how to test for effect of spatial scale

Dear Maurizio -

it sounds as though your data are what my spatial stats teacher called
"lattice" data.

The particular analysis you propose - repeatedly aggregating different
blocks of data together and then comparing across scales; sounds like
the "grieg-smith" approach to testing for spatial randomness; and the
"two term local quadrat variance (TTLQV)" approach for identifying
patches.

For more a general background on lattice (aka areal) data, try chapter
9 in Bivand et al (2009) Applied spatial data analysis with R and
section 2.6 in Fortin and Dale (2005) Spatial Analysis: A guide for
Ecologists.

It may also be that the spatially explicit nature of your data is a
red herring in this case and what you're really after is something
like rarefaction or bootstrapping.

Cheers,

Anna

[...]
My objective is to detect at what spatial scale would such a
theoretical
prediction begin to break down, i.e., if I begin to disaggregate the
data to include all the (n-1) plots close to each other, all the
(n-2),
etc., down to statistics collected independently for each plot, when
does the mean distribution calculated for all possible combinations at
that scale begin to depart significantly from the expected curve.
I cannot figure out how to approach the problem. It is not point
pattern, nor does it seem to be geostatistics (building a semi
variogram
would help but only partially). The question is really about data
aggregation at different spatial scales for plots that are close to
each
other.

Any help appreciated.

thanks
Maurizio Mencuccini
school of Geosciences
University of Edinburgh (UK)

Dr. Anna E.  Marburg

Postdoctoral researcher
Ecosystem Processes Team
Landcare Research
P O Box 40
Lincoln 7640
NEW ZEALAND

phone + 64  3 321 9729
MarbugA at landcareresearch.co.nz




Please consider the environment before printing this email
Warning:  This electronic message together with any attachments is confidential. If you receive it in error: (i) you must not read, use, disclose, copy or retain it; (ii) please contact the sender immediately by reply email and then delete the emails.
The views expressed in this email may not be those of Landcare Research New Zealand Limited. http://www.landcareresearch.co.nz