Well, you could do a vertical variogram and then use some sort of multiple of that for your horizontal variograms based on the sedimentary environment. There's probably some literature out there relating horizontal and vertical variograms to give you some guidance. I'm not sure whether the exercise is worthwhile, however. It might be better just to go with a gaussian distribution.
In the vertical direction, this is possible due to the large number of data, but in the horizontal direction, it is not possible to make an accurate estimate using the variogram for the entire reservoir.