systemic

With HAT-P-13c rapidly coming ’round the mountain, there was a very timely update on astro-ph last night. Josh Winn and his collaborators have obtained an additional slew of radial velocities which (1) demonstrate using the Rossiter-McLaughlin effect that the inner planet b’s orbit is likely well aligned with the stellar equator, (2) modify the orbital parameters, including the period of the outer massive planet, and (3) hint at a third body further out in the system.

How do these updates affect the unfolding story?

The Rossiter-McLaughlin measurement gives an estimate of the angle λ = -0.9°±8.5°, which is the angular difference between the sky-projected orbital angular momentum vector and sky-projected stellar spin vector. A non-intuitive mouthful. If we’re viewing the star edge-on, then λ = -0.9° amounts to a determination that the planet’s orbital plane is well-aligned with the star’s equator. (See this post for a discussion of what can happen if the star’s rotation axis is tipped toward the Earth). The good news from the measurement is that it’s a-priori more likely that planets b and c are coplanar — that happy state of affairs which will permit direct measurements of planet b’s interior structure and tidal quality factor. If, on the other hand, the planets b and c have a large mutual inclination, then b’s node will precess, and measurement of a small value for λ will occur only at special, relatively infrequent, times during the secular cycle. A close to co-planar configuration also increases the likelihood that the outer planet can be observed in transit.

With their beefed-up data set of out-of-transit Doppler velocities, Winn and his collaborators are able to get a better characterization of the planetary orbits. The best-fit orbital period and eccentricity of the outer planet are slightly modified when the new data are included. The best-guess center of the transit window for c has “slipped” to April 28, 2010, with a current 1-σ uncertainty of 2 days.

The later date, however, is not an excuse for procrastination! Measuring the TTV for this system is a giant opportunity for the whole ground-based photometric community, and a definitive result will require lots of good measurements of lots of transits starting now (or better yet, last month.) I’ll weigh in in detail on this point, along with the challenge posed by Mr. D very shortly…

Astronomers worldwide staggered into work this morning, some of them rudely elbowing their way to the front of the lines at the espresso machines, clear evidence that events surrounding the January 2010 ’606 holiday season have finally drawn to a close.

Hopefully the data will turn out to be of high quality! As I mentioned in yesterday’s post, ground observers in both Europe and North America were out in force for the event, collecting photometric and spectroscopic data. The action was covered from space as well. We were awarded a generous 84-hour block of time on Warm Spitzer. The telescope started collecting 4.5-micron photometry more than a day prior to the secondary transit, and ended more than two days after the periastron passage.

What do we hope to learn? By observing the run-up to the secondary transit, we should be able to establish an improved baseline temperature for the planet, which should afford a better sense of how much tidal heating is occurring. And during the days following periastron, we expect to see a near-complete drop-off in flux from the planet as the periastron nightside hemisphere rotates fully into view. The 2007 observations came to a frustrating end just as this should have been starting to occur.

In addition to the secondary eclipse and the ground-based observations, Guillaume Hebrard and his collaborators were awarded 19 hours on Warm Spitzer to observe the primary transit at 4.5 microns. Their photometric time series will enable an improved radius measurement for the planet — both because of the highly accurate photometry and because the effects of stellar limb darkening are negligible in the infrared. Their time series will establish a very precise ephemeris for the transit, which will enable future observations to monitor the system for orbital precession.

Looking forward to the results…

The long-awaited initial discoveries from the 600M Kepler mission are in!

At a scientific talk at the AAS Meeting in Washington DC this morning, and in an afternoon press briefing packed with journalists, bright lights and television cameras, the Kepler Team announced the discovery of five new transiting planets. Four are inflated hot Jupiters, and one is a hot Neptune reminiscent of Gliese 436b and HAT-P-11b. Most importantly, the Kepler satellite appears by all accounts to be performing beautifully as it continuously monitors over 150,000 stars for planetary transits.

Here’s a to-scale line-up of the Kepler starting five. Kepler-4b is so small that it’s just barely resolved at a scale where its orbit spans 480 pixels.

