[PROJ] Needs a sanity check on Vertical Transformation

[Even Rouault]

Brian,

I've been working on that very subject those last days. Pending work is in
https://github.com/OSGeo/PROJ/pull/1710
and
https://github.com/OSGeo/proj-datumgrid/pull/60

to add support for GEOID18 (as well as older GEOID versions for specialized 
purposes)
For getting a value in the grids, PROJ will use the Datum of the GeographicCRS 
indicated in the EPSG database for the transformation record between NAVD88 
and the NAD83 realizations you mention below, possibly doing a horizontal 
adjustment to go back and forth between the initial and final CRS and this 
interpolation CRS. It will not go through WGS84.

So for example (with the 2 above pull requests in place), transforming between 
NAD83(2011) + NAVD88 height and NA83(2011) (3D) shows that no horizontal 
adjustment is attempted:

projinfo -s EPSG:6318+5703 -t EPSG:6319 --spatial-test intersects -o PROJ
Candidate operations found: 2
-------------------------------------
Operation n°1:

unknown id, Inverse of NAD83(2011) to NAVD88 height (3), 0.015 m, USA - CONUS 
- onshore

PROJ string:
+proj=pipeline +step +proj=axisswap +order=2,1 +step +proj=unitconvert 
+xy_in=deg +xy_out=rad +step +proj=vgridshift +grids=g2018u0.gtx +multiplier=1 
+step +proj=unitconvert +xy_in=rad +xy_out=deg +step +proj=axisswap +order=2,1

-------------------------------------
Operation n°2:

unknown id, Inverse of NAD83(2011) to NAVD88 height (2), 0.02 m, USA - Alaska

PROJ string:
+proj=pipeline +step +proj=axisswap +order=2,1 +step +proj=unitconvert 
+xy_in=deg +xy_out=rad +step +proj=vgridshift +grids=g2012ba0.gtx 
+multiplier=1 +step +proj=unitconvert +xy_in=rad +xy_out=deg +step 
+proj=axisswap +order=2,1

Even

> GEOID18 supersedes GEOID12B in CONUS and Puerto Rico.  GEOID12B should
> be used in Alaska, Hawaii, Guam, and the Commonwealth of the Northern
> Mariana Islands.  American Samoa should not be used for anything as
> ASVD02 is no longer valid after earthquakes.
> 
> A note on Geoids, Horizontal (Geometric) Datums and Vertical Datums.  To
> get an orthometric height (aka elevation) in a particular vertical datum
> you should use coordinates in a particular geometric datum with lat,lon,
> and ellipsoid height along with a geoid model to get an orthometric
> height.  For the US here are the appropriate datums and geoid models. 
> Note that the corresponding geoid models should be used based on the
> datum of your coordinates. Also note the the original NAD 83 (1986) was
> a 2D datum while all other realizations were 3D, aka geometric.
> 
> Also note that if you are concerned about accuracy, WGS84 should be
> avoided if at all possible.   WGS84 has a 3-5 meter uncertainty so any
> time you use coordinates or use it as a pivot in transformations you
> introduce this uncertainty into your coordinates.
> 
> Table on geoids and NAD 83 realizations
> 
> NAD 83 (realization) 	Geoid Model 	Regions
> NAD 83 (1986) 	N/A
> NAD 83 (HARN) 	GEOID96/GEOID99
> NAD 83 (FBN) 	GEOID03
> NAD 83 (CORS96) 	GEOID03
> NAD 83 (NSRS2007) 	GEOID09
> NAD 83 (2011) 	GEOID12B 	HI, AK, Guam, CNMI
> NAD 83 (2011) 	GEOID18 	CONUS, PRVI
> 

-- 
Spatialys - Geospatial professional services
http://www.spatialys.com