J. Brit. Astron. Assoc., 107, 1, 1997, p. 3–4

The Great Comet Crash: the view gets clearer


July 1994: an infrared image of Jupiter shows the effect of the impact of SL9 fragment A. John Spencer (Lowell Observatory) and Darren Depoy (Ohio State University) used the 4-meter telescope at the Cerro Tololo Interamerican Observatory (CTIO) at La Serena, Chile.

In 1996 July, a conference at the Meudon Observatory, Paris, reconsidered the 1994 collision of Comet Shoemaker–Levy 9 with Jupiter. There were no major revisions of conclusions reached a year earlier (Rogers J. H., J. Brit. Astron. Assoc., 106(2), 69–81 (1996)) but some outstanding questions have been clarified, so this account will be presented as an updating of topics raised in that review.


Structure and size of the comet
Were the 20 'fragments' solid planetesimals, or re-aggregated rubble-piles? W. Benz and E. Asphaug have now ruled out the former option; after an SL9-type tidal breakup, the fragments must reaggregate into clumps by self-gravity. But these clumps could contain large solid lumps, so a compromise solution is possible. Only Z. Sekanina now holds out for a large, solid, initial object – 9km across, breaking up 2.5 hr after perijove – but as a tidally-stretched rubble-pile would be ~9km long at that time, it can satisfy the observations just as well.

More difficult is the question of secondary splitting, which does seem to require coherent and heterogeneous objects in the clumps, and may well have involved cometary activity. Sekanina, with P. Chodas and D. Yeomans, presented a complete genealogy of the fragments, in which all the off-line ones (which later disappeared or fizzled) split off from the next-following on-line fragment, some time in 1992 (or 1993 April for Q2). They proposed that these 'wimps' were small fragments flung off the rotating larger fragments during local afternoon, by centrifugal force after warming by sunlight; their motion can be explained if all the fragments retained similar rotation axes. In a mixed rubble-pile model, perhaps the same could be achieved by interactions of low-mass, mixed-density rubble orbiting or jostling around a large rotating lump.

There is now good agreement between most estimates of the sizes and energies of the fragments, although Sekanina still argues for larger sizes. The rubble-pile model predicts a progenitor with density 0.5g/ cm[^-1] and diameter 1.5km (with a two-fold uncertainty because of its possible rotation). In that case the largest fragments (G, H, K, L) would have approximately density 0.5g/cm[^-1], diameter 0.75km, mass 1×10[^14]g, and impact energy 2×10[^27] erg. Indeed, all the best estimates of the sizes of the impacts are now converging within a factor of two of these values (see Table).

Effects on the magnetosphere and ionosphere
The X-ray emission recorded by ROSAT during the K and P2 impacts was an intensification in the normal northern auroral region. This may have been caused by the cometary coma dust burning up in the ionosphere, triggering magnetic waves which spread to the opposite hemisphere. Infrared emission from H3+ and methane in the ionosphere revealed the earliest stages of an impact, as recorded by S. Miller's group at UKIRT during impact C. They found the ionosphere beginning to heat up 4 minutes before the impact, presumably due to the coma meteor storm. Within minutes after the impact, the temperature reached several thousand degrees, and faint hot methane lines were blueshifted by 25km/s – probably the initial blast wave surging over the limb and out into space. Then the main part of the plume came crashing down to dazzle the detectors in the 'main event'.

Dynamics of the impacts
The Galileo team have now worked out a temperature of ~24,000K for the bolide entry flash seen in impact Q1! For impact G, their value of 8800K probably refers to 1–2 seconds later, as the entry trail began to erupt into the fireball.

Of the two groups leading the modelling effort, the one which previously favoured large impactors (D. Crawford & M. Boslough) presented a new model which favours smaller ones, agreeing with others (see Table). The mechanism of their plume is still somewhat different, and they argue that the impactors must have penetrated at least 30km below the ammonia clouds.

It remains certain that the impacts did penetrate into the troposphere, as expected for impacts of this size. The evidence is: (i) the level of the initial fireball viewed by Galileo; (ii) models for production of 3000km-high plumes; (iii) the presence of sulphur separate from the cometary water; (iv) the vertical distribution of smoke in the cores of the sites; (v) the presence of ammonia lofted into the lower stratosphere.

