Data Set Overview
  =================
 
  This dataset contains three parts: 1) The photometrically calibrated
  images of HST ACS/HRC observations of asteroid (1) Ceres (HST
  GO-09748); 2) The albedo maps of Ceres; and 3) The shape of Ceres.
  267 images through three ACS/HRC filters, F555W, F330W, and F220W,
  centered at 535 nm, 335 nm, and 223 nm, respectively, have been
  calibrated to the standard reflectance unit, I/F, where I is the
  intensity reflected off the surface of Ceres, and pi*F is the
  incident solar flux received at the surface of Ceres.  The shape
  model of Ceres has been constructed from a subset of these images by
  (Thomas et al. 2005).  The surface albedo maps of Ceres at above
  three wavelengths have been produced from these images and the shape
  model by Li et al. (2006).  The albedo maps cover the surface of
  Ceres between +/-50 deg latitude.  For detailed descriptions of the
  calibration and data processing, please refer to Thomas et al.
  (2005), and Li et al. (2006).  Below is a brief summary for the
  photometric calibration.
 
  Observations
  ============
 
  Three HST visits was performed on 12/28/2003, 12/30/2003, and
  1/23-24/2004.  The first visit consists of 6 HST orbits, providing a
  complete and almost evenly spaced coverage for one rotation of Ceres
  with a 9.075-h period at 6.2 deg phase angle.  Other two short
  visits were at 5.4 deg and 7.5 deg phase angles, respectively.
  Totally 267 images were taken, of which 255 were in HRC-512
  sub-frame for shape model and photometric study, and 12 were in HRC
  full frame for satellite search.  Ceres was resolved to about 30
  pixels across the disk.  Both the sub-solar and the sub-Earth points
  on Ceres are close to its equator for all images.
 
  Processing
  ==========
 
  The HST images were initially processed through the standard HST
  data calibration pipeline, which applies CCD related corrections,
  bad pixel removal, bias correction, dark correction, and flat field
  (Pavlovsky et al. 2005).  The calibration history can be found from
  the relevant history keywords in the images.  Cosmic rays have not
  been removed in either the standard calibration pipeline or the
  following additional calibration, because Ceres is a moving and fast
  rotating object, for which cosmic ray rejection requires special
  processing.  The images produced by the HST calibration pipeline
  were then corrected for geometric distortion, and the celestial
  north (J2000) was rotated up.  The re-sampling is done by a
  nearest-neighbor transformation, and is thus none-reversible.  We
  have stored all images in 1024x1024 frames, and filled in the blank
  areas by zeros.
 
  The shape was produced by limb fitting from a subset of 217 images.
  The shape of Ceres was found to be an oblate spheroid.  The
  planetocentric coordinate is defined based on the pole orientation
  of Ceres, and the prime meridian is defined on a bright spot at
  about 10 deg north latitude.  For more details, please refer to
  Thomas et al. (2005) and Li et al. (2006).
 
  The photometric calibration was performed by calculating a
  calibration constant for each filter, to scale the images in DN/s to
  I/F unit directly, with the help of two HST standard photometry
  keywords, PHOTFLAM and PHOTZPT, as well as the helio- and
  geo-centric distances, and the pixel scale of ACS/HRC detector.  The
  calibration constants are listed in Table 5 of Li et al. (2006).
  However, one more special step has to be applied for the images
  obtained through F220W filter, to correct for the substantial red
  leak, which was estimated for Ceres to be ~19.8% of the total flux
  measured through this filter.  The calibrated I/F images of F220W
  filter were then scaled by a factor of 0.802.  Finally, no
  correction for the charge transfer efficiency (CTE) of the HRC CCD
  detector has been made.  But the effect of CTE has been estimated to
  be much less than 1% on the absolute calibration of these images.
 
  Starting from the calibrated images, the photometric properties of
  Ceres were modeled, and the surface albedo deviation maps of Ceres
  are constructed (Li et al. 2006).  In the photometric modeling
  process, the areas too close to the edge of Ceres were disregarded.
  Therefore with the sub-solar and sub-Earth latitude of these
  observations close to the equator, the surface coverage of these
  maps can only reach +/-50 deg reliably.  To avoid phase angle
  correction, we only used the images from the first visit to
  construct the maps.  Since the Hapke single-scattering albedo (SSA)
  averaged over the surface of Ceres is only about 7% at V-band, and
  less for shorter wavelengths, the ratio maps of the original images
  to the modeled images were considered as the albedo deviation maps
  from the global average.  Because the normal reflectance of the dark
  surface is also proportional to its SSA, these ratio maps are also
  the deviation maps of the normal reflectance.  Finally the ratio
  maps covering different areas on Ceres were projected to simple
  cylindric planetocentric longitude-latitude system, and combined to
  construct the albedo maps.
 
