Exclusive Interview:

David Koch

Kepler Deputy Principal Investigator

By Keith Cooper
Posted: March 09, 2009

David Koch is the deputy principal investigator of the Kepler mission, which successfully launched into space on 7th March to begin its three and a half year mission to seek out Earth-like planets. In an exclusive interview conducted at the end of 2008, Astronomy Now’s Keith Cooper talks

in-depth to David about how the Kepler spacecraft will go about this ambitious task.

What is different about Kepler to previous planet finders?
One mission is the COROT mission, and the big distinction between COROT and Kepler is just the overall sensitivity to detection, and it comes in several categories. First of all, how much time do you spend looking at any given starfield? COROT, I believe, is spending only 120 days on a given starfield, so that limits the orbital periods that you can detect. Basically it means you can only detect planets that are very close in to their star, with orbital periods of about a month or so. Number two is the minimum planet size that you can find. In all these missions you are limited by ‘shot noise’, that is the number of photons that you collect during a transit. That is defined basically by the aperture of your telescope. I think our aperture is three times the diameter of COROT, so that is nine times the collecting area and it also means three times lower shot noise on the star, so we can go to three times smaller area for a given star, which means we can get to smaller-sized planets. The other factor is how large is the field of view of the instrument? For COROT, I think it is a few square degrees, 2-4 square degrees; we’re looking at a hundred square degrees. So there are three parameters: the mission duration is ten times longer, the collecting area is nine times larger, and the field of view is like a factor of 20-50 bigger. So those three parameters tell you that Kepler is going to be able to detect smaller planets in wider orbits.
      From the ground, the big difference is that the atmosphere prohibits you from detecting transits smaller than one percent. So you can’t do photometry on the ground except for giant planets. You can detect things like Jupiter, but you can’t find an Earth looking through the atmosphere. If you ask about radial velocity methods, you will find that they can do Jupiters, they can do things smaller than Jupiter in short period orbits, but it is absolutely impossible to find an Earth-like planet from any kind of ground based observing.

Will Kepler be able to find terrestrial planets?
Let me first say that generally, people have referred to terrestrial planets as things up to about ten Earth masses, but going back to one Earth mass, we expect something of the order of forty to fifty Earth-like planets that we could detect with our system over three and a half years.

What kind of orbits will they be in?
They will be in Earth-like orbits as well, so we are talking about Earth-sized planets in a 1AU orbit, and we are expecting something like forty to fifty detections. That is based on the assumption that stars like our Sun have terrestrial planets.

To confirm a transit of any planet you have to monitor several transits. Will the duration of the mission be long enough to confirm Earth-like planets?
Yes. The way to look at it is if you have three transits, one year apart, the total time between the first transit and the third transit is really only two years. But that is kind of a pathological case, you need three years to see three transits that occur in any random phase, and the original plan for four years was to get another confirmation of the system, which we will get in many cases but not all cases if we just do three and a half years. So the plan is to still be able to find Earth-like planets in one year orbital periods around Sun-like stars.

Kepler is just staring in one direction of the sky. How did you decide where to point it?
The thing you want to do is maximise your results. So you want a rich starfield, with as many stars as you can get. That means you want to look into the galactic plane, and you prefer to look down a spiral arm of the galaxy rather than across the spiral arm of the galaxy. It doesn’t matter after a while because if you look too far out the stars are too faint to do the measurements. So it is only really the local part of the Galaxy that matters. The other thing is that we want to view the same piece of sky continuously for the entire duration of the mission. That basically says that you cannot look into the ecliptic plane, because the Sun will cause a problem every year for that part of the sky. So you have to look out of the ecliptic plane, and the question is how far out do you have to go? That is defined by the mechanical geometry of the telescope and the launch vehicle. It is defined by the length and the size of the sunshade that you have to use. Our sunshade has a Sun-avoidance angle of 55 degrees.

That means we have to look above 55 ecliptic latitude or below –55 ecliptic latitude. There are only two parts of the galactic plane that are above or below those values. That narrowed down the fields to select from, and then you just get out a star catalogue and start doing star counts in those two pieces of sky to figure out where do you get the most stars. It turned out to be the Cygnus region that had the highest number of stars that we could get out. It also turns out that all of our follow-up observing resources are in the northern hemisphere versus the southern hemisphere. It has nothing to do with an expectation that there is a particular part of the galaxy that has more planets than another, it is just where do you get the most stars.

