To Krige Or Not To Krige?
The best I can do is point out why nobody should work with surreal geostatistics. Geostatistocrats assume spatial dependence between measured values, interpolate by kriging, select some least biased subset of kriged estimates, smooth it to perfection, and rig the rules of real statistics with impunity. Statisticians test for spatial dependence by applying Fisher’s F-test to the variance of the set, and the first variance term of the ordered set. Interpolate between measured values only if the set displays a significant degree of spatial dependence at 95% probability. Do count degrees of freedom. A set of n measured values gives df=n-1 degrees of freedom. The first variance term of the ordered set gives dfo=2(n-1) degrees of freedom. Degrees of freedom for sets of unevenly spaced measured values are positive irrationals. ■
Thanks
Thanks Jan,
I'm quite new to this so your help is much appreciated. With a bit more reading I found out that the data does not have to be normally distributed. Data with a spatial trend can be delt with using universal kriging and defining parameters to account for the drift. This is the next part i'm trying to figure out. I know there is a trend (drift) but don't know if its significant or how to determine the parameters
I wish there was a book - Geostats for dummies 
■
Kriging Groundwater Levels
Hi,
I'm trying to optimise a groundwater level monitoring network using Kriging. I'm new to geostats so have some really simple questions. My first question are:
1) when determining the semivariogram model parameters to use for kriging groundwater levels does the data have to be normally distributed?
2) If there is spatial trends do I have to detrend or transform the data?
Thanks for your help
Simon ■