Dataset Viewer

The dataset viewer is not available because its heuristics could not detect any supported data files. You can try uploading some data files, or configuring the data files location manually.

Philippines OpenStreetMap Vector Tiles (PMTiles)

Dataset Summary

A single PMTiles archive of OpenStreetMap data for the Philippines, with 12 layers.

It merges the 12 Humanitarian OpenStreetMap Team (HOT) exports for the Philippines. Each dataset keeps its own named vector tile layer.

  • File: ph_osm.pmtiles (~946 MB)
  • Format: Mapbox Vector Tiles (MVT) in a PMTiles archive
  • Zoom range: z0–z15
  • Tiles: 155,269 (as of the 2026-10-13 HOT export)
  • Bounds: ~114.3–126.6°E, 4.6–21.1°N

Languages

Feature names come from OpenStreetMap, mostly English (name_en), Filipino (name_fil) and local names (name). Attribute names and layer names are in English.

Dataset Structure

Data Instances

The repository holds one archive:

ph_osm.pmtiles is stored via Git LFS. Each tile (z/x/y) holds every layer that has features at that address. Each feature is a point, line or polygon with OSM tags as properties.

Example feature properties from roads:

{
  "id": "way/1382563216",
  "highway": "residential",
  "name": "Example Street",
  "adm0_name": "Philippines",
  "adm0_pcode": "PH"
}

(Values are illustrative. Use the metadata command below to list the real fields.)

Layers

Layer Dataset Source size Max zoom
airports Airports 1.0 MB 12
buildings Buildings 644.1 MB 15
cultural_places Cultural places 5.3 MB 12
education_facilities Education facilities 24.9 MB 12
financial_services Financial services 1.5 MB 12
health_facilities Health facilities 3.2 MB 12
points_of_interest Points of interest 38.8 MB 12
populated_places Populated places 23.9 MB 12
railways Railways 0.2 MB 12
roads Roads 158.7 MB 13
sea_ports Sea ports 0.6 MB 12
waterways Waterways 58.7 MB 12

Layers have different max zooms. MapLibre overzooms automatically, so roads (z13) and the z12 layers stay visible when you zoom past their max zoom, while buildings continues to z15.

Data Fields

Shared by all layers

  • id — OSM feature id, e.g. way/1382563216
  • name, name_en, name_fil, name_latin — names (not every layer has all four)
  • adm0_name/adm0_pcode … adm4_name/adm4_pcode — country down to barangay, with PSGC-style codes

Dataset-specific (examples)

  • roads — highway, surface, lanes, oneway, bridge, width, smoothness
  • buildings — building, building_levels, addr_*
  • waterways — waterway, water, tunnel, width
  • health_facilities — amenity, healthcare, healthcare_speciality, capacity_persons
  • populated_places — place, population, is_in

List every field per layer with the pmtiles CLI:

pmtiles show --metadata ph_osm.pmtiles

Known Quirks

  • Mixed geometry — most layers hold points and polygons together (for example points_of_interest). Filter by $type when styling.
  • Large low-zoom tiles — the tile size limit was removed during the merge, so some low-zoom tiles are large (tile 5/26/14 is ~1.3 MB gzipped).

Data Splits

None. The dataset is one archive covering the whole country, organized by layer rather than train/validation/test splits.

PMTiles

Viewing

Browse the file in the PMTiles viewer.

Usage with MapLibre GL JS

Load the file directly from Hugging Face using the pmtiles protocol plugin for MapLibre GL JS. To self-host instead, replace PMTILES_URL; the host must support HTTP range requests.

<script src="https://unpkg.com/maplibre-gl/dist/maplibre-gl.js"></script>
<script src="https://unpkg.com/pmtiles/dist/pmtiles.js"></script>
const protocol = new pmtiles.Protocol();
maplibregl.addProtocol("pmtiles", protocol.tile);

const map = new maplibregl.Map({ container: "map", center: [121.0, 14.6], zoom: 9, /* ... */ });

const PMTILES_URL =
  "pmtiles://https://huggingface.co/datasets/bettergovph/ph-osm-pmtiles/resolve/main/ph_osm.pmtiles";

map.on("load", () => {
  map.addSource("osm", { type: "vector", url: PMTILES_URL });

  // source-layer options: airports, buildings, cultural_places, education_facilities,
  // financial_services, health_facilities, points_of_interest, populated_places,
  // railways, roads, sea_ports, waterways
  map.addLayer({
    id: "buildings",
    type: "fill",
    source: "osm",
    "source-layer": "buildings",
    filter: ["==", "$type", "Polygon"],
    paint: { "fill-color": "#9ca3af", "fill-opacity": 0.7 }
  });

  map.addLayer({
    id: "roads",
    type: "line",
    source: "osm",
    "source-layer": "roads",
    filter: ["==", "$type", "LineString"],
    paint: { "line-color": "#f59e0b", "line-width": 1 }
  });

  map.addLayer({
    id: "health-facilities",
    type: "circle",
    source: "osm",
    "source-layer": "health_facilities",
    filter: ["==", "$type", "Point"],
    paint: { "circle-color": "#dc2626", "circle-radius": 4 }
  });
});

PMTiles conversion process

The 12 HOT source archives were merged with tile-join from tippecanoe 2.79.0 and checked with the pmtiles CLI 1.31.2, both inside Docker.

The scripts used to build this archive are in the ph-osm-pmtiles tooling.

  • pmtiles merge was not usable. It only merges disjoint archives, and all 12 sources cover the same tile addresses.
  • tile-join decodes each tile and writes one tile per z/x/y containing every source layer.
  • -pk was used to skip the default 500 KB tile size limit.
  • Layer ids are unchanged, so output layer names match the dataset names in the source filenames (hotosm_phl_<layer>_osm.pmtiles).

Dataset Creation

Source Data

The 12 PMTiles files are HOT raw-data API exports on S3 (~939 MB in total, last modified 2026-10-13), published on HDX — Philippines. Underlying data is from OpenStreetMap.

Methodology

The tiles were not re-generated from raw OSM. Each source layer was copied as-is from its HOT export and combined into one archive. Feature geometry, tag values and zoom thinning are inherited from the HOT exports.

Annotations

None beyond the OSM tags and administrative attributes (adm0–adm4) already present in the HOT exports.

Personal and Sensitive Information

OSM data is crowd-sourced and public. It may include names of people or businesses in feature names and address tags. No additional personal data was added.

Additional Information

Licensing Information

The data is available under the Open Database License (ODbL) 1.0.

You may share and adapt it, provided you attribute OpenStreetMap contributors and the Humanitarian OpenStreetMap Team.

If you alter or build upon the data, you may distribute the result only under the same license.

Citation

Data © OpenStreetMap contributors, available under the Open Database License.
Exported by the Humanitarian OpenStreetMap Team (HOT) via HDX.
Downloads last month
54