Semantically enriched statewide building dataset based on the Lower Saxony 3D Building Model for urban planning applications

This dataset provides a semantically enriched, GIS-compatible representation of all residential buildings across the federal state of Lower Saxony, Germany. It is derived from the official statewide 3D building model (AdV-CityGML 1.0, Level of Detail 2 – LoD2), as published by the Landesamt für Geoinformation und Landesvermessung Niedersachsen (LGLN) via OpenGeoData.NI. The original CityGML format is highly suitable for modeling detailed and hierarchically structured 3D building geometries (e.g., roof, wall, and ground surfaces). However, its complexity and nested structure pose significant challenges for geospatial processing and large-scale spatial and data analysis — particularly in open-source GIS environments or data science workflows, which typically rely on flat, tabular data structures such as Shapefiles, GeoPackages, or CSV files. Furthermore, the highly detailed 3D geometries introduce considerable computational overhead, making statewide analysis across millions of buildings prohibitively expensive in terms of processing time and hardware requirements. To address these challenges, the original dataset was algorithmically transformed into a lightweight, analysis-ready representation. Complex 3D geometries were generalized into 2D building footprints, while key semantic attributes — including building height, roof type, usage class, and 3D surface areas — were retained in structured form. This significantly reduces data volume and enables efficient geospatial analyses even on standard computing systems. The dataset is made available in two interoperable formats: GeoPackage (.gpkg), which provides native 2D geometries with enriched attribute data and supports spatial operations such as overlays, buffer zones, or proximity calculations in tools like QGIS or Python (GeoPandas). CSV (.csv), a flat tabular version containing identical attributes, with building geometries expressed as Well-Known Text (WKT), allowing seamless use in data science environments such as Python (pandas, seaborn, matplotlib), R, or Excel. By bridging semantic richness and technical accessibility, this dataset enables full-coverage, statewide spatial analysis at scale, supporting a wide range of applications in urban planning, environmental assessment, and geospatial research.