Understanding Risk Pages
By Jon Giorgini, Senior Engineer,
Posted: 30 July 2002
When considering an asteroid listed on a risk page, the first thing to consider is the length of time over which the observations have been made.
Two or three weeks of optical data is not enough to conclusively identify an impact years in the future. Possible hazards can be flagged, but these are actually due to the lack of orbit knowledge; the asteroid could be so many places the Earth can't help but be in some of them. These listings are possible impacts, not predicted impacts.
To predict an asteroid's orbit reliably you need radar measurements or at least optical observations spanning one (preferably two) orbit periods of the asteroid. For example, for an asteroid with an orbit period of 2.29 years, that means it should be possible to usefully predict its orbit in a positive way (where it WILL be) for a few decades after two to four more years of tracking it.
By contrast, eliminating an entry on the risk page is a negative prediction; a prediction of where it will NOT be. This can be done more quickly as new measurements come in, since large initial uncertainties shrink rapidly.
A risk-page listing is not a prediction of impact, but a statement that one is possible, primarily because it is not known for sure where the asteroid will be. Of course, lots of things are possible, most of which will not happen. It's possible I will be on top of Mount Everest 60 days from now at 12:01am but I am not predicting it. And there will come a point when my being on Mount Everest at some instant can be positively excluded ("that's impossible!").
The purpose behind the JPL Current Risks Web page is to communicate possibilities to other astronomers so they know which objects require more observations. A listing is not a declaration that an object is predicted to impact, or even come close to the Earth at that time, only that the possibility has not been ruled out.
As far as the public is concerned, it just isn't worth getting worried about an object with a couple weeks of optical data showing a possible Earth encounter years from now. Sometimes, it can't even be said for certain what side of the Sun such an object will be on at the time of the listed possible impact. A few days later, additional measurements will shrink the orbit uncertainty region by a relatively large amount and the Earth will fall out of the risk zone.
When solving for an asteroid's orbit, the result isn't a single position at some time. Instead, it is a volume of space sort of like an egg where the asteroid COULD be with some probability. The further you get from the center of the region, the lower the probability that's where the asteroid really is.
The reason there is a probability region for an asteroid and not just a single answer is because the orbit solution is based on measurements that have errors due to the Earth's atmosphere and other factors. This limits how well the asteroid's position can be measured and combines with all the other measurement errors to create the original uncertainty space. This is especially true with optical data where atmospheric twinkling blurs out the positions.
A similar statement from the author first appeared in the 26 July edition of Cambridge Conference Correspondence in regards to the uproar over reporting about 2002 NT7. A/CC asked for an advisory along the same line to have on hand to refer readers to as we report on future potential hazards.
Jon Giorgini led the team that used optical data spanning half a century plus radar observations for asteroid 29075 1950 DA that together "reveals a 20-minute interval in March 2880 when there could be a nonnegligible probability of the one-kilometer object colliding with Earth." This resulted in the first ever positive Palermo Scale rating, and makes this object the most dangerous known, but there are 35 generations of scientists ahead to consider the problem and to work out a fix, if one is needed.