2

I have a lot of GML files to loop through in R, but none of them are extremely heavy (max file size is 600 Mb, but normally they range between 10-60 Mb). My problem is that readOGR function from the {rgdal} package takes too long to read them, and it slows down the loop a lot. It takes between 3 minutes up to 30 minutes to read a single file...

I have looped through .shp files and they do not take up so much time to read.

So my question is actually two:

  • Is there a faster function in R to read GML files? Have not been able to find one so far
  • If a faster reading function isnt possible, is there perhaps a quick way to convert the GML files to .shp?

Code to my readOGR:

llayer<-ogrListLayers(gmls)[1]
a<- readOGR(dsn=gmls, layer=llayer, encoding = "UTF-8", disambiguateFIDs = T)
lbusett
  • 5,801
  • 2
  • 24
  • 47
Miguel M.
  • 251
  • 1
  • 3
  • 15
  • 1
    You could try package `sf`. It's usually much faster, and I think it should handle gmls – lbusett Jun 15 '17 at 17:42
  • 1
    @LoBu have tried, it is much faster BUT apparently my computer has an issue with GDAL data, I have tried to set the environment GDAL_DATA variable in windows manually but it does not work... I'll try on my laptop and see what happens. I have also tried assigning the projection manually but it doesn't seem to work – Miguel M. Jun 16 '17 at 15:28
  • I tried today on my PC, absolutely right you were. I'll post the answer, thanks a lot! @LoBu – Miguel M. Jun 19 '17 at 16:34

1 Answers1

3

As @LoBu said in the comment section, sf::st_read() is much, much faster than readOGR. The only matter to keep in mind is that now you work with sf objects and you must either perform your operations using the sf package or convert the sf object to a SpatialPolygonsDataframe with as(file, "Spatial").

Warning, for people using work laptops: errors may occur as I mention in the comment section. Make sure you are using a laptop with full admin permits

Miguel M.
  • 251
  • 1
  • 3
  • 15