[OSRS-PROJ] Donald Elliptic Projection for AT&T V&H Coordinates?

[Duncan Agnew]

	My source for the Donald algorithm was indeed the java script
VHTransform.java that is part of the Openmap software; this also included
most of the descriptions (though not the material from John Snyder's book).
Download this from http://openmap.bbn.com/  It also includes the code for
the inverse transform.

My sources for the comparison values were

	http://www.htlt.com/products/vandh/vhnllconverter.htm
and
        http://www.colorado.edu/geography/gcraft/notes/coordsys/coordsys_f.html

(I couldn't include these in the earlier message for technical
reasons--sorry).

I'd note that the telecoms people seem to have difficulty getting agreement
amongst themselves; see
	http://www.atis.org/pub/clc/niif/nrri/100302confcall.doc
(another find by Google)

and these comments from Telecordia:
	(http://www.trainfo.com/products_services/tra/vhpage.html)
   Since their introduction, V&H Coordinates have become widely used throughout
   the telecommunications industry, not only for determining distances between
   points, but also for designating the locations of switches, transmission
   facilities, and other items of telephone plant.  However, ever since the V&H
   Coordinate system was developed, the underlying mathematical principles have
   been shrouded in mystery.  Little information has been made available
   to the general public and many of the lines of code in programs that are used
   to perform the conversions represent the result of numerous complex
   calculations that have never been publicly identified.

and they go on to say that their document will describe
                  
      C.the steps required to convert "Basic" V&H Coordinates to "Final" V&H
                  Coordinates

which suggests an additional layer, perhaps not available in the algorithm
from Openmap.

	Discrepancies at the 0.002 level in V&H correspond to about a meter:
large for a grid, but perhaps not so bad for something used largely to set
rates. The way in which the code is set up (direction cosines and lots of
precomputed constants) make me suspect that this is the original 1957
implementation, probably for an electromechanical desk calculator--so
who knows how much rounding really went on?

Duncan Agnew

----------------------------------------
PROJ.4 Discussion List
See http://www.remotesensing.org/proj for subscription, unsubscription
and other information.