Exoplanet Atmospheres: What We Don’t Know

by Paul Gilster on October 20, 2010

What happens in the atmosphere of a tidally locked world in the habitable zone of a red dwarf? We have solid work suggesting through simulations that habitable conditions could exist there, but it’s also true that we’re in the early stages of these investigations and we have no actual examples to work with. Drawing hasty conclusions is always dangerous, particularly when we’re talking about the details of atmospheric circulation on a planet no one has ever seen.

Take Gliese 581g. Assuming it exists — and there is still a bit of doubt about this, although the consensus seems to be that it’s really there — we can place it in a temperature zone that would allow life. We don’t know for a fact, though, that it isn’t a water world, covered entirely with deep ocean, a planet that migrated from beyond the snowline into its present position. And even if it is a rocky planet with a substantial atmosphere, our simulations of atmospheric circulation only represent the best that is known today. This early in the game, we should expect surprises.

Addendum: I’m evidently wrong about the consensus re Gliese 581g, as it appears that its existence really is doubtful. Check the comments to this post to learn more.

A Lesson from a ‘Hot Jupiter’

An instructive case is the planet Upsilon Andromedae b, which although hardly in a habitable region (it’s a hot Jupiter orbiting — and tidally locked to — an F-class primary in a 4.6-day orbit), has yielded useful information about conditions within its own atmosphere. The blindingly obvious notion that a tidally locked planet should have its hottest region directly in the center of the Sun-facing side turns out to be untrue, and not just by a small amount. For new work via the Spitzer Space Telescope tells us that Upsilon Andromedae b’s hot spot is fully 80 degrees away from high noon, on the side rather than the star-facing center.

Ian Crossfield, lead author of the paper on this work, has this to say about Upsilon Andromedae b:

“We really didn’t expect to find a hot spot with such a large offset. It’s clear that we understand even less about the atmospheric energetics of hot Jupiters than we thought we did.”

Upsilon Andromedae b is thus a cautionary tale, as if more were needed, about the dangers of over-extrapolation. Exoplanet atmospheric science is, in any case, an infant discipline, and aside from the simulations of red dwarf planet atmospheres (Joshi, Haberle and Reynolds’ work at NASA Ames kicked this off in a 1997 paper in Icarus), we’ve focused almost entirely on hot Jupiters, using transits where available to study atmospheric composition. We’ve found water, methane, carbon dioxide and carbon monoxide in their atmospheres in a series of remarkable investigations.

The Spitzer instrument measured the total light of Upsilon Andromedae b and its star in the infrared, which is how the unusual temperature distribution came to light. The system turns out to be brightest when the planet is at the side of the star as seen from Earth rather than when it is behind the star, showing its Sun-facing side (Upsilon Andromedae b does not transit). It’s incumbent upon theorists to come up with a solution to this latest atmospheric riddle. In the hunt are star-planet magnetic interactions and the effects of supersonic winds but one suspects we’ll be hearing of other possibilities.

The Danger of Quick Assumptions

Back to the red dwarf question. We have to be careful about making too many assumptions about Gliese 581g and any other planet discovered to be in the habitable zone of a red dwarf. I admit to being optimistic, but there is so much we don’t yet know. Suppose that Gl 581g is indeed an ocean planet. Centauri Dreams reader Dave Moore has commented in the past about a paper by Timothy Merlis and Tapio Schneider (Caltech) that offers encouraging results re tidally locked ocean planets, with moderate temperatures and a super-rotating atmosphere that maintains a mild climate (interestingly, the coldest parts of the planet are the poles, not the anti-stellar point).

Is a rocky world in the habitable zone going to be equally moderate in its temperatures, and will its precipitation patterns follow those suggested by Merlis and Schneider for a water world? We can speculate all we want about the hydrological cycle and so on, but it’s the encounter between theory (and simulations) with observational data that handily supplies the monkey wrench, as it just has in the case of Upsilon Andromedae b. Until we have observations of tidally locked red dwarf planets beyond their wraith-like appearance in hundreds of hours of radial velocity data, continued caution on habitability is the only recourse.

The Merlis and Schneider paper is “Atmospheric dynamics of Earth-like tidally locked aquaplanets,” in press at the Journal of Advances in Modeling Earth Systems (full text). The Upsilon Andromedae b paper is Crossfield et al., “A New 24 micron Phase Curve for Upsilon Andromedae b,” accepted by The Astrophysical Journal (preprint).



Chris October 20, 2010 at 11:13

I think that a lot of this exoplanet studies are going to have to be in the field of speculation for the next few decades. We really don’t have the instruments yet to get the kind of accuracy that we need to take measurements. I’m not even sure that we ever will get all of the information until we go there.

Heck, we don’t even know completely what Mars what like in its geological history and we’ve sent probes there. How are we supposed to know much about extrasolar planets?

