GEOSTATS: meaning of this semi variogram
It's a little uncomfortable trying to learn and understand what one can
and can't do in geostatistics a long way away from anyone who really
understands this stuff. So I'm hoping some kind soul with a couple of
spare minutes can help us.
We have two sets of data (average monthly soil temperature) from
1) climate station data throughout the south island of New Zealand -
always on flat sites, mostly but not only on the plains or in rolling hill
country. #data = 23
2) a set of 7 clusters of data from data loggers. Each cluster covers
approximately 10-15km and is designed to sample a range of elevations
and aspects (stratified). The clusters are located at 7 sites throughout
the south island but only on the steeplands. #data = 109
We have detrended this data for latitude, elevation, distance from coast
and aspect. The resulting semivariogram of the residuals is attached. X
axis is metres. No declustering has been done. The semivariogram for
the data from the datalogger only shows a similar pattern but with points
for the intermediate lags missing. A semivariogram from climate station
data only (but a decade earlier when there were more stations) indicates
a range of spatial autocorrelation of 100km.
1) Any comments on the validity of what we have done (there are a
number of things we are unsure about)
2) Does the point at the first lag on the semivariogram indicate that there
is no spatial autocorrelation at 10km
3) Can this semivariogram be modelled (we presume not)
Can anyone advise of any useful papers where aspect has been
included in a kriging analysis?
Thanks very much for your help - I will summarise.