[OSRS-PROJ] Pseudo X/Y values, forward and reverse

William Hersman Willy.Hersman at conocophillips.com
Fri Jan 31 14:12:08 PST 2003


Yes, let me explain a bit.  I am not a geodicist, so bear with me, I'm a
computer programmer.  We have a situation here where we need to do regional
analysis that expands across a major portion of Alaska.  There are seismic
lines which are very long, extending for hundreds of miles.  The analysis
will involve seismic data from the Canadian border well out to the Chukchi
Sea.  The software used for this analysis requires that one, and only one
projection be used for the entire project.  All of the data must be loaded
in that one projection.  It has been traditional in our work to use either
UTM or Alaska State Plane as a preferred projection.  Yes, one could argue
that there are better projections to use, I would not diagree, but it has
been decided for better or worse to go with Stateplane.

This requires that all lat/long values which are currently carried with the
data, be converted to Alaska State Plane, Zone 4.  It came to the attention
of myself and other programmers that not all software agrees on the
conversion, especially as one gets further and further away from the ASP 4
Central Meridian.  We investigated further and found that nearly none of
them, PROJ4, GCTP, ARCINFO, BLUE MARBLE, etc, etc. are able to "close the
loop," that is, given a lat/long can you calcluate X/Y and recalculate the
same lat/long.  This bothered us. Ideally one would like to be able to save
results of analysis in lat/long so that it can be used in some other
projection at some future date.

If there is a way to do as the users wish (yes, I agree with Cliff, be
careful what you wish for) then I will make a best stab at it.  If not,
I'll report that this is all foolishness, pick a better projection, pick a
smaller area, or live with the inherent weakness of standard software, or
forget doing the project.  I appreciate all your help.  I know more now
than a few days ago about the algorithm aspect, it's not just a "rounding
error" issue.

William Hersman
ConocoPhillips, Alaska



|-------->
|        |
|-------->
  >--------------------------------------------------------------------------------------------------------------------------------------------------|
  |          Duncan Agnew <agnew at bilby.ucsd.edu>                                                                                                     |
  |          Sent by: owner-osrs-proj at remotesensing.org                                                                                              |
  |                                                                                                                                                  |
  |                                                                                                                                                  |
  |          01/31/2003 12:22 PM                                                                                                                     |
  |          Any replies will be addressed to: osrs-proj                                                                                             |
  |                                                                                                                                                  |
  |           To:   osrs-proj at remotesensing.org                                                                                                      |
  |           cc:   agnew at bilby.ucsd.edu                                                                                                             |
  |           Subject:    Re: [OSRS-PROJ] Pseudo X/Y values, forward and reverse                                                                     |
  >--------------------------------------------------------------------------------------------------------------------------------------------------|





For the interested onlookers, perhaps the original poster (Mr. Hersman)
could explain why one would want to get (eg) Alaska SPCS Zone 4 coordinates
when in Zone 10? Since the grid will be very distorted, it's not as though
you can use the X&Y to get distances or direction--if you want to get
these,
why not do what is usual over large distances: use lat and long, and solve
for the geodesic?

As far as the software goes, it should probably function so that the direct
and inverse do in fact invert each other to mm precision, to avoid
confusion
--but this is not the same as their being "right" for these large distances
(that is, agreeing with an exact expression for the TM).  Does being right
matter, and if so, why? To forestall the obvious answer (which is, to agree
with other, exact, computations), I would ask, who makes such computations?
Given the rarity of TM-5-241-10, not many people, it would appear.

BTW, the Snyder's discussion of the TM equations in USGS PP 1395 clearly
states their limitations and error bounds, with references to where the
more exact formulae can be found (namely, TM-5-241-10).

Duncan Agnew
dagnew at ucsd.edu

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




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



More information about the Proj mailing list