Scientists may have discovered the first aquatic worlds

Zoom NASA, ESA, Leah Hustak

Two planets that were originally discovered by the Kepler mission may not be what we thought they were. Based on an initial characterization, these planets were thought to be rocky bodies somewhat larger than Earth. But continued observation has produced data that indicates the planets are far less dense than we originally thought. And the only realistic way to get the kind of densities they seem to have now is if a substantial amount of their volume is occupied by water or some similar fluid.

We have bodies like this in our solar system, most notably the moon Europa, which has a rocky core surrounded by a watery shell covered in ice. But these new planets are much closer to their host star, meaning their surfaces are likely a blurry boundary between a vast ocean and a vapor-filled atmosphere.

There are two main methods for finding an exoplanet. One is to watch for dips in light from their star, caused by planets whose orbit takes them between the star and Earth. The second is to determine if the light from the star periodically shifts to redder or bluer wavelengths, caused by the star moving due to the gravitational pull of orbiting planets.

Either of these methods can tell us if a planet is present or not. But having both gives us a lot of information about the planet. The amount of light blocked by the planet can give us an estimate of its size. The amount of red and blue shift of starlight can indicate the mass of the planet. With both, we can know its density. And density limits the types of materials it can be made of: low density means gas-rich, high density means rocky with a metal-rich core.

That's exactly what we were able to do with the Kepler-138 system. The data from these two methods suggests that the system contains three planets. Kepler-138b appears to be a small rocky body the size of Mars. Both Kepler-138c and Kepler-138d belonged to the category of super-Earths: rocky planets somewhat larger than Earth and considerably more massive. All orbited quite close to Kepler-138a, a red dwarf star, the most distant (Kepler-138d) orbiting at 0.15 astronomical units (one AU is the typical distance between Earth and the Sun).

In the grand scheme of things, there was nothing unusual about this system that would require a second look. But the researchers thought it was a good candidate for studying the planet's atmospheres. While the planet will block all light as it transits past its host star, a small amount of light will pass through the atmosphere on its way to Earth. And the molecules in that atmosphere will absorb certain specific wavelengths, allowing us to discern their presence.

To carry out this study, a team of researchers obtained data from the Hubble and Spitzer space telescopes, timed for when Kepler-138d transited in front of the star. And that's when things started to get weird.

With three planets crammed into a small area near the red dwarf, they are close enough to each other to influence their orbits. These create what are called "transit time variations", which means that a planet does not come in front of its host at the exact time its orbit would normally take it there. For example, one of the planets may be in a position where its gravitational pull will slow down another, causing its transit to start a little later than the calculations suggest.

It can also provide bounds for planetary mass estimates, so accurate measurements of transit time variations are good to have. And, because the Hubble and Spitzer observations came quite a long time after the Kepler data, this meant we could calculate the variations over a seven-year period.

Turns out we couldn't. If you estimated masses based on Kepler measurements and then tried to use them to predict transits in later measurements, you would fail. In fact, everything was turned upside down. "No three-planet model can simultaneously reproduce Kepler-138d's Kepler, HST, and Spitzer transit times," the researchers conclude.

That probably sounds awkward. But if a three-planet model failed, the researchers had an obvious alternative: try a four-planet model instead. And it succeeded in making sense of the data. He also provided an estimate...