kriging improvements

[kenohori]

• [x] check math
• [ ] improve figures with new code
• [x] add overview of kriging types

[edzer]

A further improvement: in figure 6.2b the sill is 140, according to the text on the same page it is 160. The latter is the usual definition, the value between 20 and 160 in the figure is often denoted as "partial sill" (the sill of the, in this case, linear model component). It would further be nice to not plot a linear-with-sill model in 6.2b ,but for instance a spherical model (as that reaches the sill, in contrast to Gaussian and exponential) because the linear-with-sill model is not positive definite in 2 or more dimensions.

[kenohori]

A further improvement: in figure 6.2b the sill is 140, according to the text on the same page it is 160. The latter is the usual definition, the value between 20 and 160 in the figure is often denoted as "partial sill" (the sill of the, in this case, linear model component). It would further be nice to not plot a linear-with-sill model in 6.2b ,but for instance a spherical model (as that reaches the sill, in contrast to Gaussian and exponential) because the linear-with-sill model is not positive definite in 2 or more dimensions.

thanks! fixed in 83d79b0