Historically, ground-warfare simulation programs have developed project-specific terrain generation systems. These stove-pipe systems satisfy a single program's requirements, involve a lot of manual editing, and employ little verification during processing. The government has invested millions of dollars developing runtime databases covering the same geographic areas. The generated databases lack correlation due to different processing tools and techniques, resulting in fair-fight issues and visual anomalies when interoperating in a confederacy.

The effort described in this paper strives to solve these problems. The goal is a means to rapidly generate high-fidelity urban terrain databases using existing applications while removing dependence on any particular tool. In addition, the solution must have the flexiblibility to evolve as programs identify new requirements. The resultant capability must import from a wide variety of sources, clean and normalize source data to a consistent representation, and deliver a correlated dataset that meets the needs of a confederacy.

This paper describes the technical challenges involved with developing an adaptable urban terrain generation framework. We'll take a look at each how the components of the system interact and discuss problems, deficiencies, and bottlenecks encountered during development. Finally, we conclude with the current state of the system and to what degree it is meeting overall expectations.