[R-sig-Geo] assigning raster cell values based on predefined criteria

Robert J. Hijmans r.hijmans at gmail.com
Thu Apr 22 19:21:34 CEST 2010


I suspect this is because the raster you are using has a
longitude/latitude CRS. This is the default if you create it using
raster()  -- I may change that.
Set it to NA (or to what it really is)

projection(r) <-  NA

And try again...

Robert


The picture suggest that you are not computing the distance to a
single point, but rather to a checkerboard (for one 'color' the black
or the white fields) like distributed set of points.

2010/4/22 Roman Luštrik <roman.lustrik at gmail.com>:
> This is the normal behavior (see image here) when I use option A-ish
> (calculating distance to a single point).
>
> Intuitively, however, I would expect the plot to look something like this
> (crudely drawn for your amusement). Can someone explain the logic behind the
> "tiled" pattern?
>
> They say a picture is worth a thousand words - hopefully, this is the case
> here and my question is clear.
>
>
>
> Cheers,
> Roman
>
>
> On Tue, Apr 20, 2010 at 6:55 PM, Robert J. Hijmans <r.hijmans at gmail.com>
> wrote:
>>
>> Dear Roman,
>>
>> I think it depends on your data. Below are two examples (both assuming
>> your polygons do not overlap).
>>
>> A) works if each cell in a polygon is closest to its own centroid (and
>> not that of another polygon). It makes a distance surface with
>> distanceFromPoints() and then uses mask() to mask out the areas not
>> covered by polygons.
>>
>> B) uses linesToRaster() to set up a raster with impenetrable
>> boundaries (the polygon outlines) and then use gridDistance()  (i.e.
>> the distance when only allowed to traverse certain cells). This method
>> would fail if you have a polygon representing several islands; or a
>> polygon with a hole with the centroid inside that hole. Also, you do
>> not have values for the outer cells of a polygon. This is perhaps OK
>> when polygons are adjacent (whose cell is it anyway?). If they are not
>> adjacent yo could use edge() with 'outer' argument on the rasterized
>> polygons instead of linesToRaster
>>
>> library(raster)
>> p1 <- rbind(c(-180,-20), c(-140,55), c(-20, 0), c(-140,-60), c(-180,-20))
>> hole <- rbind(c(-150,-20), c(-100,-10), c(-110,20), c(-150,-20))
>> p2 <- rbind(c(-10,0), c(140,60), c(160,0), c(140,-55), c(-10,0))
>> pols <- SpatialPolygons( list(  Polygons(list(Polygon(p1)), 1),
>> Polygons(list(Polygon(p2)), 2)))
>> xy <- coordinates(pols)
>>
>> #A
>> r <- raster(nrow=18, ncol=36)
>> r <- polygonsToRaster(pols, r)
>> d <- distanceFromPoints(r, xy)
>> x <- mask(d, r)
>> plot(x)
>>
>> #B
>> r <- raster(nrow=18, ncol=36)
>> r <- polygonsToRaster(pols, r)
>> l <- linesToRaster(pols, r)
>> l[l>0] <- 0
>> p <- pointsToRaster(r, xy)
>> pp <- cover(p, l)
>> gd <- gridDistance(pp)
>> gd[!is.na(l)] <- NA
>> plot(gd)
>>
>>
>> Robert
>>
>> On Tue, Apr 20, 2010 at 12:23 AM, Roman Luštrik <roman.lustrik at gmail.com>
>> wrote:
>> > Dear list,
>> >
>> > I have a polygon that I converted to raster (kudos to
>> > polygonsToRaster()). I
>> > would like to assign values (through a custom function) to cells within
>> > this
>> > polygon (now as raster). Roughly speaking, cell value will depend on its
>> > distance from the central cell (polygon centroid). How would you go
>> > about
>> > doing this? Is there a more efficient way of doing this, or will I have
>> > to
>> > hard code it, cell by cell?
>> >
>> > Cheers,
>> > Roman
>> >
>> >        [[alternative HTML version deleted]]
>> >
>> > _______________________________________________
>> > R-sig-Geo mailing list
>> > R-sig-Geo at stat.math.ethz.ch
>> > https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>> >
>
>
>
> --
> In God we trust, all others bring data.
>



More information about the R-sig-Geo mailing list