# Data attribution and licences ## Country and territory boundaries Country recognition uses the ISO point-of-view variant of Natural Earth **Admin 0 – Countries**, version 5.1.2, simplified and packaged for on-device point-in-polygon lookup. Natural Earth is a public-domain map dataset: - https://www.naturalearthdata.com/downloads/10m-cultural-vectors/ - https://github.com/nvkelso/natural-earth-vector/releases/tag/v5.1.2 The deterministic source used by `Scripts/build_country_boundaries.mjs` is the ISO point-of-view GeoJSON from that release: - `https://raw.githubusercontent.com/nvkelso/natural-earth-vector/v5.1.2/geojson/ne_10m_admin_0_countries_iso.geojson` - SHA-256: `60eb10aa951f5872507c9436937508b09be4b43dc9fa7aad7644f23ef12e1cad` The bundled transformed resource has SHA-256 `1fc8db8d7f64a91635bfc6d411b7f2b5e1151c042d962dcc20090a579ded8f5c`. The progress universe is a pinned snapshot of the 249 officially assigned ISO 3166-1 alpha-2 codes as checked on 2026-08-25. ISO permits free use of its country codes: - https://www.iso.org/iso-3166-country-codes.html The bundled `landmarks-de.json` is a transformed landmark database. Its top-level metadata declares ODbL 1.0 and this attribution: > Contains information from OpenStreetMap, © OpenStreetMap contributors ## Global Supercharger collection Collect's separate Supercharger collection is a transformed, fixed OpenStreetMap database snapshot retrieved through Overpass on 2026-08-30. The source query selects `amenity=charging_station` objects carrying either `brand:wikidata=Q17089620` or the exact `brand=Tesla Supercharger`. It excludes records explicitly tagged `access=no`, `access=private`, or `access=employees`, inactive lifecycle states, and Tesla Destination Charger records. Nearby duplicate map objects are clustered within 120 metres unless their strong Tesla references conflict. The bundled transformation contains 5,717 sites in 59 country-code buckets, including 321 in Germany. It is not a complete inventory of Tesla's network and contains no live availability, operation, access, price, output, or vehicle compatibility guarantee. Canonical identities are source-qualified OSM object IDs. Globally unique Tesla reference values are retained only as migration aliases; ten ambiguous reference values are omitted rather than used to merge distinct sites. Coordinates, charging-site identity, and source tags are from OpenStreetMap and the distributed transformed database is under ODbL 1.0: - https://www.openstreetmap.org/copyright - https://opendatacommons.org/licenses/odbl/1-0/ GeoNames `cities5000` supplies fallback place-name context for 487 otherwise unnamed sites and country context for 44 records that did not resolve through the bundled Natural Earth boundary or an accepted OSM country tag. GeoNames is licensed under CC BY 4.0; Natural Earth 5.1.2 is public domain: - https://www.geonames.org/export/ - https://creativecommons.org/licenses/by/4.0/ - https://www.naturalearthdata.com/about/terms-of-use/ The exact bundled attribution is: > © OpenStreetMap-Mitwirkende, ODbL 1.0; GeoNames geographical database, CC BY 4.0; adapted by Collect 2026-08-30 Pinned raw Overpass JSON SHA-256 values: ```text Americas 4f675c74c9c8e2ed1819a04714b36595191d3ea2f2a5130fbeac762c9519f68d Europe and Africa 2047a934ea88172db016a67d994eaf63a4dd4bdd3efd8de4416cb1098b8464b5 Asia and Pacific a99a3937c5c324dae02162ec293830bdf7adecc8ef0d0333cc4e228ed7245f40 ``` The generated `superchargers-world.json` is 1,002,408 bytes with SHA-256 `601e1f35bc7799877c42b8b0c614a2062c1fab0428fb9790cf41cd303cd2f800`. The full query, gzip archive receipts, transformation counts, and limitations are recorded in `superchargers-world.manifest.json`; the deterministic rebuild is implemented by `Scripts/build_superchargers.mjs` and checked by `Scripts/Tests/superchargers.test.mjs`. When an App Store binary conveys this OSM-derived database, the release owner must also make an unrestricted copy available outside App Store DRM, together with its database notice, transformation description, and licences. This repository is the development source copy; public release availability remains a separate release requirement. ## Subnational regions Region recognition uses a versioned, curated gameplay