Re: Geodetic precision

From: Chris Olsson (olsson@globalnet.co.uk)
Date: Mon Aug 20 2001 - 17:05:42 PDT

  • Next message: Dennis Jones: "Shuttle and ISS"

    Johnathan Wojok said:
    >If the difference in the various geoids is < 50m, and if the error range
    >in TLE's is already 500m, then I think the errors in the various geoids
    >can basically be ignored.
    
    Leaving aside the semantic issue of whether the differences between national
    datums constitutes an "error", I think it is worth noting that such differences
    tend to be more than 50 metres rather than less.  WGS84 itself is more than a
    hundred metres of Longitude from the Prime Meridian at Greenwich and many
    national datums are also more than 50 metres either from the old Prime Meridian
    system or from WGS84.
    
    Note too that these differences can become cumulative when two or more datums
    are involved in measurement of the same celestial object from different places
    on Earth.
    
    As for the error range of the TLEs being 500 metres, I'm surprised if it is
    true that precise elsets are no better than half a kilometer.  Perhaps Dr TS
    Kelso and others are wasting their time when they produce excellent
    computational software which outputs to an arc-second of declination and to a
    metre of orbital height?  Will you tell, them?
    
    
    Are you suggesting that SpaceCom somehow deliberately degrades the orbital
    parameters of satellites such as NavStar and Transit to no better than 500
    metres?  If so, it seems a rather pointless degradation as it is so easily
    corrected by the user community.  The IGEX programme for example, tracks
    Glonass and NavStar (and occasionally a few other) satellites to an accuracy of
    well under five metres and cumulatively to sub-metric precision, albeit in
    relation to the rather wacky PZ-90 datum.
    
    
    >Clock tolerance level:   10 to 30 seconds<
    
    Take a look at the SabObs on this list and you will see observation times
    listed to a precision of a hundredth of a second and declinations listed to an
    arc-second or even to a hundredth of an arc-minute.  Note that there are many
    different time datums, such as terrestrial Time, GPS time, GMT, UT, UT1, and
    UTC, but by common consent, most people have either adopted UTC or some
    specifically declared variant, such as EDT, which is easily understood by other
    observers.
    
    As for datum differences (or "errors", if you insist) being less than 50
    metres:  you are simply wrong.
    
    The local datum co-ords (AS58 on the International spheroid) of the USAF Space
    Command tracking station on Ascension Island are known to great precision. 
    That position is 68 metres from the same location when expressed on WGS84. 
    That is clearly more than 50 metres, but there are much greater differences
    elsewhere.
    
    For example; the USAF GPS tracking station on Diego Garcia in the local datum
    (ISTS072 on the International spheroid) is also known to great accuracy.  That
    position is more than 413 metres from the WGS84 figure for the same location.
    
    There are even greater differences elsewhere in the world, if you care to look
    for them.  That is why it is important to state the geodetic basis of
    apparently precise co-ordinates.
    
    
    >the various geoids can basically be ignored.
    
    I think you mean datums and spheroids, rather than geoids.
    
    Yes, you can ignore the ellipticity of the Earth, but you cannot ignore its
    three dimension-ness.  Any position, whether a location on board Planet Earth
    or that of an orbiting satellite, is a 3-D entity and therefore needs to be
    described in three dimensional terms.  
    
    You could, I suppose, replace the ellipse of rotation with a simple sphere, but
    you would still have to declare a dimension for that sphere because we still
    live in a three (at least!) dimensional universe.  You could make the radius of
    your simple sphere equivalent to the radius of a sphere which has approximately
    the same volume as the WGS84 spheroid and round off your chosen radius to 6,371
    Kilometres, for example.  
    
    What would you actually achieve by such a dumbing down of real-world geodesy?
    Nothing, I would suggest.  
    
    All you would do is add yet another "standard" to be converted to the figure of
    the Earth which other people choose to use.
    
    The Cospar standard of precision for observational location happens to be four
    places of decimal degrees.  I have no opinion whether that should be improved
    or not. I would prefer that standards are not intentionally and unnecessarily
    degraded, that's all.  
    
    All I have suggested is that we should have some common Lingua Franca when
    expressing that position.  WGS84 is one widely known standard, but there are
    others which have equal legitimacy.  Simply stating the geodetic basis of a
    stated position is enough.
    
    I can express my height as either "Five Eleven" or as "One Eighty".  Each is a
    reasonable description of my height which is approximately 5' 11" or 1.8m.  To
    make sense of my declared measurement, I need only label it with the
    measurement basis of my declared ordinate.  Similarly, if I declare a
    temperature to be 123°, I really ought to declare whether that is °K, °C, or
    °F.  
    
    There is an old joke amongst air pilots that a suitable response to an air
    trafic controller who asks "What is your height and position?" is:- "Five feet
    eleven and sitting in the front".  The joke has a serious point.  There needs
    to be some common ground between the writer and the reader of expressed
    co-ordinates, otherwise disorder displaces order.
    
    If I give a Lat/Long to an apparent precision of ±8m (eg to four places of
    decimal degrees of Lat/Long), then I really ought to declare the basis of that
    measurement if I have that information available to me.
    
    If I say that something happened sometime last Thursday, when I really don't
    know when on last Thursday it actually happened, I would might be misleading
    people if I declared that the occurrence was at 01:23:45.67.
    
    Similarly, if a geodetic position is expressed to an apparent precision of a
    tenth of an arc-second of Lat/Long, as I've seen legitimately shown on
    SeeSat-L, then the geodetic basis of those co-ordinates really ought to be
    mentioned too.
    
    I should immediately say that in four out of four cases of SeeSat-L posters
    whose co-ords I have looked into closely, all have used very well-sourced data
    and all have used perfectly justifiable levels of precision in their
    expression, despite having come from three different countries with three
    completely different national mapping datums and spheroids.  Two of them have
    since posted the geodetic basis of their co-ords and a third has privately
    confirmed the suspected national datum basis of his co-ords.  Several others
    have written privately to express a positive interest in the subject matter,
    only one has been unpleasantly negative.
    
    Yes, there are those who will exhort us to dumb down our co-ords to two places
    of decimal degrees (or worse) and there might even be some who feel that mere
    2-D cords are a suitable expression of position in their flat-Earth world.  
    
    I suggest that those who care about precision; and those who understand the
    difference between accuracy and precision: give a thought to what those
    concepts actually mean and present suitable labels upon their stated co-ords. 
    Since I have broached the subject on the List. I have noticed that several
    List-members have admirably done so and I have also received a remarkable
    number of non-hostile private communications from others who are also
    interested in making good use of the data which they have at their disposal.
    
    
    I think it was Albert Einstein who once said that:
    
    The Truth ought to be expressed as simply as possible, but never more so.
    
    
    
    
    Cheers,       Chris Olsson
    
    -----------------------------------------------------------------
    Unsubscribe from SeeSat-L by sending a message with 'unsubscribe'
    in the SUBJECT to SeeSat-L-request@lists.satellite.eu.org
    http://www2.satellite.eu.org/seesat/seesatindex.html
    



    This archive was generated by hypermail 2b29 : Mon Aug 20 2001 - 17:12:07 PDT