[Ottawa_users] New Brunswick User looking for help with GRASS and Quantum GIS - may be of interest to new users

[Scott Mitchell]

Dave Sampson has done an excellent job at providing advice for much  
of your message, so I don't have a whole lot to add.  A couple  
points, though:

On 22-Jan-08, at 16:48 , Alyre Chiasson wrote:

> My first concerns are the “unkown”s and the second is that the  
> projection is listed as “Oblique Sterographic” and not “Double”.  
> Maybe “Oblique” is actually a “Stero” projection? Somewhat  
> disconcerting for someone starting out. I confirmed the EPSG code  
> from: http://spatialreference.org/ref/epsg/2953/
>

I haven't used these specific projections, so don't have a lot to  
say, except that presumably if the EPSG code does properly match the  
projection used, OGR / PROJ should be able to take care of all the  
details?  Others on the list, and the OGR list, as Dave points out,  
are going to be much more authoritative than me on this, though.

>
> I assume the dbf file gets imported as well. What happens to the  
> labels which are in the dbf file? In the original shapefiles from  
> Environment Canada the labels are missing from the upper part of  
> New Brunswick because in the original they are over Nova Scotia  
> (image1). Will the labels be readjusted when the original file gets  
> clipped? I can’t fine anything on how label placement is  
> determined. However, the first hurdle is getting the basic maps  
> into GRASS because I don’t think Quantum can do the above.
>

Dave has addressed the fact that yes, the whole "layer" (in ArcGIS- 
speak) including the .dbf of attributes comes across in the  
translation.  But the "labels" should not be thought of as the actual  
data.  Think of the spatial entities being in one database (the  
shapefile and its friends), and the coordinates of every object  
(point, line, or area) are in that database, and there's a common ID  
value that allows attribute data in the .dbf to be associated with  
each of those objects.  There is no separate "location of the label"  
- there is just a spatial object with associated geographic  
coordinates, which is linked to various attributes, and these  
attributes can be visualized with different symbology.  So when you  
transform the coordinates in the shapefile, there is only the one set  
of coordinates involved.

In the example of point measurements, the shapefile of points would  
be transformed from one coordinate to another.  Each of these points  
also has an ID, which allows it to be joined to a table of  
attributes.  ONE option for visualizing those points is to map out  
labels.  The position of the label is a combination of the location  
of the point, plus some set of instructions to the specific GIS  
software you're using about how to display that label with reference  
to the point.  Those instructions are specific to whatever software  
you're using.  Pg 265 in the GRASS book (3rd ed.) talks a little bit  
about this, but it would be different for QGIS, because label  
placement is a cartographic question, not inherent to the spatial data.

In your example, if your layer that is driving the label drawing is  
transformed properly, the points should always be in the right location.

I hope that helps, I'm not completely sure I've understood your  
problem correctly.


> I don’t know if GRASS can do the nearest neighbour isotherm type of  
> calculation I have proposed above with some king of script but I  
> will be doing more reading or you can suggest where I might look.  
> Unless Quantum would be an easier tool to do the above, I wanted to  
> use it mainly as a viewer for the resulting product.

Perhaps an interpolation approach would be useful?  See section 6.8  
in the GRASS book.

For any kind of spatial processing work like this, the possibilities  
in QGIS are really just QGIS calling GRASS modules, so really it's  
GRASS doing the "work" and QGIS visualizing it anyway.

>

Cheers,
Scott Mitchell
Carleton University