[pdal] Regarding Reprojection Between Vertical Datums

[Greg Troxel]

Kirk Waters - NOAA Federal <kirk.waters at noaa.gov> writes:

> That's because PROJ is using a ballpark offset to convert to WGS84 (you can
> see that with projinfo -s EPSG:6318 -t EPSG:4326) with an accuracy around 2
> meters. I suspect it's using the NAD83(86) to WGS84(Transit), but doesn't
> have a way to convert NAD83(2011) to NAD83(86) because the only published
> way is a multi-grid hop. Something like
> NAD83(2011)->NAD83(2007)->NAD83(HARN)->NAD83(86), but I might have missed a
> hop in there. However, there are published Helmert transforms from
> NAD83(2011) to ITRF(2008) and ITRF(2014). The accuracy of those transforms
> will be much better.

proj is using a nop to convert NAD83->WGS84, on the theory that the
WGS84 ensemble has an intrinsic accuracy limit of ~2m.  That amounts to
a null transform from NAD83(2011) to NAD83(1986) (which isn't
unreasonable), a null transform from NAD83(1986) to WGS84(Transit)
(arguably correct) and then representing the answer as "WGS84", which is
not ok because people expect that to be "non-ancient WGS84".  (Except,
NAD83(19860 is a 2D CRS, so you can't really use it as a steppingstone
to getting WGS84 LLh from 6319.)

Or it amounts to "because the intrinsic accuracy of WGS84 is 2m, it's ok
to just use a null transform if that's known to have an error on the
order of 2m".

Which really means: if you care about < 2m, you cannot use "WGS84" as a
CRS.

Not really on topic here, but I think ensemble transforms should be done
by assuming that data in each ensemble is in the most recent ensemble
member.  Thus "NAD83->WGS84" would be done by "assume 6319, and convert
to WGS84(G2159) (which is ITRF2014)".  Transforming to NAD83(2011), then
to ITRF2014 and finally to WGS84 achieves this with the current
software, using the better transforms that you reference.

(Kirk: I know you understand all of this), but for Nicholas and others:
the basic problem is that WGS84(Transit) is quite different from later
WGS84 realizations.  NAD83 (and ~everyone else at the time) misestimated
the center of mass of the earth by roughly 2m.  Later realizations of
WGS84 have used better estimates of the center of mass, and thus more
recent WGS84 is basically the same as more recent ITRF.  But later NAD83
is, more or less, aligned with earlier NAD83.  With modern sub-meter
accuracy, NAD83 and modern WGS84 cannot be treated as equal.

Greg


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 194 bytes
Desc: not available
URL: <http://lists.osgeo.org/pipermail/pdal/attachments/20220713/2e8a7158/attachment.sig>