I come to praise Kepler, not to bury it...
The Kepler Mission is one of the little NASA spacecraft that so frequently comes along, exceeds all expectations and changes our perspective of the universe.
There is a good Quick History of the transit method and Kepler Mission concept on the website.
Otto Struve noted in his seminal 1952 note that planetary "eclipses" of their parent stars ought to be detectable by photoelectric methods, a proposal that some two decades later was quantified by Rosenblatt, and then explored in detail (including development research) by Borucki and collaborators at NASA Ames.
A space based transit mission was first proposed more than 20 years ago, and was highly rated, IF the detector technology could get to the point where very low transit amplitudes could be measured and small radii planets detected. The review also noted that there would be significant secondary astrophysical science accomplished by any such mission.
A decade later, Kepler was finally selected, the fourth time it was proposed to the Discover class medium size NASA mission. By that point the detector technology was mature, a testbed demonstrator had been built, and the concept of transit observations of exoplanets had been demonstrated both from the ground, and from space, using the Hubble Space Telescope. The latter demonstrated that very high precision relative photometry was in fact achievable from space.
Kepler launched in March 2009, just over 4 years ago.
It had a nominal mission life of three years, and a main mission goal to find earth size planets in the habitable zone of solar like stars.
After the nominal mission, Kepler was given a three year mission extension to 2016, to continue the continuous monitoring of the 150,000 or so stars in the Kepler field.
Kepler has discovered almost 3,000 planetary candidates, of which about 100 have been confirmed through a variety of techniques, and, statistically, most of the rest are likely to be real planets.
Kepler has not quite found earth like planets in the habitable zone, yet.
It is heartbreakingly close to doing so.
More time for observations is needed, primarily because the stars being observed are a little bit noisier than expected. The periodic signal from the planetary transits can be dug out of the noise, but more observations of repeated transits are needed to get the signals out as you approach the limit of detectability.
Six years of observations ought to get Kepler to its goal of detecting earth size planets orbiting stars similar to the Sun at a distance where liquid water can persist on the planet's surface.
To operate, Kepler's orientation has to be held very stably to view the stars it is looking at. To do that it uses reaction wheels:
Kepler needs three reaction wheels to stay on target.
Any less and the spacecraft drifts, losing lock on the stars.
It has thrusters but cannot use those to stay on target for any length of time before they run out fuel.
Kepler carries 4 reaction wheels. They are heavy and expensive.
That is one spare.
One reaction wheel, wheel #2, failed in 2012.
A second reaction wheel started to show symptoms of degradation a few months ago. Twice in the last few weeks the spacecraft has safed, gone to a rest pointing, while the reaction wheels were despun with the thrusters, and diagnosis tests runs.
Reaction wheels are moving parts, they wear and tear, and have finite life expectancies.
Since the reaction wheels are all the same, they are vulnerable to common mode failure.
Then, last night Kepler went into a safe mode, again.
Switching back to reaction wheel mode the diagnosis showed reaction wheel #4 had seized.
That is the end of Kepler's primary science mission.
The data is in the archives available for analysis. There will be no more.
It is just short of finding the other Earth.
So very very close..
Kepler can do some stuff with only three reaction wheels, basically driftscan observing. It is a wide band wide field optical telescope with a 1 m mirror.
Whether it is worth doing so to keep the spacecraft going will be an interesting decision.
Thanks for the nice piece on very short notice. A few points...
The original spacecraft design was for 4 years. The primary mission was shortened to 3.5 years due to budget pressures. I think that the spacecraft probably reached the four year design requirement for science operations on May 12, 2013.
It is not entirely clear that data collection can not be resumed. During the press briefing, they mentioned that they will consider various options for how to resume data collection. Yes, there's a chance it won't reach the same quality as before, but there's also a chance it could resume planet hunting at a similar precision (e.g., if the reaction wheel that failed previously were to be operable again now that it's been rested). We need to wait for the engineering work to evaluate the situation, so they can devise the best strategy and order for trying things out and determining the spacecraft's current capabilities before writing it off.
Even if no further data collection were possible, there is additional data on the ground that will take time to be reduced before it becomes available to the science team and the public via the MAST archive. There is also additional data still on the spacecraft. I have no reason to believe that data won't be downloaded to Earth, reduced, placed in the archive and lead to good science.
Finally, I'll add my own speculation. I'm not an engineer, so I don't know what is really possible (or perhaps more relevantly, what is financially feasible).
Yes, drift scanning isn't ideal for searching for Earth-size planets, but could Kepler have enough precision with two reaction wheels to conduct a survey of more stars for Neptune or Jupiter-size planets?
I would guess that even with only two reaction wheels the spacecraft could still obtain photometry of eclipsing binaires with precision good enough to search for circumbinary planets (and brown dwarfs, triple stars, etc.) via Ecipse Timing Variations. Whether it will may come down to how much it would cost to modify the software and someone at NASA weighing that against other mission operations.
Of course, it's possible that someone could propose a new and creative use. Eclipsing binary survey? Supernovae search? Near-Earth asteroids? I suspect that the project team at Ames would be interested in hearing from people who have interesting and practical ideas that could become part of a proposal for a modified extended mission for the upcoming Senior Review cycle. It will be interesting to think about these things once the engineers have assessed the remaining technical capabilities.
Quick comment: Kepler has peculiar constraints because the solar cells are on the side of the telescope - it can't point anti-solar, has to point at some oblique angle to Sun.
It also has to maintain Earth communications.
So it has a very hard time pointing now - I think it will be stuck doing a conical drift off the ecliptic.
Secondly, it can't download the whole field - with a fixed field choosing the stamps to diff and send down is easy, with a drifting field data sampling and sending will be hard.
Thirdly, the PSF is huge - about ten times the diffraction limit. This is great for relative photometry of bright stars but degrades almost all other astrophysical applications.
Then the question becomes is the residual science worth $20 million in ops? I fear the answer will be No. After a politely short interval.
Then the question is how much will go to archival research - funding for that ought to be sharply increased, but still less than the mission ops cost, to dig as deep as possible into the data and do associated modeling.
Not that I'm biased or anything...
Kepler already downloads full frame images periodically. I think it is roughly once a month.
If instead of slewing back and forth to point at the Kepler field, Kepler could point in a direction relative to Earth and Sun that allowed quickly alternating between data collection and downlinking without significant slewing, then one could get a lot of new data for new targets without having to define apertures.
In theory, one could pick just the best CCD modules to store/downlink to further cut down on data volume and increase the duty cycle. I can imagine that might be great practice for astronomers who specialize in data mining and help astronomers prepare for LSST. Would that be worth $20M/yr? Would it cost more to rewrite the software? Could it be done for less because NASA would be willing to accept a lot more risk? I don't know.
I definitely agree that NASA should maintain and even increase the funding for archival research. This is a beautiful data set, but it will be a challenge to pull out the smallest and longest-period planets. If there were only 4 years of Kepler data, then it would be even more important to develop out new statistical methods to maximize the planet detection capabilities. Once you start digging into the noise, then it becomes even more important for multiple research groups to perform independent analyses of the data to make sure that results are robust.