H. Hammel, A. Ingersoll, G. Orton, Z. Sekanina, T. Takata, R. West, P. Weissman C. Chapman, D. Gavel, M. Nolan, M. Mac Low, M. Marley, and M. Boslough
This is a compilation of individual reports lightly edited by M. Boslough.
Hammel et al. presented preliminary results from analysis of the Hubble Space Telescope imaging observations of Jupiter. They discussed observations of observed phenomena ranging from plumes to waves to atmospheric evolution. Of particular interest were the timing of the images of the observed plumes, and the correlation with other observations that occurred at or just after impact, specifically the Galileo observations. Hammel et al. also presented a summary of observed latitudes and longitudes for each observed impact site, along with an assessment of the relative size of the disturbance for each event.
Ingersoll described models of waves from the impacts, and attempted to place these models in the context of observations. The HST images revealed two concentric rings spreading out from the impact sites at speeds of about 500 and 300 m/s, respectively. Both rings spread more slowly than the slowest acoustic wave, which propagates at the altitude of the temperature minimum at 800 m/s. Gravity waves bracket the observations, but the mechanism that selects out the particular speeds that were observed has not been identified. By successfully matching the observations, we may learn more about the vertical distributions of ambient temperature and energy release from the impacts.
From analysis of the photometry of HST images, West, Karkoschka, Friedson, Seymour, Baines and Hammel found:
Chapman presented some freshly-downloaded Galileo data. The non-imaging scan- platform instruments on Galileo (the Ultraviolet Spectrometer, the Near Infrared Mapping Spectrometer, and the Photopolarimeter Radiometer) had just obtained the bulk of their data on the big G impact. And tape recorded data from the imaging sequence for the K impact were being returned in real-time right during the two-week period centered on the Comet Crash Day! So the Galileo investigators were just at the point that the groundbased and HST observers had been at during the last week of July.
Bill Merline and Clark Chapman had just two days earlier reduced the Galileo imaging data for the prompt phenomena around the time of the K impact. Their crude lightcurve (subject to further refinements) was displayed on their poster. Its overall shape is similar to that recorded by the Galileo PPR for impacts H and L, although the 50 minutes duration is a bit longer (partly lengthened by the increased sensitivity of the camera, which detects the luminous phases directly, not as an added component to Jupiter's signal). Another attribute of the K lightcurve is its "spiky-ness", including a spike peaking just 5 seconds after the impact was first seen. They interpret the first spike as the initial bolide, with the subsequent 45 seconds of luminosity reflecting the fireball phase, withOUT an intervening gap. The additional spikes might be subsequent bolides, although they could be artifacts caused by known scan-platform joggle that haven't yet been corrected.
They also displayed previously released images of the W impact, including the 3 images that show the luminous event from among the 8 images -- taken at 2 1/3 second intervals -- that have been returned to Earth. (Remaining data for W, 18 swaths of typically 8 images each, are now planned to be returned during 22 December to 9 January.) It was fascinating to compare this sequence with the HST sequence of images for W, displayed for the first time at the HVIS meeting. By extraordinary coincidence, HST snapped a picture of W within 1 second of the image taken by Galileo when the W impact flash was at its brightest. (Also coincidentally, both pictures were taken through green filters.) The fact that the Galileo sequence demonstrates that there was no visible luminosity (probably down to 0.01% the total brightness of Jupiter) prior to 5 seconds before the HST picture was taken raises a mystery: how could HST see something presumably occurring behind Jupiter's limb, probably 100 km or more below HST's horizon of visibility? One can hardly expect the fireball to rise above the limb (at 137 km above the 100-millibar level) within 5 seconds of the first bolide luminosity!
Weisman et al. summarized their NIMS observations of the G impact. The Galileo Near Infrared Mapping Spectrometer observed the impact of fragement G of comet Shoemaker-Levy 9 on Jupiter on July 18, 1994. Because of the spacecraft position, 1.6 AU from Jupiter, it had a direct view of the impact site. NIMS obtained full disk spectra of Jupiter in 17 wavelengths between 0.7 and 5.0 microns, every 5 1/3 seconds. The first indication of the impact came in a spectrum taken at 07:33:37 UT (time has been corrected to that of Earth-based observers) when a thermal source at a temperature of approximately 5,000 K and 8 km in diameter was detected. The source was still in the Jovian atmosphere, just above the tops of the ammonia clouds, as evidenced by methane absorption in the thermal spectrum. In subsequent spectra the source was seen to expand at about 3.5 km/sec for a period of almost 1 minute, cooling in the process. We interpret these observations as the fireball from the impact, rising out of the Jovian atmosphere. Further detailed analyses are now underway. This work was supported by the Galileo Project at JPL and was performed at the Jet Propulsion Laboratory under contract with NASA.
