Kevin Schofield's writings, observations, and other pointless distractions
There was good news today from NASA about the Kepler space telescope, which suffered some wear-and-tear along the way to completing its original mission, and a key mechanical fault last summer which prevents it from continuing with “business as usual.” Its new mission, dubbed “K2,” is a tribute to the ingenuity of NASA scientists.
Kepler, launched back in 2009, is a space telescope specifically designed to look for “exoplanets” or planets orbiting other stars. Since planets don’t emit light, and don’t reflect much either (especially at great distances) they are very difficult to detect. And thus historically we have had a very poor understanding of the likelihood of planets orbiting stars – and more to the point, the existence of Earth-like planets in the “habitable zone” where just the right amount of solar radiation hits the planet to allow for liquid water. Together, that would give us strong hints as to whether life might exist elsewhere in the universe.
So how do we detect whether there are any planets orbiting a star, when we can’t see the planet directly? It’s very clever, actually: we look for a “transit” when the planet crosses in front of the star (where “front” is defined as “the part of the star pointing at us”). When a planet transits its star, from our perspective the star gets slightly dimmer for the duration of the transit. Of course, the devil is in the details: an Earth-like planet transiting a Sun-like star will cause a dimming of only 95 parts per million (i.e. for every million photons sent our way from that star, 95 are blocked), so we will need a super-sensitive light sensor to be able to detect that. Also, the full transit might take anywhere from hours to months, so we need to be watching constantly for long periods of time. Plus, since there are other potential causes for a star to dim, we need to observe at least three transits in order to confirm an actual planet: that gives us two full orbits where we can observe how long it took the planet to make an orbit – and that both orbits took the same amount of time, and the transit dimming was of the same intensity and length.
Kepler was designed to do all of this very well. It has an array of 21 camera imaging chips that collectively take a 95 megapixel image every 6 six seconds with a sensitivity up to 20 parts per million of light. It takes in an image of a portion of the “celestial sphere” 105 degrees by 105 degrees, with which it can simultaneously observe over 100,000 stars.
Since one of the requirements for Kepler is to maintain observation of the same area for years (originally 3.5 years), it’s important to ensure that not only is its view not occluded, but that it’s also not affected by the gravitational pull of Earth, the moon, or other celestial bodies; as well as ensuring that light from the sun doesn’t interfere with its imaging sensors. Because of that, NASA decided that it should orbit the sun, not the earth; so Kepler sits in a heliocentric orbit nearly identical to Earth’s, trailing behind us and with a period of 371 days (as opposed to Earth’s 365.25 day orbit). It’s close to us now, but slowly falling further behind. 61 years after it was launched, we will catch up to it on the front of Earth’s orbit, and in the middle of that time it would be on the opposite side of the sun from us and probably out of contact, but right now it’s well within range of the Deep Space Network and will certainly remain so for longer than its useful life.
To understand how Kepler works, you have to understand a bit about its orientation in space. Think of Kepler as similar to a #2 pencil: a long cylinder, where the point is the camera, and 2 adjacent sides are covered with solar panels. If you were to hold the pencil pointing up and move it in a circle parallel to the floor, that would approximate Kepler’s orientation as it circles the sun: with its side to the sun and its camera always pointing up and capturing the same chunk of space (with the caveat that the diameter of an Earth orbit is inconsequential compared to the distances to the stars it’s taking pictures of). Every 3 months Kepler rotates 90 degrees so that its solar panels continue to face the sun and provide power to the telescope; the camera likewise rotates, but it never changes direction: it’s always perpendicular to the orbit, pointing at the same slice of the sky.
There is a complicated system aboard the telescope that ensures with extremely high precision that it always points the same way. It has a thruster rocket that it can use for coarse movements, and a set of four “reaction wheels” that can be controlled to spin at high frequencies to provide specific amounts of angular momentum and maintain a given direction through tiny corrections. Kepler has redundancy here: while it has four of these wheels, it only needs three in order to maintain a stable orientation in three dimensions. This is the kind of thing you do when you’re sending a $600M piece of delicate equipment to a place where you can’t ever make a service call to fix it – recall that even if we still had a working shuttle fleet, unlike Hubble the Kepler telescope isn’t in orbit around Earth – it’s approximately 1/10 of an Earth orbit away from us.
