Wednesday, July 25, 2012

Trinary Dwarfs-Data Reduction Summary

The goal of the data reduction was to take radial velocity measurements of certain observed low mass stars. In order to accomplish this we used the redspec data reduction package.

The first step in doing a reduction is combining the flat and dark files. In later steps this lets us remove influences from background sources and correct for the dark current in the telescope. This is followed by beginning the reduction by setting up how files will be used in parfile. In this step we define the file paths to the spec files to denote the target star, and the calibration star. The calibration star is used later on in the reduction to eliminate telluric absorption, and other features associated with the location of the target star. In this case, as we are working with high resolution spectra, we use an A type star as a source of calibration because they have few intrinsic lines in their spectra.

Spatial rectification with redspec is accomplished by summing an A+B pair of a calibration star to produce an image with two spectra. By finding polynomial fits to each spectral trace, and calculating guassian centroids to define the separation between the two, a curved or distorted spectra can be remapped to produce straight spectral traces with respect to the rows of the detector. Once this is done the spectral rectification step follows, where we use a polynomial fit of order 1-4 in order to fit the spectra to known arc lines or oh lines. This step matches the pixels of the picture to fitted wavelengths with regular intervals.

The last step is the main portion of the redspec. This is the step where the target and the calibrator get divided by the dark subtracted flat to remove the background sources for both the calibration star and the target star. The divided spectra gets multiplied against a normalized blackbody function corresponding to the spectral type of the calibration star to provide a relative flux calibration.

The observations for this project were made using filters N7 and N3. For the N7 observations the echelle was 63, with a cross dispersion of 35.52, and the N3 observations were made with an 62.95, and a cross dispersion of 34.08. The reductions for this project involved the use of both arc lines and oh lines for the spectral mapping portion of the reduction. The values for the oh lines came from a paper written by Emily Rice rather than the standard lines used in the echelle format simulator. The reason for this was simply greater accuracy. The N7 reduction used arc lines for the order we were interested, order 33 due to the strong methane absorption features. In the spectral mapping portion of the reduction, the fit to each individual arc line was of order 1 or 2, while the overall lambda fit used to map the arc lines to pixels was of order 1. This is because there were only 3 arc lines total on that particular echelle order when using xenon, krypton, argon, and neon line lists from keck, and due to the nature of polynomial fits, we can't use the higher order fits without introducing large errors as for example the 2nd order fit would match any 3 points with 0 error. The N3 reduction used OH lines, and we were primarily interested in order 59. For this fit, although there were a possible 10 lines to be fitted, the most accurate reduction was found using 6 out of those 10 lines, with an order 3 fit for the oh lines, and an order 2 fit for the overall lambda fit. The reason for this was that adding in additional lines caused significantly greater error even with a higher order lambda fit, and the lines themselves were not only very weak in terms of signal, but often shifted around greatly from observation to observation, relative to the 6 lines that were used in the fit.

No comments:

Post a Comment