AFTER THE 9P IMPACT                                                                                                                  Mark Kidger


(Based on a talk given at Annual Meeting of “The Astronomer” held on Saturday, 2005 October 22)



2005 will be remembered for two astounding feats in solar system exploration following on the success of the missions Cassini-Huygens, Stardust and the Mars Exploration Rovers in 2004. The Hayabusa landing on the asteroid Itokawa in November and, in July, the impact of the Deep Impact probe on the surface of comet 9P/Tempel 1. This article explores the Deep Impact mission and its results, some of which have been rather unexpected.



At 05:44:36.07UT (spacecraft time), 05:52 (Earth time) on July 4th 2005, despite dark predictions from some scientists who were concerned that the nucleus would prove to be a difficult target, Deep Impact’s impacter probe impacted on 9P/Tempel 1. The famous image on the TV screen at the JPL control centre showing the bright flare of impact was received at 05:54:39.46UT and was the first confirmation of a successful impact. A recently released series of images from the camera-spectrograph shows the impact sequence in 12 images taken over a period of 0.7s. The first pinpoint of light is seen in frame 64 from the camera. By frame 68, taken 0.281s later, at 05:44:36.30 the impact point was heavily saturated. The impact was estimated to have occurred at an angle of around 30º and this can be seen from the bright cloud of expelled material that moves down in the frame at high speed, fading rapidly; the cloud appears first in frame 68 at 05:44:36.36 and last seen faintly in frame 70, just 0.121s later when it was moving off the limb of the nucleus. The brightness of the impact flare amazed watching astronomers. No one had expected the impact to be so spectacular in the images from the mother ship. In the control room an impromptu party broke out as scientists and engineers celebrated the success of the mission, while astronomers waiting in telescopes all round the world licked their lips in anticipation of a spectacular show from the comet.


It rapidly became obvious though that the impact would be visually disappointing from Earth. Despite what was some occasionally blatant image massaging to make the difference between “pre” and post-impact more obvious in some images posted in the Internet, nothing could hide the fact that the comet had brightened much less than expected, even by the most pessimistic predictions. The series of images (left) taken by Fabiola Martín-Luis with the 82-cm IAC-80 Telescope at Teide Observatory some 8 hours before impact and then 16h and 40h after impact shows just how small and temporary the brightening was.


So, what was it that “went wrong” with the impact and how is it that the project took place in the first place?



Mission history and background

Deep Impact was included in the NASA programme of “Discovery Missions”. These are missions intended to give faster, cheaper science than the old, classical missions on a grand scale such as Voyager. To qualify as a Discovery Mission a project must cost under $500 million and take 18 months or less from go-ahead to launch. This necessarily means that the missions must be much smaller in concept and simpler than a big set-piece mission. Deep Impact was originally proposed in 1996 as a high-velocity (38km/s) impact with the asteroid (3200) Phaethon, the parent of the Geminid meteor shower and thus strongly suspected of being an extinct comet. The idea was to observe the impact and to (briefly) reactivate the asteroid as a comet. However, the review panel that scrutinised the proposal rejected it on two grounds: they were unconvinced that Phaethon was a dead comet and, what is more, sceptical that a probe could be targeted to impact. The project was revised and resubmitted in 1998 with two key changes: the target was changed from an asteroid to a low-activity comet and a targeting sensor was added to improve the probability of a successful encounter. The mission was composed of two parts: the mother ship and the impacter probe. Instrumentation was, of necessity, limited by the budgetary and mass constraints. The mother ship carried a 30-cm telescope, proudly billed as the largest telescope ever carried into deep space with a high-resolution camera-spectrograph able to take images in any one of 8 different filters, with a near-IR spectrograph to analyse the impact plume. Additionally there was a 12.5-cm medium-resolution telescope for wide-field imaging. The mother probe also had the high-gain communications antenna and the computer for control and data storage. The impacter probe, in contrast, was very basic. With a mass of 373kg the majority of the probe was constructed from copper rather than the more standard aluminium; this was done because aluminium has an extremely complex spectrum when incandescent, similar to high-pressure sodium lighting, whereas copper has just a few, narrow lines, more like low-pressure sodium. The more complex the emission spectrum of the probe when it exploded, the more difficult it would be to separate the spectrum of the impact plume from the contamination due to the probe itself. The impacter carried a 12.5-cm telescope identical to that of the mother ship. However, contrary to popular belief, it carried no explosives. The impacter was a kinetic energy weapon only; at 10.2km/s the kinetic energy of impact was equivalent to 14 times the probe’s mass in TNT; only the carrying of a nuclear device on board would have increased the energy of impact.


