I mean that these architectures resemble the well-known class of compact low-mass systems, except that they add a cool gas giant to the mix.

Recent observations have established that clusters of small planets can include rock/metal spheres with masses similar to Earth alongside planets with masses several times larger and radii puffed up by hydrogen/helium atmospheres (e.g., Kepler-20, 62, 90, 169). This diversity of composition among small planets foregrounds the importance of understanding the evolution and orbital dynamics of HD 10180 (and similar hybrids) before we can effectively assess the likelihood that these systems support Earth-like planets.

Three studies have already discussed the potential for habitable planets around HD 10180 (Lovis & al. 2011, Tuomi 2012, Kane & Gelino 2014). Unfortunately, none of them addressed the system’s formation history. Since the host star is very similar to our Sun, its habitable zone falls in the outer reaches of the gap between planet f (23 Mea at 0.49 AU) and planet g (24 Mea at 1.42 AU). This region is equivalent to the space between Mercury and Mars in the Solar System, and thus approximately co-extensive with our own habitable zone. Both Lovis’ group and Tuomi were optimistic about the possibility of an additional planet surviving there. They argued that, amid the complex web of dynamic interactions woven by the system’s packed orbits, the empty region around 1 AU was an island of stability that might harbor an Earth-mass planet. In a darker view, however, Kane & Gelino have recently argued that planet g has a substantially more eccentric orbit than previously reported. In their analysis, planet g would either prevent the formation of any habitable planets or eject such planets if they managed to form. Accordingly, the orbital gap represents a forbidden zone instead of a life zone. 

mind the gap 

Most of the systems presented in Figure 3 exhibit a similar gap between the inner and outer planets. Indeed, our own Solar System has an analogous feature: the gap between Mars and Jupiter, which extends from about 1.5 AU to 5 AU and separates our inner system of terrestrial planets from the outer realm of the giants. The only occupants of this gap are the battered objects in the Asteroid Belt, whose mass is largely confined to the region between 2 and 3.3 AU.

In Kepler-87 and 90, the gap occurs well inside the inner edge of the habitable zone, which begins beyond the outermost planet in both systems. However, in six other systems (including HD 10180, HD 219134, and Kepler-167), the gap encompasses the habitable zone. Are all these gaps truly empty, or might they hold one or more planets that have so far escaped detection? Such objects would be missed if they were slightly misaligned with the known planets (in the case of the transiting systems) or too lightweight to reach the threshold of detectability (in the RV systems). I’d love to see more research addressing this question, especially in the form of dynamical analyses of the systems already known.

It might turn out that the interesting hybrid architectures highlighted in Figure 3 are, by nature, unfriendly to habitable planets, as Kane and Gelino argued for HD 10180. In many cases, the principal antagonist would most likely be the outer giant, which would sculpt interior orbits the way Jupiter has sculpted our own system’s gap. But if they aren’t intrinsically hostile, then the configurations of HD 219134 and Kepler-167 might offer us a new architectural signpost of potential habitability. This one would supplement our existing biomarker of bona fide Jupiter analogs.

Figure 5. Fifteen potential Solar System analogs 

* Not listed in Rowan et al. 2016.

Tags: Msol = star mass in Solar units; Type = spectral type; Dist. = distance in parsecs; Mjup = planet mass in Jupiter units; a = semimajor axis in Earth units; e = orbital eccentricity; Period = orbital period in years. Selection criteria: star mass 0.7-1.2 Msol; a > 3 AU; e < 0.3, no interior giants occupying or perturbing the system habitable zone.
how rare is Jupiter? 

Compact low-mass systems are relatively abundant, both in the Kepler catalog and within a few dozen parsecs of our Sun. But their mixed-mass cousins are not. To date, however, no estimates are available for their relatively frequency in the underlying population of exoplanetary systems.

