Finding Earth’s Twin: No Easy Task

NASA’s Kepler telescope will search for Earth-like planets around distant stars.
Credit: NASA/Ames Research Center/W. Stenzel (OSC).

Does our galaxy contain twins of Earth: rocky planets, orbiting sun-like stars at comfortable distances, capable of supporting life? That is the million-dollar question for exoplanet hunters. NASA is about to launch an orbiting telescope designed to find an answer. Its name is Kepler.

Kepler, scheduled for launch on March 6, 2009, will observe a field of 100,000 stars in the Cygnus-Lyra region of the sky, searching for planets by looking for their transits. A transit occurs when a planet passes in front of the star it orbits, causing the star’s brightness to dim temporarily. The amount of dimming that occurs indicates the planet’s size; the bigger the planet, the more starlight is blocked. The frequency of the transits reveals the planet’s orbital period and distance from its star; closer-in planets have shorter-period orbits, so transits occur more frequently.

If viewed from the proper vantage point in space, Earth would appear to transit the sun once each year, dimming the sun’s brightness by about one-hundredth of one percent. (No, that’s not a typo.) Transits like these are what the Kepler mission is most interested in detecting.

“Kepler is designed to find hundreds of Earth-sized planets, if such planets are common around stars,” says Bill Borucki, a space scientist at NASA Ames Research Center in Moffett Field, California, and the science principle investigator for the Kepler mission. Most of the Earth-sized worlds it finds, however, will orbit too close to their stars, and therefore be too hot, for life as we know it. But Kepler could find as many as 50 Earth-sized planets in Earth-like orbits. “If we find that many, this certainly will mean that life may well be common throughout our galaxy,” Borucki says.

That’s if Earths are common. They may not be. But that, too, Borucki says, “will be [a] profound discovery. It will mean that Earths must be very rare; we may be the only extant life in our universe.” And what’s worse: “It’d mean there would be no Star Trek.”

Even if there are plenty of Earth twins out there, Kepler won’t find them until near the end of its three-and-a-half-year mission. “The big, massive, hot Jupiters, as they’re called, are going to roll off the Kepler assembly line” first, says Debra Fischer, a planet-hunter and assistant professor at California’s San Francisco State University. Next will come Neptune-sized worlds, and even rocky planets like Earth, but close in, with orbits much smaller than Earth’s. Closer-in planets will reveal themselves first because they will transit more frequently.

To feel confident that they have detected a candidate planet, the Kepler team needs to see three transits, three identical dips in a star’s brightness, timed at precise, regular intervals. For planets orbiting close-in, three transits could occur within the first month of Kepler’s operation.

For a planet in an Earth-like orbit, though, it will take the full length of the mission, three to three-and-a-half years, to see the requisite three transits. “We’re not going to be able to tell you very quickly whether or not there are Earths. We’re going to have to wait until we see these three transits,” Borucki says.

An artist’s rendition of Jupiter (left) and Earth (right) transiting the sun, as viewed from outside the solar system.
Credit: NASA/Ames Research Center.

And that’s only the first step, because there will be false positives: dips in a star’s brightness that look like transits but are actually caused by something else.

Noisy stars, for example. The sun is a fairly quiet star, its brightness doesn’t vary much, but others stars are quite noisy. If a star is too noisy, if its brightness fluctuates constantly, the Kepler team may not be able to extract the dimming signal caused by a transit from the background variations. It would be like trying to hear someone whisper in Grand Central Station during rush hour.

An Earth-twin world, when it transits, produces a very small signal, barely a whisper. To detect an Earth twin, says Jim Fanson, Kepler’s project manager, “we have to be able to measure the brightness change of stars down in the 20 parts per million level. It’s akin to measuring a flea as it creeps across the headlight of an automobile at night.”