The Kepler planets are primarily orbiting high-metallicity, slightly inflated, slightly evolved stars. These particular parent stars were likely selected for high-priority confirmation observations because their abundant, narrow spectral lines should permit maximally efficient, cost-effective Doppler-velocity follow-up.

Among the planets, Kepler-4b, with its composition that’s likely largely water-based, provides further evidence that the majority of short-period planets formed far from their parent stars, beyond the iceline in the protostellar disk, and subsequently migrated inward. Kepler-7b is approximately the density of styrofoam. In a conversation with a reporter, I scrambled for an analogy:

It’s like looking at a football team. You might guess from the team photo that they’re all 250 to 300 pounds. But then you find out that some of them are 25 pounds; that would come as a surprise…

Everyone is looking forward to the big-picture results that will be coming from Kepler a few years hence, as it probes into the habitable zones of Solar-type stars. In the interim, though, the veritable flood of ultra-high precision photometric data arriving via the the Deep Space Network will keep Doppler velocity follow-up observers working the late-night shifts. The parent stars of the new planets are in the V=12.6 to V=13.9 range, roughly 100 times fainter than the prime transit-bearing stars such as HD 209458 and HD 189733.

According to a S&T editor Bob Naeye, who reported on Bill Borucki’s scientific talk this morning, the first 43 days of photometric observations from the satellite generated 175 transit candidates, of which 50 were followed up in detail to extract the 5 announced planets. The Keck I telescope has been the major workhorse for the high-precision RV follow-up efforts that are required to get accurate masses. According to the Keck I Telescope Schedule, 17 nights were allocated to the Kepler team from July through December of last year. Within this time alotment, roughly 50 RV measurements for the 5 new planets were obtained. The velocity precision for Kepler-4b looks to be of order 2-3 m/s, which is excellent. Here are two thumbnails from Borucki’s talk (look carefully to read the y-axis scale):

With a slew of nights and good weather during 2010, it should be possible to get a significant number of additional planets confirmed…

I’m always impressed by the efficiency with which red dwarfs pack hydrogen, the stuff of flammable zeppelins, into such a small space: Gliese 1214 is more than twice as dense as led. The density of the Sun, on the other hand, is bubblegum by comparison.

Gliese 1214b’s orbital period is a mere 1.58 days. Its 0.014 AU separation from the system barycenter is the smallest yet measured for any planet. Yet because of the high red dwarf density, the star-planet configuration is actually rather spacious. Here’s the system to scale:

It’s interesting to compare this diagram with that of a genuinely close-in planet such as HAT-P-7b, which actually has a somewhat longer 2.2 day orbital period:

At a given period, a red dwarf fills much less of a planetary orbit than does a Sun-like star. If the occurrence rate of planets at a specified period is the same for stars of different masses, then one needs to look at $\sim(M_{\odot}/M_{\rm RD})^{2/3}$ times more red dwarfs than Sun-like stars to find a given number of transits with a particular period.

Gliese 1214b lies at enough stellar radii from Gliese 1214 that its a-priori transit probability was only about 7%. The Mearth survey currently covers only ~2000 stars, and so the fact that the discovery was made so quickly was probably not luck, but rather points to the existence of a very large number of low-mass planets orbiting small stars.

Let’s face it. The big dough goes to chase potentially habitable transiting planets. With this metric, the red dwarfs come out way ahead. If red dwarfs and Sun-like stars have equal occurrence fractions for planets with Earth’s mass and insolation, then a low-mass red dwarf has roughly four times the probability of a Sun-like star of harboring a transiting potentially habitable planet. Twice the temperature means one-sixteenth the area and the square root of sixteen is four. The red dwarfs also present a number of other advantages, see e.g. here, here, and here.

Ryan Montgomery and I have a recent paper out which foreshadows what I think is the inevitability of transit surveys that use the Mearth strategy to target true-Earth analogs the habitable zones of the lowest-mass red dwarf stars. Mearth  is itself very well-positioned to expand in this direction. I also think that a lot of effort will continue to shift toward improved Doppler-velocity capability in the near-infrared (see, e.g. this recent paper by Jacob Bean and collaborators which describes the use of ammonia gas in a glass cell to imprint a forest of fixed reference lines on a K-band stellar spectrum).

