Skip to contents

sample points on or in (sets of) spatial features

Source: R/sample.R
st_sample.Rd

Sample points on or in (sets of) spatial features. By default, returns a pre-specified number of points that is equal to size (if type = "random" and exact = TRUE) or an approximation of size otherwise. spatstat methods are interfaced and do not use the size argument, see examples.

Usage

st_sample(x, size, ...)

# S3 method for class 'sf'
st_sample(x, size, ...)

# S3 method for class 'sfc'
st_sample(
  x,
  size,
  ...,
  type = "random",
  exact = TRUE,
  warn_if_not_integer = TRUE,
  by_polygon = FALSE,
  progress = FALSE,
  force = FALSE
)

# S3 method for class 'sfg'
st_sample(x, size, ...)

# S3 method for class 'bbox'
st_sample(
  x,
  size,
  ...,
  great_circles = FALSE,
  segments = units::set_units(2, "degree", mode = "standard")
)

Arguments

x

object of class sf or sfc

size

sample size(s) requested; either total size, or a numeric vector with sample sizes for each feature geometry. When sampling polygons, the returned sampling size may differ from the requested size, as the bounding box is sampled, and sampled points intersecting the polygon are returned.

...

passed on to sample for multipoint sampling, or to spatstat functions for spatstat sampling types (see details)

type

character; indicates the spatial sampling type; one of random, hexagonal (triangular really), regular, Fibonacci, or one of the spatstat methods such as Thomas for calling spatstat.random::rThomas (see Details).

exact

logical; should the length of output be exactly

warn_if_not_integer

logical; if FALSE then no warning is emitted if size is not an integer

by_polygon

logical; for MULTIPOLYGON geometries, should the effort be split by POLYGON? See https://github.com/r-spatial/sf/issues/1480 the same as specified by size? TRUE by default. Only applies to polygons, and when type = "random".

progress

logical; if TRUE show progress bar (only if size is a vector).

force

logical; if TRUE continue when the sampled bounding box area is more than 1e4 times the area of interest, else (default) stop with an error. If this error is not justified, try setting oriented=TRUE, see details.

great_circles

logical; if TRUE, great circle arcs are used to connect the bounding box vertices, if FALSE parallels (graticules)

segments

units, or numeric (degrees); segment sizes for segmenting a bounding box polygon if great_circles is FALSE

Value

an sfc object containing the sampled POINT geometries

Details

The function is vectorised: it samples size points across all geometries in the object if size is a single number, or the specified number of points in each feature if size is a vector of integers equal in length to the geometry of x.

if x has dimension 2 (polygons) and geographical coordinates (long/lat), uniform random sampling on the sphere is applied, see e.g. https://mathworld.wolfram.com/SpherePointPicking.html.

For regular or hexagonal sampling of polygons, the resulting size is only an approximation.

As parameter called offset can be passed to control ("fix") regular or hexagonal sampling: for polygons a length 2 numeric vector (by default: a random point from st_bbox(x)); for lines use a number like runif(1).

Fibonacci sampling see: Alvaro Gonzalez, 2010. Measurement of Areas on a Sphere Using Fibonacci and Latitude-Longitude Lattices. Mathematical Geosciences 42(1), p. 49-64

For regular sampling on the sphere, see also geosphere::regularCoordinates.

Sampling methods from package spatstat are interfaced (see examples), and need their own parameters to be set. For instance, to use spatstat.random::rThomas(), set type = "Thomas".

For sampling polygons one can specify oriented=TRUE to make sure that polygons larger than half the globe are not reverted, e.g. when specifying a polygon from a bounding box of a global dataset. The st_sample method for bbox does this by default.

Examples

nc = st_read(system.file("shape/nc.shp", package="sf"))
#> Reading layer `nc' from data source 
#>   `/home/runner/work/_temp/Library/sf/shape/nc.shp' using driver `ESRI Shapefile'
#> Simple feature collection with 100 features and 14 fields
#> Geometry type: MULTIPOLYGON
#> Dimension:     XY
#> Bounding box:  xmin: -84.32385 ymin: 33.88199 xmax: -75.45698 ymax: 36.58965
#> Geodetic CRS:  NAD27
p1 = st_sample(nc[1:3, ], 6)
p2 = st_sample(nc[1:3, ], 1:3)
plot(st_geometry(nc)[1:3])
plot(p1, add = TRUE)
plot(p2, add = TRUE, pch = 2)

