MVN is the Multivariate Normal sampling that gives a value for a given point based on it's Geologic / Terrain Classification and proximity to a measured site. The MVN approach is applied to the Geology and Terrain categories independantly.
It is an alternative to the krigging approach. It has minimal differences to krigging in areas of dense data. However, in sparse areas it does not overweight the measured points for larger distances from the measured point.
The baseline model estimate and the observation are combined and the proportional influence is a function of the measurement uncertainty.