GMOS data reduction pipeline: Part 1 of 2

Home Forums Gemini Data Reduction GMOS data reduction pipeline: Part 1 of 2

Tagged: ,

This topic contains 0 replies, has 1 voice, and was last updated by  Catherine Huitson 3 years, 8 months ago.

Viewing 1 post (of 1 total)
  • Author
  • #1352

    Catherine Huitson

    Hi all,

    I’m working on a pipeline to reduce GMOS data for exoplanet atmospheres (timeseries MOS spectra of a few objects simultaneously). While the pipeline will not be useful for everyone I wanted to share some of the choices I’ve made while making it in case they are useful. The pipeline will eventually be public (with documentation) and I will also share a link to it when it becomes available. It does not use the Gemini IRAF packages and is really targeted at high-precision observations for which a custom pipeline is needed.

    There are 2 main stages in the pipeline
    1) Reduction of images from the telescope and extraction of stellar spectra
    2) Further reduction of the extracted spectra

    In this post, I will discuss the first part since that will probably have the most relevance for the greatest number of users. The second part is really only relevant for users with high-precision timeseries data and involves corrections for changes in the spectra over time.

    Here is a summary of steps in the first part of the pipeline:

    – Combine different amplifier images to make complete spectra & subtract biases & multiply by gain for each amplifier.
    – Remove cosmic rays. Since we have a long timeseries of data this is done by searching each pixel for significant outliers in time for batches of ~20 exposures at a time. We replace the outliers by the median value of the considered pixel in the batch being considered.
    – Removal of shifted columns in the south (E2V). This is no longer a problem for data taken since the Hamamatsu detectors were installed. The old E2V detector had columns in which charge was vertically shifted on the detector. We correct for this by searching for columns outside the slit which have a higher charge than the median in the flat fields and then flagging these so that they can be removed from the final analysis.
    – Identification of hot pixels by flagging outliers in the bias frames and identification of dark pixels by flagging outliers in the flat frames. The user can set the rejection thresholds.
    – Flat-fielding. We median-combine a series of flats to improve S/N to the degree needed for our data (100 ppm precision). We then average the resultant master flat in the cross-dispersion direction so that we have mean flux vs. dispersion pixel. We fit this with a smooth function of flux vs. dispersion pixel and then divide all rows of the master flat by this function to normalize (note that we do this procedure separately for each slit, which have a ~200 pixel extent in the cross-dispersion direction).
    – Tilt correction. This is done by cross-correlating the spectrum in each row of pixels in the arc spectrum with the spectrum in its central row. This then gives measured lags vs. row number which can be applied to the science spectra by interpolating them onto the non-shifted grid.
    – Extraction of 1D spectra using optimal extraction. There are various options to set including aperture width and type of background subtraction. As we have little background within the slit, we perform a linear fit for each column to measure the background but the pipeline supports higher-order functions, no background subtraction or median subtraction.

    Let me know if you have any questions or comments.


Viewing 1 post (of 1 total)

You must be logged in to reply to this topic.