[PROJ] proj8 and NAD27 <-> NAD83 transform

[Greg Troxel]

Everything Even said sounds correct to me -- and I am far less of an
expert in how proj makes decisions.

You are suffering from two problems:

  The concept of NAD83 as an ensemble, where it refers to an unspecified
  member of the set of NAD83 realizations (but only suffering a
  little).  You will notice that NCAT does not admit to this concept; it
  simply has 5 distinct NAD83 realizations.

  The EPSG database just having missing transforms.  While NCAT has
  careful transforms, the  difference around me from NAD83(2011) to
  NAD83(1986) is about 4 cm.


So IMHO the right fix is to add the missing transforms, on an
approximate basis, to the database, perhaps picking 5cm or 10 cm as the
accuracy, and saying that each NAD83 realization can be null-transformed
to NAD83(2011) with that accuracy.

Certainly it is far more sensible to assume that all NAD83 realizations
are equivalent than to assume a null transform is appropriate between
NAD27 and anything.  I wonder if proj should have this fallback at all;
it seems better to throw an error.

Also, I really have to wonder which NAD83 realization the NAD83/NAD27
grids are for.  It does not make sense that NGS would publish ensemble
grids, rather than for some specific realization.  (I can certainly see
grids where there is a different datum tag in different regions.)

It seems that it might be difficult to resolve these issues at EPSG, but
I admit to not having tried.  So probably proj should have a local set
of transforms in the repo and included with releases, that get added.

It's on my list to work on a set of local database additions, more for
mitigating the incorrect assumption that NAD83==WGS84.
-------------- 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/proj/attachments/20210602/3964555c/attachment.sig>