--- In mpml@yahoogroups.com, "Christian" <chrisk0304@...> wrote:
>
> Horizons has a new orbit solution which includes the three precovery
observations. The
> new solution gives a nominal Close Approach distance between Mars
and 2007 WD5
> on January 30.49044 of 0.000140 AU and a probability of a Mars
impact of 3.949% (or 1 :
> 25):

The reason for the discrepancy between these results and those I gave
in my previous post is that (perhaps arbitrarily) I discarded the
following two observations having residuals substantially larger than
the average, while JPL did not.
2007 12 04.22377
2007 12 05.2261
If I include the above observations in my MonteCarlo runs, I get
nearly the same results as JPL.

I am not expert in probability and statistics, but I would say that a
probability estimate which is so crucially dependent on two
observations lying among others in the middle of the observational arc
is in itself very uncertain. In other words, it seems to me that even
the error estimate is affected by a large error ... :-)

Regards
Aldo Vitagliano