Data Set Overview
  =================
    This dataset contains average Near-Infrared (NIR) reflectance spectra for
    68 main-belt asteroids that were observed at the NASA Infrared Telescope
    Facility (IRTF) from April 2001 to May 2015. These asteroids were used as
    a part of three research investigations to better constrain their NIR
    spectra, assigned taxonomies, surface mineralogies, and potential
    meteorite analogs. The asteroids in this dataset were all observed in a
    very similar manner across the entire dataset. The resulting average NIR
    asteroid spectra in this dataset were also reduced in a very similar
    manner using two different software packages.

    The research projects that utilized these average NIR spectra include: 1)
    an investigation of the spectral and mineralogical diversity of the
    M-/X-type asteroids (Hardersen et al., 2005, 2011), 2) a study to better
    define the basaltic asteroid population in the main asteroid belt
    (Hardersen et al., 2014, 2015, 2016), and 3) an investigation of (1459)
    Magnya and its spectral and mineralogical comparison to (4) Vesta
    (Hardersen et al., 2004).

    Hardersen et al. (2005) reported the first results of the M-/X-type
    asteroid study, which included six M-type asteroids, while Hardersen et
    al. (2011) provided the final results of this effort that included NIR
    reflectance spectra for 45 M-/X-type asteroids. These works reported on
    the identification of significant NIR spectral and surface mineralogical
    diversity among this group of asteroids, the widespread detection of weak
    mafic silicate absorption features for pyroxene and olivine (1-5% band
    depths), detections of possible phyllosilicate features on a few
    asteroids, and widely varying NIR spectral slopes across the entire
    spectral dataset.

    Hardersen et al. (2014, 2015) were the first results of an effort to
    better constrain the basaltic asteroid population throughout the main
    asteroid belt. This is being accomplished by obtaining NIR spectra of
    Carvano et al. (2010) classified Vp-type asteroids with Wide-Field
    Infrared Survey (WISE)-derived albedos consistent with basalt (Masiero et
    al., 2011) from a dataset compiled by Mainzer et al. (2012).

    All of the average asteroid NIR reflectance spectra in this dataset were
    observed at the NASA Infrared Telescope Facility (IRTF), Mauna Kea,
    Hawaii. The asteroids reported in Hardersen et al. (2004, 2005, 2011,
    2014) used the first-generation of the SpeX 0.7-5.3 micron spectrograph
    while the results in Hardersen et al. (2015) and the expected publication
    of Hardersen et al. (2016) used the second-generation of the SpeX
    spectrograph. All observations used the low-resolution, prism mode of the
    SpeX spectrograph (R = 94), the 0.8 arc second slit, an open dichroic, and
    an open order sorter filter. Most observations taken prior to 2008 were
    not observed at the parallactic angle while all observations after 2008
    were observed at the parallactic angle. The vast majority of the
    observations that resulted in this dataset were made in clear to mostly
    clear weather conditions at the summit of Mauna Kea.

    All observations were conducted in a uniform manner. For each asteroid, an
    associated extinction star (late F- to late G-type main sequence) was
    observed close to the asteroid on the sky (less than 5 degrees
    separation). Stellar observations were interspersed with asteroid
    observations to ensure that the extinction star airmass range exceeded
    that of the associated asteroid. The extinction star observations are
    necessary for later removal of the NIR telluric absorptions during data
    reduction. One well-known solar analog (SAO 31899, SAO 93936, or SAO
    120107) was observed each night, which was used in the reduction process
    to correct for the use of non-G2V extinction stars and to implement a
    slope correction so the final asteroid spectrum would mimic that if the
    actual reflected light from the Sun had been used.

    Data reduction for the asteroids used either SpecPR (Clark, 1980; Gaffey,
    2003) or Spextool (Cushing et al., 2004). For those asteroids reduced
    using SpecPR, the sky background signal removal was accomplished using the
    Image Reduction and Analysis Facility (IRAF), followed by importing the
    sky-subtracted data into SpecPR. SpecPR routines included derivation of
    the 1st-order extinction coefficients from the extinction stars as a
    function of wavelength and airmass for up to five sets of extinction star
    observations (10 spectra per set), sub-channel pixel offsets for alignment
    of multiple spectra, and averaging routines.  Spextool routines include a
    sky signal subtraction routine and a telluric correction that utilizes two
    sets of extinction star observations while also including the sub-channel
    pixel shifting and averaging routines. Wavelength calibration for SpecPR
    data is conducted manually by matching argon calibration emission
    lines/wavelengths to discrete channels and applying a polynomial function
    to convert the NIR spectra from channels to wavelengths. Spextool
    implements an internal wavelength calibration using one set of argon
    calibration spectra.

    For all data, each final average asteroid NIR spectrum can be
    qualitatively summarized as follows: (Asteroid/Sun) = (Asteroid/Extinction
    Star) / (Solar Analog Star/Extinction Star), where (Asteroid/Sun)
    represent the final average asteroid NIR spectrum, (Asteroid/Extinction
    Star) represents an intermediate average asteroid NIR spectrum after
    telluric and channeling shifting corrections, and (Solar Analog
    Star/Extinction Star) is the average stellar spectrum used to correct for
    observations of non-G2V extinction stars. The average (Solar Analog
    Star/Extinction Star) spectrum is smoothed to remove telluric absorptions
    as broad absorption features are not expected in stellar spectra and
    observations of solar analog stars are typically in very divergent parts
    of the sky during most IRTF observing runs.

    The data products in this dataset include the average NIR reflectance
    spectrum for each of the 68 asteroids. Each asteroid has its own text file
    that includes two or three columns of data that include wavelength,
    normalized reflectance, and error values. Most of the SpecPR-derived NIR
    spectra lack errors in the associated files while most of the Spextool
    data include errors. Where errors are reported, they are standard errors
    in SpecPR, or standard errors Robust Weighted Mean in Spextool.  Spextool
    data are normalized at 1.5 microns while SpecPR data are normalized near
    1.7 microns.

    The reported errors for the asteroid spectra that were reduced using
    Spextool are only the formal errors produced by the Spextool software. The
    reported errors result from the average of each individual asteroid
    spectrum, using the error method chosen in Spextool, where each individual
    asteroid spectrum is divided by the associated extinction star spectrum.
    The fully propagated errors are not reported, but will be reported in
    future updates to this dataset.

    The reported errors for the asteroid spectra reduced using SpecPR are
    fractions based on the variability of the point-to-point data scatter
    present in each final average asteroid spectrum reported here. Typical
    errors across a spectrum for most of the asteroids range from 1-3% with
    some asteroids having larger errors in spectral regions containing
    poorly-corrected telluric features and at longer wavelengths where SpeX
    has reduced sensitivity.

Confidence Level Overview
  =========================
    Most observations at the NASA IRTF were in clear, photometric conditions,
    but there were a few instances of cirrus present in the sky.