[GRASSLIST:3814] Re: overlay

[Radim Blazek]

It is not so bad as I thought first (I hope). 
It is known (for me) missing cleaning tool in v.overlay/library.

The problem: 
Topology rules require that 2 boundaries at the node don't have 
the same angle. It may happen, that the angle (float) for 
2 boundaries is identical beacause of representation error 
(limited number of decimal digits) even if lines are not identical.
The risk is obviously higher if the maps are similar.

The solution for today:
v.patch input=a,b output=o1
v.clean input=o1 output=o2 tool=snap,break,rmdupl,rmdac thresh=0.1
v.category input=o2 output=o3 option=del
v.category input=o3 output=o4 type=area,centroid
echo "create table o4 (cat int, a int, b int)" | db.execute
v.db.connect map=o4 driver=dbf database='$GISDBASE/$LOCATION_NAME/$MAPSET/dbf/' table=o4
v.to.db map=o4 option=cat
v.distance from=o4 to=a from_type=centroid to_type=area max=0 upload=cat column=a
v.distance from=o4 to=b from_type=centroid to_type=area max=0 upload=cat column=b

and extract from o4, but the last v.distance unfortunately crashes.

Radim


On Tuesday 06 July 2004 15:50, leo.lami at katamail.com wrote:
>  Hi,
>  I have a problem with th v.overlay command.
>
>  I have two vectors (imported by OGR from shapes) that have several
>  boundary completely overlapped because one derive from the other.
>  If I use v.overlay with "and" option I have a vector that is not the
>  correct intersection of them.
>
>  How you can see from the png file linked, same polygons of the first
>  vector (with one boundary overlapped with the boundary of the other
> vector) are not included in the intersection vector (bottom of the
> picture). In the oter case you can see that, in the same situation, I have
> the polygons included in the intersection vector (top of the picture). The
> first vector has black line and no filled area, the second has red line and
> not filled area and the intersection vector has green line and yellow
> filled area.
>
>   Someone had the some problem or does someone know the cause?
>
>   Thank you
>   Leonardo Lami