oklo

Image Source.

As I write this, it’s JD 2454425.219 (17:16 UT, Nov. 20 2007). HD 80606 b whipped through periastron a little more than 10 hours ago, and the Spitzer Space telescope is literally just finishing its 31-hour observation of the event. Next comes the downlink of the data to Earth on the Deep Space Network, and then the analysis. Definitely exciting!

The Spitzer Space Telescope is scheduled to run out of cryogen in early 2009. When the telescope heats up, we’ll lose our best platform for mid-infrared observations of hot extrasolar planets, and so there was a palpable urgency last week as everyone prepared their proposals to meet the submission deadline for Spitzer’s last general observing cycle. During the next few years, there is going to be intense development of detailed 3D radiation-hydrodynamical models for simulating the time-dependent surface flows on extrasolar planets. These models will need contact points with hard data. It’s thus vital to bank as wide a variety of observations of as wide a variety of actual planets under as wide variety of different conditions as possible. A number of fascinating exoplanet observing proposals were submitted last week by a variety of highly competent teams. I’m urging that they all be accepted!

Most of the exoplanet observations that have been done with Spitzer have focused on tidally locked transiting planets on circular orbits. HD 189733b, HD 209458b, TrES-1 and HD 149026b are the flagship examples of this class. In the past year, however, eccentric transiting planets have started turning up. Gliese 436b (e=0.15) was the first, followed by HAT-P-2b (e=0.5), and HD 17156b (e=0.67).

Drake Deming, Jonathan Langton and I decided that the most interesting proposal that we could make would be for Gliese 436 b. This is the Neptune-mass, Neptune-sized planet transiting a nearby red dwarf star. Here’s the to-scale diagram of the 2.644-day orbit:

After Gliese 436b was discovered to transit last spring, it triggered a Joe Harrington’s standing Target of Opportunity program. Both a primary and a secondary transit were observed (see this post) which confirmed the startlingly high eccentricity, and which allowed an estimate of the planet’s temperature (or, more precisely, the 8-micron brightness temperature). This turned out to be 712±36 K, which is significantly higher than the ~650 K baseline prediction.

The hotter-than-expected temperature measurement could arise from a number of different effects (or combinations of effects). By measuring the secondary eclipse, you strobe one hemisphere of the planet. If there are significant temperature variations across the surface of the planet, then a high reading might arise from chancing on the hotter side of the planet. Alternately, the effective temperature implied by measuring the energy coming out at 8-microns could be seriously skewed if the spectrum of the planet has deep absorption or emission bands at the 8-micron wavelength. Another possibility is that we’re observing tidal heating in action. Gliese 436b is being worked pretty hard in its eccentric orbit, and it should be generating quite a bit of interior luminosity as a result. If its structure is similar to Neptune, then a 712K temperature is completely understandable.

Io, of course, is subject to a similar situation. Here’s a K-band infrared photo of Io in transit in front of Jupiter:

Image Source.

Gliese 436b is in pseudo-synchronous rotation, and spins on its axis every ~2.3 days. The eccentricity of the orbit leads to an 83% variation in the amount of light received from the star over a 1.3 day timescale. This leads to a complicated flow pattern on the surface.

Here’s what Jonathan Langton’s model predicts for the appearance of the hemisphere facing Earth at five successive secondary eclipses:

Globally, the hydrodynamical model produces a statistically steady-state flow pattern that is dominated by a persistent eastward equatorial jet with a zonally averaged speed of ~150 meters per second. This eastward flow in the planet’s frame produces a light curve in the lab frame that has a ~3 day periodicity. This period is significantly longer than both the planet’s orbital period and the planet’s spin period. Our Spitzer proposal is to observe a sequence of 8 secondary transits in hopes of confirming both the amplitude and the periodicity of this light curve.

It’s certainly the case that our current hydrodynamical model is not the definitive explanation of what these planets are doing. I won’t be at all surprised if the flux variation from eclipse to eclipse is more complicated than what we predict. I’m highly convinced, however, that the model is good enough to indicate that the situation on Gliese 436b will be interesting, dynamic, and complex. The actual variation in the real observations will provide an interesting and non-trivial constraint that a definitive model of the planet will need to satisfy. The observations, if approved, will thus be of great use to everyone in the business of constructing GCMs for short period planets.

Stay tuned…

The flurry of activity surrounding the detection of the HD 17156b transits, combined with the start of the academic quarter here at UCSC, caused me to fall way behind on my stack. All of a sudden, over two weeks have passed with no oklo.org posts. I think that’s a record, unfortunately.