After a delay of two weeks a perfect launch took place on January 12th. The delay had no impact on the encounter date, which was driven by a desire to encounter the comet at perihelion (in fact, as is well known, as the comet’s perihelion passage was on July 5th, the mission team decided to advance the encounter by one day to take advantage of the coincidence in date. Unlike the Giotto fly-by of 1P/Halley, Deep Impact was a low-velocity encounter. Comet Halley has a retrograde orbit obliging the approach to he head-on; in contrast, as 9P/Tempel 1’s orbit is prograde and almost in the plane of the ecliptic causing the encounter to be a tail-chase.



Comet 9P/Tempel was chosen because it is a highly evolved object, having been in the inner solar system with perihelion distance less than 10AU for 300 000 years. Prior to 1648 the comet had a much larger perihelion distance than now, making it much fainter. A close encounter with Jupiter reduced the perihelion distance sharply and later encounters have reduced the perihelion distance still further. It is no coincidence that Ernst Tempel discovered the comet from Marseille on April 3rd 1867 shortly after a further downward drop in perihelion distance. The 1867 apparition was almost the best possible with the comet magnitude 9 at discovery and the comet observable for almost 5 months until August 27th. The comet was recovered from Marseille in 1873 by Stefan (of "Quintet" fame) and then by Tempel himself in 1879 after which a new encounter with Jupiter led to the comet being lost until Elizabeth Roemer recovered it in a single image in 1967. Since 1967 the comet has been seen at every apparition. With a 5.5-year period, alternate apparitions are very good and very bad. However, a new encounter with Jupiter in 2024 will increase the perihelion distance once more and make the comet much fainter again. An extensive ground-based observing campaign has been carried out to support the Deep Impact mission. Karen Meech has led this. This campaign started in 1999 with a programme to measure the rotation period of the nucleus. This suffered from enormous difficulties but eventually a period of 1.71 days (40 hours) was measured. The light curve amplitude of 0.6 magnitudes suggested a highly elongated nucleus with a mean diameter of 5.2±0.4km, when combined with an unknown rotation mode, this made the nucleus potentially an extremely difficult target to hit; this was one of the biggest surprises of the mission as the nucleus photographed by the probe was far less elongated than expected. In parallel with the ground based science programme there was also a targeting astrometry programme coordinated by Steve Chesley at JPL. This involved almost exclusively amateurs taking high-precision astrometry to improve probe targeting. This was successful in that the pre-release aim was within a few hundred metres of the nucleus and considerably more accurate than after the first post-release targeting manoeuvre.


The mainly Spanish “Observadores_cometas” team contributed over 5000 photometric measures in apertures from 10-60” (observations are still continuing), more than 2100 astrometric measures, 99 estimates of the total visual magnitude and some 150 images by 32 observers or groups of observers from 4 countries. The total visual magnitude estimates are fitted by a relation of m1 = 6.39 + 5 log D + 24.68 log r. This is similar to the light curve fit derived in 1999-2000. It has been suggested that 9P has been 20% less active in 2004-05 than in its previous return but, as it was a particularly poor apparition with almost no data taken within 6 months either side of perihelion we are talking about activity at high heliocentric distance (looking at the light curve from 1999-2000 the dispersion is so large that measuring a 0.2 magnitude difference in level with the current apparition looks to be difficult).


