Looping 'raster' package functions over each multi-polygon feature in R - r

Question

I am trying to loop several functions from the 'raster' package, namely, crop(), mask(), reclassify() and unstack()/as.list(). I have ten raster layers that share the same extent and data type; they correspond to land cover over 10 time-points. I want to create individual list variables for each output of crop() -> mask() -> reclassify() -> as.list(). I was able to pipe the process for 1 polygon feature, but I need to be able to loop it for each of the 10 polygon features stored in the multipolygon Shapefile, such that I can save each output list according to a specified naming convention.
Thank you and please advise. I share my code below.
EDIT: I am wondering if a for-loop is the right way to go about this, or would an lapply approach be better?
# Load libraries
library(raster) # for raster processing
library(rgdal) # for raster/vector processing
library(sf) # for Shapefile processing
# Stack 10 rasters together
raster.stack = stack(
raster("path/raster1.tif"),
raster("path/raster2.tif"),
raster("path/raster3.tif"),
raster("path/raster4.tif"),
raster("path/raster5.tif"),
raster("path/raster6.tif"),
raster("path/raster7.tif"),
raster("path/raster8.tif"),
raster("path/raster9.tif"),
raster("path/raster10.tif")
)
# Prepare reclassification codes from 9-class raster to 3-class raster
reclasscodes = c(
0,0, # no data
1,1,
2,1,
3,1,
4,1,
5,2,
6,2,
7,3,
8,3,
9,3
)
# Convert reclass codes list into n x 2 matrix
reclassmatrix = matrix(reclasscodes, ncol=2, byrow = T)
# Load multipolygon vector Shapefile
multipolygon = shapefile("path/multipolygon.shp") # Shapefile is made of n polygons
# Example subset Shapefile to polygon_1 using attribute "ID"
polygon_1 = subset(multipolygon,ID=="D-4")
# Create output for polygon_1
list_polygon_1 =
raster.stack %>%
crop(y = polygon_1) %>% # crop to bounds
mask(mask = polygon_1) %>% # mask to polygon cutline
reclassify(rcl = reclassmatrix) %>% # reclassify to 3-class
as.list() # functions the same as unstack() where raster brick is converted to list of raster layers
# I use %>% because I do not want to save any of the intermediate outputs.
# Resultant output is a variable list for polygon_1 named 'list_polygon_1' which is exactly what I want.
# Worked perfectly.
# How do I repeat this process for polygon_1 to polygon n?
# My attempt
for (i in 1:nrow(multipolygon)) {
raster.stack %>%
crop(y = multipolygon[i,]) %>%
mask(mask = multipolygon[i,]) %>%
reclassify(rcl = reclassmatrix) %>%
as.list() %>% # up till here it is the same steps as before for polygon_1
# now I want to save each list output as a separate variable according to i, e.g. list_polygon_2, list_polygon_3 etc.
assign(paste(multipolygon$ID, i, sep = '_')) # assign a naming convention for each output variable
}
# Does not work. Even without the last line of code "..assign(paste(...))" there is no output variable from the as.list() line.

Answer 1

Here is a minimal self-contained reproducible example.
Example data
library(raster)
s <- stack(system.file("external/rlogo.grd", package="raster"))
xy1 <- xy2 <- xy3 <- matrix(c(10,17, 6,10,71,60,62,71), ncol=2)
xy2[,1] <- xy2[,1] + 30
xy3[,2] <- xy3[,2] - 30
p <- spPolygons(xy1, xy2, xy3)
#plot(r, 1)
#lines(p)
What you are after
rm = matrix(c(0,100,0,100,150,2,150,255,3), ncol=3, byrow=TRUE)
out <- list()
for (i in 1:length(p)) {
x <- crop(s, p[i,])
x <- mask(x, p[i,])
out[[i]] <- reclassify(x, rm)
}
What you are saying about unstack does not make sense (and unlist does not work). I would advise against it, but you could do
out2 <- lapply(out, unstack)
I am not sure what you are really after. If you want the cell values you can make it much simpler (no need for a loop) and do
r <- reclassify(s, rm)
e <- extract(r, p)
To your question about lapply vs a loop. In terms of performance that rarely matters. lapply can be concise, but in cases like this, writing a loop is better as it is easier to read, and write, especially if you do not use %>%.