[Proj] VERTCON datum shifts scaled 1000x?

[Andy Wilson]

Hi.

I'm using the trunk version of proj to do calculate vertical datum
shifts between NGVD 29 and NAVD 88 using the vertcon files here:
http://download.osgeo.org/proj/vdatum/vertcon/

This works:

> ./src/cs2cs +proj=latlong +datum=WGS84 +to +proj=latlong +datum=WGS84 +geoidgrids=vertconc.gtx,vertcone.gtx,vertconw.gtx
-102 38
102dW	38dN -416.785
-97 29
97dW	29dN 37.597
-109 37
109dW	37dN -884.167


But the output seems to be off by a factor of 1000; these output z
values are actually millimeters. The same thing happens with feet:


> ./src/cs2cs +proj=latlong +datum=WGS84 +to +proj=latlong +datum=WGS84 +geoidgrids=vertconc.gtx,vertcone.gtx,vertconw.gtx +vunits=us-ft
-102 38
102dW	38dN -1367.401
-97 29
97dW	29dN 123.349
-109 37
109dW	37dN -2900.806


These output z values are in 1/1000 ft (millifeet?). It is trivial to
correct for this by scaling these z values down by 1000, but maybe not
the expected behavior.


For comparison, the NGS has this VERTCON site:
http://www.ngs.noaa.gov/TOOLS/Vertcon/vertcon.html
and a web form for finding datum shifts at specific lat/long:
http://www.ngs.noaa.gov/cgi-bin/VERTCON/vert_con.prl


btw, thanks for adding vertical datums to proj.4  - this is easily
most convenient way to do this sort of thing.

-andy