Thermal effects
A new theme at this meeting was the importance of dust or smoke in the distribution of the heat. These particles were responsible for the huge thermal emission seen in the splashback. During this main event, the 'dust' remained at 600K (±100K) throughout (P. Nicholson), but the gas temperatures of methane and CO climbed steadily from ~600K to 4000–5000K (K. Zahnle, R. Knacke). The reason is that dust radiates heat away more efficiently.

The dust in the ejecta was confirmed as being in the stratosphere:
– an altitude of >1000km was deduced from parallax on optical images (S. Limaye);
– initially the top was around the 0.5 mbar level, from HST image photometry (R. West);
– initially the bulk of the dust was above the 30 mbar level, from 7–13µm imaging (T. Livengood et al.), settling to the 35 mbar level over the week following the impacts.

The observed heating above the impact sites was reviewed by B. Bézard, who reinterpreted various groups' data to conclude that all the observed heating was in the stratosphere, from the 0.5 mbar level upwards, the enhancement reaching 30–60° at higher levels. This implied ~3×10[^26] ergs of heating above site L when 11 hours old – more than had been radiated away (see Table). The IRTF report of heating in the troposphere, which would require ~10[^29] ergs, may be (controversially) interpreted as due to emission from warm dust in the stratosphere. The presence of dust also explains why the stratosphere cooled rapidly – within one week for most sites. Very fine silicate particles, so fine that they remained suspended for weeks above the 0.1 mbar level, explain this behaviour.

Chemistry and long-term evolution
The Galileo Probe data have produced confusion rather than clarification, but probably make little difference to the conclusions. If the atmosphere is indeed devoid of water, K. Zahnle showed that most of the correct molecules can still be produced if the cometary material comprised 6–8 percent of the mass of the plume. But anyway, it is clear both from the sulphur in the plumes and from the energies of the impacts that the impactors did not go deep enough to encounter jovian water. In any case, S. K. Atreya argued that the Galileo results were probably due to a strong local downdraught of dried-out air in the NEBs hotspot where the Probe entered, and that the comet impact sites might have had a composition much closer to what was expected.

Cometary water was deposited high in the stratosphere, and might have been dense enough to condense into a white haze (J. Moses). So might the ammonia which welled up into the lower stratosphere. So it is interesting that HST visible-light images in late July, and mid-ultraviolet images in late August 1994 possibly showed bright clouds along the edges of the dark impact material – although these remain to be analysed. CO, CS and HCN have all remained in the stratosphere up to mid-1996, at levels of 0.02–0.4 mbar and 155K, showing little if any diminution and spreading close to the equator (A. Marten). They may last as long as 30 years, which is the theoretical time for diffusion from the stratosphere into the troposphere. However, this still seems too short for cometary impacts to be a major source of the normal jovian CO.

Amateur images showed that the visible dark belt broke up and became unrecognisable after 1995 July (J. Rogers & M. Foulkes). But dark debris was visible for longer in HST mid-UV images (R. West); as shown by the sole image at 255nm previously published, during impact week, this wavelength is exquisitely sensitive to absorption at any level in the stratosphere. Such images in 1994 August showed a spectacular broad convoluted impact belt. Images in 1995 and 1996 April showed a thinner, diffuse impact belt mainly near the limb (spreading to ~20°S but no lower, unlike the gases; most of the dust had probably now settled to ~100 mbar), and also an incomplete belt around the pole (>70°S) – possibly related to a very dark south polar belt (63–69°S) seen in I. Miyazaki's images in 1995 and 1996.

Meanwhile, infrared images at 2.3µm also showed diffuse impact haze persisting in the south at least until 1995 September (H. Hasegawa; J. Spencer); and HST images at 0.89µm still showed slight limb-brightening at the impact latitude in 1996 April (R. West).

John Rogers, Michael Foulkes & Richard McKim, Jupiter Section


Back to top of page

Return to Journal 1997 February contents page