Carl Keller October 20, 2010 at 12:07

Despite its slow counter-rotation, Venus has a distribution of heat throughout its atmosphere. It was assumed to be somewhat warmer than our own world, but the discoveries during the twentieth century exceeded all theories.

It is possible that Venus-type terrestrial planets will be found, even ‘super-Venuses’. Our second planet needs much more study, as science wishes to know why the evolution of similarly-sized neighboring worlds diverged so much. With four basic major planets and several large moons for comparison here, there are still enough variables to assert that, of terrestrial planets, those with biospheres, or conditions which promote them, can be a minority of exoplanets.

I read an account… I cannot recall in what book… that contemporaries of Eugène Antoniadi had observed a variation of features on Mercury, suggesting rotation other than tidal lock. But the piece had said that no one could refute Antoniadi’s reputation, and the ‘discovery’ of Mercury’s true rotation came as late as the 1960′s. This isn’t the only time progress would have been stifled due to culture within disciplines. Some pet theories are often based on hopes, as we all hope to find something kindred “out there’.

We stand at a time comparable to that of Cassini himself, who looked through long tubes with a small aperture objective with no color correction. Comparable. Things advance very quickly these days, considering the first exoplanets were discovered in 1992. The official planet count is already 490 http://planetquest.jpl.nasa.gov/ with hundreds or thousands more in the near future.

Lots of space for other biospheres in our galaxy. I’ll stay tuned.

Mike October 20, 2010 at 12:16

I wonder how much more we will learn from using the James Webb Space Telescope? As it is our most solidly funded new instrument we’ll have to depend heavily on it’s capabilities for the next decade or so. One hopes the next generation large ground based telescopes will also be completed in a timely fashion as well. After that, who knows what will be financed? We will need very advanced equipment to answer the really interesting questions regarding potential habitability and for examining possible evidence of biological activity.
There are some wonderful instruments in the planning stage. But will the funding be provided to build and use them? I think these very expensive projects are the only likely near-term way we will have to answer the questions that we all want answered unless the SETI people stumble unto something.
We all would like more information, more hard data to better inform our speculations. We can get it if we’re willing to pay for it. In these difficult financial times it’s unclear what will be approved and funded.

Tarmen October 20, 2010 at 12:51

Caution for sure. Also native life itself might change the climate of tidally locked places. On Earth, the forests, the grazers, the humans and the planktons are big players that modify the “natural” climate that our world would otherwise have without any life here .

(And then there’s possibly completely ‘articial’ , ‘terra’-formed climates, kept that certain way by an intelligence of some kind. ie like instinctual insectoid actions or maybe alien beavers’ workings. )

Thomas October 20, 2010 at 15:00

Why do you say that the consensus is that GJ 581 g actually exists? My impression from what I heard of the recent Torino conference is that the HARPS people demonstrated
1) If it was there they would have seen it
2) The signal Vogt et al. found vanishes if you allow the other planets in the system to be eccentric.

So far, Steve’s response seems to be “well, it’s in the HIRES data, so it’s there;” which ignores the fact that the HIRES RVs all have larger errors than the HARPS measurements.

Paul Gilster October 20, 2010 at 15:12

Interesting. Am I wrong about the consensus? Would appreciate thoughts from some of our resident astronomers in case my impression was wrong.

andy October 20, 2010 at 17:45

One of the nice things about the history of the discovery of confirmed exoplanets is that it has coincided quite well with substantial increases in the computer power available to the public. Therefore it is perfectly possible for an interested amateur to investigate the exoplanetary data. The Systemic Console is a very useful tool for doing this, and it is free to download. The data files are all in text format, so it is easy to add and modify the files to investigate data not supplied with the program.

Regarding Gliese 581, the published HARPS data (dating from 2009) is available from the VizieR service here (bear in mind that the velocities are quoted in km/s, while the systemic data files require m/s), and the Vogt et al. (2010) velocities, while truncated in the arXiv paper, are tabulated in the version of the paper available here (pdf). Systemic already includes system definition files for Gliese 581 dating from 2007, these can easily be adapted to incorporate the new velocities.

The first investigation I did was to see what the datasets on their own could find. Fitting planets to the HARPS 2009 dataset easily brings up planets b, c, d and e, but has no convincing evidence for any further planets. There are peaks near the proposed periods of f and g, but they are way down in the noise level. The Vogt et al. (2010) dataset is less effective at finding planets, I can only find evidence for planets b and c before running into a forest of alias peaks: I cannot find planets d or e in this dataset, let alone f or g. It is only when I run a fit incorporating both the HARPS 2009 data and the Vogt et al. (2010) data that I can find evidence of planets f and g. This fits with what the Vogt et al. (2010) paper says, but it immediately leads me to be suspicious that what is being seen is an artefact of the data merge rather than a genuine planet.

