Rasters

Rasters are highly memory intensive in R's already fragile memory space. However, one does not necessarily need to read a raster into memory each time. In fact, it is highly frowned upon to read in a raster in that way as it just eats up so much memory and is computationally intensive.

stack()'s beauty

It's much better to make a stack of rasters as it does not load them into memory, but opens their file location.

library(raster) A_Files <- list.files(pattern="A.tif") # files A All_A <- stack(unique(A_Files)) B_Files <- list.files(pattern="B") # files B All_B <- stack(unique(B_Files)) 

Notice, all of the elements are now in a "vectorized" form.

overlay() to calculate

From here, we can use overlay() to perform a calculation. Note, since the number of layers is equivalent between (ALL_A) and (ALL_B) we do not need to worry about an element being repeated.

x <- overlay(All_A, All_B, fun=function(x,y) x*y ) 

Export

The only downside to this approach is you will have to go through a loop to export the answers:

out <- unstack(x) outputnames <- paste0(seq_along(out), "_result.tif") for(i in seq_along(out)){ writeRaster(out[[i]], file=outputnames[i]) } 

References

For multiplying stacks see:

• overlay
• calc

For more information on stacks (see bricks as well)

• stack
• brick

Finally I saved some time using the lapply or the map function.

Here is the code with lapply:

require(raster) # raster package setwd("C:/Rasters") # set working directory A_Files <- list.files(pattern="A.tif") # files A All_A <- stack(unique(A_Files)) # stacking As B_Files <- list.files(pattern="B") # files B All_B <- stack(unique(B_Files)) # stacking Bs As <- lapply(All_A, raster) # rasters from A Bs <- lapply(All_B, raster) # rasters from B # multiplying A*B: Multiply_A_B <- lapply(seq_along(As), FUN = function(i) As[[i]] * Bs[[i]]) # saving results as "_result.tif" lapply(seq_along(Multiply_A_B), function(x) { writeRaster(Multiply_A_B[[x]], paste(All_A[x], "_result.tif", sep = ""), datatype = 'FLT4S', overwrite = TRUE) }) 

Code with the map function:

# multiplying A*B: Multiply <- Map("*", As, Bs) # the rest is the same as with the lapply example above 

The operation took 210 seconds with for loop, while with the lapply it took 190 seconds and with map it took 185 seconds.