snapshot derived from Natural Earth **Admin 1 – States, Provinces**, version 5.1.2. Natural Earth is public domain. Its Admin-1 layer is a practical cartographic source rather than a uniform official worldwide hierarchy, so Collect shows progress per country and uses the neutral umbrella term “Regionen”: - https://www.naturalearthdata.com/downloads/10m-cultural-vectors/10m-admin-1-states-provinces/ - https://github.com/nvkelso/natural-earth-vector/releases/tag/v5.1.2 - https://www.naturalearthdata.com/about/terms-of-use/ The pinned source GeoJSON is checksum-gated by `Scripts/build_subdivision_boundaries.mjs`. The generated catalogue contains 4,404 named regions in 198 country buckets. It excludes unnamed features, non-standard/custom subdivision aliases, invalid parents, and country buckets with fewer than two eligible regions. Natural Earth `ne_id` values are the stable stored identifiers; unambiguous ISO 3166-2 codes are lookup aliases: - https://www.iso.org/iso-3166-country-codes.html The generated resources are: - `CollectCore/Sources/CollectCore/Resources/subdivisions-ne-v5.1.2.json` - `CollectCore/Sources/CollectCore/Resources/subdivision-boundaries-ne-v5.1.2-.json` Catalogue SHA-256: `d30bcb5d4b6fa0482b5ca12a5b64db5ff1ceaeaea794deb77c7b505b5ac3ff5e` ## Worldwide city catalogue The city/place catalogue uses the GeoNames `cities5000` and `alternateNamesV2` dumps plus the European Commission Joint Research Centre's GHS-SMOD settlement shapes, retrieved on 2026-08-30: - GeoNames data export: https://www.geonames.org/export/ - GeoNames dump directory and format documentation: https://download.geonames.org/export/dump/ - Pinned cities source: https://download.geonames.org/export/dump/cities5000.zip - Pinned alternate-name source: https://download.geonames.org/export/dump/alternateNamesV2.zip - GHSL data package and methodology: https://human-settlement.emergency.copernicus.eu/documents/GHSL_Data_Package_2023.pdf - GHSL data access: https://human-settlement.emergency.copernicus.eu/download.php?ds=smod - GHSL citation and reuse terms: https://human-settlement.emergency.copernicus.eu/GHSLhowToCite.php - Creative Commons Attribution 4.0: https://creativecommons.org/licenses/by/4.0/ GeoNames makes the dump available under CC BY 4.0. The exact attribution embedded in Collect's generated index is: > GeoNames geographical database, CC BY 4.0; adapted by Collect 2026-08-30 The adaptation filters the 69,667-row source to feature codes `PPL`, `PPLA`, `PPLA2`, `PPLA3`, `PPLA4`, `PPLA5`, and `PPLC`; validates positive unique IDs and point coordinates; and restricts parent countries to Collect's pinned 249 ISO 3166-1 codes. The resulting 64,081 records are split across 244 nonempty country shards. This is a source-defined populated-place collection, not a complete legal inventory of the world's cities or municipal boundaries. For display, the generator streams `alternateNamesV2.txt` from its archive, rejects historic and colloquial German aliases, prioritizes the preferred flag, and prefers a non-short name within each preference tier before using the lowest alternate-name ID as a deterministic tie-break. The original GeoNames name is the final fallback. This is an adapted name selection; the upstream alternate names and source flags are not redistributed as a new general-purpose alternate-name database. Pinned input checksums: ```text cities5000.zip 1bc3ef2749ec9affdcefd53b10d77f48f296ee593f312142fb287b022fb25457 cities5000.txt e4e1d20910229a6ede5a2ab02cfc283c5cff2faf8a1531736549e681fa892a38 alternateNamesV2.zip 5fda82d1053ac5e3ddc9447497db1fbf53d56017febba8c81a18514a65c6f000 alternateNamesV2.txt (streamed, not materialized) aba6304a72c8626a1a14bd8c9187a5cf2c6fd680432cf9ac06896c825810f5fd GHS_SMOD_E2020_GLOBE_R2023A_54009_1000_UC_V1_0.zip 7254bc766e8a731ae71ef68fc180d03000c52a8f068c397b716d552202f0e381 GHS_SMOD_E2020_GLOBE_R2023A_54009_1000_DUC_V1_0.zip 5e03753a8e13bbcb2ed800e8ba8ce22b46a2c3094a649684635b4dde377b79d7 ``` GHSL publishes these data under CC BY 4.0. The generated index embeds this attribution: > European Commission, Joint Research Centre (JRC), GHSL GHS-SMOD R2023A; adapted by Collect Collect matches the GeoNames representative point by containment or at most 1,500 m to an