[PROJ] WGS72 to WGS84 transformations (was Re: PROJ Digest, Vol 8, Issue 8)

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

[PROJ] WGS72 to WGS84 transformations (was Re: PROJ Digest, Vol 8, Issue 8)

Melita Kennedy

Hello Fabian,

I knew I had heard about the difference in the scale-difference values before but couldn't remember the details. I didn't turn up anything conclusive in the documents that I have access to. I'm on the EPSG subcommittee so I asked the chairperson, Roger Lott if he remembered. He sent the question off to Noel Zinn and Roel Nicolai.

Roger answered first:

The reality is that use of either of the two Helmert CTs or the above is unlikely to be appropriate, for two reasons. Principally because there was little civilian access to WGS72 (precise ephemeris) and that, despite CRS labelling, usage was really of WGS72BE. Secondly the DMA recommendation was not to use the direct shift but instead use WGS72>local>WGS84 using the abridged Molodenski parameters for local > WGS.

Noel turned up some documents. One no longer has a source/citation attached but discusses converting WGS-72 to proposed NAD-83 (PNAD-83) and proposed WGS-84. Conversion from NWL-9D to WGS-72 has a ds value of -0.8263 ppm. A transformation between NWL-9D (NSWC-9Z2) to PNAD-83/WGS-84 has a ds of -0.6 ppm. Concatenate them together and you get +0.2263 ppm for converting from WGS-72 to PNAD-83 or WGS-84. This page has a date of December 7, 1986.

Noel then retrieved the fact that a US DOD WGS 72 report from 1974 states that between NWL-9D and WGS 72, lists delta-r as -5.27 meters. Noel reports, note that 5.27 / 6378135 =  0.82626E-6. See ds from the first document. Now see DMA 8350.2, table 2.1, converting from NSWC 9Z-2 to WGS 84, which lists a ds of -0.6 ppm, which matches the 2nd transformation from the first document. 

Noel also points out that Table 7.1 from 8350.2 (1997) lists 1.4 for delta-r between WGS 72 and WGS 84. That ends up 1.4 / 6378135 = 0.2194999E-6, or 0.219ppm to three digits. 

Roel doesn't have access to his records right now, but stated, "two agencies involved specifying the relationship between the two CRSs in different ways: one gave it as a radial correction to the orbit parameters, the other as the 7th parameter of a Helmert transformation. Working the radial correction into a Helmert ds yields this small difference. When you calculate the effect on point positioning height (= altitude), the difference was negligible at the time. Some people made a big fuss about it at the time. I think I recommended 0.219 at the time because I found the four decimal places of 0.2263 over the top."

Regards,
Melita

Melita Kennedy
Sr. Esri Product Specialist

-----Original Message-----

>From: [hidden email]
>Sent: Jun 17, 2019 12:00 PM
>To: [hidden email]
>Subject: PROJ Digest, Vol 8, Issue 8
>
>----------------------------------------------------------------------
>
>Message: 1
>Date: Mon, 17 Jun 2019 07:30:52 +0000
>From: "Lesparre, Jochem"
>To: Fabian Gross , "[hidden email]"
>
>Subject: Re: [PROJ] How to apply Helmert trafo
>
>Dear Fabian,
>
>For this transformation push/pop is not needed. That only makes sense in other situations, like when you use a geoid.
>
>If the scale factor also differs from the primary source (EPSG is a secondary source), it should be adapted in the example to avoid confusion. If EPSG differs with the primary source, EPSG should be notified that there might be a mistake in their registry.
>
>Kind regards, Jochem
>
>From: PROJ On Behalf Of Fabian Gross
>Sent: Friday, June 14, 2019 3:41 PM
>To: [hidden email]
>Subject: [PROJ] How to apply Helmert trafo
>
>Dear proj-members,
>
>I would like to apply a Helmert trafo as part of a datum shift.
>The example follows the EPSG geomatics guidance 7/2 (report 373-7-2):
>
>From WGS72 (4984) -> WGS84 (4978).
>
>There are 2 issues:
>1. How to apply Helmert-Trafo? I.e.
>
> * with or
> * without the push/pop of the ellipsoidal height.
>Only if it is not pushed/popped do the values fit to the reference and the direct/naive implementation of Helmert.
>2. The scale parameters in the example (0.219) and the EPSG catalogue of pyproj (0.2263) do not agree.
>
>The attached file has the test case.
>
>
>Kind regards
>
>Fabian Gross
>Telefon +49 (711) 648 71-995
>_________________________________________________
>sbp
>schlaich
>bergermann partner
>
>Beratende Ingenieure
>für erneuerbare Energie
>
>Stuttgart . Berlin . New York
>São Paulo . Shanghai . Paris
>
>sbp sonne gmbh
>
>Markus Balz Dipl.Ing. (FH)
>Andreas Keil Dipl.Ing.
>
>Schwabstrasse 43
>70197 Stuttgart
>Telefon +49 (711) 648 71-0
>
>www.sbp.de
>_________________________________________________

_______________________________________________
PROJ mailing list
[hidden email]
https://lists.osgeo.org/mailman/listinfo/proj