Data sources

Mapzen Terrain Tiles are powered by several major open data sets and we owe a tremendous debt of gratitude to the individuals and communities which produced them.

Attribution is required for some data providers. See the Attribution document for more information.

List of sources

The underlying data sources are a mix of:

• 3DEP (formerly NED and NED Topobathy) in the United States, 10 meters outside of Alaska, 3 meter in select land and territorial water areas
• ArcticDEM strips of 5 meter mosaics across all of the land north of 60° latitude, including Alaska, Canada, Greenland, Iceland, Norway, Russia, and Sweden
• CDEM (Canadian Digital Elevation Model) in Canada, with variable spatial resolution (from 20-400 meters) depending on the latitude.
• data.gov.uk, 2 meters over most of the United Kingdom
• data.gv.at, 10 meters over Austria
• ETOPO1 for ocean bathymetry, 1 arc-minute resolution globally
• EUDEM in most of Europe at 30 meter resolution, including Albania, Austria, Belgium, Bosnia and Herzegovina, Bulgaria, Croatia, Cyprus, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Kosovo, Latvia, Liechtenstein, Lithuania, Luxembourg, Macedonia, Malta, Montenegro, Netherlands, Norway, Poland, Portugal, Romania, Serbia, Slovakia, Slovenia, Spain, Sweden, Switzerland, and United Kingdom
• Geoscience Australia's DEM of Australia, 5 meters around coastal regions in South Australia, Victoria, and Northern Territory
• GMTED globally, coarser resolutions at 7.5", 15", and 30" in land areas
• INEGI's continental relief in Mexico
• Kartverket's Digital Terrain Model, 10 meters over Norway
• LINZ, 8 meters over New Zealand
• SRTM globally except high latitudes, 30 meters (90 meters nominal quality) in land areas

Footprints database

These source images are composited to form tiles that make up the Mapzen Terrain Tiles service. To determine exactly which images contributed to Mapzen Terrain Tiles in a particular area, you can download the footprints database and use it with a GIS program like QGIS to inspect which of these sources were used.

• GeoJSON (8.7MB, gzipped)
• PostgreSQL Dump (14.5MB, gzipped)

Source headers

To further assist in determining which sources contributed to an individual tile, the Mapzen Terrain Tiles service will respond with an HTTP header listing the sources that contributed to that tile. The value of the X-Imagery-Sources HTTP header is a comma-separated list, where each entry follows the pattern source/filename.tif.

For example, a tile might have the header X-Imagery-Sources: srtm/N39W110.tif, srtm/N39W112.tif, gmted/30N120W_20101117_gmted_mea075.tif, meaning that it was built from three source images. Two SRTM images and a GMTED image were composited together to generate the tile output. Using the footprint database dumps above you can gather more information about these source images, including the calculated resolution and footprint geometry. To find the entry in the database, look for an entry that has a matching filename attribute.

What is the ground resolution?

Ground resolution per tile pixel varies per zoom level, the given pixel cell's latitude, and input data source.

This formula generates the following table:

ground_resolution = (cos(latitude * pi/180) * 2 * pi * 6378137 meters) / (256 * 2^zoom_level pixels)

Ground resolution per zoom in meters at a given latitude:

zoom	0°	45°	60°
0	156543.0	110692.6	78271.5
1	78271.5	55346.3	39135.8
2	39135.8	27673.2	19567.9
3	19567.9	13836.6	9783.9
4	9783.9	6918.3	4892.0
5	4892.0	3459.1	2446.0
6	2446.0	1729.6	1223.0
7	1223.0	864.8	611.5
8	611.5	432.4	305.7
9	305.7	216.2	152.9
10	152.9	108.1	76.4
11	76.4	54.0	38.2
12	38.2	27.0	19.1
13	19.1	13.5	9.6
14	9.6	6.8	4.8
15	4.8	3.4	2.4

Note: Esri has documentation about converting web map zoom integers to conventional map scales.

What is sourced at what zooms?

Generally speaking, GMTED is used at low-zooms, ETOPO1 is used to show ocean bathymetry at all zooms (even bathymetry oversampled at zoom 15), and SRTM is relied on in mid- and high-zooms on land. Some countries have higher resolution data available over land sourced from other open datasets. More information about these sources is available below.

It should be noted that both SRTM and GMTED fill oceans and other bodies of water with a value of zero to indicate mean sea level; in these areas, ETOPO1 provides bathymetry (as well as in regions which are not covered by SRTM and GMTED).