A last note: Twelve-Fourteen-b is likely to become a favorite target for small-telescope observers, so I made sure to add it to the Transitsearch.org candidates table. Now that classes are done for the quarter, I’ve been going through the literature and adding or updating one or two planets a day. It’s tedious work, but I’ve noticed some interesting upcoming opportunities, which I’ll be writing about soon. For transit-themed ephemera and the latest celebrity gossip, look no further than the transitsearch twitter stream: http://twitter.com/Transitsearch.

And a postscript: In the comments, reader cwmagee points out that the implication of the post is that the HAT-P-7 and Gl1214 diagrams are to scale which eachother, but that’s not the case. He attached a version which shows a to-scale comparison of both systems:

Red dwarfs are small!

Of course, there are still 7 hours and 13 days left until the close of 2009, but I’ve got every confidence that the discovery of the decade has landed on the ground. The Mearth project has found a transiting 6.55 Earth-mass planet in orbit around the nearby red dwarf star GJ 1214. The parent star is bright enough, and the planet-star area ratio is large enough so that direct atmospheric characterization will be possible not just with JWST, but with HST. Incredible. I’m inspired, invigorated, envious. This discovery is a game changer.

The GJ1214 discovery is all over the news today. The coverage is deservedly laudatory, but interestingly, the most dramatic aspect of the detection received rather short schrift. This is easily the most valuable planet yet found by any technique, and the discovery, start to finish, required an investment of ~500K (along with the equivalent of 1-2 nights of HARPS time to do the follow-up confirmation and to measure the planet’s mass). By contrast, well over a billion dollars has been spent on the search for planets.

I’m milking that contrast for drama, of course. It’s true that GJ1214b is low-hanging fruit. The team with the foresight to arrive on the scene first gets to pick it. And the last thing I’m suggesting is a cut in the resources devoted to exoplanet research — it’s my whole world, so to speak. I do think, though, that Mearth epitomizes the approach that will ultimately yield the planets that will give us the answers we want. You search for transits among the brightest stars at given spectral type, and you design your strategy from the outset to avoid the impedance mismatches that produce bottlenecks at the RV-confirmation stage.

There’s a factor-of-fourteen mass gap in our solar system between the terrestrial planets and the ice giants, and so with the discovery of Gl 1214b (and the bizzare CoRoT-7b) we’re getting the “last first look” at a fundamentally new type of planet. CoRoT-7b is clearly a dense iron-silicate dominated object, but it likely didn’t form that way. Gliese 1214b’s radius indicates that it probably contains a lot of water. I think this is going to turn out to be the rule as more transiting objects in the Earth-to-Neptune mass range are detected.

So what next? With a modest increase in capability, Mearth is capable of going after truly habitable planets orbiting the very nearest stars. I think it’s time to put some money down…

As the decade draws to a close, it’s hard not to be amazed at the progress that’s been made on every research front related to extrasolar planets.

An area that I think is now ripe for progress comprises coordinated multi-observer checks for transits by super-Earth/sub-Neptune planets. There are now over thirty known extrasolar planets with Msin(i)’s less than that of Gliese 436b (which tips the scales at 23 Earth masses). Of these, only CoRoT-7b has so far been observed to transit, and it’s very probable that the current catalog of low-mass RV-detected planets contains one or more transiting members. Needless to say, it’d be very interesting to locate them.

To my knowledge, the lowest-amplitude transits that have been observed by amateur astronomers have been those by HD 149026b. This anomalously dense Saturn-mass planet induces a photometric transit depth of roughly 0.4%.  State-of-the-art amateur detections show the transit very clearly. Here’s an example (the observer was Luboš Brát of the Czech Republic) taken from the TRESCA database:

The identification of transits by small planets certainly won’t be a picnic. Super-Earths and  sub-Neptunes orbiting G and K stars present targets that are intrinsically much tougher than HD 149026. Unless the parent star is a red dwarf, the expected transit depths will generally be less than 0.1%, and it’ll be extremely difficult for a single small-telescope observer to obtain a definitive result.

