Cities around the world with open CityGML datasets

dataset country year Building LOD other classes textures acquisition CityGML version notes
Berlin Germany 2013 LOD2 true 2.0 Released in 2015
Brussels Belgium 2014 LOD2 Building false 1.0
Dresden Germany 2009 LOD1/LOD2/LOD3 Partially 1.0
Dutch cities Netherlands 2016 LOD1 Terrain and many other false 2.0 A few Dutch cities generated with 3dfier
Hamburg Germany 2017 LOD1 and LOD2 Cadastre footprints + LiDAR 2.0
Helsinki Finland 2016 LOD2 true 2.0
Linz Austria 2011 LOD2 false 1.0
Lyon France 2012 LOD2 Terrain, water 2.0
Montréal Canada 2009 LOD2 terrain (TIN in CityGML format) true Photogrammetry 1.0 The LiDAR dataset of the same area is also available
New York City (by TUM) United States 2015 LOD1 Roads, lots, parks, water, terrain false Photogrammetry in combination with existing public 2(.5)D datasets 2.0 article with details
New York City by DoITT United States 2016 LOD2 false Cadastre footprints + LiDAR 2.0 buildings are modeled with thematic surfaces in LOD2, however, for most buildings the geometric shape is LOD1
North Rhine-Westphalia (state) Germany 2016 LOD1+LOD2 false Cadastre footprints, LiDAR, aerial images. LOD1 is derived from LOD2 models with average roof height (details here) 1.0 Enormous datasets that covers whole NRW (NRW is the most populated state in Germany) Cities included: Düsseldorf, Essen, Oberhausen, Köln, Bonn and more
Rotterdam Netherlands 2010 LOD2 true Cadastre footprints + LiDAR 1.0
The Hague Netherlands 2011 LOD2 Terrain false 1.0 It contains terrain intersection curves!
Thuringia (state) Germany 2013 (cadastre footprints) - 2016 and earlier (LiDAR) LOD1 (full) + LOD2 (partly) false Cadastre footprints + LiDAR 1.0 Datasets that covers whole Thuringia (Thüringen); Cities included: Erfurt, Jena, Gera, Gotha
Vienna Austria 2018 LOD2 false 1.0 From the building footprints of the area multipurpose map and a surface model semi-automatically derived prototypical roof forms.

Cities around the world with open 3D datasets not in CityGML

dataset country year Building LOD other classes textures acquisition formats notes
Adelaide Australia 2015 LOD1/LOD2 true .3DS with JPEG textures, Blender, FBX terrain
Austin USA 2013 LOD2 false KMZ, Shapefile
Boston USA 2013 LOD2 false OBJ, DWG, MAX terrain
Cambridge UK LOD1 false DXF, Shapefile, COLLADA
Fredericton Canada 2016 LOD2 true KMZ
Greater Geelong Australia 2016 LOD1/LOD2 false KML terrain
Philadelphia USA 2015 LOD2 true GDB
San Francisco USA 2017 LOD1 false KML, KMZ, Shapefile, GeoJSON
Toronto Canada 2016 LOD1 Shapefile, Geodatabase, AutoCAD, SketchUp
Vancouver Canada 2009 LOD1 false Shapefile, DWG, KML


Open LiDAR datasets

Another potentially relevant list is the one of national lidar datasets on Wikipedia. These datasets can be used as input to generate 3D city models. A possibility is to use the open-source software 3dfier developed at TU Delft, or other CityGML software.