  Data
  ====
 
  The photometrically calibrated images contain all the FITS headers
  propagated from the original HST images.  Several additional
  keywords were added as well, including:
 
  PHOTIOF0  : Calibration constants applied to each image
  REDLEAK   : The fraction of red leak flux
  REDUCMAG  : The total magnitude of Ceres at both helio- and
  geo-centric distances 1 AU
  CERSUND   : Ceres-Sun distance in (AU)
  CEREARD   : Ceres-Earth distance in (AU)
  PHASE     : Solar phase angle (deg)
  NORAZ     : North pole clock angle in the image (deg)
  CENTERX   : row number of Ceres body center in the image
  CENTERY   : line number of Ceres body center in the image
  SUBSLON   : Sub-solar east longitude (deg)
  SUBSLAT   : Sub-solar latitude (deg)
  SUBELON   : Sub-Earth east longitude (deg)
  SUBELAT   : Sub-Earth latitude (deg)
 
  The reduced magnitude and the later seven keywords above about Ceres
  coordinates were only calculated for a subset of 217 images.  The
  geometry keywords were calculated from the most recently developed
  shape model and the corresponding planetocentric coordinate (Thomas
  et al. 2005).  A table listing the aspect data of these images can
  be found from this dataset.
 
  The shape of Ceres is expressed in an ASCII table by its equatorial
  and polar radii, as well as the pole orientation and prime meridian
  argument in J2000 frame.
 
  The albedo deviation maps are expressed as 720x361 images, with the
  horizontal axis corresponding to east longitude, and the vertical
  axis corresponding to the latitude.  The longitude starts from 0 deg
  at the left-most column of each map, increasing eastward, with each
  pixel corresponding to 0.5 deg increment, until 359.5 deg at the
  right-most column.  The latitude starts from -90 deg at the bottom
  row, increasing upward, with each pixel corresponding to 0.5 deg
  increment, until the +90 deg at the top row.  The pixel values in
  each map represent the deviations at the location from the global
  average at the wavelength of that map.  For example, in the 330 nm
  map, if a pixel value is 0.96, it means the single-scattering albedo
  at that position is 96% of the average value at this wavelength, so
  is the normal reflectance.  The values of the global average Hapke
  SSA and normal reflectance are listed in the FITS headers of these
  maps.
 
  Also included with these maps are the corresponding relative
  uncertainty maps.  Having the same structure as the albedo maps, the
  uncertainty maps represent the standard deviations of each
  corresponding pixel in the albedo maps, as calculated from all
  images that cover that longitude-latitude position.  Note that for
  positions that only have one or a couple of data points from
  different images, the standard deviations do not have a definitive
  meaning.  This is the case for some points at high latitude in the
  uncertainty map of 223 nm albedo map, where we did not have as many
  images to construct the albedo map as for other two wavelengths.
  Note that this uncertainty maps do not include either the absolute
  photometric calibration uncertainties, or the modeling uncertainties
  for the global average Hapke's parameter.  However, neither of them
  affects the relative uncertainties of these albedo maps.  The CTE
  uncertainties and HST imaging noise are not included in the
  uncertainty maps either, but they are both very small and
  negligible.

Confidence Level Overview
  =========================
    The uncertainties of the absolute photometric calibrations are ~2%
    for F555W images, ~3% for F330W images, and ~8% for F220W images,
    respectively.  The uncertainties for the modeled single-scattering
    albedo are ~3% at 535 nm, ~4% at 335 nm, and ~9% at 223 nm,
    respectively.  See Li et al. (2006) for more details.  The shape
    and pole orientation of Ceres have been determined very
    accurately, with 5 deg in the pole orientation, and 1.8 km in the
    semi-axes (Thomas et al. 2005).  The uncertainty for the argument
    of prime meridian is not applicable because this parameter defines
    the starting point of the longitude system, and is not a
    measurement.
 
    It needs to be noted that cosmic rays have not been rejected for
    these images, and the very small effect of CTE has not been
    removed, either.  The Ceres albedo maps are defined under the most
    recently defined body-fixed coordinate system as indicated in the
    maps.  Please refer to Thomas et al. (2005) for the definition of
    the coordinate system, and please note that east longitude is used
    in these maps.