Planet Hunters is a citizen science project that makes it possible for anyone to sieve through data taken by the NASA Kepler space mission. The Kepler spacecraft takes brightness measurements, or "light curves," of over 150,000 stars every 30 minutes. People can then hunt for planets by looking for a brief dip in brightness that occurs when a planet passes in front of the star.
Join the search at http://www.planethunters.org, or read on to learn more.
NASA's Kepler spacecraft is one of the most powerful tools in the hunt for extrasolar planets. The Kepler team's computers are sifting through the data, but we at Planet Hunters are betting that there will be planets which can only be found via the remarkable human ability for pattern recognition.
This is a gamble, a bet if you will, on the ability of humans to beat machines just occasionally. It may be that no new planets are found or that computers have the job down to a fine art. And yet, it's just possible that you might be the first to know that a star somewhere out there in the Milky Way has a companion, just as our Sun does. Fancy giving it a try?
The Kepler Public Data
On March 2009, the NASA Kepler mission was launched with the goal of using the transit technique to detect exoplanets: terrestrial and larger planets orbiting other stars. With this method, planets that pass in front of their host stars block out some of the starlight causing the star to dim slightly for a few hours. The Kepler spacecraft stares at a field of stars in the Cygnus constellation and records the brightness of those stars every thirty minutes to search for transiting planets.
The time series of brightness measurements for a star is called a light curve. The Kepler spacecraft beams data for more than 150,000 stars to Earth at regular intervals. With every download of data, the time baseline of the light curves is extended.
The project's Principal Investigator, Bill Borucki, began planning the Kepler mission in the mid-1980's and his team has been hard at work for more than a decade. To reward them for this hard work, the Kepler team has advanced access to the light curves. We at Planet Hunters are not part of the NASA Kepler team. However, NASA is releasing light curves into the public archive to encourage broader participation and we think that the public can play an important as our scientific partners in this latest Zooniverse project.
Since 1995, more than 500 exoplanets have been discovered by various techniques and it appears that roughly half of the stars in the sky have planets. There are some special challenges for planet detection with the transit technique:
- A special orientation of the orbit is required. Because the technique looks for a dimming in the brightness of the star, all of the planets with orbits that don’t pass between the star and our line of sight will be missed.
- Close-in planets are easier to detect. A transit event only occurs once per orbit; planets that are closer to their host stars race around their orbits faster than planets that orbit at larger distances. To confidently detect a transit, at least three dips in the brightness (i.e., three transit events) must occur. Thus, a planet that orbits in one year, like the Earth, requires three years of data for detection, while planets that orbit in ten days can be detected with just thirty days of data.
- Larger planets are easier to detect. The bigger the planet, the more starlight it blocks out. The Kepler mission measures the brightness of stars with such incredible precision that it is sensitive enough to detect transits of planets approaching the size of Earth.
- Planets may be harder to detect against a variable brightness background. This turns out to be less of a problem than one might think. When we look at a light curve, we're seeing how the brightness changes with time. In the Figure below, there are starspots in addition to the transiting planet (lower left spot on the star). However, starspots rotate with the star and cause relatively slow changes in the brightness of the star. Transiting planets cross the star in hours and cause quick dips in the brightness of the star. Look below to see the difference (left image). Spots cause most of the smooth and slowly varying brightness and we're learning that many stars have much larger spots than the Sun.
Humans vs. Machines
The Kepler team has been developing computer algorithms to analyze light curve data because it is not possible for them to visually inspect every light curve. While we expect computer programs to robustly identify things that they are trained to find, we are betting that there will be a number of surprises in the data that the computer algorithms will miss.
The human brain is particularly good at discerning patterns or aberrations and experiments have shown that when many people work together, the collective wisdom of the crowds can be better than an expert. Planet Hunters is an online experiment that taps into the power of human pattern recognition. Participants are partners with our science team, who will analyze group assessments, obtain follow up observations at the telescope to understand the new classification schemes for different families of light curves, identify oddities, and verify transit signals.
Planet Hunters: Sorting the Light Curves
You will be looking at changes in star brightness at a level that has never before been seen. As you sort through the light curves, you will notice different patterns. In many cases, the data scatters in a relatively flat band of points, like the cases shown in Figure 1. Most of this scatter is simply the inevitable noise that comes with any measurement. Other light curves, like those in Figure 2, are obviously variable with time. We think that most of the variability (on timescales of hours to days) is caused by starspots or pulsations. Having Planet Hunters sort families of similar light curves is part of the important scientific research.
Figure 1. Even precise measurements are not exactly perfect or reproducible and cause low-level scatter in the data. There is no visible pattern, just white noise, and so these light curves are best categorized with the middle 'quiet' icon.
Figure 2. These light curves should be tagged with the “variable” icon; then you will be asked to decide if the curve is pulsating with one cycle (like the top left curve) or regular (like the top right curve) or irregular (like the bottom curves).
It is challenging for computer algorithms to classify variable patterns so participants are making a particularly important scientific contribution in this step. We will obtain follow-up data at the telescope to understand the underlying mechanisms for these different families of variable curves and to confirm transit candidates.
Planet Hunters: Flagging Transit Events
However, the real treasure hunt is for transiting planets and these present as a relatively sharp dip in brightness in the light curve (Figure 3). A transit could appear in either a quiet or a variable curve. Indeed, it will be more difficult for computer algorithms to find transits imposed on the variable light curves, so we hope that Planet Hunters will pay particular attention to these.
Figure 3. After sorting the light curves as quiet or variable, Planet Hunters will be asked if there are any possible transits in the data. If low points are seen, then answer 'yes' and click on the icon to create a box that can be positioned over the transit features.
The size of the planet is reflected in the depth of the transit points. Earth-sized planets will exhibit a dip in brightness that is buried in the noise of the quiet light curves in Figure 1. The transit events in Figure 3 are for planets that are several times the radius of the Earth.
The time it takes a planet to complete one orbit is called the orbital period. For transiting planets, this can be determined by counting the number of days from one transit to the next. The examples in Figure 3 are fairly obvious. Planets in longer period orbits will be more challenging to detect, both for humans and for computers because a transit will not appear in every 30-day set of light curve data. Just because you don’t see a transit in the first block of data doesn't mean that there won’t be a transit in another set!
Large planets with short orbital periods are the easiest ones to detect. The most challenging detections will be small planets with long orbital periods. These will require patience and care, but are the real treasures in the Kepler data!
I See a Transit!
What happens if Planet Hunters discover a possible transiting planet? We maintain a list of transiting planets that the Kepler team announces, so the first thing that will happen is that we will check that list. If the flagged transit event is for a star that the Kepler team are already keeping an eye on, we'll let you know. If this event has not been identified and several Planet Hunters are flagging the same data, the science team will investigate. If this appears to be a new discovery, then we will follow up to obtain spectroscopic data using the Keck telescope in Hawaii. If the transit candidate passes all of the screening tests, the result will be submitted for publication. Planet Hunters who discover new transiting planets will be included as co-authors on our papers.
Experimental Data: Faking the Transits
To better understand what types of planets are detected or missed by Planet Hunters, we have created a small number of test cases for some of the light curves that participants will classify. These test cases contain fake transit events and are critical for determining the statistical completeness for planets as a function of size (depth of the transit event) and orbital period (number of transits). After participants classify a data set with a fake transit event, a message will appear notifying them that this was a test case, and the fake points will be highlighted.