[gdal-dev] Gauss conformal Lo33 on Clarke 1880 Cape Datum to WGS84 UTM36S

Tue Aug 28 03:07:50 PDT 2012

Hi Gdal-ers,
This is cross-posted on the proj list, but my query also concerns 
whether gdalwarp notices the NTv2 grid or not, as my results vary from 
the control data (point BM1) I was given.
I'm using gdal/ogr v1.9.1 on ubuntu 12.04.
Please read on, and comment:

Hi All,
This issue goes on, slower than a snail.
I have finally managed to get some source and target points from the 
original surveyor.

BM1

2872274.540m

49717.410m

7128571.659m

450273.237m

4275103.524m

I load my source image that is on Lo33 Clarke 1880 into QGIS and hover 
the mouse over BM1, a nicely painted white target.
I get:  -49717.33  -2872272.46  which is good on the eastings, but my 
source image is 2m north of this value from the surveyor. For the 
moment, so what.

I then run:
ogr2ogr -f "DGN" -dsco "3D=YES" -s_srs "+proj=tmerc +lon_0=33 
+a=6378249.145 +b=6356514.966398753 +nadgrids=./SA.gsb" -t_srs 
"+init=epsg:32736" Cl_Revised_v7_36s.dgn    Cl_Revised_v7.dgn
gdalwarp                        -s_srs "+proj=tmerc +lon_0=33 
+a=6378249.145 +b=6356514.966398753 +nadgrids=./SA.gsb" -t_srs 
"+init=epsg:32736" images.vrt               images_36s.tif

My UTM 36S WGS84 coordinate for BM1 is now:
450288.88  and  7128578.76  and this puts my transformed image 15m to 
the east and 7m to the north of the Surveyors value.

My question is:
Did I specify the s_srs and t_srs parameters correctly?
Can I be certain that the SA.gsb grid was used?

I have no idea how the original surveyor got his source and target 
values for BM1, as I have no access to this person.
Regards,
Zoltan

>>
>> I have no idea how big, or what extents are in these NTv2 grid files, as
>> I have never used them before.
>>
>> To this end, my area of interest is in Maputo (Mozambique), and
>> Mozambique apparently uses Clarke 1866 (not 1880) - but the surveyor
>> "swears blind" that the coords are on "1880" and not "1866".
>>
>> This begs the question:
>> Do the extents of this Clarke 1880 NTv2 grid stop at the South African
>> border (some 50Km west of my area of interest), or do they cover Maputo
>> sufficiently?
>>
Yes, they go up to 33E which just covers Maputo, and when I load the 
grid plus my data into QGIS, my data is inside the grid extents.
>>
>>
>>
>> On 2012/07/26 21:42, Mikael Rittri wrote:
>> > Hello Zoltan.
>> > Yes, as you say, the traditional transverse mercator projections 
>> for Southern Africa have
>> > axes positive west and south. This can be handled in Proj 4.8.0 
>> (see http://trac.osgeo.org/proj/wiki/TMSO ).
>> > However, since both your example coordinates are negative, I think 
>> you have data in a less traditional
>> > but more GIS-friendly variant, where the axes go east and north and 
>> the traditional coordinate values
>> > have been negated.
>> >
>> > Another problem is caused by the Cape datum having distortions that 
>> are not handled well
>> > by a simple Helmert datum shift ( +towgs84 in Proj.4). Such a datum 
>> shift will have errors up to 15 meters,
>> > perhaps more. You can get much better accuracy from a grid shift 
>> file. So I think your source CRS
>> > is best constructed as
>> >
>> > +proj=tmerc +lon_0=33 +a=6378249.145 +b=6356514.966398753 
>> +nadgrids=SA.gsb
>> >
>> > where SA.gsb is an NTv2 grid shift file from Cape to 
>> Hartebeesthoek94 (or to WGS84),
>> > covering South Africa. To find out where you can download it, see
>> >
>> > 
>> http://eepublishers.co.za/article/datum-transformations-using-the-ntv2-grid.html
>> >
>> > About your target CRS, I don't think there is an EPSG code for the 
>> combination of Hartebeesthoek94 and
>> > a UTM projection. But since H94 is the same as WGS84 except for 
>> tectonic motion, I think you
>> > can use WGS84 / UTM zone 36S, which can be defined as
>> >
>> >Â Â Â Â  +init=epsg:32736
>> >
>> > Hope this helps,
>> >
>> > Mikael Rittri
>> > Carmenta
>> > Sweden
>> > http://www.carmenta.com
>> >
>> > 26 jul 2012 kl. 17:14 skrev "Zoltan Szecsei" 
>> <zoltans at geograph.co.za <mailto:zoltans at geograph.co.za>>:
>> >
>> >> Hi Everyone,
>> >> I have some orthos and some 3D vector data that I need to 
>> reproject from
>> >> Clarke 1880 Cape Datum, Lo33 to WGS84 Hartebeeshoek94 datum UTM 36S.
>> >>
>> >>
>>
>

-- 

===========================================
Zoltan Szecsei PrGISc [PGP0031]
Geograph (Pty) Ltd.
P.O. Box 7, Muizenberg 7950, South Africa.

65 Main Road, Muizenberg 7945
Western Cape, South Africa.

34Â° 6'16.35"S 18Â°28'5.62"E

Tel: +27-21-7884897  Mobile: +27-83-6004028
Fax: +27-86-6115323www.geograph.co.za
===========================================

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/gdal-dev/attachments/20120828/dfaa9b41/attachment.html>
-------------- next part --------------
_______________________________________________
Proj mailing list
Proj at lists.maptools.org
http://lists.maptools.org/mailman/listinfo/proj