Glossary

We count active uses (cafes, restaurants, shops, schools, community spaces, residential, transit) within 100 m of the park edge. We subtract for hostile uses (parking lots, highways, rail, blank institutional walls). A park whose perimeter is fronted by espresso machines and front porches scores well; a park ringed by surface parking does not.

Streets touching the park, intersections nearby, paths and sidewalks within 50 m, transit stops within 400 m, an estimate of how many ways into the park exist along its perimeter. Includes a 'superblock penalty' for parks whose edges are mostly walls rather than streets.

We count distinct amenity types (playground, washroom, water, sport, garden, art, performance) inside the park polygon, drawn from the City's Parks & Recreation Facilities dataset. We reward variety, not raw count — a park with one of everything beats a park with twenty benches.

Tree canopy share inside the park (from the City's Treed Area dataset, augmented with street-tree density), ravine overlap, water surface inside the park, distance to the nearest waterbody, and a cover-diversity index. Scores high for ravines, big urban parks with mature trees, and waterfront sites. Scores low for paved plazas.

From Toronto's 3D Massing dataset: how many buildings sit within 25 m of the park edge, how tall they are, what share is mid-rise (3–7 floors) versus tower (≥ 13 floors). Mid-rise frontage scores best — it produces the 'eyes on the park' effect Jane Jacobs argued for. Towers immediately adjacent get a penalty (wind, shadow, sterile ground floors).

Adjacency (within 50 m) to highways, rail corridors, large parking lots, industrial parcels, or blank institutional edges. Higher = bad. The headline score subtracts 100 minus the risk score. Coined by Jane Jacobs in The Death and Life of Great American Cities.

Weighted average of the five forward-looking dimensions plus 100 minus the border vacuum risk. Default weights: Edge 25%, Connectivity 20%, Amenity 20%, Natural Comfort 15%, Enclosure 10%, Border Vacuum 10% (inverted). Configurable via env vars. The score is bounded 0–100 but in practice most parks fall between 25 and 70.

Per-metric: 'measured' (≥70%) when all expected source layers contributed, 'partial' (40–70%) when one is missing, 'inferred' (<40%) when the metric falls back to a placeholder. The headline doesn't shrink low-confidence dimensions — it lets them count, but flags them visually so you can read each piece at the right level of certainty.

A rule-based classification into one of twelve categories: Civic Square, Urban Plaza, Parkette, Neighbourhood Park, Athletic / Recreation Park, Waterfront Park, Ravine / Naturalized Park, Wilderness / Conservation Park, Tower-Community Green Space, Corridor / Linear Park, Destination Park, Other. Used to compare parks within their family rather than across the whole Toronto Park Catalogue.

Unsupervised k-means on the five normalised metric dimensions. Eight clusters; names derived from each cluster's most distinctive dimension versus the citywide mean, with a disambiguation tag when multiple clusters share a name. See /insights/clusters for the full breakdown.

Urban integration = the average of edge activation, connectivity, and enclosure. Natural comfort = the natural-comfort sub-score. We treat them as independent axes because the Toronto Park Catalogue is bimodal — most parks are strong on one and weak on the other. The framework Jacobs would use and the framework a forester would use produce different answers; we keep both visible. See /insights/jacobs-vs-wilderness.

Jacobs's term from The Death and Life of Great American Cities. A border vacuum is a single-use edge — an expressway, a rail yard, a large institutional campus, a parking lot — that suppresses pedestrian activity and isolates the park from its neighbourhood. We count adjacency to these uses within 50 m of the park edge.