The addressing model itself needs no geographic data — it is pure projection arithmetic over a fixed origin. Data enters for everything around it: the basemap, the landmark pages, and the boundary context that makes an address mean something to a reader.
| Source | Role here | Type | Licence / terms |
|---|---|---|---|
| OpenStreetMap | Basemap geometry via CARTO's vector tiles; landmark coordinates and neighbourhood names. | Vector | ODbL 1.0 — attribution required and given in the map's attribution control. |
| CARTO basemaps | The keyless default basemap styles (dark-matter, positron, voyager). | Vector tiles | CARTO basemap terms; attributed alongside OSM. |
| Google Maps Platform | Optional alternative basemap with better local imagery and building footprints; enabled only when an API key is configured. | Raster + vector | Google Maps Platform terms; requires the operator's own key and billing. |
| Bhuvan / NRSC (ISRO) | Reference imagery and thematic layers for the Chennai region — used to sanity-check the extent and landmark placement. | Raster | NRSC open data terms; attribution required. |
| Survey of India | Authoritative topographic reference and the National Spatial Reference Frame. | Vector / raster | Open Series Maps under the NMP. Everest 1830 datum — must be transformed to WGS84 before use; see accuracy. |
| Census of India / GCC ward data | Ward and zone boundaries for describing which administrative area a region cell falls in. | Vector | Government of India open data; attribution required. |
| CMDA | The definition of the Chennai Metropolitan Area whose extent this model covers. | Vector / documentary | Public planning documents. |
Vector, raster and the grid
- Vector — the CMA extent, ward polygons, road network and landmark points. Cell footprints are also vector: each is emitted as a true polygon whose four corners are projected individually, which is why cells tile without hairline seams even though a square in UTM is very slightly rotated in WGS84 by meridian convergence.
- Raster — the basemap tiles, and the nested grid itself, which is a regular tessellation with raster semantics: row-major indexing, a fixed origin, and a resolution per level.
- Topology — containment and adjacency are derived from the index model rather than from geometry, so they are exact and need no snapping tolerance.