The activity of the comet has been strongly asymmetric about perihelion in 2004-05 with the highest value of Afr and thus dust production about 100 days before perihelion. However, interpretation is complicated by the fact that much larger values of Afr are obtained with smaller apertures (the graph, below, was obtained with 10”, corresponding to a physical diameter of 5400km in mid-April) than with large ones. In fact, of the some 200 comets that we have studied, none has had an activity curve that was similar to this, although it is interesting that 150 days after perihelion the amount of dust production is now close to that registered pre-perihelion at the same “r” and the trend is now similar too. There is no big rise in dust emission at impact (at T-1d); in fact, it is less than 30% in a 10” aperture and smaller than the amplitude of the variations prior to impact. An interesting feature of the activity of the comet pre-impact were the well-publicised outbursts, for example, those seen by the HST on June 14th and by Charles Morris on June 23rd. Water vapour production from the SWAS instrument on SWIFT shows variations of over a factor of 2 prior to impact superimposed on an apparent downward trend, with the SWAS data showing a possible period of 8-9 days, similar to the separation between the HST and Morris outbursts and consistent with the period of the  “rippling” in the data for Afr shown above. An interesting question is how much of the observed post-impact brightening was due to impact, as both the SWAS and the Spanish data show the impact coming slightly after the bottom of the cycle, leading the SWIFT team to comment prior to impact that the comet was due to start to outburst from this intrinsic cycle approximately 1d before impact. Such a cycle would be due to a small degree of precession in the nucleus’s spin, although it is felt that the nucleus spin must be close to its minimum energy state (i.e. it has only a small amplitude of precession): even a precession by a few degrees could though lead to an active area becoming critically illuminated for a few hours every precession cycle during part of its inbound path and sparking an increase in activity in a similar way to the nucleus of C/1995 O1 (Hale-Bopp) in Autumn 1995.


Impact science


High-resolution imaging is now available of four cometary nuclei: 1P/Halley (top left – Giotto); 9P/Tempel 1 (top right – Deep Impact); 19P/Borrelly (bottom right – Deep Space 1); and 81P/Wild 2 (bottom left – Stardust). Looking at the four nuclei it is hard to credit that they all represent the same kind of object – evolved periodic comets – as the differences from one to another are so strong. Certainly, none of them look like the old Whipple model of the dirty snowball (centre). One of the principal aims of the impact was to study the internal structure of the nucleus. The classic model had a core, possibly even of solid hydrogen, wrapped in layers of ices and covered by a porous crust. More recently though three models have been favoured: a monolithic model in which the nucleus is built of a single block of homo-geneous material; a differentiated nucleus, in which the core may have different structure and composition; and a multicomponent model (that exists in different varieties and which is the favoured model), in which the different components of the nucleus may be only very weakly held together like the evident rubble pile that is the asteroid Itokawa, either by mutual gravitation, or by some binding icy “glue”. A mantle of dust then covers the nucleus, possibly with a thin outer covering of ice. Different models exist that suggest that the mantle covering the underlying pristine material may be anything from 1-10 metres in thickness; thus it was believed that Deep Impact would pass right through the dust mantle into pristine ice. By studying the formation of the crater, its size and depth, it was expected that the structure and tensile strength of the outer layers of the nucleus could be calculated.


Unfortunately, the large dust plume that was produced in the impact completely obscured the formation of the crater. Initially, contradictory information was released but, when all the images had been returned to Earth, it became evident that no information on the crater formation process could be extracted from them. The estimated size of the crater (100-m) is based on the increase in the x-ray luminosity of the comet after impact, given that the x-ray luminosity of a comet is known to be correlated with its dust activity hence, from the increase in x-ray luminosity, we can estimate the amount of dust expelled when the crater formed and thus the volume of the crater. The characteristics of the dust suggest that the impact occurred in a layer of extremely fine, highly dissecated dust like extremely dry talcum powder and that the impact itself was similar to falling into a deep drift of fresh, powdery snow. Observations from Earth have shown that there was an enhancement by a factor of 2 in gas production on impact, but that this lasted only about 15 minutes, while the central coma increased in brightness by a factor of 5. This, and the lack of reaction in the light curve of the comet suggests that the impact did not penetrate through to pristine material, suggesting that either the dust layer is thicker than expected, or that the impacter was stopped at a smaller depth than anticipated.


