Hi Alan,

There are a couple of things going on here...

First, MPC is providing MPCORB for a "standard epoch" of 6 May 2002.
Go in any direction from that date, and errors accumulate quadratically.
By the time you're seventeen years from the epoch, errors are horrible.

I've been working on some ways around this, the main one being a
program to numerically integrate MPCORB to any desired epoch, so you
aren't stuck with just MPC's default choice. (The math/physics end of this
works just fine, but there are user interface issues that I've yet to work
out.) Anyway, I integrated that orbit ahead to 1 Feb 2019 (resulting
orbit given below), and got a distance from Earth of about .2 AU.
I'm surprised your colleague got such a "near miss" result, but perhaps
he's got a numerical integration program of his own?

The .2 AU difference is presumably due to new observations (note that
the "last observed" tag is for Wednesday, 24 July). I don't have the
elements that were created from a shorter arc to get the early collision
possibility. (I fed the data from the original MPEC through Find_Orb and
got a very solid orbit, one that didn't put 2002 NT7 anywhere near the
earth on 1 Feb 2019. Again, almost certainly due to use of a different
set of observations.)

-- Bill

------------ Clip -----------
Designation: 2002 NT7
Absolute magnitude: 16.5
Slope parameter: 0.15
Orbital elements:
Semimajor axis: 1.7379966
Eccentricity: 0.5295917
Inclination of orbit: 42.38775
Argument of perihelion: 300.61031
Long. ascending node: 132.06279
Mean anomaly: 21.72270
Mean daily motion: 0.43016064 degrees/day
Epoch of elements: 1 Feb 2019
Uncertainty level: 7
108 observations made at 1 oppositions
Orbital arc: 15 days
Last observed: 2002 07 24
RMS residual: 0.59
Perturbing objects used in orbit: M-v
Reference: E2002-O38