1 Introduction

Optical positions of the outer planets (Jupiter-Pluto) are important to improve and maintain their ephemerides since modern methods of observation (radar ranging and interferometer) are not effective at the distances involved (Standish et al. 1995; Standish 1998). In particular, these data are needed to plan and support spacecraft missions to the outer solar system (Cassini, Europa Orbiter, and Pluto-Kuiper Express).

In history, astrometric positions of the outer planets (Jupiter and Saturn) were not well determined by the classical observational methods. The external mean error of such observations varied from 200 to 500 mas (Standish 1995; Pascu & Schmidt 1990). The largest systematic error was called "phase effect'' and due to an augmentation of the geometric phase defect on the planets which was difficult to account for (Fienga 1998). At present, the positional observations of the outer planets (Jupiter and Saturn) have to be made first by the measurement of their satellites, and then derived by the comparison of satellites' measured positions and theoretical positions (for example, Rapaport et al. 2002; Stone 2001; Stone & Harris 2000; Fienga 1998).

In astrometry of some natural satellites, the precise measurement of the center of their primary planet is decisive. Because in some cases, to accomplish the removal of the background gradient produced by the over-exposed image of the planet, the planetary image is supposed to be symmetric relative to the two axes of a particular reference frame, both of which contain the planet's center. As we have known, the technique of the removal of the background gradient has been used in the reduction of some outer planets' satellites, such as Jovial, Saturnian and Uranian satellites (Veiga & Martins 1995; Stone & Harris 2000; Peng et al. 2002 - hereafter, called Paper I).

Nevertheless, the direct measurement for the outer planets (such as Jupiter, Uranus and Neptune) with good precision has not been made yet.

On the other hand, their geometric centers or geometric radii can be determined precisely for some celestial bodies with big disks. For example, the most recent observations (Penna et al. 2002) show that the solar radius can be measured with the precision of 0.18 arcsec using CCD Astrolabe. Imagining the huge geometric size of the solar surface, we think the measurement with so good precision must have its measurable base. Besides, Pascu & Schmidt (1990) determined positions of Saturn by fitting 12 points around its ring obtained by manual setting with an ellipse. The internal precision due to the measurement of this planet is as good as 0.10 arcsec. These methods have a common base, i.e., the edge of the celestial body with a big disk has a precise and measurable base.

Based on this idea, an elliptic fit to the edge of Jupiter on its CCD image is tried to determine its geometric center in the present paper. The positional precision is tested by many CCD images we have obtained. The results show that as good as 0.03 arcsec of the internal precision can be reached. The remaining parts for this paper are arranged as follows: Sect. 2 gives the descriptions of our observations and instrumentation used; in Sect. 3, the positional measurement of Jupiter and its Galilean satellites is described in detail; Sect. 4 presents the calibration for the CCD field of view; Sect. 5 deals with results and analyses; the last Section will be the conclusion remarks.

Figure 1: A typical CCD image for the Jupiter and its Galilean satellites. The exposure time for this image is 2 s.