[R-sig-Geo] Replacing values in large rasters using conditional statements

[Gregovich, Dave P (DFG)]

Well, This may be somewhat remedial for some, but it is a voyage of discovery for me:

It appears that using the 'overlay' function is the analogous method I was looking for. I am not absolutely positive it works for *very* large rasters.
But it does seem to work with the 5 X 10^8-cell rasters I am currently working with:


>Hello,

>I often replace values in a raster based on values of another raster, like this:



>n.row <- n.col <- 100

>a <-raster(nrows=n.row,ncols=n.col)

>b <- a

>a[]<-rnorm(ncell(a))

>b[]<-rnorm(ncell(b))

>a[b > 0] <- 1



>What would analogous code be that works for large rasters without running into vector size allocation problems?

#First method as illustrated above.
n.row <- n.col <- 1000
a <-raster(nrows=n.row,ncols=n.col)
b <- a
a[]<-rnorm(ncell(a))
b[]<-rnorm(ncell(b))
out.1 <- a
out.1[b > 0] <- 1

#second method using 'overlay'...
out.2 <- overlay(a,b,fun=function(x,y){
                                                z <- x
                                                z[y > 0] <- 1
                                                return(z)
                                })

#check the two methods are identical
all(getValues(out.1) == getValues(out.2))

My initial indications are that 'overlay' is working really well with large, stored-on-disk rasters, and doesn't seem *too* slow
I hope this is helpful to some of you.

Thanks kindly.

__________________________________
Dave Gregovich
Research Analyst
Alaska Department of Fish and Game
Wildlife Conservation Division
Douglas, AK 99821
(907) 465-4291
dave.gregovich at alaska.gov
__________________________________


	[[alternative HTML version deleted]]