Starspots can also cause problems. A starspot, a dark region on a star’s surface, will cause the star’s light to dim as the spot rotates into view. So a starspot can look like a transit. But a starspot’s signal – how quickly the dimming occurs once it begins, and how long the dimming lasts – has a different profile than that of a transit. “It takes a considerable amount of time for starspots to rotate into view and then out of view,” says Jon Jenkins, Kepler co-investigator for data analysis. Transits happen more quickly. Part of the job of the software developed by Jenkins and his colleagues will be to distinguish between the two.

A binary star, a star system in which one star circles the other, can also produce a dimming effect that looks like a planetary transit. At least half of the stars in our galaxy are part of a binary or multiple-star system, so the Kepler team will have to weed out these false positives, as well.

That weeding will be done through follow-up studies using the radial-velocity planet-hunting technique. A planet exerts a gravitational tug on a star as it orbits, causing the star to wobble. This wobbling induces slight changes in the wavelength, or color, of the star’s light. The larger the planet, the larger the wobble; the larger the wobble, the greater the wavelength change; the greater the wavelength change, the easier it is to detect.

In a binary system, the object tugging on the target star is another star; its gravitational pull is huge, far larger than that of any planet. This is what will make it easy for radial-velocity measurements to tell a transiting planet from a binary star.

Indeed, all of the candidate planets found by Kepler will be subject to confirmation through radial-velocity studies. Until these follow-up studies are done, the Kepler team plans to hold off on announcing its discoveries. These announcements are expected to come in annual batches, near the beginning of the year, with the first batch announced in 2010. Kepler will collect data continuously and send it back to Earth once a month, but according to Bill Koch, deputy PI for Kepler, for the follow-up studies, “there’s an observing season.” The region of the sky where Kepler will be pointed is overhead during the summer. “So we’ll take data continuously, but only during the summer can you do the follow-up observations.” And then it will take a while to process the data.

There’s a catch, though: radial-velocity measurements will work for confirming planets larger than Earth, or Earth-sized planets that orbit close-in to their stars, but they aren’t sensitive enough to detect Earth-sized planets in one-year orbits. “The radial velocity people cannot detect those. If they could, we wouldn’t be flying this mission,” says Koch.

The Kepler field of view lies between Deneb, in the constellation Cygnus, and Vega, in the constellation Lyra, two of the brightest stars in the northern summer sky.
Credit: Milky Way photo by Carter Roberts/Graphic by NASA/Ames Research Center.

Does that mean the Kepler team won’t really be able to make unambiguous detections of Earth? That depends on how much you like to quibble. Once larger, closer-in planets are confirmed, Kepler’s planet-finding capability will be validated. That will boost the astronomy community’s confidence in its methodology. And quiet stars will make for more confident detections than noisy ones. In addition, the Kepler mission may be extended to as long as six years, enabling Kepler to see not three, but five or six transits for each Earth-twin candidate it finds. As each additional transit occurs on schedule, the likelihood will increase that a planet is responsible.

But there is still one astronomical phenomenon that could confound Kepler’s data analysts: a background eclipsing binary. Imagine that in the nearby background of a target star there is a faint binary pair, one star periodically eclipsing the other. As an eclipse occurs, the output of the binary pair dims. But the binary is so close to the target star that when the background eclipse occurs, it looks to Kepler like the target star is dimming. In other words, it looks like a planetary transit. And it happens on a regular schedule.

If the eclipsing binary appears far enough away from the target star, software filters will be able to tease it out and reject it. But if it appears extremely close to the target star, just on its edge, it may prove impossible to identify.

Koch isn’t overly concerned about this problem, because it’s likely to occur only rarely. “If we find dozens of Earth-like planets, then we’ll be less reticent at releasing the findings because we know that only a very small fraction of those might be contaminated by this phenomenon,” he says. But if Kepler finds only one or two Earth-twin candidates, it will be hard to know for certain whether it has found planets or background eclipsing binaries.

Koch is confident the problem will get sorted out eventually. “Even if Kepler isn’t directly able to confirm its discoveries in all cases, and even if we don’t have a space mission flying that’s capable of making these measurements now, inevitably in the future we’ll be able to detect all the false positives” that Kepler finds.

“But,” he adds optimistically, “our intent is not to have any.”