Clifford,
I'm aware that in a purely mathematical sense, there's no such thing as precision
problems with UTM and most other projections.
But I consider MapInfo's practical implementation of the projections to be an implicit
condition in all discussions on MAPINFO-L. As is ESRI's on ARCINFO/ARCVIEW/ARCGIS
lists. Unless explicitly stated that we're talking theory of course.
And we are discussing aq practical problem here, not theoretical one. My answer to Jac
has to be taken as such - a practical explanation. And you _will_ experience
distortion in MapInfo when working far outside UTM zone defintion limits.
Best regards / Med venlig hilsen
Lars V. Nielsen
Odense, Denmark
----- Original Message -----
From: Clifford J Mugnier <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Cc: <[EMAIL PROTECTED]>; <[EMAIL PROTECTED]>
Sent: Monday, September 23, 2002 2:14 AM
Subject: Re: MI-L precision of a projection
>
> Gentlemen:
>
> The "precision of a projection" is dependant on the algorithm employed by
> the programmer. In my experience, the programmers for most GIS packages
> have not been experienced Geodesists. (A certain lady at ESRI is an
> exception, but even ESRI does not appear to listen to her advice very
> often.)
>
> The Gauss-Krueger Transverse Mercator is the proper projection to use for
> UTM, but the difference in longitude from the central meridian is usually
> considered to be plus or minus three degrees, and that is based on the
> sixth derivative. There is a version of this algorithm that goes beyond
> the standard three degrees. One could go out to TWENTY-FOUR degrees on
> either side of the central meridian. That, however, requires expanding
> the infinite series of the Gauss-Krueger Transverse Mercator to the
> FOURTEENTH DERIVATIVE.
>
> The people programming algorithms at MapInfo apparently do not recognize
> practical applications that don't fit their cookie-cutter idea of what a
> GIS should be ...
>
> If the proper algorithm is employed, there is NO loss of precision for a
> Zone extension equivalent to your current problem. The "distortion" or the
> "loss of precision" is due to MapInfo; not to the Gauss-Krueger Transverse
> Mercator. You are probably working with a GIS package that originally was
> based on a sphere because nobody thought that an ellipsoidal model was
> worth the effort. Same probably goes for double-precision math ... , etc.
>
> Regards,
>
> Prof. Clifford J. Mugnier ([EMAIL PROTECTED])
> Chief of Geodesy
> CENTER FOR GEOINFORMATICS
> Department of Civil Engineering
> LOUISIANA STATE UNIVERSITY
> Baton Rouge, LA 70803
> Voice and Facsimile: (225) 578-8536
> Pager: 1-(888) 365-5180
> ================================
> http://www.ASPRS.org/resources.html
> http://www.ce.LSU.edu/~mugnier/
> ================================
>
>
>
>
> Hi Jacques,
>
> The trouble is caused by your point being way outside the UTM zone 32's
> definition area (longitude +6 - +12). Your coordinate is actually close to
> the middle of the adjacent zone (33), which _will_ cause distortion when
> reprojecting.
>
> If you try and use zone 33 instead, you'll get it to work right, I'm sure.
>
> Best regards / Med venlig hilsen
> Lars V. Nielsen
> Odense, Denmark
> ----- Original Message -----
> From: Jacques Paris <[EMAIL PROTECTED]>
> To: MIL <[EMAIL PROTECTED]>
> Sent: Friday, September 20, 2002 5:58 PM
> Subject: MI-L precision of a projection
>
>
> > I have used a table with the "Universal Transverse Mercator (WGS84) Zone
> 32
> > Northern Hemisphere (i.e. 8,104...) to build lines over region borders. I
> do
> > that by extracting with ObjectNodeX..Y of 2 successive nodes coordinates
> and
> > using the creatline() function. All that is done after setting the
> coordsys
> > to that of the table.
> >
> > I have used the program in many occasions without problem but the above
> > projection makes my program crash. The origin of the crash is that the
> > created lines are not lying exactly over the corresponding "sides" of the
> > regions. I have observed for one node (I guess it is representative of
> the
> > situation) a shift in the X direction only: a region node at
> (872411.6012,
> > 5819181.1726) is converted to a line end at (872411.6502, 5819181.1726).
> > This shift of 4 cm about is enough to make my program crash
> >
> > I seem to remember that someone wrote about a similar case, and that it
> was
> > due to a fault in this projection model during the coordinate conversions
> to
> > read in the values and to insert them back in the objects.
> >
> > Any confirmation of the problem would be greatly appreciated.
> >
> > Jacques Paris
> > e-mail [EMAIL PROTECTED]
> > MapBasic-MapInfo support http://www.paris-pc-gis.com
> >
> >
>
>
