Tuesday, February 08, 2011

Our planets are weird

The Kepler data release has delivered a huge amount of data on planets orbiting stars in our galaxy.

This survey is designed to detect planets as small as earth (or a little bit smaller), and the way it works means that it most easily detects planets that are very close to their host stars. So most of the stars in this survey orbit their host star closer than Mercury orbits the sun. Despite this, figure 1 shows that there is something very unusual about our solar system.

Figure 1. Green lines are the radii of the planets in our solar system. Earth and Venus are blended together at this scale. The red line is the probability distribution of planets in the Kepler survey.

The most common planets in the Kepler survey have a radius of about twice that of the Earth, or about half that of Neptune. There are no such planets in our solar system. In fact, this gap is what we traditionally call the dividing line between ‘terrestrial’, or rocky planets, and giant planets (the giants are then further divided into the ice giants (Neptune and Uranus), and the gas giants (Jupiter and Saturn). There is nothing in between the ice giants and the rocky planets in this solar system. And yet, that is the most common size of planet in the Kepler survey.

There are a few caveats here. Firstly, is the peak is real or is an artifact related to the difficulty in finding earth-sized planets around dim stars? We can plot the magnitude of the star vs the size of the planet to determine this (figure 2).

Figure 2. Magnitude vs. planetary radii Higher magnitude stars are dimmer.

If the Kepler abundance peak around 12600 km (2 earth radii) in figure 1 is related to difficulty in finding smaller planets around dimmer stars, then we would expect the high magnitude stars to show fewer planets with a radius between 1 and 2 earth radii (the lower right portion of the graph). This area does not appear to be under-populated. So that caveat appears to be irrelevant to this study.

Another caveat is that the planets seen so far are all close to their host stars, and may represent migrated outer planets. So the distributions may be biased by a process that did not occur in our solar system. We can look at the period vs size distribution to test this (figure 3). If the earth-sized and smaller planets are all similar to the planets in our solar system, then they may have similar orbits. If the larger planets are all scattered in from the outer solar system, then they may have a different orbital distribution.

Figure 3. Kepler planets plotted by period vs. radius. Colored dots are Kepler planets, white stars are Mercury, Venus, and Earth. Period is in days, radius is in earth radii (6372 km).

What we see is that Kepler has not yet detected any earth-sized planets in periods as long as Mercury (much less Venus or Earth). That’s OK, such a detection is not expected for a few more years. All of the earth-diameter planets are in very short orbits. However, a few of the planets with a size intermediate between earth (1 earth radii) and Neptune (4 earth radii) have orbits that are broadly similar to that of Mercury or Venus. So these planets can not be definitively identified as scattered from their orbital period alone.

So, from the data we have so far, it appears that the galaxy is full of unfamiliar planets. Or, from the galaxy’s point of view, our solar system is devoid of normal planets, and only harbors oddballs.

See part 2 for further discussion of selection effects.


Anonymous said...

I tend to agree with your general thesis, that solar systems like ours seem to be pretty rare, with the caveat that you made that Kepler won't have actually found SSs like ours yet. However, your second plot isn't very meaningful.
The fainter stars in the Kepler field mostly tend to be K and M dwarfs, and significantly smaller than the sun, and that ups the depths of transits and thus the detection probability. This probably cancels out the lower s/n given by their faintness and possibly at least partially cancels out the fact they tend to have a higher noise floor, since they're more chromospherically active, on average. A better plot would be that of spectral type against planetary size. As I've commented elsewhere, the earth sized planets detected in habitable zones (they have to all be around K & M dwarfs with the periods surveyed so far) seem to make up only 10% of ALL planets detected around those stars. Since these stars are often thought of being the best place to look for intelligence, since their lifetimes are so long combined with the fact that they're far more numerous, this may be the first step towards resolving the Fermi Paradox.

andy said...

I'm starting to wonder whether maybe the reason we have rocky inner planets is because of the Jupiter-Saturn configuration. Simulations suggest that such configurations migrate outwards rather than inwards (unlike Saturn-Jupiter configurations or Jupiter-Neptune configurations). The outward migration of Jupiter and Saturn may have prevented the inner system getting swamped with icy material and producing mini-Neptunes rather than terrestrial planets.

Would be very interesting to see what kind of planets are produced in systems without significant outer discs, e.g. Alpha Centauri where the circumstellar discs would have been truncated close to the ice-line.

Chuck said...

Dear Anonymous: Thank you for your constructive comments. An update awaits. Note, however, that the phrase "As I've commented elsewhere" isn't that useful if we don't know who you are.