Data Set Overview
    =================
 
This data set presents the different shape models of comet 81P/Wild 2,
as derived from the Stardust navigation camera images that were
obtained around the time of closest approach to the comet.  The first
model is a basic tri-axial ellipsoid shape describing the axial ratios
and their orientation.  The second is a detailed plate shape model.
Additional details and information about the ellipsoidal model can be
found in Duxbury et al.  J. Geophys. Res., Vol. 109, No. E12, E12S02
10.1029/2004JE002316 (2004) [DUXBURYETAL2004].  Additional information
about the plate model can be found in Kirk et al., 36th Annual Lunar
and Planetary Science Conference, March 14-18, 2005, #2244 (2005)
[KIRKETAL2005].
 
On 2 January 2004, the STARDUST spacecraft flew past comet 81P/Wild 2
and took 72 images. In these images, the phase angle started at 70 deg
during the approach, dropped to 3 deg near closest approach, and then
increased to 110 deg during the departure.  Most images were obtained
within a range of 2,000 km, with the closest at 237 km.  The camera
took the images using a broadband filter and a CCD that had an angular
resolution of 57 microrad/pixel.  These images are archived in the PDS
Stardust NAVCAM Data for comet Wild 2 (Data_Set_Name: STARDUST NAVCAM
IMAGES OF WILD 2, Data_Set_ID: SDU-C-NAVCAM-2-EDR-WILD2-V1.0)
 
TRI-AXIAL ELLIPSOID MODEL
 
A tri-axial ellipsoid model was fit to the limb and terminator in
all the Wild 2 nucleus images.  The model was constrained by the fact
that the viewing geometry changed by 180 degrees during the encounter
and essentially the entire illuminated hemisphere of Wild 2 was viewed
in these images (or ~50% of the surface, since there was no noticeable
rotation during the few minutes that the images were taken).
 
The axial dimensions determined from the visual fit of the model to
the limb and terminator were 1.65 x 2.00 x 2.75 km +/-0.05 km (1
sigma).  The illuminated limb is reasonably smooth, with the exception
of a few large depressions.  It is understood that half of the surface
was not illuminated and that these unseen surfaces could deviate from
the model derived from the illuminated regions.  However the model fit
to the terminator, even though quite rough due to topographic
variations, does give some evidence that the model probably
extrapolates to the unseen part of the nucleus to at least a km or
better.
 
A body-fixed coordinate system was established with the shortest axis
assumed to be the rotation pole and the longest axis used to define
the prime meridian.  The posigrade axis is defined to match the
corresponding poles determined by Sekanina (2003) [SEKANINA2003] and
Farnham and Schleicher (2005) [FARNHAM&SCHLEICH2005] and is directed
toward the position (J2000) RA=110 deg, dec=-13 deg (+/-3 deg).
 
PLATE MODEL
 
A detailed plate model was derived from the same set of images.
Pairs of these images were used to generate stereo views if the
comet and to determine the cartesian XYZ coordinates of points on
the surface.  Excel was then used to perform translations to
convert the model to coordinates of latitude, longitude and
elevation.  Results are given in planetocentric coordinates.  The
model itself consists of 6432 points forming 12514 plates,
covering the portions of the nucleus that were both illuminated
and visible during the encounter (about 50% of the total
surface).
 
A best fitting triaxial ellipsoid was fit to the plate model using a
least squares fit.  This ellipsoid had the dimensions of 1.350 x 2.002
x 2.607 km (this is slightly different from the ellipsoidal model
described above because the earlier model merely reproduced an
envelope that enclosed the visible surface).  Elevations relative to
this best-fitting ellipsoid range from -600 m to +250 m, with an RMS
range of heights of 93 m.  The orientation of the minor axis (positive
pole) of this ellipsoid is at a right ascension of 112 degrees and a
declination of -17 degrees (J2000.0).
 
Four versions of the plate model are included here.  The primary
version (WILD2_CART_VIS.TAB) is given in Cartesian coordinates based
on a body-centered system.  A second version (WILD2_PLAN_VIS.TAB) is
given in planetocentric coordinates, with distances given from the
center of the ellipsoid.  In both of these versions, only the visible
and illuminated portions of the nucleus are included.  Two additional
versions of the model (WILD2_CART_FULL.TAB and (WILD2_PLAN_FULL.TAB,
respectively) again provide the shape models in Cartesian and
planetocentric coordinates, but include full coverage of the surface
of the nucleus.  In these versions, the non-visible portions of the
surface are represented by the best-fitting ellipsoid.
 
For detailed analysis of the shape, routines such as the SURFACE and
POLYSHADE routines in IDL can be used.  To simply view the shape
model, go to the browse directory, where the shape model has been
converted to VRML format.  There are two versions, one with the
illumination that mimics that during the encounter, and one that uses
a headlight to reveal the structure in a different manner.  These
models can be viewed using any standard VRML viewer.
 
Author
------
The descriptions in this file were written by Tony Farnham.

Confidence Level Overview
    =========================
 
A visual fit was used to match the ellipsoid model to the
observations.  Uncertainties were estimated by changing the parameters
until an overlay of the model clearly deviated from the corresponding
images.  Estimated errors are 0.05 km in the axial dimensions, and 3
degrees in the angular measurements.  The plate model has a horizontal
resolution of approximately 50 meters and a vertical precision of
about 6 meters.
 
    Review
    ======
The ellipsoid model was peer reviewed on Oct 15, 2004 and accepted pending
a few minor corrections and additions, which have been applied.  The
plate model is sceduled for review on Dec 12, 2005.