Variogram in spatial package

A resubmission, as no one has replied and I still have
the problem.

Date: Wed, 17 Jan 2007 09:58:10 +0000 (GMT)
From: Hywel Jones <hywelm.jones at talk21.com>
Subject: Variogram in spatial package
To: r-sig-geo at stat.math.ethz.ch

The help page for variogram in the spatial package
leaves my slightly uncertain about a few things. I'd
be grateful for confirmation of my understanding.

If I fit a trend surface using surf.ls, and then set
the krig parameter to use that object in the
variogram
function, I'm assuming that the variogram produced
is
calculated for the residuals contained in the trend
surface object. Is that right?

I'm afraid I don't have the references to check the
following either. Using notation of Cressie, am I
right in thinking that the y co-ordinate of the
variogram corresponds to gamma (or 2x gamma)? And x:
is that h? 

And then, how does h correspond to my original data?
i.e. do I interpret it as distance calculated with
the
original x and y submitted to surf.ls, or as
distance
calculated with the rescaled x and y used within
surf.ls (I understand that the internals rescale x
and
y to -1:1).

I'd actually like to check for isotropy before using
this variogram function. Any suggestions as to
functions I might use?

Thanks in advance.

Hywel

___________________________________________________________
What kind of emailer are you? Find out today - get a
free analysis of your email personality. Take the
quiz at the Yahoo! Mail Championship. 

http://uk.rd.yahoo.com/evt=44106/*http://mail.yahoo.net/uk

___________________________________________________________ 
Inbox full of unwanted email? Get leading protection and 1GB storage with All New Yahoo! Mail. http://uk.docs.yahoo.com/nowyoucan.html
Hi Hywel,

I'm not familiar with 'spatial' but you might look at the variogram tools in 'geoR'.  This should allow you do look for anisotropy - I believe the function is vario4.

You should be able to check empirically if the variogram is on the residuals by comparing variogram with surf.ls with variogram without surf.ls but based on the residuals that you calculate yourself with lm().

Presumably 'x' is h, and you should be able to figure out the units based on the range of the 'x' axis...
I believe in geoR the output of variogram is labelled 'semivariance' so you can be sure it is gamma and not 2 gamma.
Not sure about variogram from 'spatial'.

 ----------------------------------------------------------------------------------------------
Chris Paciorek / Asst. Professor        Email: paciorek at hsph.harvard.edu
Department of Biostatistics             Voice: 617-432-4912
Harvard School of Public Health         Fax:   617-432-5619
655 Huntington Av., Bldg. 2-407         WWW: www.biostat.harvard.edu/~paciorek
Boston, MA 02115 USA                    Permanent forward: paciorek at alumni.cmu.edu
Hywel Jones <hywelm.jones at talk21.com> 01/26/07 4:50 PM >>> 
A resubmission, as no one has replied and I still have
the problem.

Date: Wed, 17 Jan 2007 09:58:10 +0000 (GMT)
From: Hywel Jones <hywelm.jones at talk21.com>
Subject: Variogram in spatial package
To: r- sig- geo at stat.math.ethz.ch

The help page for variogram in the spatial package
leaves my slightly uncertain about a few things. I'd
be grateful for confirmation of my understanding.

If I fit a trend surface using surf.ls, and then set
the krig parameter to use that object in the
variogram
function, I'm assuming that the variogram produced
is
calculated for the residuals contained in the trend
surface object. Is that right?

I'm afraid I don't have the references to check the
following either. Using notation of Cressie, am I
right in thinking that the y co- ordinate of the
variogram corresponds to gamma (or 2x gamma)? And x:
is that h? 

And then, how does h correspond to my original data?
i.e. do I interpret it as distance calculated with
the
original x and y submitted to surf.ls, or as
distance
calculated with the rescaled x and y used within
surf.ls (I understand that the internals rescale x
and
y to - 1:1).

I'd actually like to check for isotropy before using
this variogram function. Any suggestions as to
functions I might use?

Thanks in advance.

Hywel

___________________________________________________________
What kind of emailer are you? Find out today -  get a
free analysis of your email personality. Take the
quiz at the Yahoo! Mail Championship. 

http://uk.rd.yahoo.com/evt=44106/*http://mail.yahoo.net/uk

___________________________________________________________ 
Inbox full of unwanted email? Get leading protection and 1GB storage with All New Yahoo! Mail. http://uk.docs.yahoo.com/nowyoucan.html

_______________________________________________
R- sig- Geo mailing list
R- sig- Geo at stat.math.ethz.ch
https://stat.ethz.ch/mailman/listinfo/r- sig- geo

A resubmission, as no one has replied and I still have
the problem.
--- Hywel Jones <hywelm.jones at talk21.com> wrote:

Date: Wed, 17 Jan 2007 09:58:10 +0000 (GMT)
From: Hywel Jones <hywelm.jones at talk21.com>
Subject: Variogram in spatial package
To: r-sig-geo at stat.math.ethz.ch

The help page for variogram in the spatial package
leaves my slightly uncertain about a few things. I'd
be grateful for confirmation of my understanding.
The spatial package is described in more detail in Venables, W. N. and
Ripley, B. D. (2002) _Modern Applied Statistics with S._ Fourth edition.  
Springer, as the help page says.
If I fit a trend surface using surf.ls, and then set
the krig parameter to use that object in the
variogram
function, I'm assuming that the variogram produced
is
calculated for the residuals contained in the trend
surface object. Is that right?
Yes, I believe so.
I'm afraid I don't have the references to check the
following either. Using notation of Cressie, am I
right in thinking that the y co-ordinate of the
variogram corresponds to gamma (or 2x gamma)? And x:
is that h? 
The documentation is the source code. Alternatively run the same data
through similar functions in the gstat, geoR, or fields packages to
cross-check?  My reading of MASS is that the plot is of the semi-variogram
with distances in the original scaling.
And then, how does h correspond to my original data?
i.e. do I interpret it as distance calculated with
the
original x and y submitted to surf.ls, or as
distance
calculated with the rescaled x and y used within
surf.ls (I understand that the internals rescale x
and
y to -1:1).

I'd actually like to check for isotropy before using
this variogram function. Any suggestions as to
functions I might use?
I don't think anisotropy is available in the spatial package, perhaps try 
one of the other packages?
Thanks in advance.

Hywel

_______________________________________________
R-sig-Geo mailing list
R-sig-Geo at stat.math.ethz.ch
https://stat.ethz.ch/mailman/listinfo/r-sig-geo

Roger Bivand
Economic Geography Section, Department of Economics, Norwegian School of
Economics and Business Administration, Helleveien 30, N-5045 Bergen,
Norway. voice: +47 55 95 93 55; fax +47 55 95 95 43
e-mail: Roger.Bivand at nhh.no