Package 'geodist'

geodist.

Description

Dependency-free, ultra fast calculation of geodesic distances. Includes the reference nanometre-accuracy geodesic distances of Karney (2013) doi:10.1007/s00190-012-0578-z, as used by the 'sf' package, as well as Haversine and Vincenty distances. Default distance measure is the "Mapbox cheap ruler" which is generally more accurate than Haversine or Vincenty for distances out to a few hundred kilometres, and is considerably faster. The main function accepts one or two inputs in almost any generic rectangular form, and returns either matrices of pairwise distances, or vectors of sequential distances.

Convert one or two rectangular objects containing lon-lat coordinates into vector or matrix of geodesic distances in metres.

Usage

geodist(
  x,
  y,
  paired = FALSE,
  sequential = FALSE,
  pad = FALSE,
  measure = "cheap",
  quiet = FALSE
)

Arguments

x	Rectangular object (matrix, data.frame, tibble, whatever) containing longitude and latitude coordinates.
y	Optional second object which, if passed, results in distances calculated between each object in x and each in y.
paired	If TRUE, calculate paired distances between each entry in x and y, returning a single vector.
sequential	If TRUE, calculate (vector of) distances sequentially along x (when no y is passed), otherwise calculate matrix of pairwise distances between all points.
pad	If sequential = TRUE values are padded with initial NA to return n values for input with n rows, otherwise return n - 1 values.
measure	One of "haversine" "vincenty", "geodesic", or "cheap" specifying desired method of geodesic distance calculation; see Notes.
quiet	If FALSE, check whether max of calculated distances is greater than accuracy threshold and warn.

Value

If only x passed and sequential = FALSE, a square symmetric matrix containing distances between all items in x; If only x passed and sequential = TRUE, a vector of sequential distances between rows of x; otherwise if y is passed, a matrix of nrow(x) rows and nrow(y) columns. All return values are distances in metres.

Note

measure = "cheap" denotes the mapbox cheap ruler https://github.com/mapbox/cheap-ruler-cpp; measure = "geodesic" denotes the very accurate geodesic methods given in Karney (2013) "Algorithms for geodesics" J Geod 87:43-55, and as provided by the 'st_distance()' function from the sf package.

Author(s)

Maintainer: Mark Padgham [email protected]

Authors:

• Michael D. Sumner

Other contributors:

• Charles F.F Karney (Original author of included code for geodesic distances) [copyright holder]

See Also

Useful links:

• https://github.com/hypertidy/geodist

• https://hypertidy.github.io/geodist/

• Report bugs at https://github.com/hypertidy/geodist/issues

Examples

n <- 50
# Default "cheap" distance measure is only accurate for short distances:
x <- cbind (runif (n, -0.1, 0.1), runif (n, -0.1, 0.1))
y <- cbind (runif (2 * n, -0.1, 0.1), runif (2 * n, -0.1, 0.1))
colnames (x) <- colnames (y) <- c ("x", "y")
d0 <- geodist (x) # A 50-by-50 matrix
d1 <- geodist (x, y) # A 50-by-100 matrix
d2 <- geodist (x, sequential = TRUE) # Vector of length 49
d2 <- geodist (x, sequential = TRUE, pad = TRUE) # Vector of length 50
d0_2 <- geodist (x, measure = "geodesic") # nanometre-accurate version of d0

# Input data can also be 'data.frame' objects:
xy <- data.frame (x = runif (n, -0.1, 0.1), y = runif (n, -0.1, 0.1))
d <- geodist (xy)

---

geodist_benchmark

Description

Benchmark errors for different geodist measures

Usage

geodist_benchmark(lat = 0, d = 1, n = 100L)

Arguments

lat	Central latitude where errors should be measured
d	Distance in metres over which errors should be measured
n	Number of random values used to generate estimates

Value

A 'data.frame' with three columns respectively comparing the accuracy of the [Haversine, Vincenty, cheap] metrics against geodesic measures in both absolute and relative terms (as two rows of the table).

Examples

geodist_benchmark (0.0, 1.0, 100L)

---

Minimal pairwise distances between two input matrices

Description

Convert two rectangular objects containing lon-lat coordinates into an index vector of the elements in the second matrix corresponding to the minimal distance to each element of the first matrix.

Usage

geodist_min(x, y, measure = "cheap", quiet = FALSE)

Arguments

x	Rectangular object (matrix, data.frame, tibble, whatever) containing longitude and latitude coordinates.
y	Second rectangular object to be search for minimal distance to each row in the first object.
measure	One of "haversine" "vincenty", "geodesic", or "cheap" specifying desired method of geodesic distance calculation; see Notes.
quiet	If FALSE, check whether max of calculated distances is greater than accuracy threshold and warn.

Value

A integer index vector indexing elements of 'y' corresponding to minimal distances to each element of 'x'. The length of this vector is equal to the number of rows in 'x'.

Note

measure = "cheap" denotes the mapbox cheap ruler https://github.com/mapbox/cheap-ruler-cpp; measure = "geodesic" denotes the very accurate geodesic methods given in Karney (2013) "Algorithms for geodesics" J Geod 87:43-55, and as provided by the 'st_distance()' function from the sf package.

