Why we have duplicate orbits sometimes - markmac99/ukmon-pitools GitHub Wiki

The solver looks for groups of potentially matching events using a time window of a few seconds, and whose geometry is such that you can draw intersecting planes through the meteor and each camera.

Especially with long-duration events it is possible for several groups of cameras to meet the criterion separately. For example cameras in East Anglia might pick up the meteor first, creating a match. Then after it passes out of sight from East Anglia, other cameras in Somerset might pick it up, but from a different point in its trajectory, creating a second match with a slightly different start time.

Furthermore, if there's cloud about, a camera may detect an event in two parts with a gap between them. Thus you might have Searby part 1 and Searby part 2, with a few seconds difference in start times. Part 1 might match with a couple of other cameras which saw it at roughly the same start time. Part 2 might better match with different cameras which had a different view, detecting the event later in its journey. And so again you can get multiple events, but with different start times.

Finally, the solver doesn't just process data once. Each night it reprocesses the previous two days, to pick up any late-arriving data. If you have a two-station solution and then add a third camera it can sometimes change the reference time of the event, leading to a duplicate entry. While mostly, RMS camera data arives quickly sometimes it can miss the batch cutoff, and until the end of 2022 we also recieved UFO-camera data up to four months after it was captured.

At the end of 2022, we ran some analysis to 'size' the problem. A duplicate was defined as starting within one second, with start/end locations within two arcminutes, and involving at least two cameras which were the same. This analysis showed that out of 80,000 matches, we had around 400 duplicates, ie about 0.5% of the total.

If the event is a fireball, we'll normally manually reduce the data and eliminate the duplicate but for ordinary meteors there's no particular advantage to doing this.