[Proj] Setting up a proj4 specification

Mikael Rittri Mikael.Rittri at carmenta.com
Tue Feb 7 09:28:52 EST 2012


Hello,

Some months ago, I cited some published datum shifts from Arc 1960 to WGS84
in Kenya. 

Now I have tested them, and they don't work as I cited them. So here are some corrections:

I cited a three-parameter shift that was published in the Grids & Datums
column on Kenya:

   +towgs84=-179.1,-44.7,-302.6

But according to the primary source, 

   A. S. Lwangasi:
   Datum transformation parameters for the Kenya geodetic system.
   Survey Review, Volume 32, Number 247, January 1993, pp. 39-46(8).
   http://www.ingentaconnect.com/content/maney/sre/1993/00000032/00000247/art00004

this shift goes from Arc 1960 to WGS72, so it must be concatenated with a
datum shift from WGS72 to WGS84. If we use EPSG:1237 for the second step,
the result would be 

  +towgs84=-179.1,-44.7,-298.1,0,0,0.554,0.2263

I have compared this with the NIMA shift +towgs84=-157,-2,-299 (EPSG:1284),
and found differences of 3.1 m and 5.9 m on two test points, which isn't
alarming since the NIMA shift has accuracy 6 m, and the Lwangasi shift has
accuracy about 3.5 m (which must be combined with the 2 m accuracy between
WGS72 and WGS84). 
    But if the Lwangasi shift is misinterpreted as going to WGS84, the
differences were 20.4 m and 21.6 m on my test points. 

I also cited a "Molodensky-Badekas 7-parameters transformation" from the
AFREF Newsletter, and made a guess of how it could be represented in Proj.4.
However, I cannot make it work; it differs by hundreds of meters from the
NIMA and Lwangasi shifts, regardless of how I reverse signs.
   My best guess is that this datum shift is intended as a Molodensky-Badekas 10-parameters transformation. (At least, that's what I would call
it). The extra three parameters are the geocentric Cartesian coordinates of
the rotation point (evaluation point). Unfortunately, those three parameters
were not published in the AFREF Newsletter. I tried with the Arc 1960 origin
point in South Africa, but it didn't work. The rotation point is probably
the average of the observations in Kenya (I got pretty good agreement with NIMA and Lwangasi by using Nairobi as rotation point).
   
Best regards,

Mikael Rittri
Carmenta
Sweden
http://www.carmenta.com

-----Original Message-----
From: proj-bounces at lists.maptools.org [mailto:proj-bounces at lists.maptools.org] On Behalf Of Mikael Rittri
Sent: Wednesday, October 19, 2011 3:58 PM
To: PROJ.4 and general Projections Discussions
Cc: Hanlie Pretorius
Subject: Re: [Proj] Setting up a proj4 specification

PS again:

The AFREF newsletter No.5, January 2008,

http://facility.unavco.org/kb/getattachment.php?data=ODR8SkFOIDIwMDggQUZSRUYgbmV3c2xldHRlci5wZGY%3D

publishes what they call a "Molodensky-Badekas 7-Parameters transformation" 
that is "between WGS 84 and Arc 1960 datum for Kenya". I think that means that
the published transformation goes in the direction from WGS 84 to Arc 1960. 
So one would need to reverse the signs of all parameters for use in Proj.4.

Also, I think "Molodensky-Badekas" usually means that the rotation sign
convention is Coordinate Frame Rotation, while Proj.4 uses the opposite 
Position Vector Transformation.  If I am right, the signs of the rotation 
parameters must be reversed once more!  

If I am not mistaken, the Proj.4 syntax for the transformation should be 

   +towgs84=-148.5133,3.4865,-298.0307,-5.9464,7.1414,2.8544,4.3955

Regards,

Mikael Rittri
Carmenta
Sweden
http://www.carmenta.com 


-----Original Message-----
From: Mikael Rittri 
Sent: Wednesday, October 19, 2011 12:55 PM
To: 'PROJ.4 and general Projections Discussions'
Subject: RE: [Proj] Setting up a proj4 specification

PS:

Another datum shift for Arc 1960 to WGS 84 in Kenya is 

   +towgs84=-179.1,-44.7,-302.6

which is probably more accurate. See Clifford Mugnier's column on Kenya, 

http://www.asprs.org/a/resources/grids/06-2003-kenya.pdf

regards,

Mikael Rittri
Carmenta

-----Original Message-----
From: proj-bounces at lists.maptools.org [mailto:proj-bounces at lists.maptools.org] On Behalf Of Mikael Rittri
Sent: Wednesday, October 19, 2011 11:10 AM
To: PROJ.4 and general Projections Discussions
Subject: Re: [Proj] Setting up a proj4 specification

Hello, Hanlie.

[-- see  http://lists.maptools.org/pipermail/proj/2011-October/005962.html  --]
_______________________________________________
Proj mailing list
Proj at lists.maptools.org
http://lists.maptools.org/mailman/listinfo/proj


More information about the Proj mailing list