[GRASS-stats] Spatial autocorrelation of multi-spectral, uni- and mutli-temporal data sets

[Nikos Alexandris]

Greets to the statists,

I want to "describe" my multispectral (Landsat5_TM) composite datasets with 
respect to their between vs. within heterogeneity. The idea is that a 
unitemporal data set exhibits less between-axes than within-axes (spectral 
bands) heterogeneity. The opposite "should" be in the case of a bi-temporal 
dataset (in my case a pre-fire and a postfire), where the between-axes 
(spectral bands) "should" be more contrasted.

I was looking for various multivariate tests but found nothing that works 
globally on the images (i.e. without the necessity to work on samples/classes 
extracted from  the images), nothing that I can handle without the need to do 
my homework for hours first, something easy to understand, estimate and 
explain.

(I also asked in r-user but due to the nature of the question I guess it 
correctly passed unanswered.)

A friend suggested spatial autocorrelation as an option (mentioned (also) 
Moran's Index, Jaccart, MANOVA, Mixed effect model). I have a very basic 
experience on autocorrelation (reading the book Applied Spatial Analysis with 
R and having dome some exercises with a friend using climatic data). While I 
am studying potential answers, I will be eXtremely grateful for any help, 
advise, comment, hint on it (always doable within grass; R).

Milles mercis, Nikos.