Global DEM Tile Sources and Basemaps: Mapterhorn, Mapbox, MapTiler, AWS, Esri
The worldwide elevation tile sources behind the viewer (Mapterhorn, Mapbox Terrain-RGB, MapTiler, AWS Terrarium, Esri World Elevation LERC), their resolution and licences, and the satellite and map basemaps.
The Sources group in the sidebar covers two independent pickers — Terrain (the DEM driving elevation/relief/exaggeration) and Basemap (the imagery draped on top, or shown as Raster Basemap) — see Overview → Sidebar groups for where this sits relative to the rest of the app. Both accept your own sources too, not just the built-ins below — see Bring Your Own Data.
The built-ins on this page are worldwide tilesets. For the government LiDAR services bundled alongside them — 0.5 m for France and the Netherlands, 1 m for Norway, England and the USA — see National Open-Data Terrain.
Terrain
Every built-in terrain source streams raster-dem tiles (Terrarium or Mapbox Terrain-RGB encoding) — pick one from Sources → Terrain, or click its ⓘ (info) icon there for the same summary shown here plus a ready-to-run GDAL export command.
Heights are above the geoid, not the ellipsoid
Every built-in DEM here publishes orthometric heights, metres above mean sea level: Mapterhorn's Copernicus GLO-30 base is on the EGM2008 geoid, and the national lidar it swaps in stays on each country's own datum, within a metre or so of that. A GNSS receiver, a 3D Tiles model or Cesium's globe count from the WGS84 ellipsoid instead, and the two differ by the geoid undulation: from about −106 m (south of India) to +85 m (around Iceland), +45 m in the Netherlands, +48 m in the Alps. Inside this app nothing moves, since every source sits on the same plane; the offset shows up when an ellipsoidal DEM (ArcticDEM and REMA, Cesium World Terrain, Re:Earth) is compared with an orthometric one, which is what the Difference source's bias check is for; the library ships the EGM96 and EGM2008 grids as terrain sources for the conversion.
The converse, Mapterhorn on an ellipsoidal globe, is what Bert Temme's cesium-mapterhorn-provider PR 3 solves, and it is the reference for the method: a verticalReference: "ellipsoid" option samples the EGM2008 2.5′ geoid grid as a Cloud Optimized GeoTIFF from cdn.proj.org through HTTP range requests (the file is 80 MB, a view loads only the tiles under it), nine by nine points per terrain tile interpolated bilinearly, the same approach as Re:Earth Terrain, checked against its heights.json to the millimetre at four points. Should this app ever drape terrain under ellipsoidal 3D Tiles, that is the recipe.
Terrain source info dialog (Mapterhorn) — Patagonia, Chile — open in app ↗
Mapterhorn — Terrarium
mapterhorn.com · Terrarium encoding · zoom 18 · 512px tiles
An open, community-run "Protomaps for terrain" project: a global low-resolution base (ESA Copernicus GLO-30) with high-resolution national LiDAR data swapped in wherever it's available — its own coverage page maps out exactly which regions get the sharper zoom-13+ tiles.
Mapbox — Terrain-DEM v1
docs.mapbox.com · Terrain-RGB encoding · zoom 14 · 256px tiles
Global coverage at 0.1m elevation increments — a compressed, optimized successor to Mapbox's original Terrain-RGB v1 tileset. Only available through Mapbox's own SDKs (not the plain Raster Tiles API), and requires a Mapbox API key.
MapTiler — Terrain-RGB v2
maptiler.com/terrain · Terrain-RGB encoding · zoom 12 · 512px tiles
A composite of multiple high-resolution DEMs curated by MapTiler's geodata team — ~30m resolution globally, down to 5m in several European countries. MapTiler also publishes a separate "Ocean RGB" tileset for bathymetry, though this app only wires up the land-elevation Terrain-RGB one. Requires a MapTiler API key.
AWS Terrain Tiles — Terrarium (Mapzen)
registry.opendata.aws · Terrarium encoding · zoom 16 · 256px tiles
The original open Tilezen/Joerd dataset (built by Mapzen, continuing as an AWS Open Data Registry mirror after Mapzen's shutdown) — the only built-in source with genuine ocean bathymetry: land elevation (SRTM at mid/high zoom, GMTED2010 at low zoom) is blended with global ETOPO1 bathymetry everywhere water replaces a zero-elevation gap, plus regional high-resolution inserts (US 3DEP/NED, ArcticDEM, EU-DEM, Canada's CDEM, and more). No API key needed.
Esri World Elevation — LERC
Esri's global best-available blend (Vantor, Airbus DS, USGS, NGA, NASA, CGIAR, GEBCO, Land Information New Zealand, Ordnance Survey and others), in its TopoBathy form: land and seafloor as one continuous surface, so a hypsometric ramp or a 3D view reads across a coastline instead of stopping at it. The second built-in with real bathymetry, alongside AWS Terrain Tiles.
No API key. Unlike the others it is not an RGB packing at all — it streams LERC, Esri's own float raster codec, decoded in the browser by lerc://. Heights are orthometric (measured at −0.6 m mean against a reference DEM), and real data runs to about z16.
A land-only variant, with the ocean flat at 0 m rather than the seafloor, is in the Library — worth choosing when a coastline should be a hard edge, since a −7000 m trench sharing the range wrecks a ramp keyed to land relief.
These are Esri ArcGIS services. They answer without a token and with CORS open, but Esri's Terms of Use contemplate use with ArcGIS products, and consuming them from a non-Esri client is not clearly granted — worth checking before relying on one in anything published.
Which dataset is actually under my cursor?
The Sources Coverage section (Sources group in the sidebar, under Basemap) hints at exactly that for the terrain/basemap that's currently exposed at your viewport — e.g. which of Mapterhorn's composited national LiDAR datasets is active at that location (vs. its global Copernicus GLO-30 fallback), or which date/release the active ESRI imagery resolves to. Handy for knowing whether you're looking at a genuinely high-resolution local dataset or a coarser global default.
Basemap Sources
The Basemap section under Sources offers several satellite/hybrid TMS providers, all built-in with no setup required beyond an API key where noted:
- Google Hybrid — satellite + roads/labels
- Google Satellite — satellite only
- Bing Aerial
- ESRI World Imagery
- Mapbox Satellite — requires a Mapbox API key
- HERE Satellite — requires a HERE API key
- OpenStreetMap — vector-style street map, not satellite
Plus the combined Historical Imagery entry (ESRI Wayback, HLS, Google Earth Historical, Planet) — see Historical Imagery Timeline for the timeline, and Historical Sources for the providers behind it and how each resolves a date to a tile.