[GRASS-dev] v.voronoi producing wrong output

Wed Oct 14 14:20:24 PDT 2015

Thank you Maris for you reply.

I have tried the same with the 7.1svn version revision 66485 from 
OSGEO4W with no luck : still the same holes after v.voronoi.

I have managed to find a workaround which gives a hint on the problem : 
I can get the correct voronoi  output from a WGS84 location.
So it seems the problem comes from the other location in EPSG:27572

Below is the output of g.proj -p of the EPSG:27572 location. It seems ok 
to me.
-PROJ_INFO-------------------------------------------------
name       : Lambert Conformal Conic
proj       : lcc
ellps      : clark80IGN
lat_1      : 46.8
lat_0      : 46.8
lon_0      : 0
k_0        : 0.99987742
x_0        : 600000
y_0        : 2200000
towgs84    : -168,-60,320,0,0,0,0
pm         : paris
no_defs    : defined
-PROJ_EPSG-------------------------------------------------
epsg       : 27572
-PROJ_UNITS------------------------------------------------
unit       : meter
units      : meters
meters     : 1

I have the fealing it is related to some floating/precision problem, 
that is why I tried to play with the g.region resolution. But I just 
read the manual saying vector layers are not affected by this.

So I would like to try to snap all points to a 0.01 precision grid but 
do not know how to do this in GRASS. Could you please point me towards 
the right direction ? Do you have other ideas ?

Henri.

Le 14/10/2015 19:15, Maris Nartiss a écrit :
> Hello,
> I can not reproduce the issue on my ~AMD64 Gentoo system.
> GRASS versija: 7.1.svn
> GRASS SVN revision: 66489M
> Build date: 2015-10-14
> Build platform: x86_64-unknown-linux-gnu
>
> Try to install GRASS 7.1 nightly and report back if you still have
> problems with it.
> I don't see any recent changes to v.voronoi and valgrind also does not
> complain - will be hard to trace down if problem has not been solved
> in 7.1 by accident.
>
> Māris.
>
>
> 2015-10-14 15:51 GMT+03:00 Cev Hillan <cevhillan at hotmail.com>:
>> Hello all,
>>
>> As it is my first mail on this list, I would like to thank the devs and
>> the whole team for the great work done in GRASS.
>>
>> Before filing up a bug report I would like your advice on the following
>> problem : the output of v.voronoi is wrong (additionnal lines and/or
>> holes) for a set of around 4600 points in EPSG:27572 (in meters).
>>
>> These are the steps to reproduce the bug :
>> v.in.ogr --overwrite input=C:\XXXX\points.shp output=points
>> g.region -p vect=points n=n+100 s=s-100 e=e+100 w=w-100 res=0.01 -a
>> v.voronoi --overwrite --verbose input=points output=voronoi_areas
>>
>> The dataset is linked to the email, it is in EPSG:27572.
>> You will also find a picture of the (wrong) output.
>>
>> The output of voronoi command is :
>> Reading features...
>> WARNING: La couche vectorielle <voronoi_areas> existe déjà et va être écrasé
>> Utilisation du format natif
>> Voronoi triangulation for 4622 points...
>> Writing edges...
>> Writing features...
>> Writing attributes...
>> Searching for topology errors...
>> WARNING: Le nombre de centroïdes dépasse le nombre de surfaces : 4622 > 3478
>> WARNING: Nombre de contours incorrects: 5058
>> WARNING: Nombre de centroïdes en dehors de surfaces: 308
>> WARNING: Nombre de centroïdes dupliqués:   881
>> Cleaning output topology
>> Reading features...
>> Accrochage des sommets passe 1 : sélection des points
>> Accrochage des sommets passe 2 : assignation des ancrages de sommets
>> Accrochage des sommets passe 3 : accrochage sur les points assignés
>> Sommets accrochés : 663
>> Nouveau sommet : 862
>> Intersections : 54331
>> Removed duplicates: 491
>> Modifications: 0
>> Removing dangles...
>> Removed lignes : 3
>> Removed dangles: 3
>> Removing bridges...
>> Removed lines: 0
>> Removed bridges: 0
>> Construction de la topologie pour la carte vectorielle
>> <voronoi_areas at Test>...
>> Enregistrement des primitives ...
>> 72112 primitives registered
>> 139602 vertices registered
>> Construction des surfaces ...
>> 31830 areas built
>> One isle built
>> Attachement des îles ...
>> Attachement des centroïdes ...
>> La topologie a bien été construite
>> Nombre de nœuds: 35661
>> Nombre de primitives: 72112
>> Nombre de points: 0
>> Nombre de lignes :0
>> Nombre de contours: 67490
>> Nombre de centroïdes: 4622
>> Nombre de surfaces: 31830
>> Nombre d'îles: 1
>> WARNING: Nombre de centroïdes dupliqués:   114
>> v.voronoi terminé.
>>
>> This is the output of g.version -breg :
>> version=7.0.1
>> date=2015
>> revision=65800
>> build_date=2015-07-31
>> build_platform=i686-pc-mingw32
>>      ./configure  '--with-libs=/c/OSGeo4W/lib /c/OSGeo4W/apps/msys/lib
>> /c/osgeo4w/usr/src/grass701/mswindows/osgeo4w/lib'
>> '--with-includes=/c/OSGeo4W/include /c/OSGeo4W/apps/msys/include'
>> --libexecdir=/c/OSGeo4W/bin --prefix=/c/OSGeo4W/apps/grass
>> --bindir=/c/OSGeo4W/bin --includedir=/c/OSGeo4W/include --disable-x
>> --without-x --with-cxx --enable-shared --enable-largefile
>> --with-opengl=windows --with-fftw --with-freetype
>> --with-proj-share=/c/OSGeo4W/share/proj
>> --with-gdal=/c/osgeo4w/usr/src/grass701/mswindows/osgeo4w/gdal-config
>> --with-geos=/c/osgeo4w/usr/src/grass701/mswindows/osgeo4w/geos-config
>> --with-liblas=/c/osgeo4w/usr/src/grass701/mswindows/osgeo4w/liblas-config --with-sqlite
>> --with-curses --with-regex --with-nls
>> --with-freetype-includes=/c/OSGeo4W/include/freetype2 --with-odbc
>> --with-cairo --with-cairo-includes=/c/OSGeo4W/include/cairo --with-postgres
>> libgis_revision=64733
>> libgis_date="2015-02-25 01:56:29 +0100 (st, 25 2 2015) "
>> proj4=4.8.0
>> gdal=1.11.2
>> geos=3.4.2
>> sqlite=3.7.17
>>
>> I also tried on a Unix machine on which I built GRASS with the same results.
>>
>> I tried to get a small subset of points and got down to 8 points
>> producing a wrong output. Point subset and output image is also linked
>> to this email.
>> I searched existing bugs and tried to increase resolution with g.region
>> with no improvement;
>>
>> Do you have any ideas ? Should I file in a bug or am I using the tools
>> in the wrong way ?
>>
>> Thanks for your thoughts,
>> Cev.
>>
>>
>>
>> _______________________________________________
>> grass-dev mailing list
>> grass-dev at lists.osgeo.org
>> http://lists.osgeo.org/mailman/listinfo/grass-dev