Spatial kernel density estimate

A weighted or unweighted Gaussian Kernel Density estimate for spatial data

sp.kde(
  x,
  y = NULL,
  bw = NULL,
  newdata = NULL,
  nr = NULL,
  nc = NULL,
  standardize = FALSE,
  scale.factor = NULL,
  mask = TRUE
)

Arguments

x	sp SpatialPointsDataFrame object
y	Optional values, associated with x coordinates, to be used as weights
bw	Distance bandwidth of Gaussian Kernel, must be units of projection
newdata	A Rasterlayer, any sp class object or c[xmin,xmax,ymin,ymax] vector to estimate the kde extent
nr	Number of rows used for creating grid. If not defined a value based on extent or existing raster will be used
nc	Number of columns used for creating grid. If not defined a value based on extent or existing raster will be used
standardize	Standardize results to 0-1 (FALSE/TRUE)
scale.factor	Optional numeric scaling factor for the KDE (eg., 10000), to account for small estimate values
mask	(TRUE/FALSE) mask resulting raster if newdata is provided

Value

Raster class object containing kernel density estimate

Author

Jeffrey S. Evans <jeffrey_evans@tnc.org>

Examples

# \donttest{ 
 library(sp)
 library(raster)
   data(meuse)
   coordinates(meuse) <- ~x+y
       
 # Unweighted KDE (spatial locations only)        
 pt.kde <- sp.kde(x = meuse, bw = 1000, standardize = TRUE, 
                  nr=104, nc=78, scale.factor = 10000 )
#> Using extent of x to define grid
#> 
#> calculating unweighted kde
 
 # Plot results
   plot(pt.kde, main="Unweighted kde")
     points(meuse, pch=20, col="red") 
 
 #### Using existing raster(s) to define grid ####

 # Weighted KDE using cadmium and extent with row & col to define grid
 e <- c(178605, 181390, 329714, 333611) 
 cadmium.kde <- sp.kde(x = meuse, y = meuse$cadmium, bw = 1000,  
                       nr = 104, nc = 78, newdata = e, 
             standardize = TRUE, 
             scale.factor = 10000  )
#> 
#> calculating weighted kde
 plot(cadmium.kde)
   points(meuse, pch=19)
        
 # Weighted KDE using cadmium and raster object to define grid
 r <- raster::raster(raster::extent(c(178605, 181390, 329714, 333611)),
                     nrow=104, ncol=78)
   r[] <- rep(1,ncell(r))
 cadmium.kde <- sp.kde(x = meuse, y = meuse$cadmium, bw = 1000,  
                       newdata = r, standardize = TRUE, 
             scale.factor = 10000  )
#> using existing raster dimensions to define grid
#> 
#> calculating weighted kde
 plot(cadmium.kde)
   points(meuse, pch=19)
        
 # Weighted KDE using cadmium and SpatialPixelsDataFrame object to define grid
 data(meuse.grid)
 coordinates(meuse.grid) = ~x+y
 proj4string(meuse.grid) <- CRS("+init=epsg:28992")
#> Warning: Discarded datum Amersfoort in Proj4 definition
 gridded(meuse.grid) = TRUE
 cadmium.kde <- sp.kde(x = meuse, y = meuse$cadmium, bw = 1000,  
                       newdata = meuse.grid, standardize = TRUE, 
             scale.factor = 10000  )
#> using existing raster dimensions to define grid
#> 
#> calculating weighted kde
 plot(cadmium.kde)
   points(meuse, pch=19)
# }