And it’s a very good thing that NSA built in redundancy in the reaction wheels, because wheel #4 started acting up a couple of years into the mission – it was experiencing “high friction” despite having a supply of lubricant. High friction meant that its axle was rubbing, which would generate excess heat and could probably add vibration and jitter – not good. In January 2012, all the wheels were spun down in the hope that the lubricant would redistribute itself, but lubricant apparently has a mind of its own in a zero-g vacuum environment, and the problems persisted when they spun them back up.
In July 2012, wheel #2 failed. Wheel #4 continued to experience high friction, but soldiered on and allowed Kepler to complete its original 3.5 year mission in November 2012 with three operational wheels. NASA extended the mission for an additional 2 years, with the hope that wheel #4 would continue to limp onward. Last May, however, wheel #4 failed, leaving Kepler with only two operational wheels – enough to allow Kepler to continue to take lower-resolution pictures but severely degrading the value of continuing the mission. But since the vast majority of the costs of Kepler were in building and launching it, not in ongoing operation, NASA did something unique and creative: it asked people to submit ideas for new potential missions for Kepler, on the belief that just because it couldn’t continue its extended mission any further didn’t mean it was completely useless.
And thus was born “K2” – a creative proposal for how to “MacGyver” a way for Kepler to keep taking high-resolution photos of exoplanets. It came from the recognition that a constant force in a constant direction applied equally across an object can in fact help to stabilize that object. In the case of Kepler, that force is solar radiation – the photons hitting Kepler (mostly its solar panels, but other parts exposed to the sun as well) have energy and momentum which are transferred upon contact. The problem that needed to be solved was that in Kepler’s current orientation – perpendicular to its orbit – the telescope isn’t presenting a regular, symmetrical shape to the sun, and thus the force is being applied unevenly, causing jitter and spin. But if Kepler were rotated to be parallel to its orbit, rather than perpendicular, it gets a much more consistent bombardment of photons that will help to stabilize it – enough that with the two remaining reaction wheels and the occasional thruster correction, they can keep it stable enough for high-resolution work.
The downside: in its new orientation, the camera is constantly changing direction as it orbits the sun. Even if they used the reaction wheels and thruster to keep its attitude corrected to always point in the same direction, for half the year at least part of the sun will be in front of the camera, and for part of it the sun will completely occlude it. So what they have cleverly decided to do is to have the camera focused on the same chunk of the sky for 83 days at a time, then reposition it to a new chunk. Within each 83-day “campaign” they expect to get 75 days’ worth of imaging – not enough to identify multiple transits of planets in long orbits, but enough to capture close-in planets. As a point of camparison: Mercury orbits the sun in 88 days, and Venus in 223 – so K2 would almost certainly catch one transit of a Mercury-like planet, and it would have about a 1/3 chance of catching a Venus transit. All of those campaigns will be parallel to the “ecliptic plane” of our solar system, but there is certainly more than enough stars to see out there to justify it.
Kepler will use its thruster more frequently than it otherwise would, but the NASA engineers believe they can get another two years of full use out of the telescope before the thruster fuel is exhausted.
It’s important to remember that Kepler’s mission was never to catalogue the entirety of the celestial sphere around us: it was simply to survey a large enough slice to be representative of the rest, so that scientists could do statistical analysis. We know approximately how many stars there are in our galaxy, so if we can start to accurately approximate the likelihood of an Earth-like planet orbiting one of them, we can make a very educated guess about how many Earth-like planets are out there altogether. And Kepler successfully finished its original mission, so we have that data (though there is a backlog of many years in processing it all – which is why we keep hearing about new planets that Kepler discovered even though it has been inoperable since last May). Everything we additionally collect from this point on just helps us to further refine our approximations.
The big news today was that NASA officially gave its approval for K2 to move forward. I applaud the spirit and creativity of the NASA team that figured this all out, and I wish them great success in making it happen. As of today, Kepler observations have confirmed 966 exoplanets, with 3845 additional candidates awaiting confirmation. Keep going!