At 10.2km/s the approach to the nucleus was rapid and high-resolution images of the surface are only available for the last 15 minutes or so. The last image that was received was taken 3.7s before impact from 38km altitude and has a resolution of 1-m. At the time of this talk no processed images from the impacter have been made available. The last image (right) that shows the entire nucleus, taken from 3000km, 300s before impact, is particularly interesting. We can see at least four distinct types of crater, including formations that look like the mouths of geysers and bright, fresh impact craters. The impact point is marked with a small circle. What is particularly interesting is to compare the plateau-like structure marked by an ellipse, with a “meseta” structure on Mars observed by Mars Express that is thought to be due to sublimation of permafrost; the similarities between the two structures are evident.

Despite the rather disappointing visual response of the nucleus on impact, some activity was seen. HST observations show an obvious dust hood and high-resolution imaging from La Palma with the NOT shows an apparent increase in the brightness of dust jets that were visible on the evening before impact (but, once again, care must be taken with interpreting these images as the exposure time was greater in the post-impact images and the filtering used to enhance the jets does not conserve the intensity on processing, in particular, it is not obvious that all the jets should brighten so obviously post-impact rather than just the one from the impact site. The Swift, ultraviolet light curve (left) shows how rapid the post impact rise in brightness was and how quickly it started to decline. A similar pattern is seen in visible and near-IR monitoring, with the largest effects on impact logically seen in the smallest apertures.


Some of the most eagerly awaited results have been those from the mother probe’s infrared spectrometer. At the time of the talk the first results had just been released and analysis was at an extremely early stage, with less than 2% of the spectra received from the probe processed to date. The results though are interesting and cast some light on the nature of the surface. The spectra cover a range from 2-4.5 microns that contains a number of spectral bands of interest. The slit of the spectrograph was centred slightly south of the impact point. The aim was to measure the spectrum of the debris cloud that was expelled.


Two spectra are shown: the lower line is the spectrum 0.1s before impact – this spectrum is essentially featureless. The upper curve is the spectrum taken 0.6s after impact; on it is superimposed a model spectrum (the grey curve). The model supposes that there is a cloud of carbon dioxide and water vapour at 1400K that is seen superimposed on a hot spot on the nucleus that is heated to 800K by the impact. The model gives a good fit to the observed spectrum except between 3.1 and 3.7 microns where a strong emission is seen. This band corresponds to the wavelength of the carbon-hydrogen bond found in organic compounds. The implication of this spectrum is that a significant quantity of water and carbon dioxide was released on impact however, the enormous strength of the band from the C-H bond indicates that huge quantities of organic compounds were present in the surface layers and volatilised by the impact. This has confirmed the suggestions that the material of the dust mantle on the surface of cometary nuclei is rich in organic compounds; it also confirms the Giotto observation that a significant amount of relatively high molecular mass polymers were present in the dust in the coma of 1P/Halley.




The results of the impact were initially disconcerting but, despite the failure to fulfil the prime objective of the mission: to image the formation of the crater and to measure its size, the mission has been a resounding success. The fact that the initial impact was so spectacular made the lack of a strong, sustained reaction from the nucleus disappointing. Based on the best available knowledge of the structure of the nucleus it was expected that the impact would release about as much gas and dust as in as much as a week of normal activity, hence the expectation that the comet would become an easy naked-eye object. It was also expected that the crater formed would expose bare, virgin ice that could make the comet brighter and more active than usual for several returns. What actually happened is that rather than hitting a relatively solid surface the impacter fell into deep, extremely fine and dissecated dust, penetrating a considerable distance but evidently not exposing fresh ice from the sub-surface layer. The two moderate-sized craters close to the impact site are somewhat ill-defined in the images suggesting that the dust drifts at the impact point may have been especially deep compared to other areas of the nucleus and were certainly extremely depleted in volatiles. Interesting though there are some bright albedo features very close to the impact site that are suggestive of ice or frost deposits; these are also seen around the edges of the meseta north of the impact point. This suggests that not all areas of the nucleus are similar to each other in structure. Probably the results are not going to be sufficient to allow major conclusions about the internal structure of the nucleus, which is disappointment, although as analysis proceeds we are going to learn an enormous amount about the nucleus of 9P/Tempel 1 in general.