Subject: telescope data reduction From: David Cohen Date: 11/8/11 11:19 PM To: Matthew Elkins , "Imoleayo S. Abel" , Gabriella Capone , "Sierra C. Eckert" , Yuwen Wang , Seth Chapman , Adrien , Peter Weck , Sergio Rosas , Christopher Gordon West Capron , Catherine Ann Martlin , Jacob Ellington Neely , Chayanon Ruamcharoen , Samantha Leigh Pellegrino , Peng Zhao , Sizhao Wu CC: David Cohen , Eric Jensen , Ana Matković , Mary Ann Klassen , Adam Neat Hello telescope operators, Most of you have been taking data that consists of a sequence of images of a given star (or a field of stars), and you know that we often are looking at transiting exoplanets that make a star dim a bit for a period of time. And you've taken sky flats and other calibration exposures while observing. But most of you have not (yet) done anything with those data you took. I was reducing some data the other day (taken with Peng and Sergio about three weeks ago) and wanted to show you what some of the intermediate steps look like, to hopefully give you a better sense of how the data you take are being used. We also hope that at some point, some of you will learn how to do this sort of data reduction and will want to work on that (perhaps on cloudy nights when you can't observe). So - we were observing the star KC20C20793, which is a transit candidate (we haven't yet confirmed that it really has a transiting planet, because the star also seems to intrinsically vary in brightness, probably due to pulsations). Here's one of our images: http://astro.swarthmore.edu/~cohen/telescope/sample_photometry_reduction/raw_image.png (note the donut-shaped dust shadows, and the general diffuse brightness toward the center of the image) Here's the "master flat" I created by adding together about 15 sky flat images taken at the beginning of the night: http://astro.swarthmore.edu/~cohen/telescope/sample_photometry_reduction/master_flat.png That should look pretty familiar to any of you who've taken sky flats (though it's a little smoother than a single flat). Now, here's the same data image I showed you above divided by this master flat: http://astro.swarthmore.edu/~cohen/telescope/sample_photometry_reduction/processed_image.png Note that you can barely see the dust shadows, if at all, and that the diffuse light in the image is lessened and much more uniform. So, we've essentially corrected for most of the uneven "response" or "sensitivity" in the detector. Next, we want to figure out how bright our target star is in this image (and every other image we took that night). This step has a couple of parts - we can interactively define a circle around the star, count up the light (charge in the CCD pixels, really) in that circle, and also measure the "sky background" (i.e. light pollution) in a nearby annulus-shaped region of the sky. This process is shown, quantitatively in this plot: http://astro.swarthmore.edu/~cohen/telescope/sample_photometry_reduction/radial_profile.png The zero-point of the x-axis represents the middle of the star, and we count up everything to the left of the vertical line at 8 pixels. Note that when the brightness flattens out, it isn't at zero, but rather just a bit below 4000. That's the brightness of just a blank part of the sky (and presumably the location of the star, if the star weren't there) - so the brightness of the star itself is all the counts we measure in the circle with the 8 pixel radius *minus* the counts in the sky annulus (scaled by the relative areas - in pixels - of the annulus and the circle). So, after doing that, we've got a measure of the intrinsic brightness of the star. But...actually not quite, because some of the light is absorbed (or scattered) by the earth's atmosphere. And that amount is changing all the time (amount of air you look through varies with direction, light clouds or haze can come and go, etc.), so we need to compare the brightness of our target star to several other stars in the field (to take out any global trends that affect the brightness of all the stars in an image in the same way). Here I have chosen those comparison stars: http://astro.swarthmore.edu/~cohen/telescope/sample_photometry_reduction/comparison_stars.png (star #1 is the target). We then average the brightness of the other stars, and take the ratio of our target to that average, and *that* is our measurement of the target star's brightness. For that one snapshot. We do the same for every image, and plot the brightness of each image vs. the time the image was taken: http://astro.swarthmore.edu/~cohen/telescope/sample_photometry_reduction/photometry_v2-ap8.00.plot.png That's the light curve that's the final product of one night of observing. We think the decrease in brightness to about 0.15 is due to the star's pulsational variability, but the dip from there to about 0.25 is a transit, maybe. ...more study is needed, as usual. The data from this particular night is not conclusive, but I wanted to use it as a way to show you how we go from the images you've been taking (including those sky flats) to a light curve that can reveal the presence of a transiting exoplanet. (I skipped showing you the step where we measure and subtract the "dark" exposures, which is another calibration step that corrects for the charge induced by the applied voltage across the CCD (the "bias") and also the counts that are produced by thermal noise in the CCD.) Let me know if you have any questions about any of this. David