H3 spatial indexing: hexagonal grid analytics, aggregation, and visualization

[mikemcdougall]

Context

CARTO's entire analytics differentiator is H3 hexagonal grid indexing. H3 enables spatial aggregation at scale — billions of points aggregated into hexagonal cells for density analysis, movement patterns, and spatial joins. Increasingly expected for mobility, telco, logistics, insurance, and IoT analytics.

PostGIS supports H3 via the h3-pg extension. Uber's H3 is open-source and becoming a de facto standard.

Scope

Core Operations

• H3 indexing: assign H3 cell index to features at configurable resolution (0-15)
• Aggregation: group features by H3 cell → count, sum, avg, min, max per cell
• Polyfill: fill a polygon with H3 cells at a given resolution
• K-ring: get neighboring cells within K distance
• Compact/uncompact: optimize cell sets across resolutions

API Endpoints

• GET /rest/services/{id}/FeatureServer/{layerId}/h3?resolution=7&outStatistics=...
• Returns H3 cell geometries with aggregated statistics
• Compatible with GeoServices statistics response format
• OGC API endpoint: /collections/{id}/h3?resolution=7

Visualization

• H3 hexagonal grid as MVT tiles (dynamic resolution based on zoom)
• MapLibre fill-extrusion for 3D hex visualization
• Auto-generated choropleth style from aggregation values

Other Discrete Global Grids

• S2 cells (Google standard) — optional secondary index
• Quadbin (CARTO's quad-tree index) — for CARTO migration compatibility

Edition Gating

• Community: none (H3 is an analytics acceleration feature)
• Pro: H3 indexing, aggregation, MVT hex tiles, S2/Quadbin support

References

• Uber H3: https://h3geo.org/
• h3-pg: https://github.com/zachasme/h3-pg
• CARTO Analytics Toolbox H3 module
• Spatial analytics API: server-side clustering, density, spatial joins #342 Spatial analytics API (H3 aggregation extends this)