[R-sig-Geo] Cleaning small spatial polygons

[Roger Bivand]

On Wed, 4 Nov 2015, Roger Bivand wrote:

> On Wed, 4 Nov 2015, Eduardo Diez wrote:
>
>>  Is there any way of doing this or should i forget it and go on using GRASS
>>  through rgrass7?
>
> I suggest working with Emmanuel Blondel (cleangeo maintainer) to extend 
> cleangeo (also suggested in an earlier thread today, with apologies to 
> Emmanuel for picking on him!). That package already uses rgeos, and is a 
> logical place to put GRASS v.clean-like functionality (maybe even 
> encapsulating using GRASs via rgrass7 and a throwaway location).
>
> At the moment it is messy, though some rgeos internals do do something like 
> this, but it isn't exposed to users.

A possibility is to use:

set_RGEOS_polyThreshold(1e-2) # for example
set_RGEOS_warnSlivers(TRUE)

shows the remaining slivers, and:

set_RGEOS_dropSlivers(TRUE)

drops them when using for example:

> t1 <- gBuffer(<your_object>, byid=TRUE, width=0)

A buffer of zero width should not have side-effects, but your mileage may 
vary. It will only remove slivers, not dangles. I haven't tried it when it 
also finds a Polygons object under the threshold, which was your initial 
problem.

Roger

>
> Roger
>
>>
>>  Thanks
>>
>>  2015-10-19 15:03 GMT-03:00 Eduardo Diez <eduardodiez at gmx.com>:
>> 
>> >  Ok. So here's a link to a zip file that contains two shapefiles:
>> >  - pol_to_be_cleaned: the layer from which i'd like to remove small
>> >  polygons
>> >   - pol_cleaned: the layer cleaned with the function v.clean rmarea
>> > 
>> >  http://1drv.ms/1GmRWS7
>> > 
>> >  The threshold i used for cleaning was 3000 (meaning 3000 squared 
>> >  meters).
>> > 
>> >  Although i do project it before sending it to GRASS, according to the
>> >  official help page it should be able to handle it:
>> >  "Threshold must always be in square meters, also for latitude-longitude
>> >  locations or locations with units other than meters"
>> > 
>> >  Thanks
>> > 
>> >  2015-10-19 8:28 GMT-03:00 Roger Bivand <Roger.Bivand at nhh.no>:
>> > 
>> > >  On Fri, 16 Oct 2015, Eduardo Diez wrote:
>> > > 
>> > >  Dear list,
>> > > >  I'm willing to know if any knows a way of performing tha same thing 
>> > > >  i'm
>> > > >  doing through rgrass7 with GRASS when I execute the function v.clean 
>> > > >  with
>> > > >  "rmarea" as the tool argument. That is:
>> > > > 
>> > > >  "The rmarea tool removes all areas <= thresh. The longest boundary 
>> > > >  with
>> > > >  an
>> > > >  adjacent area is removed or all boundaries if there is no adjacent 
>> > > >  area.
>> > > >  Area categories are not combined when a small area is merged with a
>> > > >  larger
>> > > >  area."
>> > > > 
>> > > >  Basically i have raster of zones within a field. I convert it to
>> > > >  SpatialPolygonsDataFrame and in order to leave only the more
>> > > >  important/meaningful ones i remove the small/sliver with this tool. 
>> > > >  In
>> > > >  general it works fine but having to call an external software with a
>> > > >  specific version makes the script less portable and you have to be
>> > > >  careful
>> > > >  with updates and such. Also you have to write rasters and shapefiles 
>> > > >  back
>> > > >  and forth as GRASS can't work with in-memory objects.
>> > > > 
>> > > 
>> > >  Could you please provide an example of a built-in or contributed data 
>> > >  set
>> > >  (URL, not attachment) with the slivers you mention, so that we know 
>> > >  that we
>> > >  are addressing your problem? I don't think that:
>> > > 
>> > >  https://cran.r-project.org/web/packages/cleangeo/index.html
>> > > 
>> > >  does this, as it seems to try to repair broken geometries.
>> > > 
>> > >  Also note that you need to specify that the area threshold is in a 
>> > >  square
>> > >  planar metric - dropping slivers in unprojected geometries may be more
>> > >  complicated.
>> > > 
>> > >  Roger
>> > > 
>> > > 
>> > > >  Does someone know a way of doing this in plain R?
>> > > > 
>> > > >  Thanks
>> > > > 
>> > > >          [[alternative HTML version deleted]]
>> > > > 
>> > > >  _______________________________________________
>> > > >  R-sig-Geo mailing list
>> > > >  R-sig-Geo at r-project.org
>> > > >  https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>> > > > 
>> > > > 
>> > >  --
>> > >  Roger Bivand
>> > >  Department of Economics, Norwegian School of Economics,
>> > >  Helleveien 30, N-5045 Bergen, Norway.
>> > >  voice: +47 55 95 93 55; fax +47 55 95 91 00
>> > >  e-mail: Roger.Bivand at nhh.no
>> > > 
>> > > 
>> > 
>> 
>
>

-- 
Roger Bivand
Department of Economics, Norwegian School of Economics,
Helleveien 30, N-5045 Bergen, Norway.
voice: +47 55 95 93 55; fax +47 55 95 91 00
e-mail: Roger.Bivand at nhh.no