[GRASSLIST:5341] Re: 2 prob.: v.proj and overlapping

[Francesco Mirabella]

Hi Radim and all,

I am currently working with ivan regarding the topic of this message:

Actually, we had some problems when reprojecting from UTM to
Gauss-Boaga as Ivan wrote before. The original map was Corinne landcover
2000 (.shp).
The problem first arised after reprojection to Gauss-Boaga so we decided
to go back to the originally imported map into UTM (corinne landcover
2000 is utm).

The original UTM corinne was displayed well also when queried 
(d.what.vect), but when looking into it in detail with v.digit we found
that a lot of boundaries were duplicated. It means that each vertex snapped
over another of the same boundary, closing where the boundary start (the
start and the end of the boundary is the same!!! and also all the other
nodes).
This brings to the creation of a lot of doubled areas and their centroids are positionated exaclty over the double boundary...

To try to solve this, we decided to try to delete the double areas and
tried v.clean tool=rmarea thresh=es.1(meter) in order to have the small double areas removed.

Now the problem is that v.clean was still removing areas on the map but the process is still running after 15 hours work on a pentium 4 processor, 512 Mb ram!

many thanks for your help

Francesco



On Thu, Jan 13, 2005 at 12:09:23PM +0100, Radim Blazek wrote:
> ivan marchesini wrote:
> >Dear users...
> >I've got two problems using grass57....
> >
> >the first one:
> >I've an area vector map in UTM ED50 system (imported from a shp map)...
> >is a complex map, with a lot of boundary segments and a lot of
> >centroids... but it is quite correct because all the areas seems to
> >answer correcly to v.what.vect... and all boundaries are two areas
> >boundaries... according to v.digit..
> >I've projected it on gauss-boaga roma40 system (an italian system) using
> >v.proj...
> >now only few areas are preserved...(in the gauss-boaga system I mean)
> >and there are a lot of not correct boundaries (no area or 1 area... as
> >v.digit explain...). Consequently, also areas are not respected ....
> >can you explain me the reason of this behaviour... and how to solve the
> >problem??? thank you
> 
> That is interesting. In theory, reprojection could maybe corrupt the 
> topology, but it is not probable for UTM and Gauss-Boaga on a small 
> area. Could you send me a small example (use v.in.region and v.select)
> including both PROJ_INFO?
> 
> Radim
>