Case study
What this copy of Canada Structures stores
Diagnostic Analysis: Canada Structures field completeness
Which fields in the Canada Structures files can be used in every province, and which patterns sit in the file itself. A census of 13,760,754 structure polygons in the 13 held files.
- GIS
- Data Quality
- Contents Audit
Canada Structures is a national structure-polygon product. A held copy still has to be checked field by field. This piece counts every polygon in the 13 held GeoPackages, keeps Height zero and Height missing as separate classes, and reads type labels, land cover, overlaps, and the stored-Area gap. Height cannot be used nationally. Full methodology, results, and limitations are in the report below.
Nunavut Height recorded as zero
100%
all 11,068 Nunavut polygons; stored Height is zero, not missing
Polygons with Height missing
1,887,979
almost all sit in Ontario and Quebec
Polygons with OSM type blank
11,344,842
type is blank on most polygons in every province
Polygons with land-cover name blank
13,215,202
land-cover name is blank on almost every polygon
Interior-overlapping pairs remaining
20,180
Quebec and Ontario hold the largest pair counts
Stored Area off geometry by more than 1 percent
79.4%
typical gap is 2.9 percent; none exceed 10 percent
Real public data: thirteen provincial and territorial GeoPackages, Canada Structures version 1, Open Government Licence (Public Safety Canada). No synthetic data. No individual person or household is identified.
The Question
- Which stored fields mean the same thing in every province?
- Where does Height fail: recorded zero, or missing entirely?
- What patterns sit in area, shape, and source flags without leaving the file?
What Was Built
- Read all 13 GeoPackages. National total 13,760,754 polygons. No rows dropped.
- Separated Height zero from Height missing. Checked identifiers, source flags, and type labels.
- Measured how area, perimeter, and Height associate on polygons with a measured Height.
- Counted remaining interior overlaps and the gap between stored Area and geometry Area.
What It Found
- Every Nunavut polygon stores Height as 100% zero (all 11,068 polygons).
- Height is missing on 1,887,979 polygons. Ontario and Quebec hold almost all of them.
- OpenStreetMap type is blank on 11,344,842 polygons. Land cover name is blank on 13,215,202.
- 20,180 overlapping pairs remain. Stored Area disagrees with geometry by more than 1 percent on 79.4% of polygons.
The Analysis, in Full
Download the PDF reportStack
- Python
- geopandas
- pyogrio
- pandas
- matplotlib