Re: Geodetic precision

From: Chris Olsson (olsson@globalnet.co.uk)
Date: Thu Aug 16 2001 - 16:59:49 PDT

  • Next message: Ted Molczan: "USA 159 r observed by Paul Gabriel"

    Allen Thomson asked:
    >- Which datum is preferred today and should be used when possible?
    >(I'd guess WGS84.)
    
    
    WGS84 is becoming the de facto standard, but that does not at all mean that
    other legacy datasets should be discarded or regarded as inferior.
    
    The most important thing is to attach a spheroid and/or datum to expressed
    co-ordinates.  As I showed in the example of the Greeenwich Prime Meridian
    case, there can  be almost any number of correct co-ordinates for a given
    location, but one can only make sense of a Lat/Long doublet if one knows which
    spheroid and datum that position is referred to.
    
    Several list members, including a couple in my original home town of Edinburgh,
    Scotland, are giving perfectly respectable co-ordinates which are very
    accurate, but they are omitting the vital fact that those co-ordinates refer to
    Britain's national mapping agency datum, OSGB36.  Their stated co-ordinates are
    not wrong in any way.  They are merely capable of being misinterpreted by
    someone who is unfamiliar with their actual basis and might erroneously presume
    that the widely adopted standard of WGS84 (or perhaps the European standard
    ED50) is the spheroid/datum system to which the co-ords refer.
    
    An analogy might be the price of corn.  I could say that the current market
    price of wheat is 12.345, but that would be uninformative because it does not
    state the currency or quantity framework to which the price refers.  Similarly,
    I could state that the current tumble rate of a particular satellite is 98.765,
    but unless you know that my unstated units of measure are MilliScroates per
    MegaFarkle
    and unless you have a good idea as to the spatial dimension(s) of the Farkle
    and some calculable hold on the referred Scroate, you could not make much sense
    of the data.
    
    Latitudes, and even more so Longitudes, are necessarily referred to artificial
    datums upon universally declared spheroids, which we must specify when
    expressing a form of positional measure.  
    
    Latitude does have some kind of geophysical basis, albeit rather variable and
    squidgy, but Longitude is wholly arbitrary, so it is essential that its basis
    datum be attached, either by convention or explicitly, to any position which is
    expressed by relation to Lat/Long statements of position.
    
    A USAF targeteer, programming a cruise missile to burst through the bathroom
    window of some real or imagined enemy in a distant land, needs to ensure that
    his input co-ordinates are based upon the same 3-D geodetic datum that the
    missile's nav system will use. It is no good for him to use target
    co-ordinates,
    however accurate, if those co-ords are based on the Ali Wali Spheroid and the
    DhabDhab Datum if his missile is expecting its input co-ords in WGS84.  Instead
    of zapping the intended dictator while he is trimming his moustache in the
    morning, the missile will fly off and zap some poor sucker who happens to be at
    exactly the same Lat/Long, perhaps some hundreds of metres away, on a different
    geodetic system.
    
    
    >- If the datum is unknown, what level of accuracy should be reported and
    >used?
    
    That's a good question -- and it's one which is actually rather general in
    Science.
    
    In the context of high-quality observations of satellites or other observable
    astronomical phenomena, I would say that the expressed level of precision
    should approximate to the known or suspected accuracy of the co-ordinates which
    are being described.
    
    On early marine navigational charts, it was sufficient to place a vague marker
    at the outer extremity of the chart and annotate it with a warning: "Here be
    Dragons".  The published position was vague so the warning need not be plotted
    accurately.  The stated position was fit for purpose and did not give a false
    impression of accuracy or mislead users into associating the location with an
    inappropriate Lat/Long datum.
    
    Nowadays, it is usually possible to obtain a fairly accurate position for a
    domestic location, usually in terms of the national mapping agency's preferred
    spheroid and datum.  Another way to obtain co-ordinates of one's observation
    location is to borrow/buy a small and cheap retail GPS unit and take an average
    of several hours of readings.  In either case, the important thing is to
    annotate the spheroid and datum to which those derived co-ords refer.
    
    For rough and ready positions, such as if one is driving across a remote rural
    landscape and suddenly makes an impromptu observation of an unexpected
    phenomenon like a satellite re-entry, then I would suggest that the posted
    precision of the stated position should roughly approximate to the probable
    quality of the position itself.
    
    Examples:-
    
    Out in the boondocks, an hour's drive West of Wogga-Wogga:
    is an apparent precision of about 50 miles
    
    56.1234N   78.4321W:
    is an apparent precision of about 11 metres in Latitude and rather less in
    Longitude.
    
    56 07' 24.24"N   78 25' 55.56"W:
    is an apparent precision of about a foot in horizontal measure.
    
    
    The apparent precision is actually a quite different issue from the matter of
    stating the geodetic framework upon which Lat/Longs are stated, except in the
    crudest of positional statements.
    
    The most important thing is to keep the stated observed position, however
    accurate or otherwise, associated with an identifiable geodetic reference.
    
    Otherwise, we might waste good quality observations such as those which are
    accurately made in profusion and are often published here on SeeSat-L to an
    apparent accuracy of a hundredth of a second of time and a hundredth of an
    arc-minute of Declination.
    
    
    
    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 : Thu Aug 16 2001 - 17:04:28 PDT