Jump to content
Ian

Similarity transformation Grid to 1D array of coordinates

Recommended Posts

I've got a large number of datasets in netcdf format and need to update the coordinate reference system from GDA94 to GDA2020.  I'm new to geodesy so have been spamming geosciences Australia, who have been very helpful, but I'm stuck and need some assistance from the community if possible.

As is standard, the gridded data surfaces in our netcdfs are defined geospatially using 1D array of lat and lon coordinates. Each grid cell is ~1km x 1km.  I've tried using gdalwarp from GDAL 2.2.2 to reproject them, but the coordinates are shifted by only ~2.1x10e-15, which seems dubious.  I understand that at this resolution the shift is negligible, however, we will be using this for scaling with 30m res drone data so having the higher precision is required.

For the similarity transformation I need ellipsoidal height, so I've accessed the AUSGeoid09 .gsb file and found latitude and longitude offsets, and I've got a grid of elevation data from a DEM to approximate the AHD.  I've downloaded the GA conversion spreadsheet and coded the formulae up in python, however, this is where I'm stuck. I'm unsure how to shift 1D arrays of lat/lon coordinates using a grid of height values. Depending on what the longitude is, the height will be different for the same latitude, and vice versa.  I can do it over the full grid, but then I need to take simply unique values after rounding, and this feels dubious also.  I should say that in my tests of the code, the shift is ~ 0.5-1.5m which seems closer to correct.

I've read that there are tools in QGIS for this process, however, the NTv2 grid plugin says it only handles up to GDA94.

Can anyone shed some light on this please? 

Share this post


Link to post
Share on other sites

Hi Ian,

GDA2020 isn't really available as an "operational" datum just yet - recommend review / keep an eye on the ICSM website for information about GDA2020

http://www.icsm.gov.au/gda/index.html - have you reviewed this site for status as well as the technical and explanatory information about horizontal and vertical transformations?

Content will be updated and expanded when GDA2020 is officially "operational" as will the GA pages.  The QGIS plugins will also deal with 2D GDA94 to GDA2020 transformations. This isn't far away.

GDAL 2.2.2 doesn't implement a GDA94 <> GDA2020 transformation.

In a successful transformation, the change in horizontal coordinate values should equate to 1.5 to 1.8m horizontal difference, depending on where you are in Australia. 

I don't quite understand the summary around the use of AUSgeoid09, in particular  "the shift is ~ 0.5-1.5m"  - shift in what?

Share this post


Link to post
Share on other sites

Hi,

Thanks for taking the time to respond.  Apologies if my usage caused confusion, as you can imagine I haven't quite got the vernacular down yet. Interesting that GDA2020 isn't operational yet, the website says it was due to start usage early this year.

In answer to your question, the 'shift' I quote is an example of the difference between the original latitude and longitudes and the latitude and longitude updated using the 7-parameter transformation.  Using decimal degrees, the difference is between 3.3x10e-6 and 1.5x10e-5, so approximately 0.33m and 1.5m.  Good to know the rough estimate of what it should be horizontally.  

Interestingly, the GDAL 2.2.2 does process a gdalwarp command to reproject a netcdf to the new EPSG code.  This is the command I used

gdalwarp -s_srs EPSG:4823 -t_srs EPSG:7844 -of netCDF ANUClimate_v1-0_solar-radiation_monthly_0-01deg_1970-2016_00000000_201412.nc test2.nc

I got the EPSG:7844 from the epsg registery (though this isn't on spatialreference.org yet).

Also, the Geosciences Australia spreadsheet for this (http://www.ga.gov.au/__data/assets/file/0003/11379/transxyz.xls) doesn't include the time component, even though this is featured prominently in the matrix derivation of the transformed coordinates detailed in the Interim Release Notes.

Thanks for your help.

Ian

Share this post


Link to post
Share on other sites

Hi Ian,

IF you haven't had the need to follow things closely it is understandable that you thought GDA2020 was meant to be used since last January.  

RE the open source libraries PROJ.4 / GDAL there are still a few things that needs to be arranged to have them working appropriately with GDA2020 and the various projections used across Australia.

Yes the spatial arena is great for jargon - more of it follows below sorry.

The spreadsheet you refer to above is for transforming from the historic Australian plate-fixed (static) datums AGD66/84 to GDA94 and wont be updated to include GDA94 > GDA2020, which is also a static or plate-fixed datum. Transformations between static datums only require 7 parameters - the time element (7 rates of of change, so 14 parameters) only comes into the picture if transforming to an "earth-fixed" reference frame like ITRF or stage 2 of Australian Datum modernisation - the ATRF.  (This statement is probably clear as mud!) The introductory description is generalised to allow for both cases.

The spreadsheets dealing with the 7 parameter GDA94 to GDA2020 transformation are on the ICSM github repository https://github.com/icsm-au but i would recommend having a look at the full GDA2020 tech manual http://www.icsm.gov.au/gda/tech.html first if you have the time.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×