Another possibility is the blockPredict function in the intamap-package
(also based on gstat), which can do transGaussian kriging.
Fixing lambda = 0 is equal to lognormal kriging.
A new version with a couple of inconsistencies removed and a better
example for lognormal kriging (1.3-18) has just
Below is an example for the meuse data in sp, using gstat. I've added
this as a demo script, "lnsim.R", to gstat development tree:
https://r-forge.r-project.org/scm/viewvc.php/pkg/demo/lnsim.R?view=markup&root=gstat
On 02/05/2013 12:41 PM, Paulo Justiniano Ribeiro Jr wrote:
Indeed geoR does
Indeed geoR does not have it in an automated way.
(conditional) simulations can be obtained but then this
computations on blocks would be a sort of post-processing
along the lines Edzer describes.
I'm a bit unsure of the worth of implementing this step
for a sufficient flexiblke set of definitions
Javier,
simulation is relatively easy: simulate point support fields on a fine
grid, take the exponent of everything, compute the block mean of each
simulation (e.g. using sp::aggregate), and take the statistics of the
block means for each block. (There is just not yet a simple function
doing
Dear all,
I am trying to calculate the average value in a study area. I carried out the
sampling according to a grid, and the data are lognormal. I would like to do
with block krige (ordinary krige), I do the block krige with logarithms but I
don't know how to back-transforming.
I read Cressi
