[SOLVED] uc parameter Kriging syntax problem

Any question regarding the Interpolation method using Kriging

[SOLVED] uc parameter Kriging syntax problem

Postby xgoodyear » Wed Nov 12, 2014 8:16 pm

Hello:
I am using this code:

KrigeU <- kriging(ondgdb, grid.db,
model = data.modelXX, neigh = moving.neigh,
uc=c("2","3"), mean = NA, flag.linked = FALSE, calcul = "point",
ndisc = NA, cov.extract = NA, drift.extract = c(2,3),
flag.est = TRUE, flag.std = TRUE,
flag.colk = FALSE, flag.grad = FALSE,
ball.radius = 1,radix = "KM.All",modify.target = TRUE)

When I ran it, appears in the command line the list of authorized drift names, and then R stops.
I want to get Kriging Universal (because my data has drift in X and Y). I don`t know if the syntax of uc parameter is the problem.

Thanks.
xgoodyear
 
Posts: 5
Joined: Fri Aug 01, 2014 11:09 pm

Re: uc parameter Kriging syntax problem

Postby Didier Renard » Fri Nov 14, 2014 9:19 pm

Hello

The syntax of the kriging method that you used is not correct, as you pointed out.
As you know, the drift is specified via the 'uc' argument.
- if you want to perform an Ordinary kriging, you need to specify the universality Condition: use uc="1"
- if you want to perform a Simple Kriging, you want to discard any drift function: use uc=""
- if you want to specify a first order (internal) drift in a 2-D space: use uc=c("1","x","y")
- if you want to specify a single external drift: use uc=c("1","f1")
- if you want to specify two external drifts: use uc=c("1","f1","f2")

As you noticed, the documentation concerning the specification of the drift is not sufficient. It will be improved in 10.0.6
where a function uc.names() has been added.

Hope this will help.
Didier Renard
 
Posts: 337
Joined: Thu Sep 20, 2012 4:22 pm

Re: uc parameter Kriging syntax problem

Postby xgoodyear » Mon Nov 17, 2014 10:47 am

Didier, thank you very much for your explanation.

Only one question more what is the purpose of the parameter: drift.extract in
the function kriging ?

Regards,

Xavier B.
xgoodyear
 
Posts: 5
Joined: Fri Aug 01, 2014 11:09 pm

Re: uc parameter Kriging syntax problem

Postby Didier Renard » Mon Nov 17, 2014 7:18 pm

drift.extract and cov.extract are arguments which allow turning the kriging system into a factorial kriging (usually described only for the case of covariances).
Factorial Kriging approach. We consider that the measured variable Z is written as the sum of two underlying (and invisible) variables Y1 and Y2 (spatially independent). Then we can write:

Z = Y1 + Y2

IF we assume that Y1 is characterized by C1: Cov[Y1(x),Y1(x+h)] = C1(h)
Similarly: Cov[Y2(x),Y2(x+y)] = C2(h)
Spatially independent: Cov(Y1(x),Y2(x+h)] = 0

Then we see that:

Cov[Z(x),Z(x+h)] = C1(h) + C2(h)

In the factorial kriging, we try to estimate Y1 from the data measured in Z. This can be written in the usual kriging formalism, simply changing the right-hand side of the kriging system.

So, when using kriging() method, it suffices to:
- specify the model of Z (i.e. the one of Y1 + Y2)
- "extract" the part of Z which is relative to Y1.

Hope this will help
Didier Renard
 
Posts: 337
Joined: Thu Sep 20, 2012 4:22 pm

Re: uc parameter Kriging syntax problem

Postby xgoodyear » Thu Nov 20, 2014 3:33 pm

Thank you Didier for your clear explanation.

Regards,

Xavier B.
xgoodyear
 
Posts: 5
Joined: Fri Aug 01, 2014 11:09 pm


Return to Kriging

Who is online

Users browsing this forum: No registered users and 3 guests

cron