project: Change the coordinate reference system

projectR Documentation

Change the coordinate reference system


Change the coordinate reference system ("project") of a SpatVector, SpatRaster or a matrix with coordinates.


## S4 method for signature 'SpatVector'
project(x, y, partial = FALSE)

## S4 method for signature 'SpatRaster'
project(x, y, method, mask=FALSE, align_only=FALSE, res=NULL, 
	origin=NULL, threads=FALSE, filename="", ..., use_gdal=TRUE, by_util = FALSE)

## S4 method for signature 'SpatExtent'
project(x, from, to)

## S4 method for signature 'matrix'
project(x, from, to)



SpatRaster, SpatVector, SpatExtent or matrix (with x and y columns) whose coordinates to project


if (x is a SpatRaster, the preferred approach is for y to be a SpatRaster as well, serving as a template for the geometry (extent and resolution) of the output SpatRaster. Alternatively, you can provide a coordinate reference system (CRS) description.

You can use the following formats to define coordinate reference systems: WKT, PROJ.4 (e.g., +proj=longlat +datum=WGS84), or an EPSG code (e.g., "epsg:4326"). But note that the PROJ.4 notation has been deprecated, and you can only use it with the WGS84/NAD83 and NAD27 datums. Other datums are silently ignored.

If x is a SpatVector, you can provide a crs definition as discussed above, or any other object from which such a crs can be extracted with crs


logical. If TRUE, geometries that can only partially be represented in the output crs are included in the output


character. Method used for estimating the new cell values of a SpatRaster. One of:

near: nearest neighbor. This method is fast, and it can be the preferred method if the cell values represent classes. It is not a good choice for continuous values. This is used by default if the first layer of x is categorical.

bilinear: bilinear interpolation. This is the default if the first layer of x is numeric (not categorical).

cubic: cubic interpolation.

cubicspline: cubic spline interpolation.

lanczos: Lanczos windowed sinc resampling.

sum: the weighted sum of all non-NA contributing grid cells.

min, q1, med, q3, max, average, mode, rms: the minimum, first quartile, median, third quartile, maximum, mean, mode, or root-mean-square value of all non-NA contributing grid cells.


logical. If TRUE, mask out areas outside the input extent. For example to avoid data wrapping around the date-line (see example with Robinson projection). To remove cells that are NA in y (if y is a SpatRaster) you can use the mask method after calling project (this function)


logical. If TRUE, and y is a SpatRaster, the template is used for the spatial resolution and origin, but the extent is set such that all of the extent of x is included


numeric. Can be used to set the resolution of the output raster if y is a CRS


numeric. Can be used to set the origin of the output raster if y is a CRS


logical. If TRUE multiple threads are used (faster for large files)


character. Output filename


additional arguments for writing files as in writeRaster


logical. If TRUE the GDAL-warp algorithm is used. Otherwise a slower internal algorithm is used that may be more accurate if there is much variation in the cell sizes of the output raster. Only the near and bilinear algorithms are available for the internal algorithm


logical. If TRUE and gdal=TRUE, the GDAL warp utility is used


character. Coordinate reference system of x


character. Output coordinate reference system


SpatVector or SpatRaster


The PROJ.4 notation of coordinate reference systems has been partly deprecated in the GDAL/PROJ library that is used by this function. You can still use this notation, but *only* with the WGS84 datum. Other datums are silently ignored.

Transforming (projecting) raster data is fundamentally different from transforming vector data. Vector data can be transformed and back-transformed without loss in precision and without changes in the values. This is not the case with raster data. In each transformation the values for the new cells are estimated in some fashion. Therefore, if you need to match raster and vector data for analysis, you should generally transform the vector data.

When using this method with a SpatRaster, the preferable approach is to provide a template SpatRaster as argument y. The template is then another raster dataset that you want your data to align with. If you do not have a template to begin with, you can do project(rast(x), crs) and then manipulate the output to get the template you want. For example, where possible use whole numbers for the extent and resolution so that you do not have to worry about small differences in the future. You can use commands like dim(z) = c(180, 360) or res(z) <- 100000.

The output resolution should generally be similar to the input resolution, but there is no "correct" resolution in raster transformation. It is not obvious what this resolution is if you are using lon/lat data that spans a large North-South extent.

See Also

crs, resample


## SpatRaster
a <- rast(ncols=40, nrows=40, xmin=-110, xmax=-90, ymin=40, ymax=60, 
          crs="+proj=longlat +datum=WGS84")
values(a) <- 1:ncell(a)
newcrs="+proj=lcc +lat_1=48 +lat_2=33 +lon_0=-100 +datum=WGS84"
b <- rast(ncols=94, nrows=124, xmin=-944881, xmax=935118, ymin=4664377, ymax=7144377, crs=newcrs)
w <- project(a, b)

## SpatVector
f <- system.file("ex/lux.shp", package="terra")
v <- vect(f)
crs(v, proj=TRUE)
cat(crs(v), "\n")

project(v, "+proj=moll")

project(v, "EPSG:2169")

terra documentation built on May 29, 2024, 12:33 p.m.