calckernel: functions to calculate the sum of the product of a kernel and...

In ethanplunkett/gridkernel: Package to apply kernels to raster data with NA and edge correction

functions to calculate the sum of the product of a kernel and a patch of a grid or matrix around a focal cell.

Description

This function returns the sum of the product of each cell in the kernel and the corresponding cell in the matrix around the focal cell.

Usage

## S3 method for class 'grid'
calckernel(obj, kernel, x, y, row=y2r(y, list=obj), col=x2c(x, list=obj), ...)
## S3 method for class 'matrix'
calckernel(obj, kernel, row, col, use.old=FALSE, ...)
calckernel(obj, ...)
calc.kernel(obj, ...) # depreciated

Arguments

obj	either a matrix or an object of class "grid" containing the data
kernel	the kernel. It should be a square, odd dimensioned, numeric matrix that sums to 1.
row, col	the row and column of the focal cell. The center of the kernel will be aligned with this cell.
x, y	the x and y coordinates in map units of the focal cell.
use.old	if FALSE then Rcpp based implementation is used. Otherwise an older, pure R, version is used. This is here to facilitate testing.
...	Arguments to be passed to methods

Details

this function returns the sum of the product of the kernel and data around the focal cell. If (as required) the kernel sums to one this is the weighted average of the cells. If the kernel overlaps the edge or obj contains NA values than only the available data will be used and the kernel will be renormalized such that the sum of the used cells is one; the result will be the weighted average of the available data.

Value

A weighted average of the cells in the matrix surrounding the focal cell.

Note

Please make sure the kernel sums to 1. This could be done in the function but that would be inefficient in cases where the kernel is repeatedly applied to the matrix. Use kernel <- kernel/sum(kernel, na.rm = TRUE) to properly normalize a kernel. If the kernel doesn't sum to one than the results near edges and NA values will be inconsistent.

Author(s)

Ethan Plunkett

See Also

kernelsmooth, makegaussiankernel

Examples

m <- matrix(1:60, 6, 10)
k <- matrix(1, 3, 3)
k[2,2] <- 4
k <- k/sum(k)

kernelsmooth(m, k)

calckernel(m, k, 2, 2)

ethanplunkett/gridkernel documentation built on March 2, 2024, 9:06 p.m.