technical details regarding MCC ISS TLEs

From: Thomas Fly (
Date: Tue Aug 24 2004 - 17:40:05 EDT

  • Next message: Russell Eberst: "AUG24.OBS"

    There are hours of fun here, for anyone inclined to experiment with such stuff....
    I'm having a hard time deciding when to generate new "ISS Transit" reports, particularly now that I'm getting the
    "orbital elements files" that I've been posting under the topic "Updated ISS info from MCC;" updates to the SVPOST page
    (one of which occurred in the past 20 minutes) don't always closely follow (or follow at all) the updates I'm getting to
    those orbital elements files.
    -----Original Message-----
    Subject: Re: Iss Day 236 Orbital Elements File
    Hi Jon,
    Someone with the National Optical Astronomy Observatory told me that the TLEs posted on the SVPOST page are derived from
    a precise numerical integration technique.  Is this the case, or would I get essentially the same results by using the
    TLE in your attached iss.txt file, for a prediction several days to a week in advance?
    In other words, for predictions a week ahead or so, do the TLEs on the SVPOST page represent an improvement on the SGP4
    analytical model?
    Our web page is updated nominally on mondays and thursdays and the epoch vector for these updates is the same orbit
    determination vector that I send out on those days. In other words Coasting Arc 1 M50 ( Mean or Besselian year 1950 )
    anchor vector will be the same as the M50 vector in my orbital elements file. My software is used to transform the
    osculating inertial cartesian M50 vector to TEME which is used to generate the Mean-Mean elements and the TLE. TLE's are
    anchored just north of the next ascending node relative to vector epoch UTC time.
    Now, what we provide at the web site is Coasting Arc orbital elements files at nominal 1 day intervals ( special outputs
    include translation maneuver end UTC times and then again at one day intervals afterwards ). These "predicted orbital
    elements coasting arc M50 vectors are generated by predicting the epoch orbit determination vector ( coasting arc 1 )
    forward in time using the following scheme:
    - Equations of motion are numerically integrated using a 7th order self-starting numerical scheme using Cowells
    - Perturbations included are:
     - 7th degree - 7th order recursive gravity model ( Goddard Earth Model Version 10 - GEM10 ) to model the oblate earth
    ( non central body ).
     - sun-moon gravitational perturbations
     - atmospheric drag - 71 Jacchia-Lineberry Atmosphere is used. Marshalls predicted 50% solar-geomagnetic data is used in
    this model to compute the density. A constant vehicle area is used and is optimized by the ISS Trajectory flight
    controllers by basically minimizing downtrack or semimajor axis error between previous successive Orbit Determination
    vectors, along with other methods. ISS current mass is used.
    Again, each of these predicted M50 vectors is converted to osculating inertial cartesian  true equator-mean equinox
    (TEME) to derive the mean-mean element TLE.
    One thing to note regarding these web site TLE's. Note that the BSTAR and NDOT/2 psuedo-drag terms do not change. In
    other words they are generic (.0002 ). This is a good "average" value to use for ISS ( values can be higher or lower
    depending on atmospheric activity, ISS attitude mode, unmodelled perturbations etc... ). Thus these psuedo-drag terms
    are NOT solved for.
    The orbital elements files I send out have optimized drag terms in the TLE's. That is I wrote a routine to optimize
    BSTAR using SGP4 and NDOT/2 using SGP to minimize downtrack position between two specified previous epoch orbit
    determination vectors.
    So it might be kind of a trade off. You have to predict my TLE farther using analytic SGP4 but the drag term is more
    optimized ( especially if ISS is to continue to be in the same attitude mode ). Using the web site tle's you can choose
    one in the future derived by precision numerical theory, however when you predict it you will be using a generic drag
    term. Of course you could replace its drag term with mine or the current ISS drag term from the GSFC OIG, etc...
    You could play some analysis games and compare the previous monday predicted to the current monday and after doing many
    compares see which gives you the most accuracy.
    Hope this helps answer your original questions and let me know of any follow-up questions you may have.
    Subscribe/Unsubscribe info, Frequently Asked Questions, SeeSat-L archive:

    This archive was generated by hypermail 2b29 : Tue Aug 24 2004 - 17:45:09 EDT