Using "lapply" in R to create multiple raster files from folder with lidar data

Question

How can I read all files in a folder, perform a script and create separate outputs from all files containing the original name? I have a folder with .las files and I need to create corresponding .asc files from them. My script as below:

library(lidR)
# Path to data  
LASfile <- ("path/1234.las") 
# Sorting out points in point cloud data, keeping vegetation and ground point classes.
las <- readLAS(LASfile, filter="-keep_class 1 2") # Keep high vegetation and ground point classes
# Normalizing ground points to 0 elevation (idwinterpolation), instead of meters above sea level.
dtm <- grid_terrain(las, algorithm = knnidw(k = 8, p = 2))
las_normalized <- normalize_height(las, dtm)
# Create a filter to remove points above 95th percentile of height
lasfilternoise = function(las, sensitivity)
{
  p95 <- grid_metrics(las, ~quantile(Z, probs = 0.95), 10)
  las <- merge_spatial(las, p95, "p95")
  las <- filter_poi(las, Z < p95*sensitivity)
  las$p95 <- NULL
  return(las)
}
# Generating a pitfree canopy height modela model without null values (Khosravipour et al., 2014)
las_denoised <- lasfilternoise(las_normalized, sensitivity = 1.2)
chm <- grid_canopy(las_denoised, 0.32, pitfree(c(0,2,5,10,15), c(3,1.5), subcircle = 0.2))
# Applying a median filter, 5x5 moving window to smooth the image and remove noise
ker <- matrix(1,3,3)
chms <- raster::focal(chm, w = ker, fun = median) 
plot(chms)
library(raster)
# Writing output file
writeRaster(chms, filename="path/1234.asc", format="ascii", overwrite=TRUE) # Ändra till relevant för varje körning
citation("lidR")

I tried using lapply but I dont know how to use it in the right way. Must be something like this to read all files in the folder: list.files("path", pattern = "*.las", full.names = TRUE)

and something like this to write the output files: lapply(r, writeRaster, filename = paste0(f, ".asc"), format = "ascii")

But I cannot get it right

Answer 1

An example of my LAZ to LAS+Index conversion:

convertLAZ <- function(lazfile, outdir = "") {
  if(!dir.exists({{outdir}})) { dir.create({{outdir}}, recursive = TRUE)}
  print(lazfile)
  las <- lidR::readLAS(files = {{lazfile}}, filter = "-keep_class 2 9")
  .file <- stringi::stri_replace_all_regex({{lazfile}}, "^.*/", "")
  lidR::writeLAS(las, file = paste0({{outdir}}, "/", stringi::stri_replace_all_fixed(.file, "laz", "las")), index = TRUE)
}

f <- list.files("data/laz", pattern = "*.laz", full.names = TRUE)
lapply(f, convertLAZ, outdir = "data/las22")

You can expand it to rasterization, normalization, etc and saving as .asc. But I would encourage you to have a look on https://r-lidar.github.io/lidRbook/engine.html. In short: process your LAZ/LAS files as LAScatalog, and then tile the result raster and save to .asc.

And an example how to use parallel processing (in below example 3+1 processes - please note, it can be memory hungry, so be careful with number of workers/processing parameters like opt_chunk_buffer.

library(future)
options(parallelly.availableCores.methods = "mc.cores")
options(mc.cores = 3)
plan(multisession)
parallelly::availableWorkers()

library(lidR)

myPath <- "data/las"
ctg <- readLAScatalog(myPath)
crs(ctg) <- "EPSG:2180"

ctg@output_options$drivers$SpatRaster$param$overwrite <- TRUE
opt_output_files(ctg) <- "data/dtm2/barycz__{XLEFT}_{YBOTTOM}"
opt_chunk_size(ctg) <- 500
opt_chunk_buffer(ctg) <- 600
opt_filter(ctg) <- "-keep_class 2 9"
summary(ctg)

vr <- rasterize_terrain(ctg, 0.25, tin())
plot(vr)

Answer 2

Solved it now

.libPaths( c( "C:/Users/Public/R/win-library/4.2" , .libPaths() ) )
library(lidR) 
createASCI <- function(lasfile, outdir = "") {
if(!dir.exists({{outdir}})) { dir.create({{outdir}}, recursive = TRUE)}
print(lasfile)
las <- lidR::readLAS(files = {{lasfile}}, filter = "-keep_class 1 2 3 4 5")
.file <- stringi::stri_replace_all_regex({{lasfile}}, "^.*/", "")
# Normalizing ground points to 0 elevation (idwinterpolation), instead of meters above sea level.
dtm <- grid_terrain(las, algorithm = knnidw(k = 8, p = 2))
las_normalized <- normalize_height(las, dtm)
# Create a filter to remove points above 95th percentile of height
lasfilternoise = function(las, sensitivity)
{
  p95 <- grid_metrics(las, ~quantile(Z, probs = 0.95), 10)
  las <- merge_spatial(las, p95, "p95")
  las <- filter_poi(las, Z < p95*sensitivity)
  las$p95 <- NULL
  return(las)
}
# Generating a pitfree canopy height modela model without null values (Khosravipour et al., 2014)
las_denoised <- lasfilternoise(las_normalized, sensitivity = 1.2)
chm <- grid_canopy(las_denoised, 0.32, pitfree(c(0,2,5,10,15), c(3,1.5), subcircle = 0.2))
# Applying a median filter, 5x5 moving window to smooth the image and remove noise
ker <- matrix(1,3,3)
chms <- raster::focal(chm, w = ker, fun = median)

writeRaster(chms, file = paste0({{outdir}}, "/", stringi::stri_replace_all_fixed(.file, "las", "asc")), index = TRUE)
}

f <- list.files("C:/Lasdata", pattern = "*.las", full.names = TRUE)
lapply(f, createASCI, outdir = "C:/Lasdata/nytt")