urban-centre or dense-urban-cluster polygon. For matched places from 100,000 inhabitants, Collect converts the selected source component from the GHSL equal-area projection to WGS84, quantizes coordinates to five decimal places, removes duplicate ring closure points, and delta-encodes the resulting contour in a lazy country shard. The app combines that contour with a guaranteed 2 km centre core, a 400 m gameplay buffer, a conservative 100 m geometry uncertainty, and population-tier centre caps of 6/10/12 km. Unmatched places retain the 2/3/5/8 km population fallback. This is an adapted redistribution of selected GHSL settlement geometry under CC BY 4.0. The contours describe contiguous built-up settlement classes from a 1 km source grid; they are not legal or administrative municipal boundaries. There are 9,261 urban-centre matches, 5,736 dense-cluster matches, and 49,084 population fallbacks across the full GeoNames catalogue. Of 5,852 places with at least 100,000 inhabitants, 2,493 use 1,646 distinct matched GHSL entities; the remaining 3,359 use the population fallback. The generated version is `geonames-cities5000-2026-08-30-ghsl-zones-v3`. Its index, ordered city-shard set, and ordered zone-shard set are pinned as follows: ```text cities-geonames-cities5000-2026-08-30-index.json 3332fbcec2cfe134b6b67d1283adb8d56d970e451b213282b37ac680ecb5bf30 244 sorted country shards, SHA-256 of sorted filename:sha256 lines 17e4ef2c3b21b57876c8fe8842c7b4558bb3eb9a8199973803807e7c807a79c2 149 sorted zone shards, SHA-256 of sorted filename:sha256 lines 26ac6d9fe5b1ffb531033f3935e3d54418ee506d4f624634e4e1e44f64cc7721 All 393 sorted shards, SHA-256 of sorted filename:sha256 lines 77116fad72ba859d97a6f9eea11b026386489b5a38465f1eabd27492fb2e5b01 ``` The index and shards total 6,990,573 bytes. Rebuild or verify the resources with the downloaded ZIP files; the alternate-name member is streamed during both commands: ```sh node Scripts/build_geonames_cities.mjs \ /path/to/cities5000.zip \ CollectCore/Sources/CollectCore/Resources \ --alternate-names /path/to/alternateNamesV2.zip \ --ghsl-urban-centres /path/to/GHS-SMOD-UC.zip \ --ghsl-dense-clusters /path/to/GHS-SMOD-DUC.zip node Scripts/build_geonames_cities.mjs \ /path/to/cities5000.zip \ CollectCore/Sources/CollectCore/Resources \ --alternate-names /path/to/alternateNamesV2.zip \ --ghsl-urban-centres /path/to/GHS-SMOD-UC.zip \ --ghsl-dense-clusters /path/to/GHS-SMOD-DUC.zip \ --verify-existing ``` ## German lake water bodies The German lake collection uses **Wasserkörper-DE – Seewasserkörper** from the third cycle (2022–2027) of the EU Water Framework Directive. The data is managed by the Bundesanstalt für Gewässerkunde (BfG) in WasserBLIcK on behalf of the competent German state authorities. The pinned GML was retrieved on 2026-08-25: - Dataset metadata: https://geoportal.bafg.de/csw-client/recordbyid?id=bfg&lang=de&recordid=268d52a5-8409-4fb4-94ed-98f0c7b71d8d - Pinned source file: https://geoportal.bafg.de/inspire/download/AM/waterBodyForWFD/AM_lakeWaterBody-DE.gml - CC BY 4.0: https://creativecommons.org/licenses/by/4.0/ The metadata requires the source plus the date of retrieval to be stated. The exact attribution used by Collect is: > WasserBLIcK/BfG und Zuständige Behörden der Länder, 2026-08-25 The generated resource contains all 738 named lake-water-body records in this pinned source. Each uses the official `euSurfaceWaterBodyCode` as the durable identity, names and state assignments remain attached to that identity, and all 751 source polygons plus their interior holes are retained. The generator calculates areas, simplifies the boundaries at 0.00005 degrees, quantizes them at E5 precision, and delta-encodes the rings. Collect conservatively accounts for 12 m of possible transformation error when deciding whether a location is within the 100 m collection radius. This fixed collection is not every lake or pond in Germany. It is the national Water Framework Directive reporting set of 738 regularly monitored lake water bodies, generally representing lakes around 50 hectares or larger. Smaller standing waters are not comprehensively represented, and an official water body can be one named section of a larger physical lake: - https://www.umweltbundesamt.de/themen/wasser/gewaessertyp-des-jahres/gewaessertyp-des-jahres-2026-kleingewaesser The BfG metadata also notes that state authorities may provide newer or more detailed regional data. Collect therefore treats this as a versioned gameplay snapshot, not a live hydrological register. Rebuild command: ```sh node Scripts/build_germany_lakes.mjs \ /path/to/AM_lakeWaterBody-DE.gml \ CollectCore/Sources/CollectCore/Resources/lakes-de-bfg-wfd-2022-2027.json ``` Checksums: ```text AM_lakeWaterBody-DE.gml 2cd8d1594e7ca3d616b78daae997b447ceb01e8e5d100abeea1194b7c45214bc lakes-de-bfg-wfd-2022-2027.json 4da40b4ad3dec8d0d3b83902cc5ffe84452d079cee4a7c947fae590da0ae486c ``` ## European lake water bodies The non-German lake add-on uses the public **EEA WISE WFD Reference Spatial Datasets 2022, version 1.9** polygon service, published 2025-09-17 and retrieved 2026-08-25: - Dataset and downloads: https://www.eea.europa.eu/en/datahub/datahubitem-view/73ab1aa2-2f69-4881-941d-4429875ca715 - EEA legal notice: https://www.eea.europa.eu/en/legal-notice - Source service layer: https://water.discomap.eea.europa.eu/arcgis/rest/services/WISE_WFD/WFD2022_SurfaceWaterBody_WM/MapServer/17 The generated add-on contains exactly 26,143 reported lake-water-body records with polygon or multipolygon geometry from 28 reporting country codes. Germany is deliberately excluded and remains in the BfG shard above, producing a combined fixed universe of 26,881 records across 29 reporting countries. Each add-on identity is `EEA-WFD:`. EEA material is reused under CC BY 4.0 unless an item states otherwise; reporting-country rights and notices remain applicable. Collect attributes the adapted snapshot as: > European Environment Agency (EEA), WISE WFD 2022 v1.9, and reporting countries; adapted 2026-08-25 The Swedish source uses `NAMN SAKNAS` as a literal missing-name placeholder on 192 records. The generated resources preserve that official value and the canonical IDs; Collect presents those records as **Unbenannter See** in its German user interface. This is a Water Framework Directive reporting set, not a comprehensive list of every lake, pond, or European country. Rebuild command: ```sh node Scripts/build_european_lakes.mjs \ CollectCore/Sources/CollectCore/Resources ``` ```text lakes-europe-eea-wfd-2022-index.json d0ff3852d7bbf1e4b34f98685294882fc0c862c6522197994e496f81ea43c5db 128 lakes-europe-eea-wfd-2022-geometry-NNN.json shards ordered shard digests SHA-256: 6d3d535eb0ee267a8aa24847835db3439d142cb7bba04175dc26aeec3b2bae73 ``` The index embeds the ordered SHA-256 digest of every geometry shard; Collect verifies a shard before decoding it. The generated sharded set is 63,617,292 bytes and contains 26,143 unique non-German IDs. ## German rivers and streams The river collection uses **Waterbody-DE – HY-P Watercourse**, managed by BfG in WasserBLIcK with the competent German state authorities. The April 2026 snapshot was retrieved on 2026-08-25: - Dataset metadata: https://geoportal.bafg.de/csw-client/recordbyid?id=bfg&lang=de&recordid=7b86c167-cad5-4769-b32e-91a30f667e41 - Direct GML ZIP: https://geoportal.bafg.de/inspire/download/HY/waterbody/HY-P-Watercourse-DE_GML.zip - CC BY 4.0: https://creativecommons.org/licenses/by/4.0/ The exact attribution used by Collect is: > WasserBLIcK/BfG und Zuständige Behörden der Länder, 2026-08-25 The metadata permits commercial and non-commercial use under the federal GeoNutzV; the BfG Open Data index labels Waterbody-DE as CC BY 4.0. Only usable named records with a positive LAWA / Reverse-Pfaffstetter code are collectible. Every source line with the same code belongs to one `BFG-HY:` identity; names alone never merge features. Clear source placeholders (`NN`, `Ohne`, `- ohne -`, and `Temp`), helper-line labels, and numeric-only labels are rejected case-insensitively. A legitimate name on the same code is promoted instead; generic source names such as `Gewässer` remain valid. Reversible Windows-1252-decoded UTF-8 byte pairs are repaired conservatively without changing legitimate names that begin with an umlaut. This yields 47,816 logical waterway-code entries containing 378,215 line parts. There is no minimum length: 24,252 entries are shorter than 1 km and 39,441 are shorter than 5 km. Unnamed