Preliminary maps are required by scientists and Voyager mission planners immediately after images are received from a spacecraft for planning further data-gathering sequences and for preliminary scientific interpretations that will lay the foundation for more detailed investigations. Geometric controls have not been rigorously defined when these kinds of maps are prepared, and time is not available for thorough image interpretation. Data arrive from spacecraft in fragments. Preliminary maps allow users to view the spatial context of the new information they have received.
Final mapping is done over a period of several years. It is tied to an accurate control net, which is now being developed for the Saturnian satellites (Davies and Katayama, 1983a,b). Final maps are made at larger scales and display more meticulous interpretations of surface detail than do preliminary ones.
The maps contained in this atlas are preliminary; a second edition is planned for release in 3 or 4 years that will contain final maps from the Voyager 1 and 2 missions.
The first stage in planetary mapping after data reception is the digital processing (see app. C) of pictures to be used in the mapping. Second, image locations must be determined. Third, a coherent pictorial representation of all the available surface information is prepared in the form of a map. Finally, a set of names is selected and applied to newly discovered features. (See app. E.) The two sources of information used to determine the location of the part of a planet appearing in a given picture are the location and orientation of the planet or satellite and the location of the spacecraft. The satellites of Saturn have been observed and their orbits timed by astronomers for many years. Their orbital periods, the fact that they always present the same face to Saturn, and their location within the solar system were accurately known long before Voyager arrived. The location of the spacecraft at any given instant was determined by the character of its radio signals (the time it took them to reach Earth, their doppler shift, and so on). From these data sources alone, it is possible to draw the latitude and longitude system of the satellite in proper orientation with respect to the Voyager camera for use in making preliminary maps. These are the perspective grids that accompany each pair of Voyager pictures of each satellite. Once a grid has been drawn for each picture, preliminary maps are made by sketching image details on a map-projection graticule by manually transferring information grid cell by grid cell. This is the process used to make the preliminary maps contained in this atlas.
Final maps are controlled by precise latitude and longitude locations of a selected set of features (usually craters) determined by methods of analytical photogrammetry (Davies and Katayama, 1983a,b). The precise camera orientation at the time each picture was taken is determined as part of this process. This information is used to transform pictures to appropriate map projections in the computer (fig. C-6) so that they can be placed in photomosaics.
 The surfaces of the Saturnian satellites are portrayed with the airbrush by specially trained cartographic illustrators. These drawings show details visible in the Voyager pictures, with their correct relative emphasis, in a coherent fashion. This is possible because a human interpreter is able to examine many different pictures of the same area and to build a mental image of what that area would look like in a picture taken by a "perfect" camera. A proficient cartographic illustrator can draw an accurate representation of that mental image in the form of a map. The airbrushing technique is described in more detail by Inge (1972), Inge and Bridges (1976), and Batson (1978). The difference between preliminary and final airbrush maps is a function of the time available to make the drawings. Meticulous and detailed interpretation of all available image data is usually postponed until the final mapping phase. Preliminary airbrush maps are made at smaller scales and with more generalization of image detail than final ones; they are also made in a matter of weeks, rather than months or years.
Airbrush portrayal of the Saturnian satellites with Voyager images differs from that used for most planetary map series because of the unique character of the Voyager data set. Albedo features dominate the maps because most of the pictures have low resolution and were taken under high solar illumination. If portrayals were restricted to relief only, as is done in standard planetary mapping with more uniform coverage, only a very small part of each satellite could be mapped.
Projections and scales
The airbrush map drawings were originally compiled on conformal projections, which preserve the correct shapes of most mapped features, although map scale changes rapidly with latitude. For example, a crater at 60° latitude on a Mercator projection will appear to be twice as large as a crater of the same size at the equator. The drawings were then digitized; that is, their brightness values were measured at regular intervals and recorded on magnetic tape in the same format as digital television pictures, allowing them to be modified by digital image processing techniques discussed in appendix C. A computer program that changes one map projection to another was used to transform the original conformal projections to Lambert azimuthal equal-area projections. Forms are somewhat foreshortened near the edges of a global equal-area projection, but surface areas can be measured easily. For example, a dime placed anywhere on the Rhea equal-area map covers about 25 000 km2 (almost 10 000 mi2).
Preliminary planetary maps must be compiled in a short time period and be easily reproduced for distribution to mission scientists and engineers. The number of map sheets and total map area must, therefore, be kept to a minimum, mandating scales that are smaller than those that will be used for the final mapping.
The selection of scales for final planetary maps is based on image resolution. It is assumed that at least five picture elements, or pixels, are required to resolve a feature, and that mapped features should not have dimensions smaller than 1 mm at map scale. Consistent scales within map series are necessary for comparison of features. Scales are, therefore, selected on the basis of the availability of a few high-resolution images, with the result that some image areas are enlarged far beyond the five pixel per millimeter optimum scale.
The term "resolution" as it is used in this atlas refers to the size of a picture element on the surface of the body being imaged. The size of the pixel can be computed by multiplying the field of view of a single pixel (in radians) by the distance between the camera and the satellite (fig. D-1). True resolution is a much more complicated concept and is not discussed here. Even this simplified concept of pixel resolution is complicated by viewing and illumination geometry. For example, a pixel is square (and therefore covers a minimum area on a planet) only when the surface being viewed is perpendicular to the line of sight of the camera. The "viewing angle" of 90° is therefore optimum. As that angle becomes smaller, the....
 ....footprint of the pixel on the ground becomes elongated and covers a larger area, so less surface detail is resolved. At the limb, or edge, of the disk of a satellite or planet, the viewing angle is zero and resolution is very poor except occasionally for profiles of mountains.
Resolution of surface landforms is usually best under low-Sun illumination, because shadows add contrast to surface relief. In a picture such as figure C-8, images are most clearly resolved near the terminator, where the viewing angle is nearly 90° and features are illuminated by grazing sunlight. The resolution in that picture is poorest at the limb, where the viewing angle is small and the solar zenith angle is small. When such an image is reprojected for viewing from an overhead perspective, as in figure C-6, the resolution loss near the limb causes the images to appear to be smeared.
The resolution figures shown in the diagrams at the beginnings of parts 1 through 6 were computed by dividing the nominal pixel size by the sine of the viewing angle or by the sine of the solar zenith angle, whichever resulted in the lowest resolution value. Images are not resolved at all, obviously, when solar zenith angles are greater than 90°; that is, when it is night at the site being photographed.