Sifting for Planets with Kepler's Fine Comb

by Seth Shostak, Senior Astronomer
Here's the challenge.  Take a bare 100 watt light bulb and switch it on.  Now step back about 300 miles. Once you're in position, arrange for a friend to slowly pass a pinhead 30 feet in front of the bulb without notice or warning. Your job?  Detect the decrease in light when the pinhead gets between you and the bulb.

I suspect that's not something you do every day.  But NASA's Kepler telescope will be doing it every half-hour for the next three years and more.

Actually, not quite. Kepler will be measuring the brightness of more than 100,000 "light bulbs."

This new NASA space-borne instrument, which is now completing its shakedown cruise, is engaged in the ultimate staring contest.  Kepler will continuously monitor the luminosity of 145,000 stars in the region of constellations Cygnus and Lyra, looking for dimming of as little as 0.006 percent of a star's brightness. Unlike other schemes for finding planets around distant stars (so-called "exoplanets"), Kepler can unearth Earths. That is, it can detect worlds hundreds of light years away that are comparable in both size and orbital position to our home planet. Cousins of the Earth – and obvious candidates for life.

Kepler is perched a ten million miles into space, far beyond our world's troublesome, restless atmosphere. In truth, it's a 100 megapixel camera, taking a picture (and pretty much the same picture!) every 30 minutes. During each month of this maniacally repetitive surveillance, Kepler's most recent photo collection is sent back to terra firma.

But not all of it. Think about it: 100 million pixels every half-hour for a month?  With about a byte per pixel (usually only brightness differences are recorded), that's more than a hundred gigabytes of data – a gush of results that would be painful to transmit from orbit.  What to do?

Dealing with that question is part of Jon Jenkins' job description. Jenkins, a SETI Institute scientist who leads the Kepler Signal Processing/Detection Algorithm team, reduces the data bandwidth by taking advantage of the fact that Kepler's photos are mostly of empty space, punctuated by stars.  The stellar targets, even when slightly splattered by optical imperfections or telescope drift, brighten fewer than 5 percent of the telescope's pixels. It would be silly to send blank sky (including stars too faint to measure or background galaxies) back to the lab, over and over.

"It's like condensing a high-school yearbook," Jenkins says. "I'm only interested in my friends, so I can clip out their faces and leave most of the pages behind. Saves space."

To account for slightly smeared star images, optical changes in the telescope or drift caused by pointing error, an average of 32 pixels are allocated to each star – a patch of the CCD detector called a "stamp."  These are the yearbook faces that are returned to Earth.

The data follow a tortuous path after being telemetered from Kepler to one of NASA's Deep Space Network antennas.  They're first routed to the Jet Propulsion Lab, and then sent on to Kepler's Mission Operations Center in Boulder, Colorado. With processing taking place at every step along this yellow brick road, the information is forwarded to the Space Telescope Science Institute (think "Hubble") in Baltimore, where it's re-organized into an easily readable format. It eventually lands in the offices of Jon and the rest of the Kepler team at NASA's Ames Research Center in Mountain View, California.

That's a long trek, but fortunately the journey is highly automated. At Ames, 88 processors pipeline-process the data – combing through it in a search for planets.  Light curves – the astronomers' term for plots of brightness versus time – are generated for all 145,000 stars. Those that seem to occasionally dim – the flecks of gold in the mountains of dross – are flagged for the attention of the scientists.  Each month's data will typically have a few hundred "hits."

Not all hits will score, of course. There are always false positives requiring further scrutiny. For example, micrometeorites occasionally slam into Kepler's sunshade, kicking off dust that crosses the field, catches some sunlight, and shows up in the data as a star whose brightness has suddenly changed. Other candidates for mis-identification include so-called eclipsing binaries – double stars that happen to periodically get in front of one another.

Bottom line: Not every change in brightness is caused by a planet. So the Kepler team is eager to get its first results analyzed this summer while telescopes on Earth can verify at least some of its discoveries (the constellations Cygnus and Lyra are in the nighttime sky now). This "reality check" will give everyone confidence that – if Kepler turns up smaller, Earth-like planets – the results will be credible.

Finding those terrestrial worlds will take time – a few years, at least.  But that fact differentiates this mission from most astronomical exploration. For many forays into the unknown the excitement occurs with the first few discoveries – the first pulsars, the first quasars, the first black holes. After that, it's merely a matter of adding to the collection.

But for Kepler, the excitement waxes, rather than wanes. In its first month or two, Kepler will uncover large numbers of "hot Jupiters" – bulky planets in tight orbits. But we know about those.  Eventually, however, the small fry will surface, the rocky worlds that might be the abodes of extraterrestrial life.

"The excitement is in that glimmer of hope," says Jenkins. "The capability to find plenty of Earths – if they're there for us to find."

"It's like snorkeling over a coral reef," he says. "In the past, we did this with only our naked eyes. The view was blurred. Now we have goggles. I think we're going to see wondrous things."