The next thing I investigated was whether we would expect to see evidence for planets f and g in the already-published HARPS dataset. I created 20 fake datasets based on the HARPS observing times, reported observational errors and the Vogt et al. (2010) model of the system, including stellar jitter. What I found was that in 90% of cases, it was easy to retrieve models of the system including the four currently-known planets and at least one of planets f and g; in the most common case, occurring 60% of the time, I could find both. Obviously 20 trials is not a particularly large sample, so there is substantial uncertainty here even without including the possibilities of unmodelled systematic errors, but I think the case can be made that if these planets did exist, signs of them would have likely been seen way back in 2009 when planet e was announced. The statement from the Torino meeting that these planets still do not appear in the (as yet unpublished) new data is even more worrying.

Definitely it would be useful if more people investigated the published data, at present I personally am very sceptical about planets f and g.

Paul Gilster October 20, 2010 at 21:31

Informative indeed, andy! Thanks for this assessment, which makes me think I was wrong in saying the consensus backs the existence of planet g. The case is clearly far more problematic.

Thomas October 20, 2010 at 22:39

Hi Paul,

I actually do work in exoplanets, and I only brought this up because most people I talk to have become skeptical that 581 g exits (so I was startled to see you say that most people think it is a sure thing!). From what I know, Steve Vogt’s group has not presented a compelling rebuttal to the HARPS team’s arguments. Specifically, at the Torino meeting they presented HARPS RVs of GJ 581 over a longer time baseline than what Vogt used in his paper – with no evidence for ‘g’. Also, as I noted above, in their own simulations the HARPS people say that they can recover 581 g if they place it in their dataset, and they make the *very* good point that if you allow for eccentric orbits the planetary signal can disappear (Vogt et al. fixed everything to circular orbits).

Just aside from that though, it is a little odd that the planet is only seen with the HIRES data, since these points have larger errors and we’re pushing to the detection limits of both HIRES and HARPS. More suspiciously, Vogt et al. report ‘g’ and ‘f’ have *exactly* the same orbital velocity amplitudes – which seems odd that the mass and periods of these two planets conspired in just such a way to that possible.

Steve knows HIRES like the back of his hand, so if anyone could pull this planet out of the data it would be him, but it seems to me that right now the community is leaning towards ‘g’ being a spurious detection.

Paul Gilster October 21, 2010 at 7:42

Thanks, Thomas. The point about eccentric orbits seems telling. I appreciate the Torino update, and it’s good to get a report on where the consensus really is right now. I’ve been assuming we would hear something from Vogt’s group and wonder what direction they’ll take.

djlactin October 21, 2010 at 11:44

I can’t help but be amused: twenty years ago, “we” were still looking for exoplanets. Now, “we” are concerned about what “we” don’t know about their atmosperes! How the paradigm has changed!

Mark October 21, 2010 at 19:25

When I hear about 581 g appearing and disappearing out of the data, I’m reminded of some Star Trek episode where an advanced civilization wishes to be left alone, so they cloak their entire planet. Of course being an advanced civilization they’d have to somehow cloak the gravitational effects as well as the EM signatures. Perhaps 581 g hosts such a civilization, and they’re installing and tweaking the system. So now you see it, now you don’t.

philw1776 October 31, 2010 at 11:55

Great object lessons with 581 g on the use of stats in science. Hard #s are sometimes not. Error bars and sigmas really need to be utilized. Here and sadly in other sciences the human wish to produce a desired outcome because of philosophy or politics can sometimes produce results that aren’t really there. Vogt really went out on a limb with his pronouncements. That all said, I share his enthusiasm. Keep working at it, people. These continue to be wonderfull times.

ljk January 23, 2011 at 2:22

Hot Jupiter Radius Anomalies Explained

Authors: Konstantin Batygin, David J. Stevenson, Peter H. Bodenheimer

(Submitted on 19 Jan 2011)

Abstract: We present calculations of thermal evolution of Hot Jupiters with various masses and effective temperatures under Ohmic dissipation. The resulting evolutionary sequences show a clear tendency towards inflated radii for effective temperatures that give rise to significant ionization of alkali metals in the atmosphere, compatible with the trend of the data. The degree of inflation shows that Ohmic dissipation, along with the likely variability in heavy element content can universally explain all of the currently detected radius anomalies. Furthermore, we find that in absence of a massive core, low-mass hot Jupiters can over-flow their Roche-lobes and evaporate on Gyr time-scales, possibly leaving behind small rocky cores.

Comments: Submitted to ApJ, 7 pages, 6 figures

Subjects: Earth and Planetary Astrophysics (astro-ph.EP)

Cite as: arXiv:1101.3800v1 [astro-ph.EP]

Submission history

From: Konstantin Batygin [view email]

[v1] Wed, 19 Jan 2011 23:51:22 GMT (397kb,D)


Comments on this entry are closed.

{ 3 trackbacks }