[GRASS-dev] [grass-code I][452] v.overlay misses some vector elemnts with "and" operator

Maciej Sieczka tutey at o2.pl
Tue Oct 23 14:22:22 EDT 2007

grass-dev at grass.itc.it wrote:

> I have a set of vectory
> polygons (call it "polygons_select") and one larger
> vector polygon which almost overlaps the previous ones
> (call it "eco_clipped_central_appalachian"). Now, I would
> like to cut A by B, so I obviously do a v.overlay with
> "and". Or as it looks in reality: " v.overlay
> ainput=polygons_select atype=area alayer=1
> binput=eco_clipped_central_appalachian btype=area
> blayer=1 output=polygons_current operator=and
> olayer=1,0,0 --overwrite " Now the resulting
> "polygons_current" is missing some polygons, which should
> obviously NOT be missing! See attached screenshots: -
> polygons_select.jpg is where I start -
> polygons_select_WITH_eco_clipped_central_appalachian.jpg
> shows the second polygon (in transparent yellow) which I
> would like to overlay with the first ones. -
> polygons_current.jpg shows with the red arrow, where one
> polygon is missing.

I've done some testing. I suspect the origin of the issue
might be a bug/feature in v.overlay that prevents it from
working reliably in lat/long locations. Please anybody
verify. Details follow.

I can reproduce the problem reported by Andras [1] with his
data [2], which are a lat/long location. Namely, the output of:

$ v.overlay ainput=polygons_select atype=area
binput=eco_clipped_central_appalachian btype=area
output=polygons_current operator=and

on my machine is a vector that has the same missing polygon
as on Andras' screendump [3].

Fiddling around I reprojected Andras' data into a metric
projection (the spearfish60 location), and the v.overlay
command as above worked perfect then. Not only the polygon
missing on [3] was there, but also the artifact described by
Andras in his another report [4] were not present.

In this case, can anybody tell if v.overlay is not supposed
to be used in lat/lon locations, or is it a bug that can be



More information about the grass-dev mailing list