Question: When and why did you start to investigate the detailled mechanism behind the Leonid storms? And how did your collaboration work as you're located half an Earth apart?
Robert McNaught: Around September or October 1998 I was very frustrated by the lack of rigour in various Leonid predictions as many clearly had no theoretical basis. I emailed David with an idea that I thought, if rigorously followed, should give the correct geometry of the Leonid dust encounter. David coded it and ran it for Leonid encounters in the last 200 years. Subsequent detailed checks by myself and David on the results indicated the astonishingly close timing of these results with observed storms and outbursts. All correspondence was done by email, with occasional reference papers mailed to each other. Our collaboration was not accidental, though: We had worked together in Australia for a couple of years, and I knew with David's strong background in orbital dynamics he would take the idea seriously.
Q: In which respects is your Dust Trail model different from other models that were used for predicting Leonid storms?
Robert: The dust trail model defines the geometry of the trail with "finesse" in a simple calculation. It is basically identical with methods pioneered years ago in Russia (and largely ignored anywhere else). Our dust trail density model is a semi-empirical model of the dust trail density profile that is dependent on historical observations. The traditional approach to predicting meteor storms involved examining the correlation between comets' orbital geometry, and the historical dates and times of storms. The best this approach could do was suggest likely storm-years, and estimate peak intensity to within a day, or maybe a few hours.
To improve upon that unsatisfactory situation, it was necessary to look at how the dust particles evolve from comet-ejection to collision with Earth's atmosphere. Dust ejected as a spherical cloud each time the comet is nearest the Sun, becomes a narrow dust trail, elongated along the original comet orbit. The precise position of the dust trail in relation to the Earth can be calculated for each year, a direct hit with the a trail being necessary to produce a storm. David Asher and I have checked Earth's predicted times of approach to these dust trails, and we could back-predict the peak times to within 5 minutes of the observed times, for all years when the maxima were timed (1866, 1867, 1869, 1966 and 1969). All years with ZHR values over 500 fit dust trails, and there are no false positives either.
Q: How well then *did* your model fare in 1999?
Robert: Our timing was late by around 4 minutes (depending of what results you use; visual, radio, radar etc). The maximum rate (ZHR) determined by the International Meteor Organization is around 3500 to 4000 which is about double what I thought the upper limit on our simple model was. Any figure in the ZHR range 200-2000 would have been reasonably consistent with the historical data. However, as I already wrote before the event, the data from 1833 appeared to be underestimated, but I did not want to introduce bias into the calculations by "fudging" these historical results. If the input data is considered satisfactory, one has to accept the output as it stands.
Now that 1999 has confirmed for me the overall theory, I feel it is now reasonable to look in detail at these historical showers and reanalyse them in terms of the dust trail theory (i.e. force a fit in 1833 to the predicted time to derive the ZHR) and to consider the experience of difficulty in counting the Leonids in 1999 as it might affect previous storms (especially 1966). What we saw in 1999 strongly suggests that high rates are likely to be underestimated by a factor of several through the inability to keep count of all the meteors.
Q: What other factors might have contributed to your underestimation of the outburst?
Robert: Apart from the uncertainties in the historical reports, there was one parameter (the radial density profile) that we had assumed to be Gaussian despite the clear indication that this was far from the truth. We had no theoretical basis for using any other specific profile, but will approach a solution to this in a number of ways. In general though, knowledge of the time is the more important in the short term, until we get a better handle on the trail profile. We can locate to the best longitude, and satellite operators now know one important parameter in protecting their satellites.
David Asher: I'm sure there is some complexity that we don't understand yet, but I feel reasonably confident that 1833, 1866, 1867, 1869, 1966 and 1999 could be made quite consistent with sensible adjustments (which we haven't quantified yet) to the peak ZHR values reported in the 19th century.
Robert: This will be difficult to do without bias, but I certainly hope we can do so reasonably. I'm not interested in getting nice results that will make 1999 fit, as the errors in some of the historical data, or intrinsic variability, will necessarily mean that the data will have discrepancies.
Q: What can be said about the next three years (2000 to 2002) regarding the magnitude and times of further Leonids outbursts?
David: All the times and visibility regions we published before the 1999 storm (see our detailled websites at http://www.arm.ac.uk/leonid/dustexpl.html and http://www.atnf.csiro.au/asa_www/info_sheets/leonids.html) should be fine, though the rates we gave may need revising. But even if the ZHR predictions of 15000 to 25000 ZHRs would have to be reduced, it's obvious that they're going to remain at storm level (and higher than the 1999 storm). The ZHR values (with question marks) for 2000 Nov 18? At this stage it's hard to know: They could be just very little outbursts or we could be lucky and get something like 1999 storm level.
Robert: The results from 1999 don't change things much, especially for 2000. But the results from 2000 will be important in deriving a more precise radial density profile, as the Earth passes through the outer parts of two trails: This will place powerful constraints on the shape of the profile. Our predictions for 2000 are generally for low activity (less than 100 ZHR), but we cannot rule out storm activity with no historical data to base the predictions on for those years. Hopefully after the 2000 Leonids we should be in a position to give better ZHR predictions for 2001 and 2002. Start saving those pennies, as these should be much bigger storms than 1999 whatever values we eventually end up with.
Q: So you predict major storms for 2001 and 2002, despite the fact that it will be years after the comet's nodal crossing, when all previous predictions saw no storms whatsoever. Have there ever been storms that late in the past?
Robert: Take the outburst in 1869 as an example: Had we made a direct hit with the core of the trail in 1869, there would have been a storm. But in the last 200 years, we can state unequivocally that the dust trail model predicts NO storms that far from the comet's nodal passage, as a situation like in 2001 and 2002 simply hasn't occurred. So there is no problem here. Don't worry: The success of our model is so good, it is unreasonable to compare our results with other ideas that have demonstarted no predictive power and are not theoretically rigorous. By visualising the geometry of the comet with its "attached" dust trails, the probabilistic nature of encounters can be easily understood.
Interview conducted by Daniel Fischer