Hi Dafne
If you going to look at GSLIB (with its FORTRAN code), then this is also
useful:
Goovaerts P (2009) AUTO-IK: A 2D indicator kriging program for the
automated non-parametric modeling of local uncertainty in earth sciences.
Comput Geosci 35:1255-1270
Cheers Paul
-----Original Message-----
From: R-sig-Geo [mailto:r-sig-geo-bounces at r-project.org] On Behalf Of
Dafne Berg
Sent: 22 October 2015 10:12
To: Edzer Pebesma
Cc: r-sig-geo at r-project.org
Subject: Re: [R-sig-Geo] indicator kriging with R
Dear Edzer,
yes it is exactly what I meant. Thanks for the confirmation that it is not
yet implemented in gstat. I will have a look to gslib and the book.
Thanks a lot
Dafne
On 22 October 2015 at 09:38, Edzer Pebesma <edzer.pebesma at uni-muenster.de>
wrote:
Yes, this is what indicator kriging does.
What you want is probably going back from the [0..1] estimated
probabilities to the original Z values. Gstat doesn't help here, but
it shouldn't be too hard to write a function for this. You need to
make assumptions about the distribution between different cutoff
values, in particular about both extreme (open) classes. I think the
GSLIB and Goovaerts books would be the first ones to look into.
On 22/10/15 10:19, Dafne Berg wrote:
Dear list,
I am trying to do a indicator kriging of a continuous variable. I am
especially interested in performing the "order relation correction".
In
documentation of "predict" in gstat it is mentioned as debug level
64 (order relation violations (indicator kriging values before and
after
relation correction).
However, when running predict I can only get the values in the range
0 to 1, as expected for indicator kriging.
I am probably missing something very basic here, but I will be
grateful
any pointers in the right direction
Thank you very much in advance
Best wishes
Dafne
[[alternative HTML version deleted]]