Preparation of Wild 2 shape model The shape model of comet Wild 2 was provided by Randolph Kirk in the form of a list of cartesian coordinates of the points on the surface of the nucleus. These points were then manipulated by the PDS SBN to produce the plate model that is represented here. The process of converting the surface points to the plate model was done by Jianyang Li at the SBN, and the following describes the basic steps that were taken. Producing shape model from Kirk's data Kirk's model consisted of cartesian coordinates of points on the visible and illuminated surface. The axes are defined by the axes of the best-fitting ellipsoid to the modeled surface, with +Z in the shortest dimension (and assumed to be aligned with the spin axis), +X along the long axis (defining the prime meridian) and +Y completing the right hand system. From these cartesian coordinates, a second version was computed, in plantocentric coordinates (latitude, longitude and distance from the center of the best-fitting ellipsoid for each surface point). Redundant points There were a number of redundant points in the Kirk table (latitude and longitude each less than 0.1 deg different, and altitude less than 1 meter different). These points are situated in a long thin path, and presumably result from overlap in two segments that were combined during the modeling process. Removing these redundant data reduced the number of points from 6932 to 6432. Triangulate surface data points to find plates Using the IDL procedure TRIANGULATE, connections between the surface points were found, producing plates for the surface model. This step assumes that the sampled surface is convex on the surface of a sphere, which may not be true for small local regions, though no obvious problems are immediately apparent. A total of 12860 plates were produced to connect the surface points. These plates cover only the visible and illuminated portions of the nucleus. Plates at the edge of the visible surface The shape model doesn't have full coverage on the surface, so the automated connectivity from the last step breaks down at the edges where non-related points are "connected" over big gaps. These connections need to be removed. The criterion used to remove these points is simply to remove any connections that span a distance of more than 210 m, a value adopted because it it is the minimum distance for which there is no isolated single points left after rejecting triangles. (This does leave isolated plates and groups of plates where an elevated region was high enough to be visible at the terminator.) In this process, 562 plates are rejected, leaving 12298. Recovering rejected plates that are meaningful A visual inspection shows that for some areas, the criterion of 210 m removes plates from regions that are known to be visible, and leaves a few false connections at the edges. The removed plates were likely rejected because they lie in flat regions with few features, and so the sampling density is relatively sparse, even though the surface is well-represented. Other connectivity problems may occur in regions where there is a sharp change in elevation and the connectivity was simply not found. Real plates were recovered manually, using the Kirk LPSC manuscript as a guide. The few false connections were also removed manually. 219 plates were recovered, and 3 were removed, leaving a total of 12514 plates in the final model. Additional versions of the model Once the primary model, above, was produced, it was used to derive additional versions. As noted earlier, the cartesian coordinates were converted to planetocentric coordinates for a second version. The third and fourth versions include the same surface model as before, but also include the unseen portions of the nucleus, as represented by the best fitting ellipsoid. Again, this version of the model was created in both cartesian and planetocentric coordinates. Ellipsoid model A longitude-latitude grid was generated to cover the surface of Wild 2 from the south pole (0,-90) to latitude 40 N, with resolution 3 deg in latitude, and 3 deg in longitude on the equator. At higher latitudes, the longitudinal resolution is proportional to cos(latitude), decreasing to 1 point at the pole. The altitude for each of these points was calculated from the best-fit ellipsoid provided by Kirk. Combining the ellipsoid and the plate model To create the full plate model, the ellipsoid and plate models were combined, with the plate model taking priority. If a plate from the surface model exists at a given latitude and longitude, then it is used instead of the ellipsoid version. Furthermore, any ellipsoid point that is within 80m in space to 1 deg in lat/long is removed. Using this criterion, the ellipsoid portions of the nucleus add 2329 points, for a total of 8761 vertices. Triangulation of these points then produces a total of 17518 plates. In the versions containing the ellipsoid portions of the nucleus, an additional flag is included in the vertex and plate lists to denote whether the entries represent the derived surface points or the best fitting ellipsoid.