CCD Photometry of Galilean Satellites

Abstract Timings of Galilean satellite eclipses obtained since 1990 using CCD cameras agree very closely with recent eclipse predictions derived from Galileo spacec raft data. A comparison of CCD results with ALPO visual timings shows a consistent bias and considerable scatter among individual visual tim-ings. However, the yearly averages of visual obser-vations are consistent with CCD results, provided that the visual data are corrected for telescope aperture. CCD photometry has also led to the fol-lowing new findings: (1) there are variations of eclipse times with respect to the E2 satellite ephemeris on time scales as short as weeks; (2) Jupiter’s polar haze extends several hundred kilo-meters above its cloud tops; and (3) the collision of Comet SL-9 with Jupiter in 1994 injected extremely high altitude dust into the planet’s atmosphere. Background Over the long history of observation of the Galilean satellites visual eclipse data have accounted for most of the timings. Visual data correspond to the moment when the last noticeable light disappears during ingress of a satellite into Jupiter’s shadow or when the light is first noticed during egress. Such timings have traditionally been the basis for computing orbits for the satellites, and the string of observations is almost con tinuou s from the seventee nth ce ntury after Galil eo discovered them. Many data points were obtained during a program of visual photometry at Harvard College Observatory from 1878 through 1903 (Sampson, 1909), and the ALPO program (Westfall, 2000, and references therein) has added thousands since 1975.Attempts to improve upon the accuracy of visual tim-ings by using photoelectric photometry met with only limited success, due to scattered light from Jupiter. CCD technology overcame this problem since the background li gh t cou ld be isolated and remov ed fr om the images. The first CCD timings of Galilea n satellit e eclipses were reported by Mallama (1992a). Figure 1a (page 16) shows an example of the light curve of an eclipse disapp ea rance . These often con tain 100 or more data points and the photometric quality is typi-cally very good. We use another satellite in the same field of view for the brightness reference, and care-fully extract photometric information from the CCD images while allowing for scattered background light.In order to derive an eclipse time from an observa-tion, the observed light curve is fitted to a model light curve. The model is generated on a computer based upon satellite factors (its diameter, surface markings, limb darkening, and motion), jovian factors (the planet’s radius, atmospheric refraction, and extinc-tion due to haze) and geometrical factors (the Sun-satellite-Earth angle and the distance from the satel-lite to Jupiter). Two papers by Mallama (1991, 1992b) give more detailed information about the sat-ellite, jovian and geometric parameters. The model light curve corresponding to the observations in Fig-ure 1a is shown in Figure 1b (page 16). Once the observed and model light curves have been gener-ated they are fitt ed together by lea st-squ ares statistics, as shown in Figure 1c (page 16). This fitting gives the exact time when the center of the satellite was lined up with the limb of Jup iter and the Sun , which shou ld correspond to the mid-eclipse time given in satellite eclipse ephemerides. A complete listing of the CCD results is hosted by the American Meteor Society at http://www.amsmeteors.org/mallama/galilean/.