# Re: Incorrect prediction for Iridm 15 Dlt r

Bjoern Gimle (b_gimle@algonet.se)
Thu, 19 Mar 1998 07:03:00 +0100

```Gregory May asked:
>
>If I am correct, my elset is only a few days old.
>
>Can someone help me understand what has caused the error in the prediction?
>
>Iridm 15 Dlt r
>1 24874U 97034F   98071.68636191  .00452083  00000-0  77857-3 0  2614
>2 24874  82.7069 294.2950 0091361 155.6367 204.9969 15.94758062 38325
>
Day 98071.68 is Mar.12 at 16:30 UT, ie age of elset 3.1 days.

Alan Pickup remarked that most objects near decay can be expected to be
early, since normal prediction models do not account for the (rapid)
increase in drag.

Tyler MacKenzie pointed to an excellent way to remove this bias - use
Alan's SatEvo program, which has a more correct model for this phase.

Even without this method, there are simple ways of estimating the error.
The drag term, .00452083 orbits/day/day, is inversely proportional to
the remaining lifetime for a circular orbit - more constant for an
elliptic orbit. This has e=0.0091361, or about 120 km from perigee to
apogee, so drag should increase slower. Even using Alan's latest estimate
increased to .0050 (Tyler has .0062 for Mar.18)

The effect of (average) .0048 over 3.1 days is .0048 * 3.1 * 3.1 orbits
= .045 orbits = 4 minutes (early)

Add on top of that an average uncertainty in elsets, and irregular
variations in the atmosphere, totalling about 25%, you get 1 minute,
very close to the actual error.

Tyler's elset might have been less than a day old, so his error would
have been a few seconds, and very little gained by using SatEvo in this
case. But it helps a lot when elsets are several days old, and during
the last few days.

---------------------------------------------------------------
-- bjorn@tt-tech.se (office)  b_gimle@algonet.se (home)      --
-- 59.2237N, 18.2286E, 44 m   http://www.algonet.se/~b_gimle --