parts remain attached when another part with the same code has a usable name. The primary name is selected by greatest summed official source length; other usable names remain aliases. The generator applies an 8 m Douglas-Peucker tolerance and reduces 9,057,524 source points to 1,859,432 encoded points. Normal parts use E5 coordinates; 217 millimetre-scale source parts use a finer per-part scale so none collapse. The measured maximum source-point to encoded-segment deviation is 8.567290 m, and Collect conservatively accounts for 9 m of transformation uncertainty. That allowance covers simplification and quantization only; it is not a claim about source survey accuracy or the mapped river's bank width. Waterbody-DE supplies line geometry without dependable bank widths. Collect therefore applies a separate river-only envelope: `distance to centreline + reported horizontal accuracy + 9 m transformation allowance` must be at most 250 m. This keeps users beside broad watercourses reachable while retaining a conservative, accuracy-aware decision. It does not change the 100 m lake and landmark policies. The BfG metadata warns that the nationwide compilation can lag current state-level holdings and does not guarantee complete national coverage in every case. It describes a typical working scale of 1:25,000; source features also declare 1:50,000. Collect treats this as a Germany-only, versioned gameplay snapshot, not a live or complete hydrographic register. Source checksums: ```text HY-P-Watercourse-DE_GML.zip 9e348c783cb59d18aa952b968b6206b2c60be2518e84c2c78cc0c36cb5f74783 HY-P-Watercourse-DE.gml 29cb00db10422880190ccbec599953ada0c14078af82c9ed14c9ff113345ef63 rivers-de-bfg-waterbody-2026-04.json 9ec06c069fd9872ecb58cec9d015e88320f241388eaa26edf1ae6d922231e6a1 256 rivers-de-bfg-waterbody-2026-04-geometry-XX.json shards ordered shard digests SHA-256: 9986aedf4e4fe92143ae57ec3bf99cfd601d7e3c729eafbcd5d96dff182e2b32 ``` The metadata index is 7,273,034 bytes and the geometry shards total 18,801,101 bytes. Collect verifies the pinned index before decoding it and verifies each candidate geometry shard before loading it into a bounded 16-shard cache. Rebuild command: ```sh node Scripts/build_germany_rivers.mjs \ /path/to/HY-P-Watercourse-DE_GML.zip \ CollectCore/Sources/CollectCore/Resources/rivers-de-bfg-waterbody-2026-04.json \ CollectCore/Sources/CollectCore/Resources/rivers-de-bfg-waterbody-2026-04.manifest.json ``` ## Austrian major-river pilot The Austrian pilot uses the public **EEA WISE WFD Reference Spatial Datasets 2022, version 1.9** `SurfaceWaterBody_line` layer, published 2025-09-17 and retrieved 2026-08-30: - Dataset and downloads: https://www.eea.europa.eu/en/datahub/datahubitem-view/73ab1aa2-2f69-4881-941d-4429875ca715 - Source service layer: https://water.discomap.eea.europa.eu/arcgis/rest/services/WISE_WFD/WFD2022_SurfaceWaterBody_WM/MapServer/16 - EEA legal notice: https://www.eea.europa.eu/en/legal-notice - CC BY 4.0: https://creativecommons.org/licenses/by/4.0/ EEA material is reused under CC BY 4.0 unless an item states otherwise; reporting-country rights and notices remain applicable. Collect attributes the adapted snapshot as: > European Environment Agency (EEA), WISE WFD 2022 v1.9, and reporting countries; adapted 2026-08-30 This is a curated pilot of exactly 20 reviewed logical rivers—Ager, Drau, Enns, Erlauf, Gail, Gurk, Isel, Kamp, Lafnitz, Leitha, March, Mur, Mürz, Pinka, Raab, Steyr, Thaya, Traun, Ybbs, and Ziller—not nationwide Austrian coverage and not complete European river coverage. Donau and Rhein are deliberately excluded because overlapping BfG logical-river identities already ship in the German catalogue. Its geometry is backed by 262 official WFD `riverWaterBody` records. Those records are reporting segments, not separate collectible identities. Each collectible uses `EEA-WFD-RIVER:AT:`. A checked, checksum-gated crosswalk retains every contributing `thematicIdIdentifier`, reported source name, and reported length; names alone never become identity or merge source records. The 262 WFD bodies produce 416 polylines and 25,748 encoded geometry points. Per-body reported lengths remain in the provenance crosswalk but are not presented as a whole-river length, because