Examples

n <- 50
# Default "cheap" distance measure is only accurate for short distances:
x <- cbind (runif (n, -0.1, 0.1), runif (n, -0.1, 0.1))
y <- cbind (runif (2 * n, -0.1, 0.1), runif (2 * n, -0.1, 0.1))
colnames (x) <- colnames (y) <- c ("x", "y")
index <- geodist_min (x, y, measure = "Haversine")
# 'index' is a vector of 50 values indexing elements of `y` corresponding to
# minimal distances to each element of `x`. It is exactly the same as the
# value returned by these lines:
d0 <- geodist (x, y, measure = "Haversine")
index0 <- apply (d0, 1, which.min)
identical (index, index0)

---

geodist_vec

Description

An alternative interface to the main geodist function that directly accepts inputs as individual vectors of coordinates, rather than the matrix or 'data.frame' inputs of the main function. This interface is provided for cases where computational efficiency is important, and will generally provide faster results than the main function.

Usage

geodist_vec(
  x1,
  y1,
  x2,
  y2,
  paired = FALSE,
  sequential = FALSE,
  pad = FALSE,
  measure = "cheap",
  quiet = FALSE
)

Arguments

x1	Numeric vector of longitude coordinates
y1	Numeric vector of latitude coordinates
x2	Optional second numeric vector of longitude coordinates
y2	Optional second numeric vector of latitude coordinates
paired	If TRUE, calculate paired distances between each entry in (x1, y1) and (x2, y2), returning a single vector.
sequential	If TRUE, calculate (vector of) distances sequentially along (x1, y1) (when no (x2, y2) are passed), otherwise calculate matrix of pairwise distances between all points.
pad	If sequential = TRUE values are padded with initial NA to return n values for inputs of length n, otherwise return n - 1 values.
measure	One of "haversine" "vincenty", "geodesic", or "cheap" specifying desired method of geodesic distance calculation; see Notes.
quiet	If FALSE, check whether max of calculated distances is greater than accuracy threshold and warn.

Value

If only (x1, y1) are passed and sequential = FALSE, a square symmetric matrix containing distances between all items in (x1, y1); If only (x1, y1) are passed and sequential = TRUE, a vector of sequential distances between matching elements of (x1, y1); otherwise if (x2, y2) are passed, a matrix of lenght(x1) == length(y1) rows and length(x2) == length(y2) columns.

Note

measure = "cheap" denotes the mapbox cheap ruler https://github.com/mapbox/cheap-ruler-cpp; measure = "geodesic" denotes the very accurate geodesic methods given in Karney (2013) "Algorithms for geodesics" J Geod 87:43-55, and as provided by the 'st_distance()' function from the sf package.

Examples

n <- 50
# Default "cheap" distance measure is only accurate for short distances:
x1 <- -1 + 2 * runif (n, -0.1, 0.1)
y1 <- -1 + 2 * runif (n, -0.1, 0.1)
d0 <- geodist_vec (x1, y1) # A 50-by-50 matrix
d2 <- geodist_vec (x1, y1, sequential = TRUE) # Vector of length 49
d2 <- geodist_vec (x1, y1, sequential = TRUE, pad = TRUE) # length 50
x2 <- -10 + 20 * runif (2 * n, -0.1, 0.1)
y2 <- -10 + 20 * runif (2 * n, -0.1, 0.1)
d1 <- geodist_vec (x1, y1, x2, y2) # A 50-by-100 matrix

---

georange

Description

Calculate range of distances (min-max) between all points in one or two rectangular objects containing lon-lat coordinates.

Usage

georange(x, y, sequential = FALSE, measure = "cheap")

Arguments

x	Rectangular object (matrix, data.frame, tibble, whatever) containing longitude and latitude coordinates.
y	Optional second object which, if passed, results in distances calculated between each object in x and each in y.
sequential	If TRUE, calculate (vector of) distances sequentially along x (when no y is passed), otherwise calculate matrix of pairwise distances between all points.
measure	One of "haversine" "vincenty", "geodesic", or "cheap" specifying desired method of geodesic distance calculation; see Notes.

Value

A named vector of two numeric values: minimum and maximum, giving the respective distances in metres.

Note

measure = "cheap" denotes the mapbox cheap ruler https://github.com/mapbox/cheap-ruler-cpp; measure = "geodesic" denotes the very accurate geodesic methods given in Karney (2013) "Algorithms for geodesics" J Geod 87:43-55, and as provided by the 'st_distance()' function from the sf package.

Examples

n <- 50
x <- cbind (-10 + 20 * runif (n), -10 + 20 * runif (n))
y <- cbind (-10 + 20 * runif (2 * n), -10 + 20 * runif (2 * n))
colnames (x) <- colnames (y) <- c ("x", "y")
# All of the following returns vector of two values: minimum and maximum:
d0 <- georange (x)
d1 <- georange (x, y)
d2 <- georange (x, sequential = TRUE)
d0_2 <- georange (x, measure = "geodesic") # nanometre-accurate version of d0

---