[PROJ] Converting NAD83/NADV88 heights to HAE

[Even Rouault]

On jeudi 2 juillet 2020 08:51:00 CEST Greg Troxel wrote:
> Even Rouault <even.rouault at spatialys.com> writes:
> 
> 1) I tried to use projinfo (6.3.2) with this defintiion, thinking that
> asking proj "what do you think this really means" would be helpful:
> 
>   projinfo +proj=utm +zone=10 +datum=NAD83  +units=m +no_defs
> +geoidgrids=us_noaa_g2018u0.tif
> 
> got me
> 
>   Too many parameters: +zone=10
> 
> Is it expected that any string that's valid for cs2cs can be queried
> with projinfo?

For projinfo, you need to put the PROJ string inside double quotes, and add a +type=crs

projinfo "+proj=utm +zone=10 +datum=NAD83 +units=m +geoidgrids=us_noaa_g2018u0.tif 
+type=crs"

> 
> 2) It strikes me that once one moves away from basic 2D datums, the
> legacy syntax is dangerous, and people should be encouraged to use a
> clearer definition.  Do you agree, and if so I wonder what you think
> should be done?

The only advantage of it is its compacity, especially for simple things. Using EPSG codes or 
EPSG names for well-known objects is an alternative. Otherwise you have to use more 
verbose syntaxes with WKT or PROJJSON

> (I realize everybody but me thinks that somebody saying "WGS84" means
> they have no idea what datum they really mean and thus an error of 2m is
> ok, but this seems to be skipping the datum shift from NAD83(2011) to
> modern WGS84.)

Indeed, when using the NAD83(2011) datum, you can get more accurate horizontal 
transformations to WGS84 with the HPGN grids.

> 3) I didn't realize one caould merge two codes that way to construct a
> compound CRS.  That may just be me that doesn't know that, but it might
> help to add an example to cs2cs(1).

Was already documented for projinfo, but indeed wasn't explicitly mentionned for cs2cs. 
Should be addressed by https://github.com/OSGeo/PROJ/pull/2290

Even

-- 
Spatialys - Geospatial professional services
http://www.spatialys.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/proj/attachments/20200702/f798801c/attachment.html>