On the other hand, if a platoon of experienced observers mount a coordinated campaign on a single star, then there’s a possibility that a startlingly good composite light curve might be obtained. In theory, if one were to combine the results from sixteen independent observers, one could obtain a light curve of the equal signal-to-noise as the HD 149026b curve shown above, but for a planet with a transit depth of only 0.1%.

I spent time this weekend making sure that the transitsearch.org transit predictions for the known RV-detected low-mass planets are as up-to-date and accurate as possible. I found that HD 7924 is a good candidate star with which to test a coordinated observing strategy. The star harbors a low-mass RV-detected planet was announced earlier this year (discovery paper here):

HD 7924b has Msin(i)~10 Earth Masses, a P=5.3978d orbital period, and a 6.7% a-priori chance of being observable in transit. The (folded) photometry in the discovery paper is of quite high quality, and shows that the star is not photometrically variable. The photometry also indicates that transits with depth greater than 0.05% are probably not occurring. The parent star, HD 7924 is a K-dwarf, with a radius of something like 78% that of the Sun, which means that if the planet is a sub-Neptune it’ll have a central transit depth of order 0.075%, whereas if it is a rocky object, the depth will likely be less than 0.05%. The 1-sigma uncertainty on the time of the transit midpoint is about 0.35 days. The parent star has V=7.2, and with Dec=+76 deg, it’s circumpolar for high-latitude observers (RA=01h 21m).

Here are the next predicted transit midpoints (dates and times are UT):

HJD        Y    M  D  H  M
2455182.04 2009 12 16 12 51
2455187.01 2009 12 21 12 14
2455192.41 2009 12 26 21 48
2455197.81 2010  1  1  7 21
2455203.20 2010  1  6 16 54
2455208.60 2010  1 12  2 28

Because HD 7924b’s period is known to an accuracy of 0.0013 days (2 minutes), participating Northern-hemisphere observers can obtain data during any of the upcoming opportunities. Their light curves, once standardized, can in theory be stacked to obtain increased precision. It would be very interesting to get a sense of the practical limits inherent in such an approach. I think the best way to test the limits is to give the observations a try!

I’m nostalgic for ’97, when the discovery of a new extrasolar planet was literally front-page news. What’s now cliche was then fully viable poetic sweep. Epicurus and his multitude of worlds. Bruno burning at the stake. In that frame of mind, it’s fascinating to go back and read John Noble Wilford’s extended New York Times piece, written at the moment when the number of known extrasolar planets equaled the number of planets in our own solar system.

Some of the hyperbole still seems fresh, especially with regard to the frequency and diversity of planetary systems:

And the discoveries may be only beginning. One recent study suggested that planets might be lurking around half the Milky Way’s stars. Astronomers have already seen enough to suspect that their definition of planets may have to be broadened considerably to encompass the new reality. As soon as they can detect several planets around a single star, they are almost resigned to finding that the Sun’s family, previously their only example, is anything but typical among planetary systems.

At the recent Porto conference, the Geneva team not only reiterated their claims regarding the frequency of low-mass planets, but actually upped their yield predictions. According to a contact who heard Stephane Udry’s talk, the latest indication from HARPS is that between 38% (at the low end) and 58% (at the high end) of nearby solar-type stars harbor at least one readily detectable M<50 Earth-mass planet. This is quite extraordinary, especially given the fact that were the HARPS GTO survey located 10 parsecs away and observing the Sun, our own solar system (largely in the guise of Jupiter’s decade-long 12-m/s wobble)  would not yet be eliciting any particular cause for remark.

It also looks like planets beyond the snowline are quite common. In yesterday’s astro-ph listing, there’s a nice microlensing detection of a cold Neptune-like planet orbiting a ~0.65 solar mass star with a semi-major axis of at least 3 AU. The microlensing detections to date indicate that Neptune-mass objects are at least three times as common as Jupiter mass objects when orbital periods are greater than five years or so. Microlensing detections are an extremely cost-effective way to build up the statistics of the galactic planetary census during belt-tightening times. Much of the work is done for free by small telescope observers.

