Sunday, February 14, 2016

Accessing the Kepler Mission data

I've been asked several times over the years:
Would you be able to give advice on how to download a NASA Kepler light curve and run a quick analysis on it?

Usually the query is on behalf of an ambitious and motivated student from a high school or middle school.  I've put together a set of brief instructions to let you (and others) get started.

The NASA Kepler mission was launched in the early morning hours of March 7th, 2009, with the goal of assessing how common Earth-sized planets are around other stars.  The results have been phenomenal, with several thousand candidate exoplanets found by the mission and ~1000 confirmed/validated, many of those in multiple planet systems.

How does Kepler work? Briefly, it stared at a part of the sky 10 degrees x 10 degrees square in the constellations of Cygnus and Lyra for four straight years, pausing only to relay data back to Earth or for temporary spacecraft malfunctions (safe modes).  It has since entered a new "zombie" life as K2, but I won't get into the details of its new mission.

Every 30 minutes (or 1 minute in some cases), Kepler collected a measurement of the brightness of ~150,000 carefully selected stars in that field of view.  We call the brightness measurements as a function of time a "light curve" and/or photometric time-series.  Kepler exploits the transit method for finding exoplanets, which occurs when the planet orbiting another star "transits" in front of (eclipses) that star as seen from our vantage point here in the Solar System.

The transit method for finding exoplanets

Since the stars are bigger than planets (except in some extreme cases!), the planet does not block all of the stars light, but a characteristic percentage that is related to the square of the ratio of the size of the planet to the size of the star.  For an Earth-sized exoplanet transiting in front of a Sun-like star, the exoplanet would cause the star to dim while the planet was transiting by 0.01%, or 100 parts per million.  Jupiter sized exoplanets, ~10x bigger than the Earth, produces an easier to detect dimming of ~1%.   The transit duration (how long it lasts) and how often transits repeat are also important, but let's just stick with the dimming amount, or transit "depth" for now.

Sounds easy right?  No.  Oftentimes there is "activity" going on the surface of the star, including flares and starspots.  These can also produce irregular and semi-periodic changes in the brightness of the star, like this example, Kepler Object of Interest (KOI) 254:

Kepler light curve for KOI-254.  The wavy pattern at the top is because of starspots on the surface of the star.  Even a flare around Day 70 looks like it is visible (the horizontal axis is time in days; the vertical axis is brightness).  The Jupiter-like planet that transits in front of this star causes the short dimming events you see below a normalized intensity of 0.94.

And all this has to be done for very shallow dimming events for the smallest planets.  The noisiness of the digital cameras and within individual pixels within those digital cameras starts to matter too.

There are a number of online tools that you can use to access Kepler data, and I've organized them roughly in order of level of difficulty in terms of learning curve.

1) Planet Hunters

For the act of transit searching, there is a citizen-science project called Planet Hunters that is quite successful and found and published some planets that the official Kepler Mission team missed!

It's a great way to get acquainted with the Kepler data with an easy-to-use interface.  Many amateur astronomers have gotten involved and contributed to our body of knowledge coming from the Kepler mission.


Another great tool for visualizing Kepler light curves, is maintained at MAST, the Mulkulski Archive for Space Telescopes:

You search for your target here. It helps to have the name of your target - either Kepler Object of Interest # for candidate planet host stars, Kepler # for confirmed planet host stars, or Kepler Input Catalog (KIC) # for any one of the host stars Kepler monitored, or the coordinates (Right Ascension and Declination).  More on getting those down below.

The website will return any matching results in a table.  Please bear in mind that these websites are coded by small teams of engineers and scientists who don't have the resources of thousands of web programmers that places like Google and Facebook have to make their complicated (and slick) search interfaces.


Then, clicking on one of these links will take you to a page that lets you plot the light curve in your web browser:

In the above example you can easily see the dips in brightness from the transiting planet, and you can use your mouse to zoom in to see a single transit event:

There it is, raw Kepler data!  You can even hover over the plot and see the values of the data being plotted.    What I like to do with my students is have them manually measure the transit depth of a Jupiter-sized planet (since easy to see) and compute the radius of the planet from the square root of the transit depth.  You can work this up in a simple spreadsheet like this one:

3) NASA Exoplanet Archive

The NASA Exoplanet Archive lets you find, visualize and compute periodograms (find significant repeating patterns) for all Kepler light curves here:

All you would need would be the KIC ID of the target, or you can search by Right Ascension and Declination.  Fortunately, the NASA Exoplanet Archive also maintains lists of Confirmed and candidate Kepler exoplanets, which gives you more naming information (KIC#, KOI#, Kepler #), coordinates, and model stellar sizes, to name a few important properties:


For example, on the Kepler light curve search page, you can enter in the Kepler ID (KIC #):

8561063, and Click "View"

This is the KIC # for the multi-planet system Kepler-42, also known as KOI-961.  Kepler-42 is one of my favorite Kepler systems, discovered by some of my colleagues.  It's got three planets that transit, between the size of Earth and Mars, and they orbit a dim red dwarf star!

In the results table, if you click on the link:
(PDCSAP Time Series)

for any entry, it will take you to the light curve visualization tool with all the Kepler time-series for that star (Kepler data is split into "quarters" typically of ~90 day length).
This particular source has a LOT of Kepler short cadence data, so it takes a long time to load the interface.  But eventually you will be rewarded with seeing the actual Kepler data that was used to discover this fascinating multi-planet system around a cool red dwarf star.

I was responsible for the scientific input for a lot of these tools when I used to work at the NASA Exoplanet Archive.  So maybe I'm a bit biased!

From there, you can refer to the User Guides to learn more about everything that can be done to manipulate the data within the browser, including computing some advance time-series analysis tools called "periodograms".  If you want to do more, you can download the light curves too and play with them offline with your own favorite software like Microsoft Excel.

4) PyKE
Another far more complex tool to download is called Pyke: 

I wouldn't recommend it for beginners, and it helps to already be familiar with the Python programming language and/or IRAF astronomer tools.  But PyKE allows you to interact with the Kepler light curves (and individual detector pixels) at a very low-level.

No comments:

Post a Comment