From OpenStreetMap Wiki
Jump to: navigation, search


The purpose of this import is to add high-quality building footprints for the city of Burlington, Vermont USA to Openstreetmap.

Data source and approval

Maintained by the city office of planning and zoning since 1979, the Burlington building footprints layer is officially in the public domain, along with other citywide datasets. Infrequent updates meant that the dataset was not reflective of all existing buildings at the time of import; some omissions are visible.

When asked for approval to include this data in OpenStreetmap under the terms of the ODbL, city GIS manager Jay Appleton replied:

"City data are in the public domain. They are the people’s data."

He also specified that he was not in a position to license the original dataset as ODbL, but he was clear that its public domain status allowed such use by third parties like OSM.

Dataset preprocessing

Original Format

The dataset was provided in shapefile format, NAD83 State-Plane (meters) Vermont projection.

QGIS-based processing

  • The building polygons were geometry-checked and stripped of all nonessential attribute fields, then reprojected to EPSG 4326.
  • Conflict detection was performed by importing the city OSM coverage via the QGIS OSM tool. Any footprints that intersected existing OSM features (other than POIs) were removed from the import set.
  • At the request of the imports community, two ancillary data features were added to each footprint:
    • Maximum building height in meters - this was derived from a ground-normalized, lidar-based DEM produced by the UVM Spatial Analysis Lab
    • Address of the containing city parcel - this was performed using a spatial join. In some cases multiple buildings are listed with the same address and apartment numbers are not included, but this was the most-comprehensive way of adding addresses from the public domain.


An unaltered version of the Paul Norman's script was used to convert the processed shapefile to .osm XML. Note that no special translation file was used, and the "attributes" were converted directly to tags.

JOSM-based processing

  • All buildings were orthogonalized, which in some cases altered non-orthogonal buildings. These were manually repaired (e.g. the Burlington PD building)
  • Geometry was validated
  • Tags were standardized to OSM norms. These were included in the final list:
    • addr:city
    • addr:housenumber
    • addr:street
    • building
    • height
    • name
  • The city was scanned block-by-block for conflicts with the recent bing imagery, and in 10 cases building footprints were removed where the structure had been torn down since the last dataset update. Several instances of new construction were also observed, but not added as part of this import.
  • Most buildings associated with the University of Vermont were also removed to prevent overlap with a concurrent import effort by Andrew Guertin.

Import process

  • The entire dataset was uploaded (just under 10,000 ways) simultaneously under the contributor name btv-prints
  • If any errors arise, a full revert will be applied and the dataset will be repaired in JOSM or QGIS as appropriate

Future Steps

Some in the community raised the question of how these footprints will be kept in sync with the city data layer, and at the moment I don't have a systematic plan for it. Specifically, the city is also without a plan for regular updating (largely without funding, for that matter), so it does not seem necessary at this time to set up a recurring import process. In fact, given the active OSM community, it seems more likely that the city might periodically update its dataset from OSM. Any such plans will be made in consultation with the city GIS director.

With a comprehensive building layer for the city, we can now begin adding POI information directly to structures.