[GRASS5] problems with [rsv].proj

[Eric G. Miller]

On Thu, May 30, 2002 at 09:26:50AM -0400, Frank Warmerdam wrote:
> I would be happy to incorporate some sort of support for error estimation
> support in PROJ.4 if practical.  However, in general returning regions in
> which a projection is valid is very difficult.

I'd have to look at the geotrans source again... My recollection is the
range errors are probably only applicable to certain projection types.
And, the error estimation seemed to be sort of hard coded...

> I would also note that PROJ.4 *does* contain datum shifting based on grids,
> (like US nadcon grids, and Canadian NTv1 grids) as well as 3 and 7 parameter
> datum shift approximations.  The 3/7 parameter support was derived from
> Geotrans.  Hopefully this can be properly incorporated into GRASS in 5.1. I
> was a bit weirded out that datum shifting was grafted into GRASS late in the
> pre-release countdown sequence.

Yea, you've put a lot of great stuff into proj4...  There are several
recent changes that should've waited, IMHO (locales, readline, this...).

> I would be interested in including some sort of geoid conversion support in
> PROJ.4 but I have very little experience with it, and am I am not clear on
> the need for it.  Hopefully others with the need can help drive such work.

The geoid height is really only useful iff you have WGS84 data, collect
via GPS maybe, and want it's height component to better approximate the
mean sea level height (and possibly vice versa).  Better GPS software
already has this option in data export, but the cheaper units probably
are exporting ellipsoid heights (which can be pretty far off the
surface, not to mention the higher error in GPS heights...). 

-- 
Eric G. Miller <egm2 at jps.net>