In many cases, a graphics screen provides the most efficient form of data entry. A scientist can manipulate graphics objects on the computer screen to create or modify data sets. Assume that a correspondence is to be made between the numerical value of some model parameter and the color of a pixel on a computer screen. The number of pixels on a workstation is roughly a million. Visualization software permits a modeler to ``pan'' the screen over even larger data sets or to ``stack'' data sets, one screen behind another. In this fashion a computational scientist is capable of setting many millions of data values by simply ``painting'' the screen. For example, with just one pass around the screen drawing the boundary of a section of the aquifer, the modeler can specify which finite difference cells lie within the model region and which cells are exterior. Similarly, by simply using a paint program, the modeler can assign colors to hydraulic properties and quickly characterize a million point planar region of the model. Section 5 of this chapter includes examples of such object oriented data base manipulation. Finally, before the model is run, the modeler can query the data base in a graphical manner. This often permits the modeler to spot errors or inconsistencies in massive data sets that would be difficult to inspect by any nongraphical method.