[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: nested variogram models
Michael Bock wrote:
> is there a way to use 'nested' varigrams for kriging-interpolation in
> gstat? 'Nested' means for example to use a gaussian model for lag 1-4
> and an exponential one for the farer lags.
Micheal, your explanation is not the usual one for "nested variograms".
In fact, the gstat implementation (using sums of partial models) is.
Try for instance plotting the variogram model
1 Sph(200) + 1 Lin(2000)
It looks as if the linear structure is absent at lags smaller than 200,
and present at larger lags. But it isn't.
There are good reasons for using this concept. Try some text book, and
look for positive definiteness of semivariance functions.