And it’s not as if there’s been no new exoplanet news. The past two weeks have seen the announcement of two more new transiting planets from Gaspar Bakos’ HATnet project. “Yo, what up TrES?” HAT-P-5b has a period of 2.79 days, and looks good in a 400-pixel wide everything’s to scale diagram:

There are enough transiting extrasolar planets now, so that it’s interesting to look for trends in the planetary properties. At Jean Schneider’s exoplanet.eu site, there’s a nifty set of php routines that make it very easy to dial up all the different correlation diagrams. With the inclusion of HD 17156b, HAT-P-5b, and HAT-P-6b, a plot of planetary radii vs. stellar metallicity is pointing to an interesting trend. It’s quite apparent that the metal-rich stars tend to harbor smaller planets. This seems to be indicating that metal-rich disks are yielding planets that have highly enriched heavy element fractions, which in turn is giving us an important clue into the planet formation process.

If we ignore Gliese 436b, which is far smaller in both mass and radius than all the other known transiting exoplanets, then there’s a fairly obvious hint of two separate sequences in the diagram — a large-radius sequence, and a small-radius sequence. Feel free to voice your opinion in the comments section…

It’d certainly be nice to get more transits by planets orbiting bright parent stars. To that end, it’s important to stress that literally every single planet transiting a V<13 parent star is located north of the the celestial equator. It’s pretty clear that the southern hemisphere Doppler-wobble planets have not been fully followed up with photometric campaigns. I’m thus keen to get the Southern-Hemisphere Transitsearch.org corps out on the sky during the coming austral summer. First on the list is HD 76700b. This 0.20 Jupiter-mass planet orbits with a period of 3.970985 days, and has an extensive and fairly recent set of published radial velocities. I just updated the orbital fit, and found that the transit windows are still quite narrow. A simple bootstrap analysis shows that the uncertainty in the time of the transit midpoint is not much wider than the expected duration of the transit itself. The star is just coming visible in the early morning, and so it should be straightforward to either confirm or rule out a transit for this particular planet.

Image Source.

I’m happy to report that HD 17156b is observable in transit, and that Transitsearch.org observers played the key role in the discovery.

Regular oklo.org readers are familiar with HD 17156 b. This planet has an orbital period of 21.2 days, which is nearly four times longer than any other known transiting planet, and an eccentricity of e=0.67 (even higher than HAT-P-2b’s eccentricity of e=0.5). The geometry of the eccentric orbit has a periastron angle of 121 degrees, which means that the planet is quite close to the star as it perforates the plane containing the line of sight to Earth. Here’s a scale model showing the planet (at equally spaced intervals), the star, and the orbit:

Early last week, I got e-mail from Mauro Barbieri, an Italian post-doctoral researcher who is working on the CoRoT satellite, and who’s based at LAM in Marseille, France. In his spare time, Mauro has been writing articles for popular astronomy magazines in Italy, and has worked to coordinate Italian amateur astronomers for participation in Transitsearch campaigns. On the night of Sept. 9/10, he recruited four Italian amateur astronomers to monitor the transit window.

Two observers in Northern Italy were clouded out prior to the start of the transit, but two others, D. Gasparri and C. Lopresti, were able to observe through much of the night in Central Italy. Their data looked promising, showing clear ingresses at the same time as the ingress was observed by Jose Almenara in the Canary Islands. Ron Bissinger, observing from Pleasanton, California, and seven hours farther west, was able to start observing just after the transit ended.

As it happened, Roi Alonso, another CoRoT postdoc, is good friends with Jose Almenara, who made the Canary Island observations on Sep 9/10. Barbieri and Alonso did a careful analysis of the three transit-bearing data sets, and concluded that the transit is present at 3-sigma, 5.3-sigma, and 7.9-sigma, respectively. Almenara’s data, in particular, is excellent, despite of the fact that high winds occluded part of the mid-transit time series. By last Friday, on the strength of the detections, we had begun drafting a paper that discusses the discovery.

The Almenara egress is particularly convincing (the bottom time series shows the result of subtracting out the best-fit transit signal):

Barbieri’s and Alonso’s fit to the data from Sep. 9/10 implies the following properties for the planet and the transit. The model consistently takes into account the eccentric character of the planetary orbit:

The fit to the data suggests that the radius of HD 17156b is just a bit larger than the radius of Jupiter. This is fully in line with our theoretical models of the planet. HD 17156b experiences strong tidal forces during its periastron passages. This tidal heating might be observable in the form of excess infrared radiation, but it is not serving to inflate the planet beyond its expected radius.

