O-D catchment methodology — primary and secondary trade areas

The Woodfine co-location platform defines trade areas for each cluster using an Origin-Destination (O-D) model based on crow-flies distance rings over a hexagonal spatial grid. Each cluster is assigned two catchment zones that determine which population and spend data is attributed to it. The trade-area inputs to the deterministic ranking system and the V3 catchment ranking methodology flow from this model; population and spend layers are documented in trade-area data sources.

[edit]Spatial Framework

Trade areas are computed using the H3 global hexagonal grid at resolution 7. Each H3 resolution-7 cell covers approximately 5.16 km² with a centre-to-centre spacing of approximately 2.11 km. The grid is continuous and consistent worldwide, enabling direct comparison between clusters across all countries in the current dataset.

[edit]Catchment Zone Definitions

Primary catchment: All H3 resolution-7 cells whose centre point falls within 35 km (crow-flies) of the cluster centroid. This zone represents the immediate trade area where the majority of regular shopping trips originate.

Secondary catchment: All H3 resolution-7 cells whose centre point falls between 35 km and 150 km (crow-flies) of the cluster centroid. This zone captures the wider regional draw, including occasional shoppers and cross-regional trips.

The 35 km primary boundary is a provisional parameter based on established retail geography conventions. It is subject to refinement once empirical origin-destination data becomes available.

The 150 km outer boundary aligns with the platform's data collection radius, ensuring that every cell contributing to a cluster's catchment has been ingested and verified.

[edit]Distance Method

All distances are calculated as the crow-flies (great-circle) distance using the haversine formula. No drive-time routing is used. This approach is:

Reproducible without map routing infrastructure
Consistent across urban and rural geographies
Computationally efficient over millions of H3 cells
Suitable as a baseline before empirical O-D data is available

H3 ring traversal identifies candidate cells efficiently (17 rings ≈ 35 km; 72 rings ≈ 150 km at resolution 7), with haversine as the definitive distance measure.

[edit]HOME and AWAY Perspectives

The platform distinguishes two perspectives on catchment population.

HOME: Population counts derived from residential data (WorldPop 2026). Represents where people live within each catchment zone. This is the default view and is fully implemented.

AWAY: Population counts representing daytime or workplace population. Workplace distribution differs from residential distribution — concentrated in commercial districts and employment centres rather than dispersed across residential areas. The AWAY perspective is planned; the data source is pending.

[edit]One Cell, Multiple Clusters

A single H3 cell may fall within the catchment of multiple co-location clusters. This is intentional: trade areas are not exclusive territories. A household within 35 km of two competing clusters contributes to both clusters' primary catchment populations. This reflects the competitive retail landscape and is foundational to the cross-cluster comparison methodology; cluster boundary handling at the same parking lot is documented in cluster deduplication threshold.

[edit]Application

Catchment zone membership is the basis for:

Population aggregation (census data by zone)
Spend aggregation (grocery, hardware, wholesale spend by zone)
Cross-cluster competitive ranking (see: Catchment Ranking Methodology V3)

The catchment polygons displayed on the map are generated from the same 35 km / 150 km crow-flies radii, visualised in two distinct colours to distinguish primary from secondary zones.

[edit]See also

[edit]References

Catchment area — Wikipedia, accessed 2026-06-14
Trade area — Wikipedia, accessed 2026-06-14
H3: Uber's Hexagonal Hierarchical Spatial Index — H3 Geo, accessed 2026-06-14
WorldPop Global High Resolution Population Denominators Project — WorldPop, University of Southampton, accessed 2026-06-14

Wikipedia content reproduced under CC BY-SA 4.0.

Cluster centroids from which catchment distances are measured are derived from OpenStreetMap POI records. OpenStreetMap data © OpenStreetMap contributors, licensed under ODbL.