How to iterate through directories to process raster files without a loop with the apply family in R

I have raster files (.tif) inside different folders (those are my models). I need to import them into individual objects and process them for creating an array (I'll use "abind" for that).

I do have a large number of models thus creating a huge array dataset. I've manage to create a function for reading and processing the models at once. The catch is that maybe is not such a good idea growing an array inside a for loop due to the complexity of the dataset, which could incore in some error.

How to build a function for that making use of the apply R's family functions?

Here is the function I've made

require(raster)
require(abind)
require(rgdal)

myFunction <- function (x)
{
directories <- list.dirs( x, full.names = TRUE)[-1]
e <- extent(xmin, xmax, ymin, ymax)
rcp <- NULL

for (i in 1:length(directories))
{
models_raw <- stack(list.files(directories[i],pattern = ".tif$", full.names = TRUE))
models_e <- crop( models_raw , e )
val <- values (models_e)
coord <- xyFromCell(models_e, 1:ncell(models_e))
models <- cbind(coord, val)
models <- na.omit(models)
rcp <- abind (rcp, models, along = 3)
}

return(rcp)
}

scenario <- myFunction( x = ".//data//models//")


1 Answer
1

Simply generalize your processing into a defined function with directory as a parameter, all to be called with lapply. At the end use a do.call to combine all models. There is no need for individual objects.

lapply

do.call

model_process <- function(dir) {
model <- stack(list.files(dir, pattern = ".tif$", full.names = TRUE))
e <- extent(xmin, xmax, ymin, ymax)
model_e <- crop(model, e)
model_val <- getValues(model_e)
coord_model <- xyFromCell(model_e, 1:ncell(model_e))
model_final <- cbind(coord_model, model_val)
model_final <- na.omit(model_final)

return(model_final)
}

directories <- list.dirs(".//data//models//", full.names = TRUE)[-1]

model_list <- lapply(directories, model_process)

scenario <- do.call(abind, c(model_list, along=3))

# ALTERNATIVELY:
apn <- function(...) abind(..., along=3)
scenario <- do.call("apn", model_list)


model_final <- rasterToPoints(model_e)

do.call

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.