gkde: Geographic Kernel Density Estimator using linear or Haversine...

In rsh249/rasterExtras: Parallel computing with rasters

Description

This function calculates a kernel density estimation for raster objects.

Usage

gkde(grid, points, parallel = TRUE, nclus = 4,
  dist.method = "Haversine", maxram = 4, bw = 200)

Arguments

grid	A raster object to match.
points	A two column data frame in the form (lon,lat) or (x,y)
parallel	TRUE or FALSE, should this code be executed in parallel.
nclus	IF parallel==TRUE then how many cores in the cluster.
dist.method	Which distance should we use? Haversine for lat/long projections,or Pythagorean for flat images and/or small areas.
maxram	Maximum theoretical RAM usage. Will be divided by nclus for parallel jobs.
bw	Bandwidth. Numeric bandwidth. See bw.calc for help

Examples

require(raster)
grid = raster::raster(nrows=81, ncols=162, xmn=-180, xmx=180, ymn=-90, ymx=90, vals=NULL)
grid = raster::setValues(grid,values=(as.vector(seq(1:raster::ncell(grid)))))
points = cbind(
      c(seq(xmin(grid), xmax(grid), length.out=1000), 
               seq(xmax(grid), xmin(grid), length.out=1000)), 
      c(seq(ymin(grid), ymax(grid), length.out=100), 
               seq(ymin(grid), ymax(grid), length.out=100))
               )
plot(grid); points(points);
den = gkde(grid, points, parallel=TRUE, dist.method='Haversine', bw= 10)
plot(den)
points(points)

rsh249/rasterExtras documentation built on June 7, 2019, 7:38 a.m.