Data sources and accuracy

Elevation

Terrain comes from Mapterhorn, an open terrain tile set that uses the best open elevation model available for each place:

• Copernicus GLO-30 (about 30 m) everywhere on land — a newer and more accurate global model than SRTM.
• National elevation models where they are open — more than a hundred of them, in much of Europe, North America, Japan, Taiwan and elsewhere, down to about 1 m. The full list is on the Mapterhorn site.

If Mapterhorn does not answer, TOPOLUMA falls back to Terrain Tiles on AWS (SRTM and national data, assembled by Mapzen). Empty patches in any tile are rebuilt from coarser data rather than shown as holes, and you are told when that happens.

TOPOLUMA reads the tiles at up to zoom 15 and interpolates between source samples. Smoothing is applied before contouring so that a 30 m grid does not show as stair-steps. Both Copernicus GLO-30 and SRTM are surface models: in dense forest and cities the ground can read several metres high.

What the accuracy means for your project

• Good for: understanding the shape of the land, orientation and slope, early massing, competition and feasibility drawings, site models, and context around a surveyed plot.
• Not good for: retaining walls, terraces, kerbs, small embankments, drainage design or anything you will build on. Vertical errors of several metres are possible in steep or forested terrain.
• Always survey the plot itself before design development or a permit application.

Buildings, roads, water and vegetation

Map layers come from OpenStreetMap, queried live through the Overpass API and cached for seven days. When Overpass is overloaded, TOPOLUMA falls back to OpenStreetMap vector tiles from OpenFreeMap, which contain the same buildings, roads and water in slightly simplified form but no individual trees.

Coverage of OpenStreetMap is excellent in most cities and varies in rural areas. Building heights are tagged for a minority of buildings worldwide; the studio tells you how many buildings in your area have a height and uses your default for the rest.

Generated trees

Most forests are mapped in OpenStreetMap as outlines rather than as single trees, and many places have no woods mapped at all. To give 3D views, sections and models realistic tree cover anywhere in the world, TOPOLUMA generates trees from two sources:

• ESA WorldCover 2021, a global land cover map at 10 m resolution derived from Sentinel-1 and Sentinel-2. Tree cover and mangroves become trees; shrubland becomes low bushes.
• OpenStreetMap woods, orchards and parks, each with its own spacing and height range. Where OpenStreetMap maps any vegetation, including lawns and meadows, it takes priority over the land cover map.

Trees are kept clear of buildings, water, roads and railways. Their positions are spread evenly through the tree cover and their heights are estimated by type: they show where trees are, not each individual tree. Trees mapped individually in OpenStreetMap are always used as they are.

Background maps

• Street map: OpenFreeMap styles built on OpenMapTiles.
• Satellite (on screen only, never in exports): Esri World Imagery (resolution and date per place from Esri's metadata); Sentinel-2 cloudless 2016 by EOX (10 m); NASA GIBS HLS Sentinel-2 (30 m, one day's pass); USGS The National Map orthoimagery (1 m NAIP, conterminous US); and OpenAerialMap drone and aerial surveys where they exist. The comparison tool shows each source's resolution, date, coverage and credit.
• Place search: Photon with Nominatim as a fallback.

Licences and attribution

Commercial use of your exports is allowed under these licences.