[R-sig-Geo] Questions about spatio-temporal modelling

[Luca Candeloro]

Starting from environmental and metereological data, I have defined an
annual count variable (number of favourable events, raster type object).
The purpose of the analysis is to predict the next year favourable events
raster, given the time series (last 15 years).
Working at the pixel level, it would be possible to make a Poisson
regression, but treating pixels independently, would loose spatial
effects...
Which is, in your opinion, the best approach?
Is there a spatio-temporal model for this kind of data that could usefully
combine spatial effect  with time series analysis?
Thanks for any suggestions

	[[alternative HTML version deleted]]