Happily, the landscape of Jupiter analogs is now emerging from the haze, emboldening both W16 and R16 to estimate the true occurrence rate of Jupiter-like planets in our Galactic neighborhood. W16 calculate a frequency of about 6%, while R16 find 1%-4%. Since both groups used a more generous definition of their target than I allow, I believe they would find a substantially lower frequency if they refocused their sights on cool gas giants that tolerate habitable planets. These might occur in only 1%-2% of planetary systems.

Such a population share is much lower than I imagined a few years ago. If my new guess is accurate, then systems like ours are rare – possibly as rare as Hot Jupiters, for which recent studies have calculated a prevalence of about 1% or less around Sun-like stars (Bayliss & Sackett 2011, Wright & al. 2012, Wang & al. 2015).

So there’s another reason to appreciate the Earth (not to mention the Jupiter) we already know about. 

Bayliss DD, Sackett PD. (2011) The frequency of Hot Jupiters in the Galaxy: Results from the SuperLupus survey. Astrophysical Journal 743, 103.
Kipping DM, Torres G, Henze C, Teachey A, Isaacson H, Petigura E, Marcy GW, Buchhave LA, Chen J, Bryson ST, Sandford E. (2016) A transiting Jupiter analog. In press. Abstract: 2016arXiv160300042K
Kane SR, Gelino DM. (2014) On the inclination and habitability of the HD 10180 system. Astrophysical Journal 792, 111.
Lovis C, Ségransan D, Mayor M, Udry S, Benz W, Bertaux J-L, & al. (2011) The HARPS search for southern extra-solar planets. XXVIII. Up to seven planets orbiting HD 10180: probing the architecture of low-mass planetary systems. Astronomy & Astrophysics 528, A112. Abstract: 2011A&A...528A.112L
Motalebi F, Udry S, Gillon M, Lovis C, Ségransan D, Buchhave LA, Demory BO, Malavolta L, Dressing CD, Sasselov D, et al. (2015) The HARPS-N Rocky Planet Search I. HD 219134 b: A transiting rocky planet in a multi-planet system at 6.5 pc from the Sun. Astronomy & Astrophysics 584, A72. Abstract: 2015A&A...584A..72M
Raymond SN. (2006) The search for other Earths: Limits on the giant planet orbits that allow habitable terrestrial planets to form. Astrophysical Journal 643, L131–L134. Abstract: 2006ApJ...643L.131R 
Rowan D, Meschiari S, Laughlin G, Vogt SS, Butler RP, Burt J, Wang S, Holden B, Hanson R, Arriagada P, Keiser S, Teske J, Diaz M. (2016) The Lick-Carnegie exoplanet survey: HD 32963, A new Jupiter analog orbiting a Sun-like star. Astrophysical Journal, 817:104.
Tuomi M. (2012) Evidence for 9 planets in the HD 10180 system. Astronomy & Astrophysics 543, 52.
Vogt SS, Burt J, Meschiari S, Butler RP, Henry GW, Wang S, Holden B, Gapp C, Hanson R, Arriagada P, Keiser S, Teske J, Laughlin G. (2015) A six-planet system orbiting HD 219134. Astrophysical Journal 814, 12. Abstract: 2015ApJ...814...12V
Wang J , Fischer DA, Horch EP, Huang X. (2015) On the occurrence rate of Hot Jupiters in different stellar environments. Astrophysical Journal 799, 229.
Wittenmyer R, Tinney CG, O’Toole SJ, Jones HRA, Butler RP, Carter BD, Bailey J. (2011) On the frequency of Jupiter analogs. Astrophysical Journal 727, 102. Abstract:
Wittenmyer R, Butler RP, Tinney CG, Horner J, Carter BD, Wright DJ, Jones HRA, Bailey J, O’Toole SJ. (2016) The Anglo-Australian planet search XXIV: The frequency of Jupiter analogs. Astrophysical Journal 819, 28.
Wright JT, Marcy GW, Howard AW, Johnson JA, Morton TD, Fischer DA. (2012) The frequency of Hot Jupiters orbiting nearby Solar-type stars. Astrophysical Journal 753, 160.