Re: [Geotools-devel] Cassini-Soldner projection

[Andrea A. <aa...@op...>]

Mauro Bartolomeoli ha scritto:
> 
> 
> Andrea Aime wrote:
>> Mauro Bartolomeoli ha scritto:
>>> We are working on an Italy Cadastre importer, and we are thinking to 
>>> use udig to view the produced data, so we are going to need a CRS to 
>>> attach to Cassini-Soldner projected data.
>>> It's not widely tested and not very accurate (the error is in the 
>>> order of meters, compared to commercial software we use to convert to 
>>> the same data). 
>>
>> Interesting. Wondering if it's the math or the towgs84 params.
>> Italy has the very sorry situation (not sure how common in other
>> countries) that in order to have sub-meter precision on a local area
>> transformation you need the TOWGS84 params _for that area_, usually
>> contained in a document that can be bought by the militars.
> For the math, I think we could investigate if the formulas used by PROJ 
> are different from those found on the "OGP Surveying and Positioning 
> Guidance Note number 7, part 2 - August 2006" you can find on the EPSG 
> site (if I have enough time I'll try to compare the two set of formulas).
> 
>>
>> The 3 towgs84 params contained in the epsg database do not allow
>> for a precision better than 4 meters, and unfortunately geotools
>> is just picking one of them, I believe the one valid for Sardinia,
>> so if you're reprojecting on the continental part, you'll
>> get even bigger errors. (I'm reporting by memory, so I may
>> be completely off on the magnitude of the errors or about the
>> chosen towgs84 set of params).
> I think this could one of the inaccuracy causes (maybe not the only 
> one). The points I used for testing are in the continental part (Urbino 
> zone), quite far from Sardinia, so the several meters error is probably 
> due to these parameters.
> The test I did uses the following CRS:
> PROJCS["unnamed",
>    GEOGCS["unknown",
>        DATUM["unknown",
>           SPHEROID["Bessel 1841", 6377397.155, 
> 299.1528128,AUTHORITY["EPSG","7004"]]],
>               PRIMEM["Greenwich", 0.0],
>                UNIT["degree", 0.0174532925199433],
>                AXIS["Longitude", EAST],
>                AXIS["Latitude", NORTH]],
>                PROJECTION["Cassini_Soldner"],
>                PARAMETER["latitude_of_origin", 43.72513523],
>                PARAMETER["central_meridian", 12.63626379],
>                PARAMETER["false_easting", 0],
>                PARAMETER["false_northing", 0],                           
> UNIT["m", 1.0],
>               AXIS["x", EAST],
>               AXIS["y", NORTH]]
> 
> Do you think it's correct? The latitude of origin and central meridian 
> are the WGS84 coordinates of the Urbino reference point (got from 
> www.fiduciali.it).
> Mauro

The problem is not in the latitude of origin, but in the lack of the
TOWGS84 parameter. The projection library won't be able to make a proper
datum change.
If you look at the srs list of GeoServer you'll see that the Bessel
ellipsoid is used in many many projections, and each one has a different
set of TOWGS84 parameters depending on how the ellipsoid is 
located/rotated relative to the wgs84 one (which is geocentric).
You'll have to find the proper parameter for that area of Italy...
unfortuantely the EPSG database does not seem to contain any.

Here is a mail which contains some for Rome40, and seems to imply
the ones for the old Monte Mario 1875 won't be that different:
http://lists.maptools.org/pipermail/proj/2003-June/000789.html

Some more information on the topic is here:
http://grass-italia.como.polimi.it/dispense/georef_grass.pdf

Hope this helps
Cheers
Andrea