The SIM2D Procedure

The SIM2D procedure produces a single output data set: the OUTSIM=SAS-data-set. The OUTSIM= data set contains all the needed information to uniquely identify the simulated values.

The OUTSIM= data set contains the following variables:

• LABEL, which is the label for the current SIMULATE statement

• VARNAME, which is the name of the conditioning variable for the current SIMULATE statement

• MODSVAR, which is the name of the input item store variable associated with the current correlation model in an unconditional simulation

• _ITER_, which is the iteration number within the current SIMULATE statement

• GXC, which is the coordinate for the current grid point

• GYC, which is the coordinate for the current grid point

• SVALUE, which is the value of the simulated variable

If you specify the NARROW option in the PROC SIM2D statement, the LABEL and VARNAME variables are not included in the OUTSIM= data set. This option is useful in the case where the number of data points, grid points, and realizations are such that they generate a very large OUTSIM= data set. The size of the OUTSIM= data set is reduced when these variables are not included.

In unconditional simulation tasks where no input data set is specified, the VARNAME variable is excluded from the OUTSIM= data set. In unconditional simulation tasks where you have specified an input data set, the VARNAME variable is included but given a missing value. In the case of mixed conditional and unconditional simulations (that is, when multiple SIMULATE statements are specified, among which one or more contain a VAR= specification and one or more have no VAR= specification), the VARNAME variable is included but is given a missing value for those observations that correspond to an unconditional simulation.

The MODSVAR variable is included in the OUTSIM= data set only when you specify an input item store with the RESTORE statement, and it indicates the presence of that store. This variable helps you identify the output of different unconditional simulations when the model input comes from an item store; it is not suggesting that a simulation task is conditioned upon it.

Specifically, the MODSVAR variable has a missing value for conditional simulation tasks. The variable also has a missing value for unconditional simulations for which you specify a correlation model explicitly, either with the FORM= option or with the MDATA= data set in the SIMULATE statement. In all other cases, the MODSVAR variable indicates the input item store variable that is associated with the store model used for the current unconditional simulation task.