CODEWITHSUNDEEP
rrasterr-spterra
Jannes
Jannes

Reputation: 23

R terra - most efficient way to assign an array to a SpatRaster object that does NOT contain all cell values

Let's say we have an array modeldata (data comes from a terrestrial model), whose dimensions are:

> dim(modeldata)
[1] 67420   518

The first dimension includes the cells of the grid, the second a timeseries from 1500:2017
The unusual length of the first dimension is due to the presence of the terrestrial cells alone to save space.

In the raster package I was dealing with it the following way:

> coords
        [,1]   [,2]
[1,] -179.75 -16.25
[2,] -179.75  65.25
[3,] -179.75  65.75
[4,] -179.75  66.25
[...,]  ...     ...
[67420,] 179.75  71.25

> wgs84 <- sp::CRS("+proj=longlat +ellps=WGS84 +datum=WGS84 +no_defs")
> modeldata_spdf <- sp::SpatialPixelsDataFrame(coords,
                                               data = data.frame(modeldata),
                                               proj4string = wgs84)

> modeldata_brick <- raster::brick(modeldata_spdf)

Please do not judge me for this way,
I am more interested in an comparable (performant) approach using the terra package.
Another approach which is also be fine would be to use a SpatRaster mask layer instead of the coordinates.

Thanks :-)

Upvotes: 1

Views: 1502

Answers (1)

Robert Hijmans
Robert Hijmans

Reputation: 47491

The approach below is for situations where you compute values for a subset of locations, the cells of interest, on a known regular raster. For other cases see values<-, rast(, xyz=TRUE) and rasterize.

Example data

library(terra)
r <- rast(res=10)
set.seed(0)
cells <- sample(ncell(r), 5)
xy <- xyFromCell(r, cells)
xy
#        x   y
#[1,] -165 -25
#[2,]   25  55
#[3,] -135 -55
#[4,] -155 -45
#[5,]  -75   5

In this case we have 5 locations, and the model data below, md, has 2 time steps (layers) of 5 values each

md <- cbind(A=1:5, B=5:1)

# compute cell numbers (not needed here as we have them already)
cells <- cellFromXY(r, xy)

# create empty raster
x <- rast(r, nlyr=2)

# assign values
x[cells] <- md

In your case it may be more efficient to keep the cells numbers instead of the coordinates.

The above should also work for raster with minor modifications.

Upvotes: 0

Related Questions