CODEWITHSUNDEEP
rpolygonrasterr-raster
Nadav Bendavid
Nadav Bendavid

Reputation: 33

Trouble converting a large raster to polygon using R

I have a rather large raster (384 MB) that I am trying to convert to a polygon shapefile in R. The rastertoPolygons function from the raster package doesn't seem to be able to handle this, as I tried running it but gave up after it was going for 7+ hours.

I also tried to use gdal_polygonize.py from GDAL in python via this function by John Baumgartner but after letting the function run for 30+ minutes I still have nothing. Am I simply not letting it run long enough? I was under the impression that gdal_polyonize.py was supposed to be very quick, i.e. seconds.

Here's a link to my raster file.

Any guidance would be greatly appreciated.

Upvotes: 3

Views: 2742

Answers (1)

Robert Hijmans
Robert Hijmans

Reputation: 47051

terra does this much faster than raster (but not faster than GDAL because that is what it uses)

library(terra)
r <- rast("top6loss.tif")

Note that you have 22 billion cells (that is a lot by most standards, and that is why it takes a while):

ncell(r)
#[1] 21989436765

It finishes in 10 minutes on my laptop

system.time(p <- as.polygons(r))
#   user  system elapsed 
# 562.34    3.54  568.77 

p
#class       : SpatVector 
#geometry    : polygons 
#dimensions  : 6, 1  (geometries, attributes)
#extent      : -13.54777, 12.33558, -6.134633, 9.781491  (xmin, xmax, ymin, ymax)
#coord. ref. : +proj=longlat +datum=WGS84 +no_defs

These are the 6 values

as.data.frame(p)
#  top6loss
#1     2254
#2     5418
#3    13623
#4    14344
#5    15885
#6    19654

You can save the file with

writeVector(p, "cells.shp")

Upvotes: 4

Related Questions