General Information
The id attribute provides a unique value for each building composed of the longitude and latitude attributes serving as a unique identifier of the building. Attribute latitude refers to the latitude of the centroid of the building footprint polygon. Similarly, longitude refers to the longitude of the centroid of the building footprint polygon.
Attribute geometry provides the definition of the building footprint polygon as acquired from the source data set, while attribute area_in_meters provides the pre-computed size of the building surface area in square meters.
Footprint Source Information
Attribute footprint_source distinguishes the source the building footprint was obtained from. It has three possible values: google, microsoft and osm referencing Google’s Open Buildings, Microsoft’s Building Footprints and Open Street Maps respectively.
For those buildings, which are obtained from Google’s Open Buildings data set attribute vida_confidence provides the percentage value how confident Google is that the polygon represents a building. In the case the building is obtained from Microsoft’s Building Footprints or Open Street Maps the value is set to zero.
Open Street Maps provides a wide range of interesting attributes for each building, such as its name, type and building classification, which are ingested into the dataset alongside the Open Street Maps provided unique building identifier as attributes osm_name, osm_type, osm_building and osm_id. Additional tags for buildings may be provided in the osm_other_tags attribute. These attributes are filled only for those buildings, which are directly sourced from Open Street Maps and are left empty for buildings from other sources.
Building Height
For each building couple of different height estimates are provided. The maximal estimated height of the building can be found in the height_max attribute. Similarly, the estimated median height and the estimated mean height of the building can be found in the height_median and height_mean attributes respectively. The height attribute provides the estimated height of the building in terms of storeys. By definition a single storey of a building is defined to be three meters tall and all buildings require to have at least one storey. This value is computed by rounding the estimated mean height to multiples of three.
SMOD Layer
Each building is cross referenced with Global Human Settlement’s Settlement Model grid. The ghsl_smod attribute represents the GHS classification of the area the building belongs to based on its location. Where the layer does not provide any value the Very low density rural grid cell value is used. These values are simplified into Urban, Suburban and Rural categories provided in the urban_split attribute.
Classification
Each building is classified into residential and non-residential, denoted by the classification_type attribute using values res and non-res respectively. This classification is done by a machine learning model for buildings, which have over 20 square meters of area. The name of the machine learning model is stored in the ml_model attribute and its confidence with the classification is stored in the related ml_confidence attribute. The fact, that a building is classified by a machine learning model is denoted by providing classification_model value as the classification_source attribute. In case the building has lower building area than 20 square meters the building is classified to be a residential building by default, a fact denoted by setting the classification_source attribute to area.