Gavel et al. presented High Resolution Observations of the Event Using a Speckle Imaging Camera at Lick Observatory
During the week of the comet collision, July 16-22, they imaged Jupiter at high resolution using a visible light CCD coupled with speckle image processing. They were able to obtain images with 0.3 to 0.2 arcsecond resolution (1200 km on the surface) using the Lick Observatory 3 meter telescope. The images clearly show the comet fragment impact spots, the dark halos of ejecta, and other morphological detail.
Although they did not image any of the impacts as they occurred, they obtained 3 to 4 hour contiguous sequences of images of the planet with the purpose of observing dynamic behavior such as shock waves or other after effects of the impacts. The data from July 19 shows the G-D spot on its third rotation. The expanding ring seen by HST on the first rotation seems to have dissipated by that time. The L spot was imaged on its first rotation on July 20 and so far our data shows no evidence of an expanding ring, although more of this data needs to be processed. HST tracked rings expanding at about 600 m/s on four of the impact spots, but does not have L spot data. The nature of this phenomenon is still under debate. It may possibly be a sound wave or gravity wave trapped in the tropospause.
Comet Impact Network Experiment (CINE)
Mike Nolan presented some preliminary observations by Steve Larson's Comet Impact Network Experiment (CINE) of a visible-wavelength plume from the H-impact. This plume was observed beginning 4 minutes after the time reported by the Galileo PPR instrument. The plume was observed 3000 km above the limb of Jupiter using the 4.2 m William Herschel telescope at La Palma, Spain.
IRTF Seismic wave team (Marley, Walter, Dayal, Deutsch, Fazio, Hoffmann, Hora, Hunten, Sprague, Sykes, Wells) reported on the search for seismic waves from the IRTF. Weather limited the seismic wave search to the aftermath of only impact R. Given the actual experimental sensitivity, waves of temperature amplitude exceeding about 0.7K should be detectable. No waves have yet been detected. Assuming the estimate from Zahnle and Maclow that about 15% of the impactor energy should couple to acoustic waves is correct, the non-detection places an upper limit on R impactor energy of a few times 1E28 ergs. Further analysis should better refine this limit.
All five of the U.S. groups that modeled the impacts through the fireball stage were represented in Santa Fe.
The models are not yet converging; at least when it comes to hard conclusions about how big the fragments were and how deeply they penetrated. That is not necessarily bad news--it means that there is a menu of working hypotheses to choose from. There is good agreement among various modeling groups about much of the phenomenology; particularly about the high fireballs and plumes that were predicted before the impacts by all five groups (for sufficiently large impactors).
Some of the disagreement among the modelers comes directly from the computational simulations themselves: issues such as equations of state, numerical resolution, 2-D vs. 3-D, and initial condition assumptions. As was the case before the impacts, much of the disagreement centers around interpretation of the simulations: What in the simulation leads to something observable? What is physically significant and what is an artifact? What is a reasonable way to extrapolate the simulations to later times?
Another factor causing a dispersion among the models is a lack of consistent interpretation of the actual impact data. Each group of modelers is attempting to interpret the data within the context of its own simulation. Only as small subset of the observational data has actually been seen by the modelers. As the data becomes more available, the various working hypotheses will either have to be modified or rejected.
Also, the series of impacts do not appear to have been a "clean experiment". Some of the impacts may have been multiple, and some of the flashes of light were possibly scattered off of dust in the coma. It is not immediately obvious what the various sources were, but the modelers think that they can provide the framework for interpretations. There was general optimism that, as more data become available, a consensus would be possible. Two end-members of the current distribution of opinion are described.
