We want a standardized set of simulations. The goal is NOT to allow for infinite flexibility.
Instead, we want a (smallish) set of options to ensure utility acroos many models.

Need a fixed set of options to allow options for:
1) multivariability (one, two, three)
2) non-stationarity (none, low, high)
3) isotropy (none, low, high)
4) smoothly varying parameters (smooth, sharp)
5) grid (T/F)
6) space/space-time
7) number of points (easy/med/large)
8) gaussian/non-gaussian fields (transformation or distribution based)
9) number of simulations/realizations
10) covariance that allows for scientifically relevant covariates
11) specifying the amount of holdout observations

Notes:
1) parameters that aren't specified are all generated
2) ability to resimulate the same realizations for reproducibility
3) could make a generate a large set of datasets and put them on a site
4) output in sp object (and geoR object)

To-do:
1) Martin needs some input (this week!) for nonstationary model choices (model and parameters choices)
2) We need to have some people with different methods test it before it's released.