[Proj] Similar datum transforms for MGI differ too much.

[Mikael Rittri]

Hello,
I think I have found confirmation for my theory on Wikipedia:

> According to the European longitude adjustment of Theodor Albrecht 
> (ca. 1890) the Ferro meridian is 17° 39' 46.02" west of the Greenwich meridian. 
> But for the geodetic networks of Austria, Germany and Czechoslovakia, the 
> value 17° 40' 00" was adopted in the 1920s ... 
> For the geodetic networks of Hungary and Yugoslavia, the value of 
> Albrecht was used prior to the switch to the Greenwich prime meridian.
( http://en.wikipedia.org/wiki/El_Hierro#The_.22Meridian_Island.22 ) 

That Albrecht's value is used in Yugoslavia, is also confirmed by the 
University of Pittsburgh: http://corona.eps.pitt.edu/Website/yugoslavia.html 
(except that they say 17° 39' 36", but I think the 36" must be a typo 
and 46" was intended). 

So, the MGI (Ferro) datum was surveyed in the Austro-Hungarian Empire 
in the 1800s, and covered what is now Austria and former Yugoslavia.  
This was a single, continuous geodetic datum. 

But when Greenwich replaced Ferro as the prime meridian in the 1920s, 
Yugoslavia was politically (and geodetically) independent from Austria, 
so the countries used different values for the Ferro longitude.  A "Ferro curtain"
was lowered between Austria and Yugoslavia, so that MGI (Greenwich) in Austria
differs from MGI (Greenwich) in Yugoslavia by 14 arc seconds of longitude (or
possibly only 13.98 arc seconds).  

I think it is wrong of EPSG to say that MGI (Greenwich) is a single datum that 
covers Austria and former Yugoslavia. A datum can have distortions, but an 
abrupt 300 meter discontinuity at a political boundary...?

Anyway, if a datum shift is intended for MGI in Austria, but is used for MGI 
in former Yugoslavia, the results are wrong by about 300 meters.  This means 
that the current TOWGS84 parameters are bad in the OGC WKT on the following 
pages on www.spatialreference.org : 

   MGI / Balkans zone 5, EPSG:31275, http://www.spatialreference.org/ref/epsg/31275 

   MGI / Balkans zone 6, EPSG:31276, http://www.spatialreference.org/ref/epsg/31276  

   MGI / Balkans zone 7, EPSG:31277, http://www.spatialreference.org/ref/epsg/31277 

   MGI / Balkans zone 8, EPSG:31278, http://www.spatialreference.org/ref/epsg/31278

   MGI / Slovenia Grid, EPSG:2170, http://www.spatialreference.org/ref/epsg/2170

   MGI / Sloven National Grid, EPSG:3787,  http://www.spatialreference.org/ref/epsg/2170

Because they use TOWGS84 parameters that are intended for MGI in Austria.
Also, the OGC WKT for 

   MGI, EPSG:4312, http://www.spatialreference.org/ref/epsg/4312

does the same thing, so it can be used only in Austria. 
And the OGC WKT for 

   MGI (Ferro), EPSG:4805, http://www.spatialreference.org/ref/epsg/4805 

gives the rounded value for the longitude of Ferro, which is used only in Austria.

EPSG suggests some good datums shifts for MGI in former Yugoslavia. 
However, note that EPSG's general "MGI (Ferro) to MGI (1)" datum shift, code 1757, 
uses the rounded value of the Ferro longitude, so it is applicable only in Austria,
not in former Yugoslavia as EPSG claims. 

I wonder if there are other European datums that have been divided by the "Ferro curtain"? 
 
--
Mikael Rittri
Carmenta AB
SWEDEN
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: den 18 februari 2009 16:18
To: PROJ.4 and general Projections Discussions
Cc: kundenservice at bev.gv.at; info at bev.gv.at; tomaz.petek at gov.si
Subject: Re: [Proj] Similar datum transforms for MGI differ too much.

Hello again,
I think I have had a flash of insight, but I don't have any hard evidence.
So I hope someone can confirm my theory, which is as follows: 

<theory>

The MGI datum, used in Austria and former Yugoslavia, was originally based on the Ferro meridian, not Greenwich.  This is the datum that EPSG calls "MGI (Ferro)", code 6805. 

When the mapmakers in Austria and Yugoslavia decided to adapt MGI to the Greenwich meridian (to get "MGI", EPSG code 6312), they used different values for the longitude of Ferro.  In Austria, they assumed that Ferro was at 
    
    17° 40' 00" West of Greenwich,

which is the value given by EPSG. But in Yugoslavia, they assumed that Ferro was at 

    17° 39' 46" West of Greenwich,

which is the value given Table 4.1, http://www.map-reading.com/ch4-1.php . 
This would mean that, although the MGI (Ferro) graticule is continuous, the MGI (Greenwich) graticule is discontinuous at the Austrian-Sloven border.

In other words, MGI (Greenwich) is not a proper datum.  We should talk of Austrian MGI (Greenwich) and Yugoslavian MGI (Greenwich) as separate datums.

The results of my tests showed a difference in the MGI transforms for Austria and Slovenia, and I said that the difference was 299.7 meters, almost exactly east-west.  But expressed in longitude degrees, the difference is 14.092 arc seconds, which corresponds almost exactly with the 14" difference between the two Ferro longitudes.
(The remaining 0.092 arc seconds corresponds to 2 meters at this latitude, which  can be explained by the limited accuracy of the datum transforms.)  

</theory>

So, can anyone confirm this theory? 

--
Mikael Rittri
Carmenta AB
SWEDEN
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: den 18 februari 2009 11:48
To: proj at lists.maptools.org
Subject: [Proj] Similar datum transforms for MGI differ too much.

Hello,
here is small mystery.  
[-- see rest of post at http://lists.maptools.org/pipermail/proj/2009-February/004397.html --] _______________________________________________
Proj mailing list
Proj at lists.maptools.org
http://lists.maptools.org/mailman/listinfo/proj