Figure 9
Since the SL-9 impacts occurred on the leading side of Jupiter, but on the back side, beyond the limb, as seen from Earth, Galileo was in the unique position of being able to see the impacts as they occurred, rather than observing the effects several minutes later, as was the case for all Earth-based observations (See Fig. 1, Section 1).
Normally when designing a science observation, there is little if any uncertainty about the desired time of the observation; the uncertainty is in target ephemeris and instrument pointing accuracy. A common solution to this is to design a mosaic that covers an area large enough to insure that the target is observed. The comet impact observation sequence design problem was effectively the opposite of this-- the target (Jupiter) ephemeris was not a factor, and instrument pointing accuracy needed to be considered, but was not a significant issue, but the time of each event was uncertain to tens of minutes. This posed a significant challenge in the design of these sequences, especially in view of the fact that the sequences were very complex and could not be fully updated as the impact time estimates changed. The measurement strategy development faced the additional challenge that temporal scale and intensity of the impact phenomena were generally uncertain by several orders of magnitude. Most all the observation data had to be stored on the tape recorder, which can only hold the data equivalent of about 150 SSI full frame images. Due to the 10 bps telemetry limitation, only about 5% of the tape can be returned in the time available. Even after the impacts had occurred and were recorded, the impact times still would not be known to better than a few minutes, and maybe more, depending on how much was seen by ground-based observers, so the problem was how to know where on the tape to go to retrieve the data for playback.
Five of the eleven instruments on the Galileo spacecraft were deemed suitable for observing the comet impacts. The instruments that make in-situ measurements are not usually suitable for making observations at a distance of 1.6 AU. Four of the five instruments were the remote sensing instruments mounted on the scan platform - SSI, NIMS, PPR, and UVS (See Ref. 4 for instrument description). The fifth was the PWS, included because of the possibility that it could detect radio frequency emissions caused by the impacts. A sixth instrument, the Dust Detector Subsystem (DDS), will watch for changes in the dust streams from Jupiter, which take 1 - 2 months to reach Galileo.
A new capability for the Galileo orbital mission that was implemented early for SL-9 is on-chip-mosaicing. This refers to the ability to make multiple exposures of the target on the CCD before copying to tape. The significant advantage of this capability over a single exposure per frame was that up to 64 Jupiter images (an 8x8 array resulting from 64 multiple exposures) could be stored for a tape space cost of one frame-one exposure per frame would have limited us to the aforementioned 150 images making it virtually impossible to capture even a single impact because of the timing uncertainties. An example array of images is the one used to observe the W fragment impact shown in Figure 9. SSI observing strategies varied over the six events it covered - for W, the platform was moved and an exposure shuttered every 2-1/3 seconds to provide images giving a time/intensity history of the impact response.
Figure 9
Generally the observing strategy was for a single instrument to be prime for a given impact, and other instruments could ride along where practical. This was driven both by the need to limit the complexity of the sequences, as well as, the conflicting requirements of the different instruments; SSI wanted to move in many small steps to build up the on-chip mosaic, NIMS wanted to sweep across Jupiter to account for pointing inaccuracies, PPR wanted to point in a single fixed direction. For all but one observation, the approach used for the PPR data was to store the instrument data directly in the spacecraft central computer in a buffer that was then played to ground on a nearly daily basis. This was feasible because the PPR data output rate is relatively low, and had the advantage of providing a near real-time return for determining impact times. The utility of this in support of recorded data return was limited by the fact that the PPR was not observing the same impacts as were being recorded by the other instruments. An identical approach was used for the PWS, which observed continuously from before the first impact through the entire sequence. Again, a low data output rate made this feasible.
The sequencing strategy used for the recorded observations to solve the problem of changing impact time estimates was to design an observation window for each event of about two hours duration, during which the instrument would be pointed, operate, and take data. Then, a moveable record window of about one hour duration was placed in this observation window during which the data would be recorded. The shorter record window was necessary to avoid overflowing the tape, and since its placement could be updated quite easily late in the process, it provided the flexibility to respond to late changes in the impact time estimates. Figure 10 shows which fragment impacts were observed by which instruments and the approximate times of impact.
Figure 10
Figure 11
Figure 12
Figure 12. Galileo SL-9 K Event
Figure 13
Figure 13. Images of Jupiter at 2 1/3 sec intervals showing the night-side impact of fragment W of Comet Shoemaker-Levy 9
Playback searches for the purpose of locating the data of interest on the tape will be complete by late September, and from then through January of next year, the DSN tracking allocated to Galileo will be virtually dedicated to the return of the impact observations. The D and E observations were missed due to a modeling error in the ground software used to design the spacecraft sequence. Indications are that the balance of the Galileo observations were very successful, and will contribute new and unique information not available from the vantage point of Earth.