CASPIR produces near-infrared spectra in both long slit and cross-dispersed formats. This section deals with the reduction of long-slit grism data. These observations normally consist of a number of object and sky frames acquired through the execution of a DO file, or via the nodding observing mode. Typical observing sequences would:
1) Record off-source sky spectra for extended objects, and object spectra with the object placed at different positions along the slit to counter ghosts and bad pixels.
2) Record nodded pairs of frames with an unresolved or slightly resolved object placed at two positions along the slit to permit sky subtraction.
3) Record sets of frames with the object placed at a number of positions along the slit to permit sky subtraction and to counter ghosts and bad pixels.
A spectroscopic dataset is therefore naturally defined as a set of related observations of a given object with a particular grism. The reduction of most long-slit grism datasets will follow the path:
1) Create BIAS and DARK frames, and linearize object and sky frames.
2) Subtract sky background from object frames, and combine all object frames in the dataset to a single spectral image.
3) Create FLAT frames for each grism, and remove pixel-to-pixel sensitivity variations.
4) Fix known and random bad pixels.
5) Apply a geometrical transformation to align the dispersion direction with image columns, align the spatial direction with image rows, and perform the wavelength calibration.
6) Correct for non-uniform illumination along the slit.
7) Extract one dimensional spectra from the combined spectral images with appropriate background subtraction in the slit direction to remove residual sky features.
8) Flux calibrate the one dimensional spectra.
9) Correct for absorption features in the standard star spectrum, inexact cancellation of terrestrial atmospheric absorption features, and slit losses in the standard star measurement.
Users are cautioned that infrared spectroscopy presents greater challenges than optical spectroscopy. The infrared sky makes a significant contribution to deep infrared spectroscopic observations and must be removed using frequent sky observations. Terrestrial atmospheric absorption is also strong in the near-infrared (see Figure 60 to 63 in Appendix M) and must be removed through division by a measurement of a featureless star obtained close to the object measurement in both time and position on the sky. The quality of the final spectrum frequently depends on the accuracy of this correction. Spectroscopic observations will normally use long integration times that increase the importance of dark current as a noise source. Pixels that saturate due to their high dark current are not removed by sky subtraction and must be treated as bad pixels.
The reduction procedures described here use the local MSSSO CASPIR package running in IRAF. Refer to §6 for instructions on how to obtain this package. Users are advised to familiarize themselves with the general procedures available within IRAF for spectral reduction by reading the documents A User's Guide to CCD Reductions with IRAF and A User's Guide to Reducing Slit Spectra with IRAF. Compressed postscript versions of these documents are available via anonymous ftp to iraf.noao.edu in the files iraf/docs/ccduser2.ps.Z and iraf/docs/spect.ps.Z.