Project Data Tranformation Issue

[Anonymous User]

Original User: Nikolaev_d

Hi ESRI and all

I think I have an important question to ask and I am sure that sharing this information will be a great benefit to everyone.

I have discovered a huge spatial difference in results when reprojecting data (project command) using different transformation methods available from the standard ESRI transformation list.
This is what I have done in ArcGIS10.

I have generated a point, which is located in Papua New Guinea using GDA94 MGA Zone56 coordinate system.

I have projected original point to AGD66 Zone56 using NTv2 transformation and my output point is shifted only about 1m from the original location, which is a bit suspicious as I know that the difference between AGD66 and GDA94 in Papua New Guinea should be about 180m-200m.

Also I have projected the original point to AGD66 Zone56 using
AGD_1966_To_GDA_1994
AGD_1966_To_GDA_1994_2
AGD_1966_To_GDA_1994_3
AGD_1966_To_GDA_1994_4
And all my results are significantly spatially distributed (far away from each other). For instance the difference between AGD_1966_To_GDA_1994_2 and AGD_1966_To_GDA_1994_3 transformation results is about 17 metres!

Is anybody out there can explain me what is going wrong with my transformations? I am aware that all transformations are supposed to produce slightly different results but 17m... I think it is too much.

I have attached an illustration for a better understanding of this issue and the original point as a shapefile.

Thanks
D

[MelitaKennedy]

Hello,

I think you should take a look at the listings in technical article 21327 which lists areas of use and accuracy in the ArcGIS 10 version, and parameter values in the earlier versions.

None of these transformations are listed as valid for Papua New Guinea.

_2 is designed for ACT
_3 is designed for Tasmania
_4 is designed for NSW

AGD_1966_To_GDA_1994_7 might be possible because it's for the Northern Territory, but we don't provide that grid file. Its information is included in the national version used in AGD_1966_To_GDA_1994_11_NTv2. This NTv2 transformation does extend to -10 latitude (only), but that doesn't guarantee that the values are reasonable. I've come across grids that 'smooth' out to the edges.

Melita

[Anonymous User]

Original User: Nikolaev_d

Thanks for your reply Melita

I realise that different translations are designated for different areas. I simply did not realised that the difference may be so big.

I was hoping that I would able to find a transformation, which is suitable for Papua New Guinea. For some reasons all other countries around the world have a dedicated coordinate system (even Fiji which has a population of only 860K) but Papua New Guinea (with a population of 6.5 million) has not got a coordinate system listed in a standard ESRI list.

Could you please advise me if I can find a proper coordinate system and translation(s) suitable for PNG in available ESRI resources?

Thanks
D

[MelitaKennedy]

We added PNG94 in version 10, I believe. There's a transformation, but it has zeroes for the parameter values. According to the EPSG online registry, when PNG94 was defined, it was considered equivalent to the current version of WGS84.

Both EPSG (I'm on the subcommittee that maintains the dataset/registry) and Esri add what people ask us to add.

PCGIAP only lists PNG94, and no transformation information. There is an email contact for the National Mapping Bureau. Perhaps they might have some more information?

I wish I could be of more help.
Melita

[Anonymous User]

Original User: Nikolaev_d

OK, thanks a lot Melita

It looks like Papua New Guinea is out of action for ESRI users...

PS PNG94 is not included in version 10

Thanks,
D

[MelitaKennedy]

Oops! Sorry about that.

I was at home and didn't have access to check. We have added the coordinate systems for 10.1 (to be released next year).

Melita

[Anonymous User]

Original User: bodiska

Dear Melita,

Currently I use ArcGIS Desktop 9.3.1 SP 1, License Type: ArcInfo.

I tried to do transformation between two Lambert systems, but
had no result because of "Undefined geographic transformation".

Avoiding the not always accurate solution of the 'on-the-fly
projection' and then 'export' I used the tool of the
ArcToolBox, Data management Tools, Projections and Transformation,
Project (Or Batch Project), Geographic Transformation (optional).

