terra (version 1.7-55)

zonal: Zonal statistics

Description

Compute zonal statistics, that is summarize values of a SpatRaster for each "zone" defined by another SpatRaster, or by a SpatVector with polygon geometry.

If fun is a true R function, the <SpatRaster,SpatRaster> method may fail when using very large SpatRasters, except for the functions ("mean", "min", "max", "sum", "isNA", and "notNA").

You can also summarize values of a SpatVector for each polygon (zone) defined by another SpatVector.

Usage

# S4 method for SpatRaster,SpatRaster
zonal(x, z, fun="mean", ..., w=NULL, wide=TRUE,
		as.raster=FALSE, filename="", overwrite=FALSE, wopt=list())
# S4 method for SpatRaster,SpatVector
zonal(x, z, fun=mean, ..., w=NULL, weights=FALSE,
		exact=FALSE, touches=FALSE, as.raster=FALSE, filename="", wopt=list())
# S4 method for SpatVector,SpatVector
zonal(x, z, fun=mean, ..., weighted=FALSE, as.polygons=FALSE)

Value

A data.frame with a value for each zone, or a SpatRaster, or SpatVector of polygons.

Arguments

x

SpatRaster or SpatVector

z

SpatRaster with cell-values representing zones or a SpatVector with each polygon geometry representing a zone. z can have multiple layers to define intersecting zones

fun

function to be applied to summarize the values by zone. Either as character: "mean", "min", "max", "sum", "isNA", and "notNA" and, for relatively small SpatRasters, a proper function

...

additional arguments passed to fun, such as na.rm=TRUE

w

SpatRaster with weights. Should have a single-layer with non-negative values

wide

logical. Should the values returned in a wide format? This only affects the results when nlyr(z) == 2

as.raster

logical. If TRUE, a SpatRaster is returned with the zonal statistic for each zone

filename

character. Output filename (ignored if as.raster=FALSE

overwrite

logical. If TRUE, filename is overwritten

wopt

list with additional arguments for writing files as in writeRaster

weights

logical. If TRUE and y has polygons, the approximate fraction of each cell that is covered is returned as well, for example to compute a weighted mean

exact

logical. If TRUE and y has polygons, the exact fraction of each cell that is covered is returned as well, for example to compute a weighted mean

touches

logical. If TRUE, values for all cells touched by lines or polygons are extracted, not just those on the line render path, or whose center point is within the polygon. Not relevant for points; and always considered TRUE when weights=TRUE or exact=TRUE

weighted

logical. If TRUE, a weighted.mean is computed and fun is ignored. Weights are based on the length of the lines or the area of the polygons in x that intersect with z. This argument is ignored of x is a SpatVector or points

as.polygons

logical. Should the results be merged with the attributes of z?

See Also

See global for "global" statistics (i.e., all of x is considered a single zone), app for local statistics, and extract for an alternative way to summarize values of a SpatRaster with a SpatVector. With aggregate you can compute statistics for cell blocks defined by a number of rows and columns.

Examples

### SpatRaster, SpatRaster
r <- rast(ncols=10, nrows=10)
values(r) <- 1:ncell(r)
z <- rast(r)
values(z) <- rep(c(1:2, NA, 3:4), each=20)
names(z) <- "zone"
zonal(r, z, "sum", na.rm=TRUE)

# with weights 
w <- init(r, "col")
zonal(r, z, w=w, "mean", na.rm=TRUE)

# multiple layers
r <- rast(system.file("ex/logo.tif", package = "terra")) 
# zonal layer 
z <- rast(r, 1)
names(z) <- "zone"
values(z) <- rep(c(1:2, NA, c(3:4)), each=ncell(r)/5, length.out=ncell(r))

zonal(r, z, "mean", na.rm = TRUE)

# raster of zonal values
zr <- zonal(r, z, "mean", na.rm = TRUE, as.raster=TRUE)


### SpatRaster, SpatVector
x <- rast(ncol=2,nrow=2, vals=1:4, xmin=0, xmax=1, ymin=0, ymax=1, crs="+proj=utm +zone=1")
p <- as.polygons(x)
pp <- shift(p, .2)
r <- disagg(x, 4)

zonal(r, p)
zonal(r, p, sum)
zonal(x, pp, exact=TRUE)
zonal(c(x, x*10), pp, w=x)


### SpatVector, SpatVector

f <- system.file("ex/lux.shp", package="terra")
v <- vect(f)[,c(2,4)]

p <- spatSample(v, 100)
values(p) <- data.frame(b2=1:100, ssep1=100:1)

zonal(p, v, mean)