x = st_sfc(st_polygon(list(rbind(c(0,0),c(90,0),c(90,90),c(0,90),c(0,0)))), crs = st_crs(4326))
plot(x, axes = TRUE, graticule = TRUE)
if (compareVersion(sf_extSoftVersion()["proj.4"], "4.9.0") >= 0)
  plot(p <- st_sample(x, 1000), add = TRUE)

if (require(lwgeom, quietly = TRUE)) { # for st_segmentize()
  x2 = st_transform(st_segmentize(x, 1e4), st_crs("+proj=ortho +lat_0=30 +lon_0=45"))
  g = st_transform(st_graticule(), st_crs("+proj=ortho +lat_0=30 +lon_0=45"))
  plot(x2, graticule = g)
  if (compareVersion(sf_extSoftVersion()["proj.4"], "4.9.0") >= 0) {
    p2 = st_transform(p, st_crs("+proj=ortho +lat_0=30 +lon_0=45"))
    plot(p2, add = TRUE)
  }
}

x = st_sfc(st_polygon(list(rbind(c(0,0),c(90,0),c(90,10),c(0,90),c(0,0))))) # NOT long/lat:
plot(x)
p_exact = st_sample(x, 1000, exact = TRUE)
p_not_exact = st_sample(x, 1000, exact = FALSE)
length(p_exact); length(p_not_exact)
#> [1] 1000
#> [1] 1000
plot(st_sample(x, 1000), add = TRUE)

x = st_sfc(st_polygon(list(rbind(c(-180,-90),c(180,-90),c(180,90),c(-180,90),c(-180,-90)))),
 crs=st_crs(4326))
# FIXME:
#if (compareVersion(sf_extSoftVersion()["proj.4"], "4.9.0") >= 0) {
#  p = st_sample(x, 1000)
#  st_sample(p, 3)
#}
# hexagonal:
sfc = st_sfc(st_polygon(list(rbind(c(0,0), c(1,0), c(1,1), c(0,0)))))
plot(sfc)
h = st_sample(sfc, 100, type = "hexagonal")
h1 = st_sample(sfc, 100, type = "hexagonal")
plot(h, add = TRUE)
plot(h1, col = 'red', add = TRUE)

c(length(h), length(h1)) # approximate!
#> [1] 97 99
pt = st_multipoint(matrix(1:20,,2))
ls = st_sfc(st_linestring(rbind(c(0,0),c(0,1))),
 st_linestring(rbind(c(0,0),c(.1,0))),
 st_linestring(rbind(c(0,1),c(.1,1))),
 st_linestring(rbind(c(2,2),c(2,2.00001))))
st_sample(ls, 80)
#> Geometry set for 4 features  (with 1 geometry empty)
#> Geometry type: MULTIPOINT
#> Dimension:     XY
#> Bounding box:  xmin: 0 ymin: 0 xmax: 0.08531849 ymax: 1
#> CRS:           NA
#> MULTIPOINT ((0 0.01737041), (0 0.8441574), (0 0...
#> MULTIPOINT ((0.03337024 0), (0.03010541 0), (0....
#> MULTIPOINT ((0.04159846 1), (0.06942899 1), (0....
#> MULTIPOINT EMPTY
plot(st_sample(ls, 80))

# spatstat example:
if (require(spatstat.random)) {
  x <- sf::st_sfc(sf::st_polygon(list(rbind(c(0, 0), c(10, 0), c(10, 10), c(0, 0)))))
  # for spatstat.random::rThomas(), set type = "Thomas":
  pts <- st_sample(x, kappa = 1, mu = 10, scale = 0.1, type = "Thomas") 
}
bbox = st_bbox(
c(xmin = 0, xmax = 40, ymax = 70, ymin = 60),
  crs = st_crs('OGC:CRS84')
)
set.seed(13531)
s1 = st_sample(bbox, 400)
st_bbox(s1) # within bbox
#>       xmin       ymin       xmax       ymax 
#>  0.1809029 60.0231214 39.9875925 69.9900375 
s2 = st_sample(bbox, 400, great_circles = TRUE)
st_bbox(s2) # outside bbox
#>       xmin       ymin       xmax       ymax 
#>  0.2031083 60.2048778 39.9612431 69.9107718