Some thoughts on datum realizations and transformations

1081
1
10-19-2016 03:53 PM
Labels (1)
MelitaKennedy
Esri Notable Contributor
4 1 1,081

A customer had some questions on why Esri has "null" or "bookkeeping" transformations as well as why there are transformations which have the same parameter values, but convert between multiple geographic coordinate systems. Here's what I wrote off the top of my head.

 

  • WGS84 is actually not accessible to normal consumers because it’s a military system. Yes, GPS reports WGS84, but in a degraded state. RTK/post-processing are actually linking to ITRFxx or local GCS/datum like NAD83 (2011).
  • A lot of data is labeled NAD83 but is really another realization like NAD83 HARN or CORS96 or worse, a mix of several realizations as the dataset has been edited over the years.
  • Similarly, there’s tons of data that’s labeled WGS84 that really isn’t.
  • Some states had one HARN realization, other had 2 or 3. NGS only published conversions between NAD83 (1986) and the first HARN realization, not the later ones.
  • Transformations didn’t exist between HARN, NSRS2007, and 2011 until a few years ago. We’ve put in transformations for GEOCON and GEOCON11 v1.0. There’s a new NGS version in beta that further differentiates between the various HARN realizations, by calling later ones FBN.
  • The NGS CORS website has published some 14 parameter transformations (time-based coordinate frame which has 3 translations, 3 rotations, plus a scale difference plus 7 more parameters that have time components) between ITRFxx or IGSxx (basically equivalent) to NAD83 (realization). Esri has incorporated these as 7 parameter coordinate frames by dropping the time components. That means the transformations occur at the ‘reference epoch.’
  • Knowing there’s a mish-mash of data out there, I’ve put equivalents where I duplicate an ITRF-to-NAD83 with WGS84-to-NAD83 versions. I’ve also added null or “bookkeeping” transformations (where the parameters are zeroes) to get between GeoCRS (geographic coordinate reference systems) where there’s no other transformation or at a certain accuracy level, these can be considered equal.
  • I have not been consistent about it, and have been putting in fewer as time goes on and I learn more about geodesy and as the accuracy has been improving on the more recent realizations.
  • EPSG (http://www.epsg.org and http://www.epsg-registry.org) plan to put in the multiple realizations of WGS84 and Canada’s NAD83 CSRS shortly. We’ll follow, probably for 10.5.1. That’ll make everything even more confusing!
1 Comment
About the Author
I've been at Esri for over 25 years, working on map projections, coordinate reference systems and geographic/datum transformations. Although I have a master's degree in geodetic science from Ohio State University (go Bucks!), I'm not a geodesist. Last great new book that I read: Network Effect by Martha Wells