Needless to say, we were quite excited by the quality of the fit. Everything seemed to hang together quite well, but confirmation was essential. The full transit would be visible across the United States and Canada. I wrote to the transitsearch.org mailing list, urging observers to monitor the star through the night of Sep. 30/Oct. 1. Dave Charbonneau of Harvard had been following oklo.org, and was impressed by the Sep 9/10 data. He worked very hard to organize and coordinate observers, and he’ll be leading a follow-up paper that uses data from all the transits to improve the planetary and the orbital characterization. Dave was very generous in offering to notify us if a transit was confirmed by the cadre of observers that he’d recruited.

Sunday morning was perfectly clear in Santa Cruz. Ron Bissinger lives in Pleasanton, just forty miles to the Northeast, and was ready to observe. His pipeline is quite automated, and so if he could observe, I knew that we would rapidly rule out or confirm a transit.

Late Sunday afternoon, I went running, and noticed that a gloomy bank of clouds was visible over the Pacific to the west:

At dusk, it was still clear, but the weather forecast did not look good. The bands of clouds that had stayed offshore to the Northwest were beginning to slide across the skies. By midnight, it was evident that the Bay Area would not be producing useful photometry. Furthermore, all of Arizona and New Mexico were rained out. Observer after observer reported in to say that they had not gotten data. The only positive report of clear skies came from Dave, graduate student Philip Nutzman, and postdoc Jonathan Irwin, who had set up a small telescope on the roof of a Harvard/CfA building in Cambridge MA.

By Monday morning, however, it was clear that several observers had indeed managed to obtain data. In addition to the Harvard roof observations, Bill Welsh and Abhijith Rajan had obtained usable photometry from the Mt. Laguna Observatory run by San Diego State University. Dave reported to me that on Sunday, while Dave was visiting the Zoo with his daughter, Bill had called with the news that they had managed to secure a night on the telescope and were at that moment driving up the mountain. In addition, a report came from Tim Brown of Las Cumbres observatory that while their Hawaii site had been weathered out, observations had been successfully made from parking lot of the observatory headquarters in Santa Barbara. And in addition, Don Davies, a Transitsearch.org observer in Torrance California had obtained a 10,000 CCD frames under good sky conditions.

On Tuesday evening, I got a phone call from Dave. The transit was clearly visible in both the Cambridge and the Mt. Laguna data. Thirty minutes later, I got an e-mail from Don Davies, who, in the early stages of analysis was seeing a clear transit-like signal at the expected time. We signed on Davies as a co-author, added Dave’s personal communication to the paper draft, and submitted the discovery. The paper will be showing up on astro-ph today. Here’s a link to a .pdf version:

Barbieri et al. 2007, AA, submitted (157 kb)

As for HD 17156b itself, the transit should present a number of exciting opportunities for follow-up observations. The large orbit leads to a 26-fold orbital variation in the amount of flux received from the parent star. This should drive complex weather on the surface, and indeed, even the night side of the planet should be glowing from its own radiation. Here’s a frame from Jonathan Langton’s most recent simulation of the planet which shows the night side hemisphere:

And here’s a one-orbit 1 MB animation of the surface flow patterns, glowing and roiling with their own emitted heat.

Image Source.

Tonight, Sept. 30/Oct 1, is the night to follow-up to confirm whether HD 17156b can be observed in transit. Earlier this morning, I sent the following e-mail to the Transitsearch Observers list:

Hello Everyone,

I’d like to alert you to an important follow-up opportunity TONIGHT to observe HD 17156 for a possible transit by its companion planet. North American Observers are best situated for the event.

HD 17156 b has been the topic of several blog posts on oklo.org, see: [1], [2], [3], [4], [5].

Photometry taken by Jose Manuel Alemenara Villa on the Sept. 9/10 opportunity was suggestive of a possible transit with duration 169 minutes, a photometric depth of 0.007 magnitudes, and a mid-transit time of HJD ~ 2454353.614. These values are all quite close to what one would expect if HD 17156b is really transiting.

If the event observed by Alemenara Villa is due to a transit, then the next transit will be centered at HJD~2454374.83 (CE 2007 October 01 07:55 UT Monday) with the transit beginning at about 06:30 UT.

Observing should start as soon as possible this evening, and observers are encouraged to take photometry for as long as possible.

My fit to the published radial velocities predicts a transit midpoint centered at HJD 2454374.87 (CE 2007 Oct. 01 08:52 UT Monday), with a +/- 0.3d uncertainty in the time of central transit. The Alemenara Villa event sits nicely inside this window.

Thanks very much!
best regards,
Greg

It looks like much of the Southwest is clouded out, and although the skies outside here in Santa Cruz are currently cobalt blue, it’s predicted that clouds and even rain will materialize after midnight. SoCal, however, and many locations in the midwest and east look good to go. Here’s a selection of California predictions from the clear sky clock. This is a cool graphical tool for use in scheduling observations. Dark blue is good, white is bad.

