Abstract: Due to collaborative efforts and advances in data acquisition technology, a large volume of geometric models describing urban buildings has become available in public domain via "Digital Earth" software like ESRI ArcGlobe and Google Earth. As a consequence, almost every major international city has been reconstructed in the virtual world. Although mostly created for visualization, we believe that these urban models can benefit many applications beyond visualization including video games, city scale evacuation plans, traffic simulations, and earth phenomenon simulations. However, before these urban models can be used in these applications, they require tedious manual preparation that usually takes weeks, if not months. In this paper, we present a framework that produces disjoint 2D ground plans from these urban models, an important step in the preparation process. Designing algorithms that can robustly and efficiently handle unstructured urban models at city scale is the main technical challenge. In this work, we show both theoretically and empirically that our method is resolution complete, efficient, and numerically stable.