## Re: AI-GEOSTATS: "strength" of autocorrelation

Expand Messages
• Volker I don t understand how you can describe a curve as shallow if you consider the sill to be irrelevant. How do you measure shallowness without a
Message 1 of 2 , Feb 9, 2004
• 0 Attachment
Volker

I don't understand how you can describe a curve as
'shallow' if you consider the sill to be irrelevant.
How do you measure 'shallowness' without a reference
to the sill?

I woudl also like to say that your hard line approach
to the nugget effect is not shared by those of us who
work at the 'sharp end'. I have investigated cases
where the nugget effect is clearly visible as part of
the structure of the phenomenon and have had the
opportunity to research into controlled trials on
contiguous sampling on several occasions.

No matter how close you take your sample, no matter
how well you sample, there will always be a difference
between two contiguous samples. This difference may be
minimal, in which case you will see no nugget effect.
Sometimes it is a significant part of the total
variation of the sample values.

Some of the nugget effect will be errors introduced by
sampling, but only a full balanced statistical study
will convince me that all of the nugget effect is
sampling error.

Isobel Clark
http://geocities.bizland.com/whatsnew.htm

--- Volker Bahn <lochapoka@...> wrote: > Dear list
members,
>
> I would like your opinions on my ideas on the
> definition of "strength" of
> autocorrelation as interpreted through a variogram.
> There seem to be two
> important parts: a) the range and b) the shape of
> the variogram. (Whereas
> the sill and the nugget are in my opinion not of
> significance to the problem
> as the former is only an expression of the
> variability in the data and the
> latter should theoretically not exist at all and is
> most likely a
> measurement error or distance bin artifact.) So for
> a constant range, the
> more shallow the variogram the stronger the
> autocorrelation and for a
> constant curve, the longer the range the stronger
> the autocorrelation. Could
> I maybe capture both notions by extracting the area
> under (i.e., intergrate
> over) a correlogram (let's say up to where it
> crosses zero)? Has anyone
> invented or successfully used such a measure? Does
> it make sense to you?
>
> Cheers
>
> Volker
>
> -~^/\^~^/\~/\^~^^/\/\~^~*
>
> Dept. of Wildlife Ecology - Rm. 210
> University of Maine
> 5755 Nutting Hall
> Orono, Maine
> 04469-5755, USA
> Tel: (207) 581 2799
> Fax: (207) 581 2858
>
>
> --
> * To post a message to the list, send it to
> ai-geostats@...
> * As a general service to the users, please remember
> to post a summary of any useful responses to your
> questions.
> * To unsubscribe, send an email to majordomo@...
> with no subject and "unsubscribe ai-geostats"
> followed by "end" on the next line in the message
> body. DO NOT SEND Subscribe/Unsubscribe requests to
> the list
> * Support to the list is provided at
http://www.ai-geostats.org

___________________________________________________________