CODEWITHSUNDEEP
rraster
J. Win.
J. Win.

Reputation: 6761

Changing a single cell value in a large raster

I've got a rasterbrick, with layers of climate data on islands, surrounded by NA ocean cells. I'd like to set a few of the ocean cells to the value of the coast nearby. Having a surprising amount of trouble doing this. In my mind the code should be something like below, but I run out of memory. The rasters are large approx 2500x2500.

# dummy stack with three layers
slogo <- stack(system.file("external/rlogo.grd", package="raster")) 
# let's say cell 5 is the one I want to change
newvals <- 1000*(1:3)
for(i in 1:nlayers(slogo)) slogo[[i]][5] <- newvals[i]

Subsequent step will be writing the rasters to separate ascii files; so an alternative hacky solution might be snipping into the text file and replacing single values...

EDIT: Perhaps there's a solution someone can recommend by resetting memory allocation limits? Here's the error message I show after running on my large raster. Error: cannot allocate vector of size 504.1 Mb In addition: Warning messages: 1: In readBin(raster@file@con, what = dtype, n = nc, dsize, dsign, : Reached total allocation of 3979Mb: see help(memory.size) 2: In readBin(raster@file@con, what = dtype, n = nc, dsize, dsign, : Reached total allocation of 3979Mb: see help(memory.size) 3: closing unused connection 4 (C:\Users\jcw\AppData\Local\Temp\R_raster_jcw\r_tmp_2015-08-16_130255_2000_11926.gri) 4: In getBilData(object, r = startrow, nrows = nrows, c = startcol, : Reached total allocation of 3979Mb: see help(memory.size) 5: In getBilData(object, r = startrow, nrows = nrows, c = startcol, : Reached total allocation of 3979Mb: see help(memory.size) 6: In getBilData(object, r = startrow, nrows = nrows, c = startcol, : Reached total allocation of 3979Mb: see help(memory.size) 7: In getBilData(object, r = startrow, nrows = nrows, c = startcol, : Reached total allocation of 3979Mb: see help(memory.size)

Upvotes: 2

Views: 992

Answers (2)

Robert Hijmans
Robert Hijmans

Reputation: 47071

You can use the raster::update function to change values in raster files on disk (do make backups!)

Upvotes: 0

Forrest R. Stevens
Forrest R. Stevens

Reputation: 3485

You were pretty close, it's just an indexing issue:

library(raster)

# dummy stack with three layers
slogo <- brick(system.file("external/rlogo.grd", package="raster")) 

##  Check values of all three bands at cell 5:
slogo[5]

Yields:

     red green blue
[1,] 255   255  255

##  Set values of all three bands to zero:
slogo[5][1:3] <- 0

slogo[5]

Yields:

     red green blue
[1,]   0     0    0

An alternative that might get the OP past the memory issues of processing all of the bands at once:

##  Alternatively, load each band into memory separately, do the
##    replacement and write each band back to disk as a separate GeoTiff:
for (i in 1:nbands(slogo)) {
    r <- raster(slogo, layer=i)
    r[5] <- 0
    writeRaster(r, file=paste0("band_", i, ".tif"), format="GTiff")
}

Upvotes: 1

Related Questions