I have a point vector for the entire globe at 0.25 degree resolution.
library(terra)
library(rnaturalearth)
# create point data
ref_grid <- terra::ext(-180, 180, -90, 90)
ref_grid <- terra::rast(ref_grid)
res(ref_grid)<- 0.25
values(ref_grid)<-1 #dummy values
projections <- "+proj=longlat +datum=WGS84 +no_defs +ellps=WGS84 +towgs84=0,0,0"
crs(ref_grid) <- projections
pts <- terra::as.points(ref_grid, values=TRUE, na.rm=TRUE)
# download world shapefile
world_shp <- rnaturalearth::ne_countries()
world_shp <- terra::vect(world_shp)
world_shp <- terra::project(world_shp, projections)
I want to extract those points that fall within a landmass i.e. remove all those points that are in the ocean. So I did this:
e <- terra::extract(world_shp, pts)
But it's been two hours now and this is still running. Is there any faster way to do this?
You should not transform raster data to vector data if you want efficiency. In this case, you can do the following in a second or so:
library(terra)
library(rnaturalearth)
ref_grid <- terra::rast(res=0.25)
world_shp <- rnaturalearth::ne_countries(returnclass="sf") |> vect()
ref_grid <- rasterize(world_shp, ref_grid)
p <- as.points(ref_grid)
(but perhaps, depending on what you do next, the last step should also be avoided)