GDA2020 transformation issues

Christopher,

I have verified the differences observed but am not surprised by them. The transformation from GDA94 to GDA2020 probably relies upon a complex system and transformation through 7 parameter transformation or NTv2 grid shift files is an approximation. The NTv2 file format allows for arbitrary densification of the data, and more precise grid shift files could have been designed by the Australian Government, yet, denser grid shift files result in files of greater size. The size of the Australian file is 83 megabytes which is already relatively large. The governement probably limited the density of the grid to keep the file size acceptable. A difference of 1 cm is not too bad. A bit large for cadastral purposes in dense urban area but acceptable for most purposes otherwise.

I have encountered NTv2 file as large as 1 gigabytes for provinces of Germany and probably these can result in millimetric accuracy yet the size is unmanageable and Bentley System does not distribute those files which must be installed manually by clients.

If the Australian government wants to they can design denser files and distribute them.

Christopher,

I have verified the differences observed but am not surprised by them. The transformation from GDA94 to GDA2020 probably relies upon a complex system and transformation through 7 parameter transformation or NTv2 grid shift files is an approximation. The NTv2 file format allows for arbitrary densification of the data, and more precise grid shift files could have been designed by the Australian Government, yet, denser grid shift files result in files of greater size. The size of the Australian file is 83 megabytes which is already relatively large. The governement probably limited the density of the grid to keep the file size acceptable. A difference of 1 cm is not too bad. A bit large for cadastral purposes in dense urban area but acceptable for most purposes otherwise.

I have encountered NTv2 file as large as 1 gigabytes for provinces of Germany and probably these can result in millimetric accuracy yet the size is unmanageable and Bentley System does not distribute those files which must be installed manually by clients.

If the Australian government wants to they can design denser files and distribute them.

Hi Alain,

Thank you for verifying this and for the information provided.

We will liaise with Landgate and/or GeoScience Australia if indeed higher accuracy conversions are required.

Thank you,

Chris.

Alain Robert said:

A difference of 1 cm is not too bad. A bit large for cadastral purposes in dense urban area but acceptable for most purposes otherwise.

I have encountered NTv2 file as large as 1 gigabytes for provinces of Germany

1cm for cadastral purposes seems really good even for us crazy Germans

It's not really cadastral surveys that has us concerned - it's the large scale control networks for various infrastructure projects.  If we can not match published values, we are in trouble before we even start.

We have received the following information from Landgate;

Hi Christopher,

Thank you very much for sharing this very good observation. This is as expected though. The GDA2020 SSM coordinates are the product of the national adjustment and the GSB NTv2 (conformal + distortion) grid was derived from the differences between this national adjustment (GDA2020 coordinates) and the state adjusted GDA94 coordinates. To model the distortion part of the grid the least squares collocation technique was applied. The technique applies weighted distortion at surrounding survey marks to estimate distortion at grid nodes. The approach involves development of a distance-dependent covariance function to represent the spatial behaviour of distortion in two-dimensions (East and North). Without going into details, the grid is the best estimation of the difference between GDA2020 and GDA94 coordinates but won’t produce exact match. Please note the GSB NTv2 grid has associated transformation errors (in East and North) and they are provided in field 3 and field 4 of the GSB file.

Having said that the GSB NTv2 conformal + distortion grid is still the most accurate way of transforming coordinates between GDA94 and GDA2020 but we need to understand that GSB won’t give the exact values as provided by national adjustment and reported on GOLA SSMs report. Perhaps the best way to validate your transformation results is to compare its results to this online software provided by Geoscience Australia:  http://positioning.fsdf.org.au/

For more information regarding transformation please refer to our website: https://www0.landgate.wa.gov.au/business-and-government/specialist-services/geodetic/gda2020

Just to let you know ICSM is finalising a document regarding GDA94 to GDA2020 transformation grid development and I will let you know once this document is publicly available.

If you need more explanations please do not hesitate to contact me.

Thank you.

Cheers,

Irek

Irek Baran | Principal Consultant – Spatial Positioning

Location Services

Landgate, 1 Midland Square, Midland WA 6056