re: Elset refresh rate
Walter Nissen (dk058@cleveland.freenet.edu)
Fri, 26 Jun 1998 10:31:30 -0400 (EDT)
Rob Matson writes, in a very insightful message:
> but these represent a small fraction of
> the ~8200 RSOs for which regular TLEs are maintained.
I don't know what RSO means, but I would guess this is the entire catalog
from US SPACECOM.
> Just a quick response to Walter's remark: "Many elsets could be updated
> every month or two, or more, with no loss of useful information." That
> all depends on what you're trying to do.
Rob and I are both oversimplifying some to keep it simple. I suspect our
opinions on this subject are very, very close.
As a technical matter, I assert that my statement is correct without any
"depends". Intelligent distribution could vastly reduce the amount of
data transmitted. But Rob is correct that whether you need frequent
updates does indeed "depend". I would join him, saying, some important
elsets need frequent, or even, very frequent updating.
A few people think I make no attempt to shorten my messages, but that's
not true. When I posted my message mentioning the huge waste of bandwidth
in updating elsets, I deliberately left out a dismissive clause saying
that there was something to be said to the contrary. For this
forbearance, we were all rewarded by Rob's fine message detailing a number
of justifications for frequent updates of elsets.
> If you are just trying to acquire a satellite visually, I agree.
> As long as you're willing to start looking a few minutes early and
> bracket the predicted time of the pass on either side, you'll probably
> see it.
Dan Laszlo writes on page 72 of the July S&T that "...predictions may
remain good for weeks". I claim that this is very, very true; that it is
an understatement. And that Rob also is unduly pessimistic about how
large the error is likely to be.
I would imagine that Rob and I are in close agreement on the usefulness of
frequent updates, because one doesn't want to spend a lot of time looking
at blank sky (from light polluted areas, anyway), e.g., when pursuing the
objects in lower orbits and powered objects. This includes the objects
people care about most, like Mir and the shuttles. The shuttles are a
special case because most shuttle elsets contain bad data, particulary bad
drag factors, which make them unsuitable for long term use, tho I have had
some success in simply substituting somewhat reasonable drag factors
bodily into the elset.
But not all elsets need to be updated frequently; many remain good for
weeks and months. These are the high, unpowered objects. Low objects and
powered objects are a different story.
I ran an unscientific, but very suggestive, experiment.
For a 100 minute interval beginning with the Sun 11 or 12 degrees down on
the evening of 19980622, I ran the first 1000 elsets (if I'd chosen the
last 1000, my results might be less supportive of my position) from
Molczan's files for 19980522 and 19980619 into QuickSat, using my own
slightly peculiar QUICKSAT.MAG, asking for all illuminated passes visible
from my location with an altitude above 15 degrees and a magnitude
brighter than 8. I eliminated a couple of dups, choosing to retain the
higher pass in each case, and compared the times generated by QS with
elsets 3 days old or more to those 31 days old or more. There were a
total of 235 passes by 235 objects. 203 culminated (mostly, a few were
not illuminated at culmination) within 15 seconds of the time predicted by
the earlier elsets. That's 86%. 32 culminated with more than a 15 second
difference. That's 14%. I reiterate that most people obtain vastly more
elsets than they really need to. Intelligent distribution could vastly
reduce the amount of data transmitted. Mostly because of the other
reasons I cited in my earlier message, but for this reason also. Some of
these elsets are good for months on end.
Here is the number of objects for given differences in seconds:
difference # of
in seconds objects
0 31
1 40
2 28
3 18
4 14
5 18
6 14
7 7
8 9
9 3
10 2
11 4
12 5
13 3
14 3
15 4
16 2
17 2
18 2
19 1
21 2
22 1
23 1
24 1
26 1
29 2
31 1
32 1
38 2
43 1
54 1
58 1
63 1
78 1
89 1
98 1
100 1
109 1
143 1
145 1
262 1
1183 1
----
235
The 262 second difference is for Meteor 2-18. I think this might be due
to a bad value in one of the elsets, rather than any ordinary uncertainty.
Bjoern and I, and others, have noted before that an occasional elset will
contain bad data.
QuickSat's "TIM" parameter (output value) is strongly, but not perfectly,
correlated with the long-term accuracy of the predictions.
Cheers.
Walter Nissen dk058@cleveland.freenet.edu
-81.8637, 41.3735, 256m elevation
---
Wanna see more sats? QuickSat, with Molczan input, is your best nightly menu.
Need help using QuickSat? Ask Mike McCants or here on SeeSat-L.
(My apologies to the 2 who've said they'd prefer the formal "want to").