Messages

1

General / Re: AUSgeoid20 - Horizontal Datum Correct?

« on: May 24, 2019, 08:36:57 AM »

Interesting. If I run it on AUSGeoid2020_20180201.gsb I get GDA94. See below.
https://s3-ap-southeast-2.amazonaws.com/geoid/

Where did you get that TIF file from? I only see the  GSB/DAT files at the address above. I can't find any description of what the difference between the 2017/18 series files is. Both are available in the S3 bucket. The header data is the same apart from name/date changes but there are differences in content.

Looks like their web app is using a 2017 version, but the sparc (Solaris i guess) version. Not sure what the difference is. Byte order perhaps?
https://bitbucket.org/geoscienceaustralia/ausgeoid/src/master/geoid-core/src/test/resources/

My best guess is some confusion creeping in due to the NTv2 format being intended for horizontal, not vertical grid shifts. GDA94 is specified in the NTv2 header, but that's for the vertical shift, not horizontal, and it's being interpreted at some step as referring to the horizontal datum. If I generate a TIF myself with gdaltranslate, it defaults to WGS84 for the horizontal datum, unless I force it to, say, GDA2020 with

Code: [Select]

gdal_translate -a_srs EPSG:7844 -b 1 -of gtiff AUSGeoid2020_20170908.gsb AUSGeoid2020_20170908.tif

In that case I get similar header to your tif file.

Code: [Select]

$ntv2_file AUSGeoid2020_20170908.gsb
filename AUSGeoid2020_20170908.gsb
========  hdr ===================================
NUM_OREC 11
NUM_SREC 11
NUM_FILE 1
GS_TYPE  SECONDS
VERSION  1.0.0.0
SYSTEM_F GDA94
SYSTEM_T AHD_1971
MAJOR_F  6378137.0
MINOR_F  6356752.314
MAJOR_T  6378137.0
MINOR_T  6356752.314

num-recs         1
num-parents      1
parents        0 [ AUSGEOID ]
conversion  0.0002777777777777778  [ SECONDS  ]

========    0 ===================================
SUB_NAME AUSGEOID
PARENT   NONE
CREATED  01012010
UPDATED  01012010
S_LAT    -219540.0      [ -60.98333333333333 ]
N_LAT    -28800.0       [          -8.0 ]
E_LONG   -626340.0      [ -173.9833333333333 ]
W_LONG   -334800.0      [         -93.0 ]
LAT_INC  59.99999999996559 [ 0.01666666666666572 ]
LONG_INC 59.99999999996559 [ 0.01666666666666572 ]
GS_COUNT 15454800

name        AUSGEOID
parent      NONE
num subs    0
num recs    15454800
num cols    4860
num rows    3180
lat min     -60.98333333333333
lat max     -8.0
lat inc     0.01666666666666572
lon min     93.0
lon max     17398333333333333
lon inc     0.01666666666666572


Code: [Select]

ntv2_file  -h /data/metashape-data/buchanv1/AUSGeoid2020_20180201.gsb
filename /data/metashape-data/buchanv1/AUSGeoid2020_20180201.gsb
========  hdr ===================================
NUM_OREC 11
NUM_SREC 11
NUM_FILE 1
GS_TYPE  SECONDS
VERSION  1.0.0.0
SYSTEM_F GDA94
SYSTEM_T AHD_1971
MAJOR_F  6378137.0
MINOR_F  6356752.314
MAJOR_T  6378137.0
MINOR_T  6356752.314

num-recs         1
num-parents      1
parents        0 [ AUSGEOID ]
conversion  0.0002777777777777778  [ SECONDS  ]

========    0 ===================================
SUB_NAME AUSGEOID
PARENT   NONE
CREATED  01012010
UPDATED  01012010
S_LAT    -219540.0      [ -60.98333333333333 ]
N_LAT    -28800.0       [          -8.0 ]
E_LONG   -626340.0      [ -173.9833333333333 ]
W_LONG   -334800.0      [         -93.0 ]
LAT_INC  59.99999999996559 [ 0.01666666666666572 ]
LONG_INC 59.99999999996559 [ 0.01666666666666572 ]
GS_COUNT 15454800

name        AUSGEOID
parent      NONE
num subs    0
num recs    15454800
num cols    4860
num rows    3180
lat min     -60.98333333333333
lat max     -8.0
lat inc     0.01666666666666572
lon min     93.0
lon max     17398333333333333
lon inc     0.01666666666666572


2

General / AUSgeoid20 - Horizontal Datum Correct?

« on: May 23, 2019, 09:59:52 AM »

Hi,

I'm trying to transition to a workflow based on GDA2020 using the ausgeoid2020 vertical offset grid.

The AUSGeoid2020.tif on the Agisoft website has it's horizontal CS defined as GDA94. You can see this with gdalinfo for example.

That seems odd, as the GDA2020 technical manual (http://www.icsm.gov.au/sites/default/files/GDA2020TechnicalManualV1.1.1.pdf) specifies that only co-ordinates in GDA2020 should be transformed using it, presumably as it's aligned to GDA2020. Ausgeoid 09 is aligned to GDA94. I can manually set the horizontal CS to GDA2020 when installing geoid20, but I'm not clear on what the behaviour will be in that case.
https://www.ga.gov.au/scientific-topics/positioning-navigation/geodesy/ahdgm/differences

Without knowing what process was used to generate that geotiff file from the source grid files it's hard to know what Agisoft's intention was here.

I did look at the ausgeoid2020 transform grids obtained from ICSM, and they provide them in NTv2 format. These add to my confusion as their header describes them as using GDA94 for the source vertical datum and AHD as the target. Given the same GRS80 ellipsoid is used in GDA94 and 2020 that seems confusing but doesn't change any results. The CRS for their grids meanwhile is set as WGS84. No mention of which IRTF frame that refers to.

I have to assume that their documentation, not the metadata, is correct and these are really expecting GDA2020 co-ordinates. I didn't get as far yet as trying to verify with their web portal the correct way to interpret all of this, as, given the 1' by 1' vertical grid cell size and the small shift between GDA94 and GDA2020, there are only small regions that would actually see the effects of that misalignment.

In any case, can anyone add any clarity to what is going on in the geoid2020 TIF that Agisoft is supplying?

Thanks,

Rob