Yet another dispatch pointing toward a profusion of planets comes from an article posted last week on astro-ph by Brendan Bowler of the IfA in Hawaii. Work that he’s done with John Johnson and collaborators indicates that the frequency of true gas giant planets orbiting intermediate-mass stars (former A-type stars like Sirius that are now in the process of crossing the Hertzsprung gap) is a hefty 26% within ~3 AU.

An embarrassment of riches? Certainly, the outsize planetary frequency means that the cutting-edge of the planet-detection effort will be shifting toward the Sun’s nearest stellar neighbors, as these are the stars that offer by far the best opportunities for follow-up with space-based assets such as HST, Spitzer, JWST et al.

As competition for ground-based large-telescope RV confirmation of run-of-the-mill planet transit candidates orbiting dim stars heats up, the threshold magnitude (at a given bandpass) at which stars become largely too faint to bother with will grow increasingly bright. We’re talking twelve. Maybe nine. Pont et al., in their discovery paper for OGLE-TR-182b refer to this threshold as the “Twilight Zone” of transit surveys:

The confirmation follow-up process for OGLE-TR-182 necessitated more than ten hours of FLAMES/VLT time for the radial velocity orbit, plus a comparable amount of FORS/VLT time for the transit lightcurve. In addition, several unsuccessful attempts were made to recover the transit timing in 2007 with the OGLE telescope, and 7 hours of UVES/VLT were devoted to measuring the spectroscopic parameters of the primary. This represents a very large amount of observational resources, and can be considered near the upper limit of what can reasonably be invested to identify a transiting planet.

A recent e-mail from Bruce Gary prompted me to pay a return visit the Exoplanet Transit Database (ETD) which is maintained by the variable star and exoplanet section of the Czech Astronomical Society. I came away both impressed and inspired. The ETD is really leveraging the large, fully global community of skilled small-telescope photometric observers.

There are hundreds of citizen scientists worldwide who have demonstrated the ability to obtain high-quality light curves of transiting extrasolar planets. I’ve developed many contacts with this cohort over the past decade through the Transitsearch.org project, and small-telescope observers played a large role in the discovery of the two longest-period transits, HD 17156b, and HD 80606b.

Once a particular planet has been found to transit, there is considerable scientific value in continued monitoring of the transits. Additional perturbing planets can cause the transit times to deviate slightly from strict periodicity, and a bona-fide case of such transit timing variations (TTVs)  has become something of a holy grail in the exoplanet community. A perturbing body will also produce changes in the depth and duration of transits as a consequence of changes in the orbital inclination relative to the line of sight. Moreover, for favorable cases, a large moon orbiting a transiting planet can produce TTVs detectable with a small telescope from the ground.

New transiting planets are being announced at a rate of roughly one per month. The flow of fresh transits continuously improves the odds that systems with detectable TTVs are in the catalog, but also makes it harder for any single observing group (e.g. the TLC project) to stay on top of all the opportunities.

The Exoplanet Transit Database maintains a catalog of all publicly available transit light curves. At present, there are 1113 data sets distributed over 58 transiting planets. The ETD site provides a facility for photometric observers to upload their data, and also provides online tools for observation scheduling and automated model fitting. Simply put, this is a groundbreaking resource for the community.

The ETD also provides concise summaries of the state of the data sets. Light curves are divided into five quality bins, depending on the noise level, the cadence, and the coverage of the photometry:

It’s interesting to go through the summary reports for each of the transiting planets. Here’s the current plot of predicted and observed transit times for Gliese 436b, the famously transiting hot Neptune:

The data show no hint of transit timing variations. (So what’s up with that e?)

