[R-sig-Geo] convert from spatstat to sp; RFC about generic behaviour

[Robert J. Hijmans]

Dear Edzer,

In raster the default approach for extracting raster values with
polygons is that a polygon has to cover the center of a cell, for the
cell to be included. However, you can use argument "weights=TRUE" in
which case you get, apart from the cell values, the percentage of each
cell that is covered by the polygon, so that you can apply, e.g., a
50% area covered threshold, or compute an area-weighted average.

In the case of lines, any cell that is crossed by a line is included.
For lines and points, a cell that is only 'touched' is included when
it is below or to the right (or both) of the line segment/point
(except for the bottom row and right most column).

I agree that this is very elegant ( I do not find it very intuitive,
but that might change when I get used to it! ) For its elegance I have
added it to raster (v 1.6-9) ( also in the spirit of, where possible,
trying to have similar approaches in spatial packages that Jeremy
recently called for
http://www.mail-archive.com/r-sig-geo@stat.math.ethz.ch/msg09015.html
). I have also added raster[polygons] <- value  , a wrapper around
polygonsToRaster(raster, value, update=TRUE) )

Robert



On Sat, Oct 30, 2010 at 3:24 AM, Edzer Pebesma
<edzer.pebesma at uni-muenster.de> wrote:
> I like that solution; it seems intuitive, and as this seems to be a FAQ,
> I would like to copy this behaviour to sp classes, such that
>
> obj[geom,] # in case there are attribute data in obj, or
>
> obj[geom] # in case obj has only geometry
>
> return the items in obj that are within the geometry of geom. This would
> liberate us from selecting the non-NA indexes returned from overlay (or
> in case of a grid, assign NA to everything outside).
>
> A problem occurs when obj is pixels or a grid (or raster or im), and
> geom lines or polygons, I can see two difficult cases:
>
> - the grid cell centre does not, but part of the grid cell does overlap
> with geom or cross geom,
> - the grid cell touches geom, but does not overlap / cross.
>
> Robert, how does raster deal with these two cases? Does it have fixed
> behaviour, or can the user choose?
>
> Rolf, how does spatstat deal with these two cases?
>
> On 10/30/2010 12:23 AM, Turner Rolf wrote:
>>
>> ________________________________________
>> From: r-sig-geo-bounces at stat.math.ethz.ch [r-sig-geo-bounces at stat.math.ethz.ch] On Behalf Of Peter Larson [pslarson2 at gmail.com]
>> Sent: Saturday, 30 October 2010 8:28 a.m.
>> To: r-sig-geo at stat.math.ethz.ch
>> Subject: [R-sig-Geo] convert from spatstat to sp
>>
>> Hello all,
>>
>> Is there any way to convert from spatstat to sp?
>>
>> Specifically, I am looking to clip a density plot produced in spatstat
>> to include only points within a polygon boundary in a shapefile. if
>> there is a way to clip in spatstat, this would also be acceptable.
>>
>> Clipping in spatstat is done simply using [] --- let X be your density plot (image; object of class "im")
>> and let W be an object of class owin determined by the polygonal boundary that you have in mind.
>>
>> Just do
>>
>>     Y <- X[W]
>>
>> to get X clipped to the window W.  To find out how to make ``a polygon boundary in a shapefile'' into
>> a window, see the vignette on handling shapefiles in spatstat:
>>
>>     vignette("shapefiles")
>>
>>         cheers,
>>
>>               Rolf Turner
>> _______________________________________________
>> R-sig-Geo mailing list
>> R-sig-Geo at stat.math.ethz.ch
>> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>
> --
> Edzer Pebesma
> Institute for Geoinformatics (ifgi), University of Münster
> Weseler Straße 253, 48151 Münster, Germany. Phone: +49 251
> 8333081, Fax: +49 251 8339763  http://ifgi.uni-muenster.de
> http://www.52north.org/geostatistics      e.pebesma at wwu.de
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>