2006gy_reduction_description.txt

…slit by offsetting from a nearby star. The location of the supernova was identified by
matching a simulated cross-dispersion (XD) profile from WFC3 observations (day 2303) with
the XD profile of the STIS observations.
The WFC3 image was first rotated to the same orientation as the STIS observations.
The XD profile for the WFC3 data was simulated for a number of possible slit positions by
summing the WFC3 pipeline calibrated files (_drz.fits) in the dispersion direction over ~5
pixels representing the STIS slit width on the WFC3 detector. The STIS spectrogram was
also collapsed, however, due to the degradation of the STIS CCD detector, we found it
better to median stack rather than to sum the columns in the cosmic-ray rejected images
(_crj.fits).  The stacked columns were rebinned to the WFC3 plate scale to create a XD
profile. We compared WFC3/F814W to STIS/G750L as these observations most closely matched
in wavelength coverage and included the H-alpha line which is the strongest line in the
spectrum. The slit location in the WFC3 image was identified by finding the offset in the
cross-dispersion direction that creates the best match, in both shape and scale, between
the WFC3 and STIS XD profiles for each virtual slit position. Figure 2 highlights the slit
position and orientation on the
WFC3/F814W image (left). On the right the XD profiles of the WFC3 observation (blue) and
the HST/STIS/G750L observation (green) both normalized to the peak flux. The similarity
of these two profiles...

The 1D spectrum for each observation was extracted using the CalSTIS custom extraction
software stistools.x1d
(http://ssb.stsci.edu/doc/stsci_python_x/stistools.doc/html/x1d.html?highlight=x1d#module-stistools.x1d).
The default extraction parameters for STIS are defined for an isolated point source. For
both the G430L and G750L the default extraction box width is 7 pixels while the background
extraction box width is 5 pixels. Figure 3 shows the 2D spectrogram for G750L. While
the 2D image is dominated by the galaxy nucleus, the SN trace is faintly visible below the
galaxy in the region surrounding H-alpha. Due to the closeness of the SN to the bright
galaxy core and its faintness, we reduced the extraction box size to 3 pixels to optimize
the S/N. The optimal extraction location and the background were determined by stepping the
extraction box up the detector in the region near the supernova. The extraction with the
highest S/N set the SN extraction location. The 3 extractions above and below the SN
location were median combined to estimate the background as close to the SN as possible. This
lead to a background extraction box height of 5 pixels above the SN extraction box and 5
pixels below the SN extraction box.
All of the STIS spectra are combined to produce a single spectrum using the splice tool in
STSDAS (http://stsdas.stsci.edu/cgi-bin/gethelp.cgi?splice). As recommended we ignore the
edge columns by setting their data quality flags to 4. Figure 4 plots the final,
background subtracted spectrum.