Over the next ten weeks (well, nine weeks from now as this post is one week late), I'll be working in Professor John Johnson's Exolab with Tim Morton at Caltech's astrophysics department. The Exolab is interested in the study of extrasolar planets.
Background: Transits and False Positives
An extrasolar planet is commonly detected through studying its transit across a star. Generally, astronomers find this planet by plotting the flux of a star with respect to time and finding the transit signal. If there is a dip in the amount of flux, then something could be blocking the light of the star; hopefully that something is a small planet moving across the star's surface from our point of view. Typically, however, the amount of light blocked is very small. The value is called delta, which varies with (Rplanet/Rstar)^2. For a Jupiter-sized planet, delta ~ 0.01, and for an Earth-sized planet, delta ~10^-4!
However, there are other things that can cause the flux to dip. Binary star systems can also cause the flux to lower, though delta is usually much greater for binary stars in comparison to planets (delta ~ 0.1). Also, for binary star systems, if we plot the flux vs time graph long enough, there should be a second dip as the other companion gets blocked behind the first star. Under certain conditions, however, binary star systems can be misinterpreted as extrasolar planets transiting across stars due to the way their transit signal looks, or the way the light is collected, leading to what we call a false positive.
Solution (Where I come in)
In order to avoid interpreting false positives as transiting planets, we need to be able to better plot the light from each star as a function of its distance. Thus, for my project, I will be analyzing CCD images taken by the Palomar 60'' telescope fitted with the Robo-AO and plotting the brightness across each CCD image. If there are any anomalies in the light curve, such as a sharp spike in brightness far away from the star, we can conclude that there are other sources contributing to the amount of light in the CCD image. This light can then be subtracted out (along with background noise) to get the light curve of the main star. This technique is called photometry, or the measurement of the star's flux.
In Tim's words, what I'm doing in a nutshell is determining "with a certain degree of confidence" if there are or aren't any bright objects in a patch of sky that can influence the transit signal.
I will be doing photometry using CCD's/FITS files and writing scripts with Python.
A CCD is a device that contains an array of "wells" capable of holding around ~40k photons per well. As a certain number of photons hits the well, an electron will be emitted and registered onto a pixel. These pixels, along with some other information, form an image called a FITS file. The ratio of electrons emitted (in ADU's, or Astronomical Data Units) to the photons absorbed (or counts) is called the Gain of the CCD.
Though a star is considered as a point source of light, due to atmospheric effects, or seeing, the light of the star is smeared out across a circle of pixels determined by the point spread function. The point spread function should theoretically be determined by the resolution (wavelength/diameter of the telescope); however due to seeing it is not. Thus, the FITS files we obtain of the stars will have a bright center surrounded by rings of increasingly dim pixels.
We can plot the intensity of the light with respect to radius and obtain a bell curve. Significant deviations from the bell curve can be marked as potential sources of light and subtracted out. However, a trickier problem is finding the brightest an object can be and yet not be detected at five times the noise (which we will call sigma). This has to do with measuring and subtracting the background noise of the sky.
The final plot I will make will be a plot of the magnitude of the star as distance from its center increases. If there are no other significant sources of light, this plot should be smoothly declining.