In modelling a variogram I find that often the closest pairs start above the semi-variance and then it dips below the semi-variance at longer ranges. This makes it very difficult to model. What would cause this?

Similar questions and discussions