Mac Low and Zahnle made the argument for small impactor sizes. They presented five major observational results that can be consistently explained by impactors with size of order 1/2 km and impact energies of order 1e27 ergs. The first two observations are the lack of atmospheric water and the abundance of sulfur in the plumes. Their analytic and numerical models show that small impactors will release most of their energy in the ammonium hydrosulfide cloud layers, raising much sulfur and no water. The third observation is that the fireballs were quite dim--no visual observations of fireballs were made from Earth, and Galileo saw fireballs less than 10 percent the brightness of the planet. Models of the lightcurves show that objects bigger than about 1 km should have produced fireballs visible from the Earth and much brighter in the Galileo observations. The fourth observation is the preimpact length of the train of the fragments. Tidal breakup models appear to agree that the observed length of the train can be explained by a parent body of diameter 1.5 km, leading again to the conclusion of small impactors. The fifth and last observation is the spot sizes. Small impactors appear quite capable of producing 10 000 km spot sizes with the observed structure.
The big theoretical question appeared to be how fast and how high plumes rise from the explosion of an impactor of any particular size. There are unresolved problems with every plume simulation. Mac Low and Zahnle are using an analytic model, while most other groups are attempting to directly transfer entry models onto a larger grid. Mac Low believes that the low resolutions used in the tails of the other entry models are suppressing Kelvin-Helmholtz instabilities that should couple the entry trail to the surrounding atmosphere efficiently. This could explain the difference seen in depth of energy deposition. The subsequent transfer of the trail onto an even lower resolution grid then cools it substantially. In Mac Low's models, the plume rise is extremely sensitive to the initial temperature, so this could be the determining factor. The temperatures may not be exactly right, because of the lack of a detailed equation of state, and the analytic model does not resolve the shock structure in the trail any better than the low resolution numerical models. However, Mac Low says they have better control over their temperatures, and think they are forcing temperatures into at least the right regime while everyone else is much too cold.
Crawford, Boslough, Trucano and Robinson (Sandia Labs) argue for significantly larger fragments, with an upper bound on the largest between 2 and 3 km. This could be two orders of magnitude greater in mass and kinetic energy than the other extreme if the density were the same, but the energy deposition curve in the upper part of the atmosphere depends more on diameter than mass. The Sandia model is still open to large, low-density fragments or clusters. The big difference is in the fraction of energy that goes into the fireball. If the large fragment is a solid piece, then most of the mass and energy in the Sandia simulation goes deep and stays deep. The Sandia modelers attempted to explain much of the observed phenomena in terms of their model:
There is still substantial uncertainty in the size estimates from the Sandia model, because precise, unambiguous arrival times for the fireball at the limb have not been available. The early-time fireball trajectory depends on the fragment size, but an uncertainty of 15 seconds in limb arrival time can make an order-of-magnitude difference in estimated mass.
The Sandia team believes the reason that they are getting deeper penetrations than Mac Low for a given fragment is that the Sandia simulations include the velocity field in the insertion of the entry wake into the fireball calculation. The material in the lower part of the wake is still moving downward at very high velocity. It continues to do so, carrying mass and energy with it, while the fireball is exploding out the top.
Comparison of theoretical to observed radiances and plume heights
Takata, Ahrens and others used plume height and observed radiances to infer fragment and progenitor size for the comet. Pre-impact diameters of Comet Shoemaker-Levy 9 fragments (SL9) were computed by comparing observed maximum height of plumes and their radiance with detailed Smoothed Particle Hydrodynamic (SPH) modeling of impact in three dimensions [Takata et al., Icarus, 109, 3-19, 1994; Ahrens et al., GRL 21, 1087-1090 and 1551-1553, 1994a,b]. Comparison between preliminary absolute peak radiances of the plume from impact of fragment R observed by the NASA/Infrared Telescope Facility SL9 Team with the MIRAC2 camera (Hoffmann et al., Infr. Phys. Tech. 35, 175, 1993) at 7.85, 10.3 and 12.2 microns with SPH power versus wavelength yields (using logarithmic interpolation at 10 microns) a diameter of ~1.5 km. In contrast, maximum plume height after impact is directly proportional to fragment energy, for example, inferred from images of the G impact recorded by the Hubble Space Telescope (Hammel et al. 1994, Science, in prep.) yields a height above the ammonia cloud deck of ~3300 km. Interpolating previously maximum plume height calculated for 0.4 and 2 km diameter fragments yields a fragment diameter of ~2 km (assuming a density of 1 g/cm3). Using these values to calibrate the relative sizes of fragments [Weaver et al., Science, 263, 787, 1994] yields an estimate of the progenitor SL9 diameter of ~4 km and a total impact energy of 6 x 10**30 ergs or ~ 10**8 Mton of TNT equivalent impact energy for the entire SL9 fragment chain. The SL9 progenitor diameter is less than the seven short-period comet nuclei in Luu's [Pub. Astron. Soc. Pac. 106, 425, 1994] catalog. Comparisons between other aspects of the SPH model, such as the penetration depth, will be made as new observational data become available.
