There are several sources of information that can be used to estimate the actual impact times of the major fragments of Comet Shoemaker-Levy 9. Astrometric data has been used to determine updated orbits and these orbits have been used to determine the predicted impact times when the computed position of the fragment enters the 1 bar (atmospheric pressure) level of Jupiter's atmosphere. Values for Jupiter's mean radius and obliquity were taken from Reference 1. Because the astrometric data nearest the impact times themselves are the most powerful for reducing the error of the predicted impact times, we were particularly fortunate in receiving recent astrometric data from the European Southern Observatory (Richard West, Olivier Hainaut and colleagues) that were reduced with respect to the Hipparcos star catalog. Extremely valuable sets of astrometric data were received from the U.S. Naval Observatory in Flagstaff (Alice and Dave Monet), McDonald Observatory (A. Whipple, P. Shelus and colleagues), Spacewatch (J. Scotti and colleagues) and a number of other observatories. The final pre-crash impact predictions were sent out on July 16, 1994. As we received astrometric data from Dave Jewitt and Dave Tholen on a few of the trailing fragments on July 19, a revised subset of the July 16, 1994 predictions were issued just before midnight on July 19, 1994. For completeness, the final predicted times of impact are given in column 1 in the table below.
Impact times were determined by Andy Ingersoll and Reta Beebe using Hubble Space Telescope information on the location of the northwestern edge of the dark spots resulting from some of the impact events. The longitudes determined for these spots were compared to the longitude predictions given by Chodas and Yeomans and the differences in longitude were converted to time differences between actual and predicted impact times. The errors associated with this technique are a estimated to be a few minutes. For some fragments there are more than one impact time estimate determined from different frames from the Space Telescope. We put the most weight upon those determinations made from the frame taken closest to the impact time.
For fragments H and L, the Galileo PPR instrument observed the flash phase of the bolide entry so that for these two cases, we have impact time data that are accurate to +/- 5 seconds. However, since we do not know whether the PPR times correspond to the initial impact or to a subsequent flash, we have assigned uncertainties of +/- 1 minute. There is also a hint of a signal in the PPR data corresponding to the Q1 impact. These data were provided by Terry Martin. By comparing the H and L impact times determined from the PPR data with the respective predicted impact times, we note that the PPR estimate is 6.1 minutes later than the ephemeris prediction for the H fragment and 7.9 minutes later than the prediction for the L fragment. The average of these two differences is 7.0 minutes and this average, when added to the predicted impact time, will give a rough determination of the true impact time.
For fragments B, D, K, Q1, and R, we have estimates of both the initial flash times and the subsequent first plume observation. We only considered those plume observations seen in the 2-3 micron region. The time differences between initial flash times and first plume observations were respectively +6, +5, +6, +6, and +8 minutes. An analysis by Andy Ingersoll and John Clarke suggests that for fragment G, there was an 8 minute lag between impact and the rise of the arc-shaped plume to where it could be seen in sunlight. In the absence of other information, the first plume observation minus an average of these values (+6.2 minutes) would give an estimated impact time. From the two fragment impacts observed by the GLL PPR, there is also evidence that the initial flash, as observed by ground-based telescopes, comes about 1 minute after the flash seen by the PPR instrument.
The impact times are UTC times received at Earth (light time corrected).
In setting forth the accepted impact times given in the final column of the following table, the priority of the various available techniques as as follows:
1. GLL PPR timing (Fragments H & L) 2. When definitive flash times are available, with subsequent plume observations noted about 6 minutes later), we generally took the impact time as one minute before the flash time since the PPR instrument recorded its first signals about one minute before the reported flash times. (Fragments D,G,Q1,Q2,R,S,V, and W) 3. Estimates determined from HST longitudes 4. Estimate determined from first plume observation minus 6.2 minutes 5. Chodas/Yeomans prediction with empirical adjustment of + 7 minutes
The impact times for fragments A,C,E,K, and N were determined by considering the ephemeris prediction error (about 7 minutes early for most fragments), the times determined from the HST longitude estimates (uncertainty = 3-4 minutes or more) and the times determined from plume observation times (impact time = plume observation time less 5-8 minutes). An effort was made to consider and balance these three factors and the uncertainties on the estimated impact times reflect our confidence level. For fragment F, the impact time was determined using the ephemeris prediction and the Lowell Observatory estimate of when the F spot was seen on the terminator. In the absence of any quantitative impact time observations for fragments P2, T, and U, only the ephemeris prediction was used (plus 7 minutes). The impact time estimate for fragment B is based upon observatory reports and is relatively uncertain because the impact time occurs before the ephemeris prediction and well before the estimate determined from the HST longitude estimate.
References: 1. Explanatory Supplement to the Astronomical Almanac. University Science Books, 1992, p. 404. Accepted Fragment Date Prediction Impact Time July (HH:MM:SS) & 1-sigma error A 16 20:00:40 20:11:00 (3 min) B 17 02:54:13 02:50:00 (6 min) C 17 07:02:14 07:12:00 (4 min) D 17 11:47:00 11:54:00 (3 min) E 17 15:05:31 15:11:00 (3 min) F 18 00:29:21 00:33:00 (5 min) G 18 07:28:32 07:32:00 (2 min) H 18 19:25:53 19:31:59 (1 min) J 19 02:40 Missing since 12/93 K 19 10:18:32 10:21:00 (4 min) L 19 22:08:53 22:16:48 (1 min) M 20 05:45 Missing since 7/93 N 20 10:20:02 10:31:00 (4 min) P2 20 15:16:20 15:23:00 (7 min) P1 20 16:30 Missing since 3/94 Q2 20 19:47:11 19:44:00 (6 min) Q1 20 20:04:09 20:12:00 (4 min) R 21 05:28:50 05:33:00 (3 min) S 21 15:12:49 15:15:00 (5 min) T 21 18:03:45 18:10:00 (7 min) U 21 21:48:30 21:55:00 (7 min) V 22 04:16:53 04:22:00 (5 min) W 22 07:59:45 08:05:30 (3 min)