Search code examples
rexportrasterr-rasterkriging

Problem in exporting a kriging map as Raster

I created a kriging map which was created by using the kriging() and image() functions of the kriging package (table is the data with the coords and values):

krig <- kriging(table@coords[ ,1], table@coords[ ,2], response = table@data$Joined.l_8, model = "spherical", lags = 3, pixels = 100)
krig_raster <- image(krig, main = NULL, xlab = "X coords", ylab = "", col = heat.colors(100))

The result can be viewed in the picture attached.

enter image description here

My problem is that I cannot export this map as a raster. When using the following command of the raster package:

writeRaster(krig_raster, "/home/stathis/Desktop/test.tif", format="Gtiff", overwrite = TRUE)

I get the following error:

Error in (function (classes, fdef, mtable)  : 
  unable to find an inherited method for function ‘writeRaster’ for signature ‘"NULL", "character"’
Solution

  • When you do:

     krig_raster <- image(krig, main = NULL, xlab = "X coords", ylab = "", col = heat.colors(100))

    it plots the image and returns a NULL, so writeRaster is trying to write a NULL and gives you an error saying exactly that. You need to convert the output from kriging to a raster object first. Reading to docs for kriging tells me the output predictions are in the $map part of the returned object, and in a three-column format that should be okay to feed into raster::rasterFromXYZ. If I use the example in ?kriging and do this with the kriged object:

    > r = rasterFromXYZ(kriged$map)
> plot(r)

    I see a plot of a raster class object of the USA which I can then save using writeRaster via writeRaster(r, "usa.tif")

    As a note I'd be wary of using a kriging function that doesn't return prediction variances or let you first examine the variogram before proceeding - check out the gstat package for more thorough kriging procedures.