# Data licence

지도 데이터 © OpenStreetMap 기여자, ODbL 1.0. 이 폴더(`data/`)의 파일은 OpenStreetMap 데이터에서 만든 파생 데이터베이스이며
아래 경로에서 그대로 내려받을 수 있습니다.

## OpenStreetMap — Open Database License (ODbL) 1.0

Every data file under `data/` (the folder that holds this notice) is a Derivative Database of OpenStreetMap data,
© OpenStreetMap contributors (https://www.openstreetmap.org/copyright), and is made available under the Open Database
License (ODbL) 1.0 (https://opendatacommons.org/licenses/odbl/1-0/). Some files also contain the other sources listed
below. The OpenStreetMap-derived place, street and station names in the game code (`src/data/places-seoul.js`,
`src/data/subway-lines.js`) and the landmark footprints measured on OpenStreetMap are covered by the same notice.

The Derivative Database consists of these files (paths relative to this file):

| Path | Contents |
|---|---|
| `manifest.json` | index of the engine data below (sizes, hashes, attribution) |
| `ground.json`, `vectors.json`, `river.json`, `street-trees.json`, `park-trees.json` | ground cover, road / rail / water vectors, river outline, street and park trees (engine core) |
| `terrain-near.json`, `terrain-far.json` | terrain heights (Terrain Tiles, see below) |
| `tiles/buildings/*.json` | building footprints and heights of the engine core |
| `neighborhood/*.json` | regional building packs (`index.json`, `E1.json`, `E2.json`, `E2-index.json`; Overture / Shi et al. / building register, see below) |
| `nodeul-roads/*.json`, `parked/*.json`, `riverside-*/*.json` | road overlay, parked-car slots, riverside paths |
| `north/*.json` | the northern city centre (Gwanghwamun, Jongno, Gyeongbokgung, Bukchon, City Hall) |
| `seoul/index.json`, `seoul/graph.bin`, `seoul/tiles/*.bin`, `seoul/tiles2/*`, `seoul/districts.json`, `seoul/pois.json`, `seoul/overview.json` | Seoul-wide roads, buildings, land use, districts and places (SR containers; the deployed `.bin` files may be gzip-compressed) |
| `rg2-cache.bin`, `boot-cache/*.bin` | precomputed geometry derived from the files above (gzip) |

The engine files (`*.json` with an `hkb1` / `hkb2` payload) store the binary tables as gzip + base64; `manifest.json`
lists each file's size and SHA-1 / SHA-256.

**Offer (ODbL §4.6):** the entire Derivative Database is the set of files above, offered in machine-readable form at
the paths given, next to this notice, to everyone who can open the game. The pinned OpenStreetMap extracts they were
made from (Overpass API responses of 2026-10-01, 2026-10-02, 2026-10-12 and 2026-10-13 and a Geofabrik extract of
2026-09-26) and the bake scripts that produce them are available from the publisher of this site on request.

## Other sources

Terrain: Terrain Tiles by Mapzen / Tilezen via AWS Open Data (https://registry.opendata.aws/terrain-tiles/); SRTM data
courtesy of the U.S. Geological Survey; at low zoom levels GMTED2010 and ETOPO1 may contribute (see
https://github.com/tilezen/joerd/blob/master/docs/attribution.md). Used in `terrain-near.json`, `terrain-far.json`,
`river.json` and `vectors.json`.

Building roofprints (E1 Pungnap–Cheonho–Amsa low-rise fill, `neighborhood/E1.json`, desktop tier): Overture Maps
Foundation buildings theme, release 2026-09-23.1, licensed under ODbL 1.0
(https://docs.overturemaps.org/attribution/#buildings). The machine-learned roofprints in it come from Shi et al., East
Asian buildings dataset (https://doi.org/10.5281/zenodo.8174931), CC BY 4.0
(https://creativecommons.org/licenses/by/4.0/). Footprints are historical ML roofprints (imagery date unknown); heights
are typology estimates, not measurements.

Building heights (Seoul building register): 서울특별시, 서울시 건축물대장 표제부 (OA-22424,
https://data.seoul.go.kr/dataList/OA-22424/S/1/datasetView.do), 공공누리 제1유형 (KOGL Type 1, source attribution). Used for
(a) the heights in `tiles/buildings/*.json` of OSM footprints whose exact road address matches one register record 1:1
(`b.hsrc` 5 = register height, 6 = register floor count x register storey of its use class), (b) the statistical
floor-count prior of the E1 low-rise fill in `neighborhood/E1.json`, and (c) since 2026-10-02 the same 1:1 /
one-footprint address heights on 29 observed OSM buildings of the E1 regional pack (regional `b.hsrc` 5 / 6). An
address match is not a survey of the footprint; the excluded and estimated buildings are listed in the bake ledger.

E2 Seongsu / Ttukseom / Seoul Forest building fill (`neighborhood/E2.json`, `neighborhood/E2-index.json`): observed
OpenStreetMap building outlines (© OpenStreetMap contributors, ODbL 1.0; pinned Overpass extract of 2026-10-01) and,
desktop tier only, machine-learned roofprints from the Overture Maps Foundation buildings theme, release 2026-09-23.1
(ODbL 1.0), which come from Shi et al., East Asian buildings dataset (https://doi.org/10.5281/zenodo.8174931,
CC BY 4.0). Heights: OSM tags, 1:1 road-address matches in the Seoul building register (OA-22424, 공공누리 제1유형), else
a labelled statistical draw from the register rows of the footprint's 법정동 (not a record of that building). ML
roofprints are historical (imagery date unknown); no height is surveyed. Since 2026-10-02 the same block also covers
the south bank between the core bake and the Jamsil pocket (Apgujeong / Cheongdam / Sinsa / Samseong edge) and the
hills north of Seongsu (Haengdang / Wangsimni / Sageun / Majang / Yongdap), from two more pinned Overpass extracts
(osm_base 2026-10-02T05:40:59Z) and the same Overture release; park structures inside parks are drawn on the evidence
of a name / facility tag or a recent imagery-traced OSM edit (OSM API history, read-only).

Seoul-wide generated buildings (the last `n.bg` buildings of each `seoul/tiles/*.bin`, height source 3) are estimated
from OSM roads and land use; they are not real building positions.

Third-party code, models, motion data and fonts: see `../THIRD_PARTY_NOTICES.txt`.
