Factors that currently limit the development of such systems include the lack of terrain and environmental data at the level of detail required for ground-based simulations, the magnitude of the scene generation process, and the inadequacy of input devices that allow a user to interact with the virtual world. In this paper, we address two important issues in automating the construction of large scale virtual worlds. First we describe the process by which various sources of spatial data can be brought together into a common representation framework. We then describe how this data can be automatically transformed into an efficient representation of the underlying virtual world suitable for graphical presentation. This includes the integration of man-made features such as buildings, roads, and bridges with terrain and surface material information into a common `terrain skin'. Several examples are given using large scale spatial databases developed for distributed interactive simulation (DIS) applications.