CCD observations are rapidly replacing photographic observations of the satellites. For faint satellites, the reasons are quite clear; the high quantum efficiency of the CCD has made it possible to obtain solid observations of satellites with poor observational histories. And for those faint satellites embedded in the halo of the primary, the linear response of the CCD has made it possible to easily remove the astrometric and photometric effects of the halo.
What is not so clear is why the CCD is being applied to the astrometry of bright satellites. Certainly digitized images are easier to work with, and from a practical perspective, it is much easier and more economical to use a CCD for astrometry than to obtain and maintain a large plate-measuring machine. There is, however, no astrometric advantage of the CCD for bright images. On the contrary, there are several important astrometric problems to overcome.
Since the CCD is a metric device as well as a detector, it must be calibrated in the same manner as the plate-measuring machine. It should be determined to what level the rows and columns are evenly spaced, the same width, and to what level they are orthogonal. Also, it must be ascertained to what level the chips are flat; most are bowed or crinkled which may change with time and temperature. These calibrations are more important for satellite astrometry than for other astrometric applications, such as parallax astrometry.
In addition, the small field of the CCD makes it difficult to solve the classical problems of scale, orientation and coordinate origin. Unless focal length is sacrificed for sky area, one will be limited to a scale and trail reduction of the observations. But one cannot expect to obtain a scale value accurate to 1 part in 10,000 from catalog stars (whose typical relative errors are +/-0.4 arcsec) over a CCD chip just 400 arcsec in size! Likewise, the orientation of the CCD chip cannot be determined with the same precision as a photographic plate, with either trails or stars. Coordinate origin is generally not a problem if intersatellite observations are sought. If, however, the position of the primary is also desired, it will be necessary to build a CCD mosaic.