In other cases, however, there are hints that either the best-fit orbital period needs adjustment, or that, more provocatively, the TTVs are already being observed. TrES-2 provides an intriguing example:

In sifting through the database, it looks like XO-1, CoRoT-1, Hat-P-2, OGLE-TR-10, OGLE-TR-132, OGLE-TR-182, TrES-1, TrES-3, and WASP-1 are all worthy of further scrutiny.

Over the past year, as a result of Stefano Meschiari’s efforts, the Systemic Console (latest version downloadable here) has been evolving quite quickly behind the scenes. Stefano and I are working on a paper which illustrates how the console can be used to solve the TTV inverse problem through the joint analysis of radial velocity and transit timing data. In the meantime, it’s worth pointing out that the ETD database lists transit midpoints in HJD for all of the cataloged light curves. These midpoints can easily be added to the .tds files that come packaged with the console.

As is usually the case, there’s been little or no shortage of interesting developments in the field of extrasolar planets. The biggest recent news has been the announcement at the Barcelona conference of a definitive mass for the ultra-short period transiting planet CoRoT-7b. It weighs in at a mere 4.8 Earth Masses (copy of the Queloz et al. preprint here).

Recall that CoRoT-7b caused quite a stir earlier this year with its weird properties. The planet’s year is a fleeting twenty hours and twenty nine minutes, and it induces a tiny transit depth of 0.03%. Unfortunately, the parent star presents a less-than-ideal target for high-precision radial velocity work. It has spots that come and go, and its stellar activity produces frustratingly noisy Doppler measurements. As a result, at the time of CoRoT-7b’s initial announcement, there was no definitive measurement of the planet’s mass.

That’s changed, however, with an unprecedentedly all-out deployment of the HARPS spectrograph. From the Queloz et al. preprint:

A total of 106 measurements between 30 and 60 minute exposure time each were obtained over 4 months, and with sometimes 3 measurements being taken on the same night.

Now in my notoriously biased opinion, such observational enthusiasm is perhaps best reserved for stars such as Alpha Cen B, but a fair argument can be made that the massive investment of time did pay off. Remarkably, the radial velocity data set shows that there are two short-period planets in the CoRoT-7 system. The outer companion, which doesn’t transit, has a period of 3.7 days and at least eight Earth masses. Most dramatically, by combining the mass and radius measurements of CoRoT-7b, one arrives at a density of 5.5 grams per cubic centimeter, essentially identical to that of the Earth, suggesting that the planet is largely composed of refractory materials. (I hesitate to apply the term “rocky” to the CoRoT-7c landscape for the same reason that I’d refrain from describing the Amazon Delta as “icy”.)

In a very real sense, the HARPS campaign on CoRoT-7b has given us our last first look at a fundamentally new category of planet — that is, a world lying in the factor-of-fourteen mass gap spanned by Earth and Uranus. And, from exo-political point of view, the stakes surrounding this discovery were very high. The first density measurement of a planet in this category could just as easily have been made by teams combining high-precision Doppler measurements with either (1) Warm Spitzer, (2) ground-based photometry, (3) Kepler, (4) MOST, (5) HST, or (6) CoRoT. So I can imagine that there was a certain impetus underlying the scheduling of that huge block of HARPS time.

The discovery could, however, still be waiting to be made. Despite all the effort with HARPs, there remains a hefty 70% error on the density determination. This means that there’s a ~16% chance that CoRoT-7b is actually less dense than Neptune.

I’ll go out on a limb: CoRoT-7b’s density will turn out to be anomalously high. More than 90% of “super Earths” will turn out to be “sub-Neptunes” as far as their density is concerned.

Full-resolution Poster-sized .pdf of the above.

The next HD 80606 transit is coming up this week. While the sky position of the star will be much more favorable during the coming January event, observers across the US have an opportunity to get photometric measurements of the ingress early Thursday morning.

The transit begins just after 11 AM UT on Sept. 24, and will unfold over the next 12 hours, meaning that observers in Japan and East Asia will be able to catch the egress.

Josh Winn of MIT is organizing a repeat of the successful June campaign (detailed in this post). If you’re a capable photometric observer, and if you’re interested in participating in the campaign, definitely get in touch with him.