The Open Database of Buildings (ODB) Metadata document: concepts, methodology and data quality

1. Overview

For the purpose of exploring open data for official statistics and to support geospatial research across various domains, work was undertaken to create an accessible and harmonized database of buildings containing open data by various levels of government within Canada. This document details the process of collecting, compiling, and standardizing the individual building datasets that were used to create the Open Database of Buildings (ODB), which is made available under the Open Government License - Canada.

In its current version (version 3.0), the ODB contains approximately 14.4 million buildings up from approximately 4.4 million building features in version 2.0. The database is expected to be updated periodically as new open datasets on buildings from government sources become available. This vector dataset is provided as zipped GeoPackage files at the provincial or territorial level.

Access to the dataset is available through the Statistics Canada Linkable Open Data Environment website.

2. Data Sources

Hundreds of data sources were used to create the ODB. The data providers, which include municipal, regional, provincial, and federal levels of government, are provided in Appendix A, as well a list of sources can be found in the ODB_v3_data_providers.csv file that’s bundled with each zipped regional ODB dataset. The data sources provide the name of the source, name of the dataset, download URL for where the data was collected, the data licence, licence version, the licence URL, licence attribution, and download date. For further information on the individual licenses, users should consult directly with the information provided on the open data portals for the various data providers.

3. Reference Period

Ideally, the reference period would be the period from which the underlying building data refers to. Such information, however, was usually not available from the open data portals. In lieu of this, Appendix A provides the date each dataset used in the ODB was downloaded. Data were gathered between November 2024 and April 2025 from open data portals for version 3.0 of the ODB. Users are cautioned that the download date should not be used to indicate the reference period of the data. If specific information concerning the reference period of data is required, users should contact the appropriate data providers shown in Appendix A.

4. Target Population

The goal of the ODB is to create a centralized and harmonized repository of buildings available from open data government sources across Canada. As work on ODB progresses, definitions and thresholds will evolve. Users are reminded that unedited data can always be obtained directly from the open data portals of the various data providers.

5. Methodology

The compilation of the ODB buildings dataset followed a seven-step process, including: 1) data collection 2) ingestion, 3) enrichment, 4) cleaning, 5) standardization, 6) deduplication, and 7) dissemination. This section of the metadata document will provide an overview of each of these steps used to generate ODB version 3.0.

Data Collection:
The first step in this process was the identification of potential building datasets which detail the shape and location of the building features but also datasets which could be used to enrich the context of those features. For each dataset identified, a copy of the original data was downloaded, and a record of the dataset’s characteristics was stored. This record included: source and licensing information, the download date, the day the data was last updated by the provider, and identification of which building features and attributes to copy over to the ODB.

Ingestion:
With open building datasets identified in the data collection stage, each source record was processed in the ingestion step with the goal of collecting and preprocessing the data into a provincial/territorial collections of building records. This step begins with loading each open dataset, filtering out records outside the scope of ODB, and then sub-setting record attributes to only include data columns which are relevant to ODB and renaming these columns in a consistent manner.

With only the relevant portions of the building datasets selected, each dataset is provided the same coordinate reference system, Statistics Canada Lambert Conformal Conic (EPSG:3347), and provided a unique ID for the record based on the records, geometry, source, and dataset name.

Lastly each feature is conflated to the Statistics Canada’s Census Subdivision spatial boundary file and associated with Province or Territory. Using this associated information records are split into provincial/territorial subsets and exported.

Enrichment:
For each provincial/territorial collection of open building data records, many features represent the same building structure. To provide a richer building dataset, records which were provided by the same source and used the same license, were combined when possible. This was performed by finding all point features which intersected or were within 5 metres of a building polygon and filling in missing building polygon attribute data when it was available in a related building record.

Following the combining of building features, all point features were subsequently dropped.

Cleaning:
Following the enrichment process, a cleaning process was employed to provide a more consistent final product. This included, dropping any features outside of the Canadian study area, filtering out any ingested records which were deemed unsuitable for inclusion into ODB, updating column names, and re-ordering columns.

Standardization:
Attribute values frequently vary by data provider but can also represent the same value. To provide a more cohesive product, similar attributes were made more consistent in value during the standardization step. For example, if the building types “LIBRARY”, “Library”, “library”, and “Library building” were all included in datasets, a more consistent value of “Library” would be applied to ensure that only one label be applied to the same feature type. For a complete list of all the substitutions applied during the standardization step, see Appendix B.

Deduplication:
Following standardization of building attribute values, duplicate building polygons are identified and dropped from the final output. This deduplication step, includes different methods of removing duplicate records, including a simple deduplication of all identical records, deduplication of features which have the same unique id, and spatial deduplication. The spatial deduplication drops any building polygons which overlap other buildings features, with the building polygon with the richest attribute data being retained over the other.

Dissemination:
The final step in the ODB pipeline is to prepare and export the data for consumption. This includes filling in empty cell values with Statistics Canada standard values, translating the datasets for English and French versions, and providing associated documents such as a list of the data providers with the output provincial/territorial datasets.

6. Data Dictionary

This data dictionary below describes the variables contained within the ODB.

Variable - id

Variable – source_id

Variable – source

Variable - dataset

Variable - Census subdivision unique identifier (csduid)

Variable – Census subdivision name (csdname)

Variable - prov_terr

Variable - name

Variable - type

Variable - address

Variable - height

Variable - floors

Variable - units

Variable – sq_ft

Variable – year_built

7. Data Accuracy

All building features in the ODB were collected from open data government sources. Only subsets of these input data sources were included in the ODB. In situations where multiple building feature records from the same source were collected, attempts were made to combine this data into a single richer harmonized record. Although the final ODB building features may consist of multiple record details, attempts were made to minimize the changes made to the original attribute data. In some cases, attribute data was cleaned up, standardized, or combined with other values, but spatial geometries were not modified from the original source.

During the review of the data, features which were deemed inaccurate or not representative of building features were dropped from the final output. Overlapping geometries were also dropped, with the features with richer records being kept over those with less details.

The ODB represents open data from government sources that was found at the time of compilation and should thus not be interpreted as a complete or objective "ground-truth" of what buildings exist in Canada. The current coverage across Canada of the ODB is incomplete with data primarily existing for larger cities. The gaps in the database reflect areas where open data on buildings from government sources was not found.

8. Geographic Representation

The Open Database of Buildings is available on the Statistics Canada website in the following geographic representation:

• Projection: Lambert conformal conic
• False easting: 6200000.000000
• False northing: 3000000.000000
• Central meridian: -91.866667
• Standard parallel 1: 49.000000
• Standard parallel 2: 77.000000
• Latitude of origin: 63.390675
• Linear unit: metre (1.000000)
• Datum: North American 1983 (NAD83)
• Prime meridian: Greenwich
• Angular unit: degree
• Spheroid: GRS 1980

The North American Datum of 1983 (NAD83) is an adjustment of the 1927 datum (NAD27) that reflects the higher accuracy of geodetic surveying.

Appendix A: Data Providers

Appendix B: Standardization Value Substitutions