What should I select if I need to transform data

FROM THIS:
---------------------------------------
PROJCS["ETRS_1989_LAEA_L52_M10",
GEOGCS["GCS_ETRS_1989",
DATUM["D_ETRS_1989",
SPHEROID["GRS_1980",
6378137.0,
298.257222101]],
PRIMEM["Greenwich",
0.0],
UNIT["Degree",
0.0174532925199433]],
PROJECTION["Lambert_Azimuthal_Equal_Area"],
PARAMETER["False_Easting",
4321000.0],
PARAMETER["False_Northing",
3210000.0],
PARAMETER["Central_Meridian",
10.0],
PARAMETER["Latitude_Of_Origin",
52.0],
UNIT["Meter",
1.0]]
--------------------------------------

TO THIS:
--------------------------------------
Projection: Lambert_Azimuthal_Equal_Area
False_Easting: 0.000000
False_Northing: 0.000000
Central_Meridian: 9.000000
Latitude_Of_Origin: 48.000000
Linear Unit: Meter (1.000000)

Geographic Coordinate System: GCS_Sphere_ARC_INFO
Angular Unit: Degree (0.017453292519943299)
Prime Meridian: Greenwich (0.000000000000000000)
Datum: D_Sphere_ARC_INFO
  Spheroid: Sphere_ARC_INFO
    Semimajor Axis: 6370997.000000000000000000
    Semiminor Axis: 6370997.000000000000000000
    Inverse Flattening: 0.000000000000000000
--------------------------------------

Input and Output coordinate systems are defined properly,
but the transformation is blocked because of
"Undefined geographic transformation".

Earlier I used the easy PROJECT function of ArcInfo, but now I need
to project shapes and also using projections not supported
by ArcInfo Workstation anymore.

Could you help?
Thank you in advance.

Best regards,
Katalin Bodis

[MelitaKennedy]

Hi Katalin,

This comes up pretty frequently on the forums. The issue is that the target (or sometimes the source) coordinate system is based upon a sphere. We don't have any transformations to/from a sphere because they're not a true, geodetic geographic coordinate system/datum.

In this case, it is easier to use ArcMap. Ignore the transformation warning, and export the data. Basically, the input/output GCS differences are ignored and internally, the data is redefined as the output GCS if no transformation is available. That's what was happening in ArcInfo Workstation. The on-the-fly projections in ArcMap are not less accurate than using the Project tool, unless no transformation is set when it's needed. In this case, it's not needed.

If you instead want to set up a custom transformation, use Longitude_Rotation for the method which doesn't have any parameters. This is emulate the 'don't do a transformation.'

Hope this helps,
Melita

[Anonymous User]

Original User: bodiska

Dear Melita,
Thank you for the clarification.
Best wishes for the New Year,
Katalin

[PaulSmith2]

Hi All,

I would like to way in on this issue with PNG94 and AGD66.

I would firstly like to direct anyone to the Quickclose Pty Ltd website and Richard Stanaway.  http://www.quickclose.com.au/

Also to his number of excellent Publications on this very topic http://www.quickclose.com.au/publications.htm

Some Points to Note:

There is no Standard transformation between AGD66 and PNG94 for Papua New Guinea.

Papua New Guinea has the added complexity which Australia does not have which is internal tectonic movement. Ie there is a number of different plates moving in different directions and velocities.

PNG94 is a Static Datum (Epoch 1994.0)  And for all purposes aligned with WGS84 in Epoch 1994.0   (Not Now)

WGS84 is a Dynamic Datum (Constantly Moving) - Take a reading 10 years ago and take one now on the same physical spot in PNG and you will get a different coordinates.


So... How to deal with this...

Richard has developed a set of AGD66 to PNG94 transformation parameters for our local work area. And the same would need to be established if you want to use old AGD66 coordinates with any accuracy.

All WGS84 coordinates need to be tied to a local monument on the same plate as the area you are working. You then need to apply a velocity shift to convert to PNG94 (Epoch 1994.0)

Richards Publications explain this in more detail.


This is a very complex area and I am sure I am getting a few greys because of it.


Regards,

Paul Smith