Probably this will be the worst reentry prediction ever made (excluding the ISRO’s prediction), but I’m pretty confident that the following trajectories are fairly accurate (for a 570-day prediction):
The earliest date is what is called “outlier”, because it has nothing to do with the others and should be discarded. The average of the remaining dates is 2019-07-15 and should be taken as a possible reentry date with an high uncertainty (statistically speaking, there is an 87.5% probability that the reentry date is inside the interval 2019-03-06 and 2019-11-24, assuming no knowledge of the distribution of the dates).
The following graph shows the high variability of the reentry dates as a function of the TLE epoch:
see, for example, the big difference for the TLEs near 17337 (day 337 of the year 2017). It means that the TLEs are very inaccurate for this object (I don’t know why). This is confirmed also by the next graph:
that shows the unrealistic variability of the radius vector at perigee obtained from the TLEs (not via simulation).
I must admit that it seems a bad prediction also to me and hence I would be interested in an analytical way to calculate a possible reentry date, starting from a given state (position/velocity vectors). Does anyone know how to try some calculations?
The earliest date is what is called “outlier”, because it has nothing to do with the others and should be discarded. The average of the remaining dates is 2019-07-15 and should be taken as a possible reentry date with an high uncertainty (statistically speaking, there is an 87.5% probability that the reentry date is inside the interval 2019-03-06 and 2019-11-24, assuming no knowledge of the distribution of the dates).
The following graph shows the high variability of the reentry dates as a function of the TLE epoch:
see, for example, the big difference for the TLEs near 17337 (day 337 of the year 2017). It means that the TLEs are very inaccurate for this object (I don’t know why). This is confirmed also by the next graph:
that shows the unrealistic variability of the radius vector at perigee obtained from the TLEs (not via simulation).
I must admit that it seems a bad prediction also to me and hence I would be interested in an analytical way to calculate a possible reentry date, starting from a given state (position/velocity vectors). Does anyone know how to try some calculations?