GDA2020 and Web Mercator

[MarkChilcott]

Hi Peoples,

Environment : ArcGIS Pro 2.1

If our data is in GDA2020 and we transform to Web Mercator, the software first does a datum transformation to GDA94, and then to Web Mercator.  The Geographic Transformation is chosen for me, and I cannot alter it, nor can I delete it – as shown below.

This is what is occurring:

• GDA_1994_To_GDA2020_1
  • A datum transformation to GDA94
• GDA_1994_To_WGS_1984
  • NULL datum transformation to WGS84

Effectively– this is not the GDA2020 realization of web Mercator - It transformed to GDA94 first.  It is the GDA94 realization of Web Mercator.

Are other people aware of this?

Implications?

Cheers,

Mark

[MarkChilcott]

Hi Peoples,

I followed up the issue with Esri.  They have confirmed the behavior described, and are looking toward a solution.

It is not so straight forward.  The EPSG codes were not available so a work around was done.  This would appear to have been an oversight of the ICSM and thus affects all software vendors.

Right now there is not a datum definition that goes between GDA2020 and WGS84 web Mercator. It is being ratified by EPSG / ICSM. As a work around the way Esri have implemented the migration to WGS84 Web Mercator is to project back to GDA94 first.

Esri are working to provide a GDA2020 to WGS84 Web Mercator but until this is in place by the international bodies then Esri cannot implement a full datum transformation. Until this time the solution is to go through GDA94 first.

Users implementation plans for proving web services based on the GDA2020 realization of web Mercator may well be affected by this.

Cheers,

Mark

[KylieMiddleton]

Mark, have you had any further updates on this from Esri? from their blogs message is still the same. The ICSM wrote to Esri about this issue (Ive seen the letter). Would like to know how customers/software client would know when consuming data as WGS84 (4326 or 3857) if it is the GDA94 or GDA2020 realisation, as GDA2020 => WGS84 is null transformation.

[MelitaKennedy]

Hi Kylie, Mark, 

GDA2020 is a null transformation to WGS84 (G1762) or ITRF2014 AT 2020.0, not to generic WGS84. Although EPSG does have a null transformation (parameters are zeroes), EPSG:8450, its (in)accuracy is up to 3 m. 

Because we already have a null/zeroes transformation between GDA94 and WGS84, we don't want to add the same for GDA2020 to WGS84. It could create a 'cycle' with data going through the different transformations and being labeled as GDA94 or GDA2020 when it's really the other one. Plus overlaying two WGS84/Web Mercator layers which are actually GDA94 and GDA2020 will cause problems. We're still discussing it here.

Melita

[KylieMiddleton]

Thanks Melita,

Can you recommend the best datum to store data in when moving to GDA2020 and publishing web services or this that part of the ongoing discussion? Services that will be used in Esri products along with Esri base maps? Currently, we store and publish mostly vector data in 3857 due to performance issue we have experienced when publishing in 4283 and building Esri Maps and Apps. Our current transformation path is GDA94 (4283)>3857, however, our Imagery will be moving to GDA2020 (currently 4283). Cheers Kylie.