I played around with things some more and found that the variability is coming from the ter rebuild and the ter size. One model I don't have to scale the ter up to get the desired results, however, I have to rebuild it with more points. In another model I have to scale it up and rebuild it. To complicate things more how many points you add in the rebuild affects things. Given all the different combinations of possible things going on, there is going to be a lot of variability, no way around it.