Data Set Overview
  =================
    This data set contains raw spectral images of Comet 9P/Tempel 1
    obtained with SpeX, the low- to medium-resolution near-IR
    spectrograph and imager at the NASA Infrared Telescope Facility
    (IRTF) on Mauna Kea, Hawaii. In addition to the spectral images of
    Tempel 1, numerous flat field and Argon lamp images are included
    that are required to process the comet data.
 
    The SpeX instrument is fully described in papers by Rayner et al.
    (2003,2004) [RAYNERETAL2003, RAYNERETAL2004].  For the Comet
    9P/Tempel 1 observations, a number of slit and prism / grating
    combinations were used, producing spectra with resolving powers
    ranging from 100 to 1500, and covering the two spectral regions
    from 0.8 to 2.5 microns and from 2 to 5.5 microns.  The lowest
    resolution spectra, taken in the LowRes Prism mode, cover the
    wavelength interval from 0.8 to 2.5 microns in a single,
    continuous spectrum.  The higher resolution modes SXD (covering
    the interval 0.8 to 2.5 microns), LXD 1.9 (1.9 - 4.2 microns), LXD
    2.1 (2.1 to 5.0 microns) and LXD 2.3 (2.3 - 5.5 microns) use
    gratings and prism cross-dispersers to separate overlapping
    spectral orders.  In these higher-resolution modes, there is
    sufficient wavelength overlap between adjacent orders to produce
    spectra with continuous coverage over the entire interval from 0.8
    to 5.5 microns.  Diagrams showing the spectral position on the
    array for LowRes Prism (LRes15_spec_layout.jpg), SXD
    (SXD_spec_layout.jpg), LXD1.9 (LXD19_spec_layout.jpg) and LXD2.3
    (LXD23_spec_layout.jpg) modes are provided in the documents
    directory.
 
    The spectrograph utilizes a 1024x1024 Aladdin 3 InSb array.  The
    spatial (along-slit) image scale for all of the slit / grating
    combinations of the spectrograph is 0.15 arcsec / pixel.  The gain
    is fixed at 13.0 electrons / ADU.  The array is linear to about 5
    percent up to 4000 DN per read.  Each fits image contains the sum
    of all non-destructive reads and coadds acquired during the
    exposure, and thus will normally contain count levels much higher
    than 4000 DN.  To obtain the average counts per read, a
    normalization value is provided in the image header (keyword =
    DIVISOR) that is the product of the non-destructive reads and
    coadds.
 
    Observations were typically obtained as A - B nodded pairs, in
    which the telescope position was shifted about 7.5 arcsec along
    the direction of the 15-arcsec long slit, producing an offset in
    the target spectrum between the A- and B-beam images.  These image
    pairs allow for mutual sky subtraction between the A - B and B - A
    image combinations.  The beam position for each image is
    integrated into the fits file name, with the extensions of
    '_a.fit' and '_b.fit' referring to A- and B-beam positions,
    respectively.  While most of the comet spectra were taken with
    slit widths of 0.5 and 0.8 arcsec, some of the standard star
    observations were made using wider slit widths, ranging up to 3.0
    arcsec, as a means of determining accurate flux calibrations.
 
    The following standard stars were observed for the purpose of flux
    calibration and telluric (atmospheric) corrections.  Catalog
    abbreviations are:  HD = Henry Draper Catalogue,  HR = Harvard
    Revised Bright Star Catalogue.  The V magnitudes listed are from
    various catalogues.  References are given (when available) for the
    spectral classification.
 
    Star                    V mag   Spec_type
    HR 4689                  3.89     A2IV   [COWLEYETAL1969]
    HR 5107                  3.40     A3V   [COWLEYETAL1969]
    HR 5255                  5.76     A0Vs   [COWLEYETAL1969]
    HR 5332  (= HD 124683)   5.53     A0V   [HOUK&SMITH-MOORE1988]
    HR 5959                  5.53     A0Vs   [COWLEYETAL1969]
    HR 6787                  4.36     B2IV   [LESH1968]
    HD 116960                8.00     A0V   [WOOLLEYETAL1969]
    HD 122749                8.31     A0V   [HOUK&SMITH-MOORE1988]
    HD 123309                9.40     A0V   [HOUK&SMITH-MOORE1998]
    HD 144873                8.50     G5
 
    On the nights of July 2 - 4 UT, low-resolution (prism-mode)
    spectra were rapidly recorded in 'movie mode'.  For these
    observations, the telescope was not nodded (images acquired in the
    A-beam only), a sub-frame readout of the array was used in order
    to reduce the read time overhead, and the image sequences were
    stored in the form of 3-D fits image cubes.  Each image cube has
    the dimensions 512x122x16 pixels.  The fits header associated with
    each image cube contains the start and end times for the first and
    last frames contained in that cube.  A text file associated with
    each image cube gives the GPS-derived time (minutes and seconds)
    at which the last pixel is read for each image plane stored in the
    cube.  A near-infrared light curve of Comet 9P/Tempel 1 has been
    derived from the movie-mode data taken on July 4 (Fernandez et al.
    2007) [FERNANDEZETAL2007].
 
    The document directory also contains log files (plain text) for
    each night that spectra were obtained.  These logs contain
    descriptions of the sky conditions and list the circumstances for
    each image frame (or image cube) including target name, start and
    end times, airmass, slit and grating.  The last column contains a
    flag (Y or N) indicating whether spectroscopic and imaging data
    (with the guider array) were recorded simultaneously.  When data
    were recorded simultaneously, the spectra can be directly compared
    to the on-slit images contained in the companion dataset, IRTF
    Near-IR Imaging of Comet 9P-Tempel 1 V1.0, in order to derive
    spectrophotometric results.  By comparing the off- and on-slit
    images of the Comet and Standard stars in the imaging dataset, one
    can constrain the flux fraction passing through the spectrograph
    slit as a function of time.  This is particularly useful in
    analyzing the spectroscopic 'movie' sequences taken around the
    time of the DI impact event.
 
    While various spectral reduction packages exist, an IDL program
    called Spextool was developed specifically to reduce and extract
    spectral data taken with SpeX.  The Spextool program is described
    in Cushing et al. (2004) [CUSHINGETAL2004].  The software is
    publicly available and can be download from the IRTF website.
 
    These data were obtained through a coordinated effort by the
    following observers:
       Michael S. Kelley
       Neil Dello Russo
       Alan Tokunaga
       Carey Lisse
       Yan Fernandez

Confidence Level Overview
  =========================
    SpeX is a very stable instrument that produces consistently high
    quality data within the limits of instrument sensitivity.  The
    spectrograph array is relatively clean of bad pixels, except for a
    cluster of hot pixels centered near array coordinate X = 605 Y =
    770.  In addition, there are two diagonal lines of dead pixels,
    roughly 2 pixels wide, that traverse the array.  These lines can
    be masked and corrected for as part of the data reduction process
    (a procedure to deal with these lines is built into the Spextool
    reduction algorithm).  The time stamps given in the image headers
    are accurately derived from a GPS signal.