A Data Model for Geometric, Semantic and Spatial Topology Validation of CityGML Datasets

Abstract. 3D city models are gaining importance across applications such as urban analytics and digital twins, where geometric and semantic data quality directly determine downstream usability. While tools for validating the geometric and semantic consistency of individual CityGML features already exist, support for validating the spatial topology between distinct features remains absent from current validation tools. 
This paper presents ongoing research extending the data model of CityDoctor2 (CD2), an open-source Java library for CityGML validation and repair, to enable spatial topology validation at city-model scale. We describe CD2’s existing data model and detail the architectural adaptations required to support topology checks: replacing the original streaming-based parser with a persistent, database-backed storage layer complemented by an in-memory cache, enabling spatial indexing, multi-threaded validation, and candidate-pair queries via bounding-box predicates. Performance evaluation on a CityGML test dataset shows the new architecture incurs a considerable increase in runtime when compared to the previous streaming approach, in exchange for the ability to perform spatial topology validation on datasets too large to hold entirely in memory.