Of the ten small inner satellites of Uranus discovered with Voyager II in 1986, eight were recovered with the WFPC2 of the Hubble Space Telescope (HST). Thirty three exposures were taken on 14 August 1994 with the Planetary Camera (PC1) in the BVRI filters. Measurable images of Ariel and Miranda were also obtained on the same CCD frames with those of the faint satellites.
The PC1 focal plane has been reported to be affected by wavelength - dependent distortion, which is sizable at some distance from the center of the CCD. Astrometric reductions were made using a linear model for the focal plane as well as several published distortion models. Our strategy was to use accurate JPL Miranda - Ariel ephemerides to calibrate the scale and orientation of the PC1 frames and to evaluate the reduction models.
While filter f555wV has been recommended for astrometry, and has, thus, been the most widely calibrated, we found that detection of the faint satellites near Uranus was most effective in the I filter. This makes the model of Trauger et al. (1995, Calibrating Hubble Space Telescope: Post Servicing Mission, ed. Koratkar and Leitherer, p. 379) most valuable, since it accommodates the bandpass. But even in that model, both filter and geometric differences were detected relative to the JPL Miranda - Ariel ephemerides.
While a linear field model would have been sufficient for milliarcsecond astrometry in the region of the inner satellites, our technique of internal calibration required the full distortion correction for milliarcsecond precision. For full-well images, such as Ariel and Miranda, the astrometric precision, 2-3 mas, was limited by the accuracy of the distortion model and the distance from center, not by centroiding precision. For the fainter satellites, the astrometric precision varied from 7 mas for Puck (V=20 mag) to 50 mas for Bianca (V=23 mag) and was related to background planetary halo involvement as well as signal-to-noise in the images.
Shortly, we expect to obtain HST observations of the inner Neptunian system. While most reduction procedures are expected to be similar, the astrometric problems are more severe. Their resolutions will be discussed here.