[R-sig-Geo] rasterize in parrallel

[Robert J. Hijmans]

Jerome,

rasterize is indeed a bit slow, it is #1 on the list of functions that
need a rewrite for speed. However, I think your assumption is wrong.
You can rasterize all fields in one step, and it is pretty quick for
the example data:

library(raster)
polyg <- shapefile("polyg.shp")
grain <- 50
nsc <- 4
r <- raster(polyg, res=c(grain, grain))

x <- rasterize(polyg, r)
# perhaps followed by
y <- deratify(x)


So clustering is inefficient here, but I think the below shows how you
would do athat, using much simplified code (whenever you think you
need a construction like "eval(parse(text=paste(", you are wrong).

library(raster)
polyg <- shapefile("polyg.shp")

library(doParallel)
cl <- makeCluster(4)
registerDoParallel(cl)
grain <- 50
nsc <- 4
r <- raster(polyg, res=c(grain, grain), crs=crs(polyg))

rr <- foreach(i= 1:nsc, .packages="raster", .combine=c) %dopar% {
     rasterize(polyg, r, field=paste0("sc", i))
}

stopCluster(cl)



Best, Robert

On Mon, Jul 13, 2015 at 3:03 PM, Jérome Mathieu <jerome.mathieu at upmc.fr> wrote:
> Dear all,
>
> I need to rasterize the same shapefile according to different fields stored
> in the @data of the shapefile. So I need to loop trough the fields (named
> sc1 to sc27) to rasterize upon each field. With a regular sequential loop
> it takes a very long time (much longer than in ArcMap), so I would like to
> parallelize it, but I didn't succeed so far.
>
> I tried :
>
> # read shapefile
>  library(maptools)
>  polyg<-readShapePoly("D:\\polyg.shp")
>  polyg at proj4string <- CRS("+init=epsg:2154")
>
> library(doParallel)
> cl <- makeCluster(4)
> registerDoParallel(cl)
>
> grain<-50
> nsc<-4
>
> foreach(idxsc = 1 : nsc , .packages="raster", polyg=polyg) %dopar% {
>
> eval(parse(text=paste("RRpolyg_",idxsc,"<-raster(polyg,res=c(grain,grain))",sep="")))
> # create raster template
> eval(parse(text=paste("projection(RRpolyg_",idxsc,") <-
> proj4string(polyg)",sep="")))      # projection
> eval(parse(text=paste("RRpolyg_",idxsc,"<-rasterize(polyg,RRpolyg_",idxsc,",field=polyg at data$sc",idxsc,")",sep="")))
> # rasterization
>
> }
>
> stopCluster(cl)
>
> and I get the following error :
>
> Error in { :
>   task 1 failed - "unable to find an inherited method for function 'raster'
> for signature '"numeric"'"
>
>
> the data are here
>
> http://we.tl/7qqegfLoBA
>
>
> Any help would be greatly appreciated !
>
> Jerome
>
>
>> sessionInfo()
> R version 3.1.3 (2015-03-09)
> Platform: x86_64-w64-mingw32/x64 (64-bit)
> Running under: Windows 7 x64 (build 7601) Service Pack 1
>
> locale:
> [1] LC_COLLATE=French_France.1252  LC_CTYPE=French_France.1252
> LC_MONETARY=French_France.1252 LC_NUMERIC=C
> LC_TIME=French_France.1252
>
> attached base packages:
> [1] parallel  stats     graphics  grDevices utils     datasets  methods
> base
>
> other attached packages:
>  [1] rgeos_0.3-11     rgdal_0.9-2      maptools_0.8-36  lattice_0.20-31
> foreign_0.8-63   raster_2.3-40    sp_1.1-0         doParallel_1.0.8
> iterators_1.0.7  foreach_1.4.2
>
> loaded via a namespace (and not attached):
> [1] codetools_0.2-11 compiler_3.1.3   grid_3.1.3       tools_3.1.3
>
>
>
>
>
>
>
> --
> Jérôme Mathieu
> Université Pierre & Marie Curie
> Institute of Ecology and Environmental Science (Paris)
>
> Bât. A - 7ème Etage, porte 715
> 7 quai St.-Bernard
> F-75252 Paris Cedex 05
>
> tel: 01 44 27 34 22
>
>         [[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