The error ellipse doesn't necessarily have to shrink onto one fixed position.

Very interestingly, 2013 TV135 impact probability still increasing, in spite of the small number of new added valid observations (only 7 in last 24 hours):
(last update on Oct,30 - 20 UT)
At this moment, however, only NEODyS impact data are updated while JPL didn't refresh impact data after Oct,28; even stranger, JPL orbital data shows a significant uncertain increase in last day! Do someone knows possible explanations of these facts?
PS: asteroid is now 32 million km (20 million miles) away and surpassed visual magnitude 18; I suspect we are definitively loosing it in very few days and, unfortunately, no close approaches are forecast until 2032...

Edit: new JPL risk estimate is now published (plot updated)

I'd think, that by the increasing distance the determination of the absolute position will fluctuate more and more, and that way add larger terms to the mean square error sum faster than increasing sample number can divide them away.

The orbit looks a bit like close to a 19:5 resonance with Earth.

Gerald, to be honest only JPL data show such worsening in orbital data uncertain. Moreover, the new orbital parameters and impact table just released uses a different set of observations compared to NEODyS; indeed, for the first time, number of used observations now diverges in the two sources (459 vs 463), with also very different fit RMS values (0.36 vs 0.46 )...
It seems that JPL operated a stronger outliers removal in last assessment and probably this increased also the difference in estimated risk: as showed below, now the ratio between the two impact probabilities increased by 50% if compared to previous days!
Updated on Oct,31 - 18.30 UT

New record in the probability and Palermo scale values from NEODyS!
Updated on Oct,31 - 18.30 UT
now P = 1 : 4300 and PS = -0.58 (26% of background) with 485 observation...
Added: now also JPL probability is rising and is close to NEODyS (1:6250) - look also to the "ratio plot" in the previous post

Looks like it's going to a DA14 style close shave, at least. With this size (14 times bigger or so), it's gonna be a lot more spectacular in the radar views?

Not necessarily, nominal distance for 2032 close approach remain high (0.105au from NEODyS)...

Yeah, I was assuming a pass of the same distance, though the size has to count for something.
Let's wait for more data to come in...

I decided to change plots layout in order to make them more readable (I hope so):
Last update : Nov,02 - 07 UT
(added NEODyS orbit uncertains - probability on Oct,19 arisies from this source)

In the next days I will update these plots without posting new images and eventually adding only notes here below.

I calculate 429 orbits (with Find Orbit - Monte Carlo method - noise 0.3) from 501 observations (arc 24 days) to Epoch 1 Nov. 2013
and then integrate all orbits to Epoch 26 Aug. 2032. That integrated MPCORB I use in Celestia soft (mpcorb2ssc can make ssc file).

Here is screen from Celestia and 429 virtual 2013 TV135 asteroids at time of potential impact.
The most of this asteroids are some distance from Earth, but some are very close.



Here is SSC file to Celestia
 mpc.zip ( 16.54K ) Number of downloads: 242

Very impressive Adam! I would like to know how many solutions falls within a certain distance, let's say 0.1 au from Earth...
Is the simulation software (and observations) freely available?

Observations are in minorplanetcenter
http://www.minorplanetcenter.net/db_search...t_id=2013+TV135



Please download the Celestia
http://www.shatters.net/celestia/download.html

and my zip (mpc.zip), extract file to folder extras then run Celestia.


The date of potential impact is the date when Earth cross paths of one of this virtual objects. At 26 Aug 2032 distance of 2013 TV135 (nominal object/orbit) from Earth will be about 0.3 AU.

Closest approach of 0.3 AU is not close?????

Not exactly, see example on NEODyS site

2013TV135 - IMPACTOR TABLE
http://newton.dm.unipi.it/neodys/index.php...;n=2013%20TV135
Impact date 2032/08/26.359 (but this date is a date when Earth cross one of virtual orbits)

and 2013TV135 - CLOSE APPROACHES
http://newton.dm.unipi.it/neodys/index.php...;n=2013%20TV135
0.1315990 AU (closest) at 2032/08/06.09107

0.3 AU is the distance of ONE virtual object (without noise produced by Monte Carlo method), all other virtual objects may be in different position at orbit path.

Just small difference in position at now become big difference at 2032 year.

Adam, I do not see any 2032 close approach in this table (perhaps content has changed in the last hours)... anyway, thanks for all precious infos from you and Greenish about this. Now I understand why, sometimes, in JPL/NEODyS tables there are "strange" forecasts of multiple close approaches or impacts separated by relatively short time intervals, they correspond tho different virtual impactors...!

Added: meanwhile, probability has been drastically reduced in both sources and now Palermo Scale classification is well below -1:
Update: Nov,03 - 14 UT
Note 1: For some reason, on Nov,02 JPL decided to remove first 4 days from observations; this reduced arc-lenght and rms but increased uncertain in orbital parameters...
Note 2: New JPL estimate (in same day) added new observations, with further shrinkage of P.I. and orbit uncertain...
Note 3: JPL condition code changed from 7 to 6, confirming significant improvement in orbit knowledge!