Sources and lineage
At launch the address base is the Statistics Canada National Address Register. Postal code points are derived from it. Parcels shown on the catalogue map come from provincial and municipal sources and are display-only.
Why NAR first
It is the only source with consistent national coverage and a consistent schema. Starting there means the register behaves the same way in Nunavut as in Toronto — thinner, and visibly so, but structurally identical.
Municipal open data is better where it exists: higher positional accuracy, faster to reflect new construction, because the municipality is the body that assigns the address. It is also uneven, absent in smaller municipalities, and published in a dozen incompatible schemas that change without notice.
Blending the two well is real work. Doing it badly produces a file that is better in some municipalities and quietly inconsistent across boundaries, which is worse than a uniform baseline. So: NAR at launch, municipal sources layered in afterwards, one municipality at a time, with the change log recording each.
What we do to it
Parse. Source address strings are split into civic number, suffix, unit, street name, type and direction. The original string is retained in source_address so any parse can be audited.
Standardise. Street names, types and directions are normalised against a national gazetteer.
Attribute spatially. Province, census division and census subdivision are attached by point-in-polygon, never by name matching. Name matching fails on every municipality that shares a name with another, and there are more of those than people expect.
Assign persistent identifiers. address_id is minted once and carried forward, so a join written this quarter still resolves next quarter.
Transform to WGS84. Once, on ingest, with the NAD83 to WGS84 shift applied. See projections.
How postal points are derived
Every address record in the National Address Register carries a postal code. For each code present on one or more addresses, the addresses are grouped and a representative position is calculated for the group. That point, the count of member addresses, and the distance to the furthest member are what ship.
The geometry is calculated from address positions in a federal government dataset. It is a derivative product.
The advantage of deriving rather than sourcing is that every point is backed by addresses that demonstrably exist. The limitation is that coverage is inherited: a code with no matched addresses produces no point, rather than being interpolated from a neighbour.
That is a deliberate choice. An absent point is a visible gap; a guessed one is an invisible error.
Address Register is not affiliated with, endorsed by, or sponsored by Canada Post Corporation. The points are not certified for mailing or address validation.
Parcels
Displayed on the catalogue map as context. Not downloadable, not part of any extract, and not covered by any licence granted here.
Parcel data licensing varies by province and by municipality, and several sources permit display but not redistribution. Sorting that out source by source is ongoing.
When the base improves
As municipal sources are added, existing records will change position and attributes without changing address_id. That is the point of the identifier. Those changes appear in the diff as moved or attribute_change, never as added, so a growth analysis is not polluted by data improvement.