[Spice_discussion] Re: question on coordinates of date (Bryan Butler)

[Jon Giorgini]

#1) The 0.001" J2000 astrometric difference is probably due to using DE421.
Horizons still uses DE405, and will until a suitable new long-term planetary 
is completed (forecast: summer 2009). 

DE421 is something of an interim improvement to the modern era solution,
primarily for spacecraft missions. The coming improvement will allow extension
to the more distant eras Horizons has to support for historical research.

#2) The difference in the TOD output from SPICE/SOFA is very likely Horizons 
relativistic stellar aberration. Horizons "apparent RA/DEC" is not just wrt 
the TOD coordinate system, but includes other effects on the perceived position
relative to the coordinate system.

You might not get exact agreement between SPICE & Horizons due to other issues 
like light-bending+, but requesting stellar aberration in your "SPKEZR" call
should remove most of it.

-------------------------------------------------------------------------------
Jon Giorgini                       |  Navigation & Mission Design Section
Senior Analyst                     |  Solar System Dynamics Group
Jon.Giorgini at jpl.nasa.gov          |  Jet Propulsion Laboratory
-------------------------------------------------------------------------------

>Date: Tue, 27 Jan 2009 21:01:20 -0700
>From: Bryan Butler <bbutler at nrao.edu>
>Subject: [Spice_discussion] question on coordinates of date
>To: spice_discussion at naif.jpl.nasa.gov
>Cc: Bryan Butler <bbutler at aoc.nrao.edu>
>
>
>hello all,
>
>first off - i would have searched the listserv archives for an answer to 
>this, but it gives me an error every time i try to access them.  so, 
>i'll just see if anybody is still reading these and can help :).
>
>i'm trying to calculate coordinates of date, and running into some problems.
>
>as an example, consider the geocentric position of the moon.  i'll 
>calculate it in J2000.0 and also on 2009Jan30 at 0hUTC:
>
>       CALL FURNSH ('naif0009.tls')
>       CALL FURNSH ('de421.bsp')
>       CALL FURNSH ('tod.tk')
>
>...
>
>       YEAR = 2009
>       MONTH = 1
>       DAY = 30
>       HOUR = 0
>       MINUTE = 0
>       SECOND = 0
>       WRITE (UTCSTR,1) YEAR, MONTH, DAY, HOUR, MINUTE, SECOND
>       CALL STR2ET (UTCSTR, ET)
>
>... as a check get the J2000 coordinates
>
>       CALL SPKEZR ("MOON", ET, "J2000", "LT", "EARTH", STATE, LT)
>       CALL RECRAD (STATE, RANGEMOON, RAMOON, DECMOON)
>
>... then try the true, of date coordinates
>
>       CALL SPKEZR ("MOON", ET, "ECI_TOD", "LT", "EARTH", STATE, LT)
>       CALL RECRAD (STATE, RANGEMOON, RAMOON, DECMOON)
>
>
>the result is:
>
>J2000:   23:23:44.8371 -00:00:36.156
>of-date: 23:24:13.6409 +00:02:28.349
>
>the first comparison i did was with the SOFA routines, calling 
>iau_PNM80() on the STATE returned from the J2000.0 call.  this results 
>in exactly the same 'of date' coordinates as above, so i'm pretty sure 
>i've got the frame file (tod.tk) defined properly.
>
>then i thought i'd compare this with the positions in horizons 
>(http://ssd.jpl.nasa.gov/horizons.cgi), which gives:
>
>J2000:   23:23:44.8370 -00:00:36.157
>of-date: 23:24:12.6659 +00:02:23.032
>
>which is slightly different in J2000.0 (a few milliseconds of arc in 
>right ascension and one millisecond of arc in declination), but quite 
>different in of date (15 seconds of arc in RA, 5 seconds in dec).  this 
>is also true of other dates i try - the errors are small in J2000.0, but 
>quite large in of date coordinates.
>
>so why is it that horizons gives such a different answer than SPICE for 
>of date coordinates?
>
>	-bryan butler
>	 scientist, EVLA computing division head
>	 national radio astronomy observatory