In the last post, I pretty much wrote off HD 17156 b, which was the subject of last week’s transitsearch.org photometric follow-up campaign. Ron Bissinger observed the star during the latter part of the transit window, and saw no evidence of a transit. Tonny Vanmunster wrote with the news that Belgium was clouded out.

Soon after the post went up, however, Jose Manuel Almenara Villa of the Instituto de Astrofisica de Canarias posted a comment:

Hi Greg,

I observed HD17156 in the transit window. Unfortunately the night was windy, affecting the small telescope so the photometry is not so clear as we would wish. Anybody else observe?

It’s possible that I have a central transit. I can show you some plots if you want. I will try to observe again on December 3 (I think that is my next opportunity).

Regards,
Jose

On Saturday, Jose sent me his photometric plots, I should point out that he emphasized once again that the night was windy. In his plots (I’ve rewritten the labels in illustrator so that they show up better on the narrow blog-page format) the black dots are individual observations (R filter, 7 s exposures), the red dots bin 6 observations, and the blue dots bin 12 observations.

On the night before the night of the transit window, he got baseline photometry which shows considerably less scatter, and which does a nice job of showing his excellent photometric technique:

He fit a simple trapezoidal transit template to his data. The resulting fit has a duration of 169 minutes, a depth of 0.007 magnitudes, and a mid-transit time (HJD) ~ 2454353.614. These values are all quite close to what one would expect if HD 17156 b really is transiting. The possible event ends just prior to the start of Ron Bissinger’s time series.

So what to think? It’s most important to reiterate Jose’s point that the weather was not particularly good, and that a block of critical data is missing during the event itself. I myself have contracted transit fever several times in the past, and have built up sufficient immunity to refrain from getting too excited. I think a conservatively realistic assessment would say that there’s still an 11% chance that HD 17156 b transits are occurring, and that the uncertainty in the window has been narrowed down significantly. Over the long run, if transitsearch.org is going to succeed, then its important to stay cautiously optimistic. The good thing about a transit is that it repeats with clockwork regularity (barring the unlikely, but tough-luck situation where dynamically induced precession of the node induces transit seasons.) The next chance to observe HD 17156 during the transit window falls to North America on Oct. 1, where hopefully there’ll be multiple observers on the sky. We’re bad – We’re Nationwide…

To end on a heartfelt note, I think that the global collaborative efforts that go into these transitsearch campaigns have been both fun and inspiring, even when the result is the high-probability flat-line light curve. It would be exciting, though, if Jose ends up leading a discovery paper with the other participating observers as co-authors.

Image Source.

It’s not looking good for transits by HD 17156 b. Ron Bissinger of Pleasanton, California obtained a block of photometric data that covered a significant chunk of the transit window. His time series lasts from JD 2454353.68 through 2452353.88, and shows no hint of an event:

His observations were taken just after the peak of the transit midpoint histogram:

No word yet on whether anyone in Europe or the eastern US were able to observe during the first half of the window. If you got data, let me know.

Also, the Gliese 176 window has opened up. If you’ve got a telescope, a CCD, and a free evening, you know what to do!

Image Source.

It’s 01:58 UT Sep. 10, and HD 17156 has moved into its transit window. Hopefully photometric transit observers across Europe have clear skies. If you’re collecting data, drop us a note on the comments page!

Sep 09, 2007 Europe Satellite Map Source.

Most of California looks pretty good for catching the latter part of the transit window once it gets dark tonight. I was up on Mt. Hamilton last night, and even though it was clear, there was a strong smell of smoke in the air. Bits of gray ash from the nearby forest fires were floating down like snow, and so they couldn’t open the dome of the 36-inch.

Image Source.

The odds of a HD 17156 transit are 10.9%, so its best not to get hopes up too high. Its always good to have the next candidate ready to go, and as luck has it, there’s another good one in the hopper.

Endl et al. have published a preprint describing the discovery of a Neptune-mass planet orbiting the nearby red dwarf star Gliese 176 (aka HD 285968). This discovery is further evidence in favor of the core-accretion prediction that Neptune planets should be common around low-mass stars whereas Jovian-mass planets should be relatively rare. Endl et al.’s new planet has an orbital period of 10.24 days, an a-priori transit probability of 3%, and an expected transit depth of 0.4%. This is a low-amplitude signal, but it is nevertheless accessible to many experienced amateur astronomers. The discovery paper makes no mention of a photometric transit search, making this planet a very attractive Transitsearch.org candidate. The star is located at RA 04:43, Dec +18:57, and the next transit window is centered on Sep. 15, 2007.