Spatial Interpolation -- Notes, Overheads and R Code
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

The file spatial-notes.pdf (8 pages) are an introduction to the
spatial abilities available in R packages.

Other files relate to a talk given (April 15) at the course described
at: http://www.geostat-course.org/GeoSciences_AU_2011

interp-ohp.pdf: Discusses and demonstrates the use of Generalized
Additive models to fit smooth curves and surfaces, on their own or in
conjunction with kriging, for spatial interpolation.

spatial-notes.pdf: Brief notes on spatial interpolation using R.
[added May 31, 2011]

The file meusesp.Rdata in this directory holds the object meusesp 
of class "SpatialPointsDataFrame".  An object that has all the same
columns except ahn can be created thus:

  library(gstat)    # sp is a dependency, and will also be attached
  data(meuse)
  meusesp <- meuse
  meusesp$Soil <- meuse$soil
  levels(meusesp$Soil) <- c("1","2","1")
  proj4string(meusesp) <- CRS("+init=epsg:28992")
  coordinates(meusesp) <- ~x+y

The easiest way to see how the functions cv.gamkrige() and
compare.gamkrige() work is to run them, using default arguments:
1) Create meusesp as above, or load or attach meusesp.RData
from this directory into your workspace.
2) Type cv.gamkrige()
3) Type compare.gamkrige()  [this calls cv.gamkrige()]

Now experiment, if you are so inclined, with models that are
alternative to the defaults used in the calls to the functions
cv.gamkrige() and compare.gamkrige().

The file slidefuns.R has code for the functions used to create
the graphs in the overheads spatial.pdf

NB: Attaching or loading meusesp.RData from this directory will
give an object that has the columns Soil and ahn (this is height
above sea level, where elev is height above river bed).

John Maindonald
April 10 2011.