How will you follow-up on interesting planets?
Well, there are two parts to the follow-up. One is that you have to eliminate false positives, and the other is what interesting things can you learn about the planet that you have discovered. We have to do follow-up observing to eliminate false positives caused primarily by background eclipsing binaries. That is, you are looking at a star that is, say, twelfth magnitude, but in the background is a binary system that is five magnitudes fainter, or even more, and when that eclipses, you get a small dimming of that fainter object, and it is blended with the target star that you are looking at. It gives you the appearance of a terrestrial-sized transit. What you have to do then is go to a ground-based telescope, and figure out, ‘hmm, was there a background binary near the target I’m looking at’, and secondly start doing some moderate-precision spectroscopy to look for radial velocity shifts in the lines. What you will find is that if it is a binary system, then the radial velocity shifts are very high, they are like kilometres per second, as opposed to if you are looking at a gas giant, it is a few metres or tens of metres per second. You do some imaging and you do some spectroscopy, and in that way you eliminate any potential false positives that get into the system, and we expect to see a lot of them. There is also the false positive case where you don’t have a background eclipsing binary, you have a background star with a Jupiter transiting, so we expect we detect lots of Jupiter’s in the background when we are actually trying to look for a transiting terrestrial planet in the foreground star. Those will be interesting detections too but they won’t be what we’ve built the mission for.

Once we have a real detection, then first of all you want to ask yourself, ‘tell me everything you can about that star’, tell me its mass, tell me its radius, tell me its temperature. From the radius of the star I get the radius of the planet because the transit just tells me the ratio; I need to get the mass of the star so I can plug it into Kepler’s law and get the distance that the planet is from the star; if I’ve got the distance of the planet and the temperature of the star, I can then calculate where the habitable zone for that star is. Those are the things that we can get easily. The next question is, can I get the mass of the planet? It is doable if it is a giant planet; Jupiter is 300 times the mass of the Earth. The Swiss group that has the HARPS instrument are building a copy of that for the William Herschel Telescope in the Canary Islands and a team from Harvard-Smithsonian are funding the development of that instrument and with it we will then be able to measure the masses of, first, if it is an Earth-like mass in a short period orbit we can actually measure the mass, and if you have the size and the mass you can get the density. If you have the density then you can ask if it is rock, if it is a gas giant, is it an icy planet, so getting the density is the next major step. The next thing you would like to know is something about its atmosphere, and as you know that has been done for giant planets like HD209458, but it is a lot harder to do that for terrestrial planets. We are hoping that of the terrestrial planets that we find, there will be a number of them around bright enough stars that we can then do follow-up spectroscopy of the atmosphere with things like the James Webb Space Telescope.

Will Kepler be able to see random microlensing events?
Technically it is possible, there is a chance that it could happen, but in general I would not think that we would see any, and the reason is that the CCDs that we have produce a lot of data. We do not send down [to Earth] a full image of the 100 square degrees. We only have pixels of interest around the stars of interest, so unless the star happens to be a target star on our list, we wouldn’t be able to see a lensing event even if it were in our field of view because we wouldn’t get the data.

How far away will you be able to see transits?
The typical distance for stars is somewhere around 200 parsecs up to a kiloparsec. There really aren’t a whole lot of stars closer than 200 parsecs, hence the probability of detecting anything is slim. Beyond a kiloparsec the stars are so faint that you won’t be able to see a transit, except perhaps for things like gas giants. It is a good sample of our Galaxy.

From that sample will you be able to get an idea of how typical our own Solar System is?
The intent is to understand the frequency of Earth-sized planets, and the way you do that is that in essence you look at an unbiased sample of stars and you measure how many of the various types of planets you can find, different sized planets in different orbits, and what you try and understand is, is the orbital distribution of planets similar to what we have in the Solar System, is the size of planets similar, is the number of planets similar. We expect, for example, that ten percent of the time we will see things like both an Earth and a Venus, because if you detect one planet that means you are looking in that system’s ecliptic plane, and if the planets’ orbital alignments are distributed the way they are in our Solar System then you have a fairly good probability of detecting multiple planets in that solar system, since the orbital inclinations are pretty much co-aligned. We expect to find not only single planets but multiple planets, and then the other part of the follow-up programme that I skipped over was, once you’ve found an Earth, you then go the guys on the ground and say, ‘now go to that star and see if you can find a Jupiter, go find out what else is in there’. If it is a Jupiter with a five year period, we wouldn’t see it in our mission as transiting.

How difficult is it separating multiple planets?
That’s the fun part, I’m looking forward to those cases! They are easy to separate for several reasons. One is that you are looking for something that is perfectly periodic, and if I give you two things at two different periods, with several transits you should pick that out. Number two is that for each planet, being in a different orbit, the transit duration is different. The transit duration depends upon the orbital period and the size of the star. So if the orbital period is different, the duration is going to be different. And if the planets have a different size, then the depth of the transit is going to be different. So you have the period, the duration and the size of the planet all as signatures that will make it lots of fun to separate out multiple planet systems. Man, I’m looking forward to that day!

The mission is scheduled to last three

and a half years; is an extension possible? We could have an extension of two and a half years, so we could go for a total of six years.

For more information about the mission be sure to read Kulvinder Singh Chadha's feature Kepler's quest for other Earths in the March issue of Astronomy Now, and stay tuned to www.astronomynow.com for updates throughout the mission!