As everyone now knows, the collision of comet Shoemaker-Levy 9 with Jupiter did not bring about the end of the world (although the first impact occurred just 24 hours before the start of the World Cup final, which Italy lost - so there may have been some validity to claims that it would portend some sort of human catastrophe). But, while failing to live up to the expectations of mad prophets, the event surpassed those of the countless astronomers who prepared to observe it with just about every major astronomical facility in the world. At the JCMT, the 10-day encounter was observed by a large consortium of UK, Canadian, US and French astronomers.

In the atmospheres of the giant planets, spectral lines can arise either due to emission in the stratosphere producing narrow lines (typically 20 MHz wide), or due to absorption in the troposphere against the background thermal emission from the planet's interior, producing strongly pressure broadened absorption lines (typically 10 GHz wide). Our aims were to use the JCMT heterodyne receivers to search for possible stratospheric emission lines, and the University of Lethbridge Fourier Transform Spectrometer (FTS) to look for the much broader tropospheric absorption features. Prior to the run, no near-millimetre spectral features had been observed from Jupiter in its normal condition. We were awarded 11 JCMT shifts, covering the period of bombardment and a couple of days on either side.

Our initial expectations were, frankly, not very high. We were concerned that any anomalous spectral structure from the small impact sites might be at a very low level, and could be swamped by the enormous continuum from Jupiter, which we could not avoid with the relatively crude angular resolution of a mm/submm telescope. Moreover, fluctuations due to tracking errors could lead to varying and non-linear baselines, preventing any lines from being detected. In addition, providence had decreed that Jupiter would only be observable from around 3.30-10.30pm each day, which includes a large chunk of afternoon time when the atmospheric noise is usually at its worst. Finally, we were also afraid that the weather would be bad throughout the run; only in this respect were we completely right.

Line observations:

For line observations, we adopted an observing procedure of beam-switching 60" at 4 Hz, and alternating between the impact site on the southern hemisphere and a corresponding reference point on the northern hemisphere. Not knowing what molecules to look for, we chose CO, HCN and H2S as likely candidates. Observations using receiver B3i in the two days preceding the first collision showed no evidence of emission from either hemisphere.

The first impact (A) occurred at 9.30am on July 16th, and the second (B) at 4.50pm on the same day - during the observing shift. We looked at both sites in the HCN 4-3 line (354 GHz), and saw nothing, even though we caught B as it rotated into view just after the event and tracked on it for over an hour. Undeterred, we set up to observe the site of impact C in the same line. The impact occurred at 9.07pm, and the site became visible about 20 minutes later. We carried out a cycle of integrations on site C and its reference position over 20 minutes, and this time we were rewarded with a clear detection of the HCN 4-3 line with an antenna temperature of about 1 K and a FWHM of 8-10 km/s (see Figure 1). Subsequent observations did not result in a confirmation of the detection, but Jupiter was setting and the system temperature rapidly increasing. So at that stage we were not able to say whether or not the effect was short-lived.

On the following night, July 17th, we succeeded in detecting HCN 4-3 from the 3.5-hour old impact F site. We looked at the position of F in CO 3-2, but did not detect anything to a noise level of 150 mK with TBD km/s resolution. We were looking forward to catching the biggest collision of them all, impact G, due to occur at 9.30pm; but then it started to rain.

On the 18th, we managed to observe G (now 21 hrs old), detecting a very strong HCN 4-3 line (see Figure 2). We also detected the 12-hour old H impact in the same line. So it was clear that the HCN was persisting for at least a time on the order of 1 day, but the rate of decline of the signal from a single impact site was yet to be measured.

For the next few days, the weather was generally miserable as hurricane Emilia and tropical depression Fabio passed by to the south of the Big Island. Nevertheless, we observed when we could, and confirmed HCN 4-3 at various other positions. Impact sites were now piling up to such an extent that several sites were usually contained within our 14" beam, making it difficult to tell which siteswere dominating the emission. We also used receiver A2 to make two measurements of the HCN 3-2 (266 GHz) line on the position of impact R, 48 hrs apart (see Figure 3).

At the time of writing (July 24th - our last night), all the impacts have occurred, and attempts to get follow-up data on previous measurements and monitor variability have been largely prevented by bad weather.

FTS observations:

Although the FTS is ideally suited to the detection of broad tropospheric absorption features, observations proved very difficult due to the unstable atmospheric conditions. The line receivers were given precedence whenever the weather was poor, which was most of the time. On one occasion, at the precise moment when the FTS was fired up and ready to take over from the line receivers, tropical depression Fabio reached the summit and the telescope shut down for the remainder of the shift. As a result of the generally poor conditions, only a few results were obtained.

The observing strategy for the FTS component of the run was to point at the central meridian of the planet, offset towards the south pole to observe the impact sites rotating through the field of view at -44 degrees S latitude. An equal number of spectra were taken in this position and in a corresponding position in the northern hemisphere.

Spectra in the 1100-mm band of UKT14 revealed a repeatable absorption feature in the southern hemisphere of the order of 5% of the continuum background. This feature is centred on coincident lines of HCN and PH3, and at the time of writing we are not able to confirm which molecule is responsible. Measurements in the 850-mm filter may enable us to distinguish between them by revealing (or not) the 4-3 HCN absorption, however at the time of writing these spectra have not yet been analysed.

Spectra in the 750-mm band of UKT14 did not reveal any H2S absorption features. Several transitions occur in this filter. We infer that tropospheric H2S was not generated by the impacts.

What does it mean?

Although it is far too early to come to any definite conclusions, here are some preliminary thoughts.

1) HCN is not normally present at detectable levels in Jupiter's atmosphere (in fact, it has never been detected before). Given the small size of the impact regions, it is clear that it was produced in large quantities by many of the impacts.

2) The narrow line widths (generally less than 10 km/s) imply that the emission originates from the upper stratosphere rather than from deeper levels in the atmosphere.

3) The origin of the HCN is uncertain. It is thought to be unlikely that it is being dredged up from the deep atmosphere. It could be brought in by the comet itself, or created by chemical reactions caused by the impact. It will require careful analysis of these and other observations, and detailed chemical modelling, to answer this question.

4) Our observations show that the HCN persisted over timescales of days following the collisions. The two 3-2 spectra of the fragment R site show that the line width and shape remained unchanged over a 48-hr period, while the intensity decreased by a factor of more than two. This suggests changes in the stratospheric abundance of HCN rather than temperature or pressure of the emitting gas.

Conclusions:

Despite the bad weather, it was a successful campaign. Our JCMT observations of stratospheric HCN include the only submillimetre detections of the SL9-Jupiter collision. We were able to formulate a picture (albeit incomplete) of the decay of the HCN emission in the days following the impacts . It will take some thought and consideration of our results in the context of other observations to sort out the implications for Jovian and cometary chemistry. The detection of tropospheric HCN or PH3, whichever it turns out to be, is certainly unique and will also have significant implications for the chemical models. On the whole we were excited to be part of this event, despite the weather; but having spent the last two weeks at HP, we be feel happy to let somebody else do it next time.

Matt Griffin, Queen Mary Westfield College,

Andre Marten, Observatoire de Meudon,

David Naylor & Greg Tompkins, University of Lethbridge,

Gary Davis, University of Saskatchewan,

Henry Matthews & Wayne Holland, JAC

Byron Han, Institute for Astronomy, University of Hawaii, Honolulu