WFD groups can include side arms and modified reaches. Collect records a conservative 12 m geometry-processing allowance and applies the shared 250 m river collection envelope. The allowance covers service generalization and coordinate encoding; it is not a claim about source survey accuracy or river width. Pinned contract and resource hashes: ```text Version eea-wise-wfd2022-rivers-at-major-1.9-2025-09-v1 Crosswalk e4fe3af984c8131360bfa4824f6b091fb344a8b846475bd368732840d0ac26a2 Resource 320afed6732f1b34f7e6eeecac322b097dcd50075a93e9632de5713f8e5cb98b Inventory 20 rivers, 262 WFD bodies, 416 polylines, 25,748 points ``` Rebuild command: ```sh node Scripts/build_austria_major_rivers.mjs \ CollectCore/Sources/CollectCore/Resources ``` ## United States hydrography pilot The US region pack uses the U.S. Geological Survey **3D Hydrography Program 3DHP_all** ArcGIS service, published through The National Map. The service was refreshed 2026-08-05 and the pinned pack was retrieved 2026-08-25: - 3DHP data access: https://www.usgs.gov/3d-hydrography-program/access-3dhp-data-products - Pinned service: https://3dhp.nationalmap.gov/arcgis/rest/services/usgs_3dhp_all/FeatureServer - 3DHP_all service specification: https://www.usgs.gov/ngp-standards-and-specifications/3d-hydrography-program-products-3dhpall-2023-service-specification - USGS copyrights and credits: https://www.usgs.gov/information-policies-and-instructions/copyrights-and-credits USGS states that the service data are open and non-proprietary; USGS acknowledgment is appreciated. The pinned service credit is: > USGS TNM / NGTOC – 3D National Hydrographic Program (3DHP). Data refreshed August 5, 2026. This pack is `usgs-3dhp-all-2026-08-05-pilot-dc-v1`. It uses a documented Washington, DC-region seed envelope of `[-77.12, 38.79, -76.91, 39.00]` and is not national, state-complete, or territory coverage. Every named GNIS identity intersecting that envelope is expanded to all current service parts sharing the selected GNIS ID. As a result, an included identity can have geometry outside the seed envelope; that does not broaden the pack's completeness claim. The stable collection ID is `USGS-GNIS:`. Current-service `id3dhp` and `OBJECTID` values remain only as reversible source crosswalks and are not used as long-lived collection identity. The pack includes every GNIS-named flowline in feature types 1 through 5 except elevation-breaching and hydro-unenforced connectors, plus GNIS-named waterbodies of type Lake. There is no length cutoff for named streams and no area cutoff for named lakes. Unnamed features do not count toward progress. The exact pack contains 5 lakes, 54 rivers, 5 lake polygons, 2,053 river polylines, and 116,587 encoded geometry points. E6 coordinate encoding adds at most 0.1 m uncertainty. Because the rolling national service mixes source generations and publishes no single nationwide positional-accuracy value, Collect separately deducts a conservative 50 m reserve in collection decisions. The 50 m value is a gameplay safeguard, not the encoding error and not a claim about USGS source accuracy. Pinned contract and resource hashes: ```text Service contract SHA-256 2495ee843c9e051f99c84ea52fe9a53e1527188b1255b010f046a44b579d06e6 Flowline contract SHA-256 24eaf06b194d017a813c575c66a9be5021850a398cc925d37064fa3547c47b2a Waterbody contract SHA-256 984cf8fa61e1f87d0d5c0be8385280fdf92ece9a5c74da92388432d30bc05325 hydrography-us-usgs-3dhp-2026-08-05-pilot-dc-index.json be9d7247936906028031e4bf0369f4f871fb6b36e8a94b06b329a4a2a8b19a6c hydrography-us-usgs-3dhp-2026-08-05-pilot-dc-geometry-dc.json 668382315b5e6d79d40d009b54fbfa713f8bb1bf6f1899ba4710e34e663eda7a ``` Rebuild and offline verification commands: ```sh node Scripts/build_us_usgs_3dhp.mjs \ --output-dir CollectCore/Sources/CollectCore/Resources \ --coverage pilot-dc \ --cache-dir .cache/usgs-3dhp node Scripts/build_us_usgs_3dhp.mjs \ --output-dir CollectCore/Sources/CollectCore/Resources \ --verify-existing ``` The generator also defines a resumable `nationwide-us50` scan, but no such pack is bundled. Its existence is not evidence of completed national coverage. ## Canadian hydrography pilot The Canadian region pack combines named geometry from Natural Resources Canada's **Canadian Hydrospatial Network (CHN)** with the stable Feature Id identity used by the **Canadian Geographical Names Database (CGNDB)**: - CHN dataset: https://open.canada.ca/data/en/dataset/ae385105-e48c-4b54-bd0f-dfb7303301cb - Pinned CHN service: https://maps-cartes.services.geo.ca/server_serveur/rest/services/NRCan/chn_en/MapServer - CGNDB dataset: https://open.canada.ca/data/en/dataset/e27c6eba-3c5d-4051-9db2-082dc6411c2c - CGNDB API: https://geogratis.gc.ca/services/geoname/en - National Hydro Network fallback source recorded by the resumable pipeline, but not bundled in this pilot: https://open.canada.ca/data/en/dataset/a4b190fe-e090-4e6d-881e-b87956c07977 - Open Government Licence - Canada: https://open.canada.ca/en/open-government-licence-canada The required attribution is: > Contains information licensed under the Open Government Licence - Canada. The bundled version is `nrcan-cgndb-ca-2026-08-25-chn-01BB000-2025-06-20-pilot-v1`. Its only geometry coverage is CHN work unit `01BB000`, **Kedgwick**, last validated 2025-06-20, grade `AABA`, with a published area of 1,573.206748 km². It is not nationwide Canadian coverage. Neither nationwide metadata nor nationwide geometry is bundled, and the pipeline deliberately did not download the national NHN archives. CGNDB representative points are metadata only and can never make an identity proximity-collectible without matching bundled CHN/NHN geometry. The source-qualified collection ID is `CA-CGNDB-FEATURE:<32 lowercase hexadecimal Feature Id>`. Every named geometry-backed identity in this work unit is eligible; there is no minimum length cutoff for named streams. The 1,960 named source flowline features and 68 named source waterbody features group into 95 collectible identities: 26 lakes and 69 watercourses. There are zero metadata-only entries in this pilot. E6 encoding error is 0.1 m and the recorded processing allowance is 2 m. The separate 50 m conservative collection reserve is used because this CHN pack does not expose a suitable per-feature positional-accuracy value. The reserve is deliberately not presented as coordinate error or source accuracy. Pinned resource hashes: ```text hydro-ca-nrcan-2026-08-pilot-index.json 9047a02b5ddcbd67784df9692668536452e6f27d5a62eb05082f585f6483b85e 16 hydro-ca-nrcan-2026-08-pilot-geometry-XX.json shards ordered geometry-set SHA-256: dfc4ffaca223da818d080fc5673a91945d7dfd0b09ae43027c9c578c5d257785 ``` The index and shards total 517,600 bytes. Individual shard sizes and SHA-256 hashes are pinned in `CollectCore/Sources/CollectCore/Resources/hydro-ca-nrcan-2026-08-pilot.manifest.json`. Rebuild command: ```sh node Scripts/build_canada_hydrography.mjs \ CollectCore/Sources/CollectCore/Resources \ --cache-dir .cache/canada-hydrography \ --work-unit 01BB000 \ --snapshot-date 2026-08-25 ``` The generator supports bounded cache, download, output, total-size, and free space hard stops, and can plan a work unit without publishing output. Those capabilities make later region packs resumable; they do not imply that any other Canadian work unit has been bundled or validated. ## Landmark catalogue The landmark catalogue has two separately licensed source databases plus the hand-curated seed list. ### German landmark source Sources used by the German catalogue include: - [OpenStreetMap](https://www.openstreetmap.org/copyright) — coordinates and categories, licensed under the Open Database License 1.0. - [Wikidata](https://www.wikidata.org/wiki/Wikidata:Licensing) — fame signals and some descriptions, CC0. - [GeoNames](https://www.geonames.org/export/) — city and administrative context, CC BY 4.0. - [BKG](https://www.govdata.de/dl-de/by-2-0) — German state polygons, dl-de/by-2-0. - Some notable/legendary descriptions are adapted from the exact Wikipedia article identified in each landmark's `descriptionSource`, under CC BY-SA 4.0. The app links the item-level source from the collection detail view. The transformed German database is present at: `CollectCore/Sources/CollectCore/Resources/landmarks-de.json` When an App Store binary conveys this OSM-derived database, the release owner must also make an unrestricted copy available outside App Store DRM, with the applicable database notice, transformation