GDA94 - GDA2020 Transformation Methods
There are three different accurate methods for transforming data between GDA94 and GDA2020, although they only potentially produce two different outcomes. The methods are:
7-parameter similarity transformation: A mathematical equation that relates two datums by origin translations of each the three axes, a rotation around each of the three axes, and a scaling (i.e. a total of seven parameters). It is a conformal transformation, meaning it preserves the shape of spatial data - so a square shape in the source datum will be a square shape in the target datum.
Image shows horizontal GDA94-GDA2020 shift at Tasmanian GNSS CORS sites. The 7 parameter similarily transformation effectively models this displacement in conjunction with the constant 0.09m offset in ellipsoidal height between the two datums
Conformal only NTv2 transformation grid: This method uses a grid of latitude and longitude shift vectors to transform coordinates and the grids shifts are derived entirely from the 7-parameter similarity transformation method. That is, it gives exactly the same results as that method. It is provided as an alternative to support the requirements of some users for a grid-based solution.
Conformal and distortion NTv2 transformation grid: This method also uses a grid of latitude and longitude shift vectors but the final grid shifts are comprised of the conformal transformation element combined with a grid of vectors that model the distortion in GDA94 coordinates present in the survey control network.
The third transformation method will generally produce a different result to the first two (in Tasmania the difference varies between 0 and 0.12m, see below) and so it is essential that the transformation method is stored as metadata with any transformed dataset. Refer to Preparing metadata to accompany spatial datasets for more information about metadata, including examples of how to record the transformation details.
Further information on the different methods is available from the ICSM website, including two fact sheets:
https://www.icsm.gov.au/sites/default/files/DatumMattersFactSheet4.pdf
https://www.icsm.gov.au/sites/default/files/DatumMattersT1FactSheet.pdf
Choosing a Transformation
Whether a Conformal or Conformal and distortion transformation is the appropriate method to apply to a spatial dataset depends upon a number of factors, including the origin of the data coordinates (e.g. derived solely from survey control marks or GNSS CORS networks or an aggregation of the two), the intended use of the data (e.g. critical engineering construction or in support of state wide mapping operations) and the magnitude of the distortions.
Within Tasmania, the mean distortion in GDA94 coordinates across the 3500 survey control stations modelled in the Conformal and distortion NTv2 transformation grid is 0.026 metres. Only 230 had a modelled distortion of greater than 0.05 metres and at only 25 of those was the distortion 0.1 metres or greater. The largest consistent distortion in a populated area is approximately 0.08m in Queenstown, which is specific to the urban area and does not extend further.
The small magnitude of the distortion modelled in relation to the accuracy of most aggregated spatial datasets, in conjunction with the fact virtually all spatial software and tools will default to a conformal transformation unless the conformal plus distortion transformation option is specifically chosen by the user, has led to Land Tasmania adopting the 7-parameter similarity transformation as the default method.
Details of the distortion vectors at each survey control mark used to generate the Conformal and distortion NTv2 transformation grid is available from Land Tasmania upon request.