[OSGeo Africa] QGIS and projection names
gavin at afrispatial.co.za
Thu Apr 17 14:19:04 PDT 2014
Hi Hanlie / Magriet
On 17/04/2014 14:03, Hanlie Pretorius wrote:
> I created a shapefile in QGIS and selected South African CRS:
> HBK_NO_23 as my CRS.
> When I open the file in QGIS, the metadata for the projection reads:
> USER:100001 - * Generated CRS (+proj=tmerc +lat_0=0 +lon_0=23 +k=1
> +x_0=0 +y_0=0 +ellps=WGS84 +towgs84=0,0,0,0,0,0,0 +units=m +no_defs)
> Furthermore, this 'user specified' CRS now exists in my list of
> projections. I find this rather annoying because I end up with a
> growing list of identical projection definitions.
QGIS will add a user specified definition whenever you open a file with
a definition that doesn't match one in the QGIS CRS database. HBK_NO_23
has '+axis=enu' which won't make a difference but is just not a close
enough match. Although if you created it in QGIS I imagine it should be
writing the +axis=enu out in the projection file. If this isn't behaving
as you'd expect I suggest you create an issue at http://hub.qgis.org/.
> The .prj file reads:
> and the .qpj file reads:
The definition of "degree" in the .prj and .qpj is different in your two
examples above. This should not be the case! It looks like this calls
for another issue in http://hub.qgis.org/ if they were both generated by
QGIS. I've just tested it and indeed QGIS is writing different degree
values in the qpj and prj files. afaik QGIS will then read the qpj in
future when you open that file, while other software will open the prj
> I'm just wondering if all these different CRS specifications may be
> causing problems when using the shapefile with other programs, such as
> ArcGIS. I am having problems with shapefiles looking correct in QGIS
> but shifted in ArcGIS. The shift is about 70 meters in the x direction
> and 255 meters in the y direction.
The definitions themselves are just different ways of saying the same
thing in WKT. I'm sure the error is coming in because of that difference
in "degree" value between the prj and qpj files.
> I found this web page (http://www.georeference.org/forum/t115330)
> saying that the difference between WGS84 and Hartbeeshoek 1994) is
> approximately 0.2 m and 0.3 m in Lo y and x coordinates respectively.
I don't know where that comes from and I have never seen any reference
that there is a shift between WGS84 and Hartebeesthoek94. The proj4
definition of Hartebeesthoek is 'towgs84=0,0,0,0,0,0,0' which means that
there's no transformation, i.e. coordinates on those datums are the same.
> I see in ArcGIS the transformation reads dx=0, dy=0, dz=0. So if the
> two datums are identical for most purposes, what else could be causing
> a shift in positions between the two programs?
same comments as above.
More information about the Africa