description, and licences. This repository is the source copy during development; it is not yet a deployed public download. ### Worldwide curated landmark baseline The additional worldwide shard is a transformed subset of the GeoNames `allCountries` gazetteer and is licensed under CC BY 4.0: - source: [GeoNames export dump](https://download.geonames.org/export/dump/) - format and reuse terms: [GeoNames readme](https://download.geonames.org/export/dump/readme.txt) - licence: [Creative Commons Attribution 4.0](https://creativecommons.org/licenses/by/4.0/) - attribution: `GeoNames (www.geonames.org), world gazetteer snapshot 2026-08-30` The `allCountries.zip` snapshot was retrieved on 2026-08-30. Its compressed SHA-256 is `d3611a5bacee47be394193695255e1319772569ba5b091de12fbbd013710c431`; the extracted `allCountries.txt` SHA-256 is `cbb7b7f11ee443d09972324e1d3fdba58e2e630a6dc7f8caf4b513e67a14a3dd`. The pinned English feature-code definition has SHA-256 `d44f129c1cffdeabf759c99fe7ea5b84f615c7f3e7366fb76ccf990f408f6af7`. `Scripts/build_world_landmarks.mjs` processed 13,463,895 rows, found 31,429 eligible records before seed exclusion, removed 48 likely seed duplicates, and selected 3,441 point landmarks across 192 countries and territories. Germany is excluded because its deeper OSM-derived catalogue remains authoritative. The transformation allowlists visitable feature codes, validates coordinates and ISO country codes, applies feature-specific alternate-name thresholds, deduplicates, limits records per country, and balances category and spatial coverage. Alternate-name count is selection evidence only; it is not a claim of objective importance. GeoNames also does not guarantee that every record is complete, current, or suitable as access guidance. For Austria only, Collect enriches the bounded GeoNames starter selection with state metadata by running each point against the already bundled Natural Earth 5.1.2 Admin-1 Austria geometry. Of 40 source landmarks, 38 resolve to seven states. The exact-boundary records `world:geonames:2778624` and `world:geonames:6697201` remain unassigned; the transform does not guess a side or imply state-balanced national coverage. The receipt pins the subdivision catalogue and Austria-boundary hashes alongside these counts. Generated files: - `CollectCore/Sources/CollectCore/Resources/landmarks-world.json` - SHA-256: `b2e700af7eae42cacbdee31c8d6d1f7da6191f7749ced3be7e6326a377542f64` - `CollectCore/Sources/CollectCore/Resources/landmarks-world.manifest.json` - records source hashes, selection policy, counts, and output hash The worldwide shard uses stable source-qualified IDs in the form `world:geonames:`. It is a compact discovery baseline, not a claim of complete landmark coverage in any country. The 110 hand-curated `seed:` landmarks remain the only runtime **Legendär** tier and take precedence over nearby source duplicates. ## Landmark medal identity references The Legendary landmark medal pilot uses freely licensed photographs only as architectural identity references. OpenAI ImageGen transformed them into stylized, text-free medal reliefs using the separately authored Brandenburg medal as a material and composition reference. The generated results are modified works and do not imply endorsement by a photographer, landmark owner, or public institution. Exact revisions and checksums are recorded in `Design/LandmarkMedalPilot/manifest.json`. - Kölner Dom: anonymous 1909 photograph, public domain. https://commons.wikimedia.org/wiki/File:Cologne_Cathedral,_west_front,_1909.jpg - Schloss Neuschwanstein: Wilfredor, CC0 1.0. https://commons.wikimedia.org/wiki/File:Neuschwanstein_Castle_2024.jpg - Reichstag: Ansgar Koreng, CC BY-SA 3.0 Germany. Modified into generated medal artwork; attribution and ShareAlike terms are retained. https://commons.wikimedia.org/wiki/File:Reichstagsgeb%C3%A4ude_Westfassade.jpg - Heidelberger Schloss: Immanuel Giel, public domain / PD-self. https://commons.wikimedia.org/wiki/File:Gesamtansicht_des_Heidelberger_Schlosses.jpg - Wartburg: Photo by CEphoto, Uwe Aranas, CC BY-SA 4.0. Modified into generated medal artwork; attribution and ShareAlike terms are retained. https://commons.wikimedia.org/wiki/File:Eisenach_Germany_Wartburg-Castle-00.jpg