Dimensions and fragmentation of nuclei.
Sekanina presented dimension and fragmentation estimates of comet nuclei. Central regions on the digital maps of 13 nuclear condensations of Comet Shoemaker-Levy 9, obtained with the Planetary Camera of the Hubble Space Telescope on January 27, March 30, and July 4, 1994, have been analyzed with the aim to identify the presence of distinct, major fragments in each condensation, to deconvolve their contributions to the signal that also includes the contribution from a surrounding cloud of dust (modeled as an extended source, using two different laws), to estimate the dimensions of the fragments and to study their temporal variations, and to determine the spatial distributions of the fragments as projected onto the plane of the sky. The deconvolution method applied is described and the results of the analysis are summarized, including the finding that sizable fragments did survive until the time of atmospheric entry. This result does not contradict evidence of the comet's continuing, apparently spontaneous fragmentation, which still went on long after the extremely close approach to Jupiter in July 1992 and which, because of the jovian tidal effects, may even have intensified in the final days before the crash on Jupiter. On plausible assumptions, the largest fragments are found to have had effective diameters of ~4 km as late as March and even early July 1994. In most condensations, several sizable companion fragments (~1 km across) have been detected within ~1000 km of the projected location of the brightest fragment and the surrounding dust cloud has been found to be centered on a point that is shifted in the general direction of the tail, probably due to effects of solar radiation pressure. Since the developed approach is based on certain premises and involves approximations, the results should be viewed as preliminary and the problem should be a subject of further investigation.
Various comments of participants:
One BIG RECCOMENDATION coming out of the HVIS meeting is that the comet session at DPS should begin with a DEFINITION OF TERMS! It seems everybody had to put a lot of effort in to defining just what they meant by fireball, plume, thermal emission, etc.
The big question is still the size of the impactors. There are good arguments for both small and large objects, but no definitive answers are yet at hand.
I perceive a general trend toward larger dimensions for the incoming fragments, the largest of which obviously were >> 1 km. It seems that of the modelers, only Zahnle et al. still stick with subkilometer-sized bodies.
The puzzle that still remains to be untangled is which observations refer to the penetration phase of the events and which to the resulting post-explosion phase. Surprisingly, one of the most controversial points is a discrepancy among the reported timings of the various observations, something that should not be a problem. Consistent interpretations of the events observed virtually simultaneously from the Galileo spacecraft and from the Hubble Space Telescope, which will probably be critical in these efforts, still were not available at the meeting.
The splat is an attractive alternative for the source of the strong IR flux.
The sources of the shadowed radiation observed by HST could be (a) thermal, (b) light reflected along the coma (tail) by the bolide, or (c) light reflected along the dust remaining in the atmosphere from the breakup.
The impact sites are glowing in the infrared at 7.85 microns, showing that the stratosphere was warmed (this much I knew), and Marley's careful registering of frames showed me that the warm areas are quite extended around impact sites and that the shape of the warm areas is similar (the same, I suspect) as the visually dark areas. I infer from this that the stratosphere was heated up not only by the impact, but from above by the heated ash.
The Galileo data suggest (if radiation is thermal in original) that the impact is immediate, color temperatures are in the range of 7000 - 10 000 K and emitting areas have an equivalent diameter of 1 - 5 km at the start. NIMS data from G suggest early expansion/rise rates of 3.5 km/sec, which is smaller than the velocity needed to reach the WF/PC2-observed altitudes on the order of 3000 - 4000 km, so the rate obviously needed to accelerate.
I'm not comfortable with the near burial of cometary stuff, because we see spectra of metals in the impact sites even hours later, and these just DO NOT come from Jupiter.
What is the bright spot observed by HST in Jupiter's shadow? Is it really emission, and from what?
Why can't we figure out the timing? Has a bolide been observed?
Comet Shoemaker-Levy Home Page