Visualization is especially useful in allowing modelers to interrogate, in great detail, the data sets produced by their models. Three dimensional, time dependent data sets are particularly difficult to digest. The computational resources required to produce such data sets are often very large. In research mode, a computational scientist will frequently have to recompute a set of results, using different input parameters, initial conditions and boundary conditions to expose certain details or features in the mathematical model. Visualization helps the modeler ensure that all of the detail available from a particular set of computational results is exposed before undertaking the often expensive task of recomputing.