Data sources per zoom:

zoom	ocean	land
0	ETOPO1	ETOPO1
1	ETOPO1	ETOPO1
2	ETOPO1	ETOPO1
3	ETOPO1	ETOPO1
4	ETOPO1	GMTED
5	ETOPO1	GMTED
6	ETOPO1	GMTED
7	ETOPO1	SRTM, NRCAN in Canada, with GMTED in high latitudes above 60°
8	ETOPO1	SRTM, NRCAN in Canada, with GMTED in high latitudes above 60°
9	ETOPO1	SRTM, NRCAN in Canada, EUDEM in Europe, with GMTED in high latitudes above 60°
10	ETOPO1, NED Topobathy in California	SRTM, data.gov.at in Austria, NRCAN in Canada, SRTM, NED/3DEP 1/3 arcsec, data.gov.uk in United Kingdom, INEGI in Mexico, ArcticDEM in latitudes above 60°, LINZ in New Zealand, Kartverket in Norway
11	ETOPO1, NED Topobathy in California	SRTM, data.gov.at in Austria, NRCAN in Canada, SRTM, NED/3DEP 1/3 arcsec and 1/9 arcsec, data.gov.uk in United Kingdom, INEGI in Mexico, ArcticDEM in latitudes above 60°, LINZ in New Zealand, Kartverket in Norway
12	ETOPO1, NED Topobathy in California	SRTM, data.gov.at in Austria, NRCAN in Canada, SRTM, NED/3DEP 1/3 arcsec and 1/9 arcsec, data.gov.uk in United Kingdom, INEGI in Mexico, ArcticDEM in latitudes above 60°, LINZ in New Zealand, Kartverket in Norway
13	ETOPO1, NED Topobathy in California	SRTM, data.gov.at in Austria, NRCAN in Canada, SRTM, NED/3DEP 1/3 arcsec and 1/9 arcsec, data.gov.uk in United Kingdom, INEGI in Mexico, ArcticDEM in latitudes above 60°, LINZ in New Zealand, Kartverket in Norway
14	ETOPO1, NED Topobathy in California	SRTM, data.gov.at in Austria, NRCAN in Canada, SRTM, NED/3DEP 1/3 arcsec and 1/9 arcsec, data.gov.uk in United Kingdom, INEGI in Mexico, ArcticDEM in latitudes above 60°, LINZ in New Zealand, Kartverket in Norway
15	ETOPO1, NED Topobathy in California	SRTM, data.gov.at in Austria, NRCAN in Canada, SRTM, NED/3DEP 1/3 arcsec and 1/9 arcsec, data.gov.uk in United Kingdom, INEGI in Mexico, ArcticDEM in latitudes above 60°, LINZ in New Zealand, Kartverket in Norway

Sources native resolution

You might be wondering why we source from different data sets at different zooms. Besides bandwidth reasons, it's helpful to know the native resolution of each data set which is expressed as a nominal arc resolution which maps roughly to a web map zoom for "best displayed at".

In more practical terms, this results in some datasets being "oversampled" for a given zoom and map location.

• In the water, most bathymetry values at zoom 15 for a pixel that has a ground resolution of 5 meter will actually be showing an oversampled zoom 6 ETOPO1 value (nominally 2.5 km).

• On the land, most elevation values at zoom 15 for a pixel that has a ground resolution of 5 meter will actually be showing an oversampled zoom 12 SRTM value (nominally 30 meters).

This formula generates the following table:

ground_resolution = (cos(latitude * pi/180) * 2 * pi * 6378137 meters) / (256 * 2^zoom_level pixels) / 30.87 meters per arc second

Arc resolution per zoom and data sources, per pixel:

zoom	meters at equator	arc seconds	nominal arc degrees minutes seconds	source	nominal ground units
0	156543.0	5071.0	1.5 arc degrees
1	78271.5	2535.5	40 arc minutes
2	39135.8	1267.8	20 arc minutes
3	19567.9	633.9	10 arc minutes
4	9783.9	316.9	5 arc minutes
5	4892.0	158.5	2.5 arc minutes
6	2446.0	79.2	1 arc minutes	ETOPO1	2.5 km
7	1223.0	39.6	30 arc seconds	GMTED2010	1km (not used)
8	611.5	19.8	15 arc seconds	GMTED2010	500m (not used)
9	305.7	9.9	7.5 arc seconds	GMTED2010	250m
10	152.9	5.0	5 arc seconds
11	76.4	2.5	3 arc seconds	Canada	90m
12	38.2	1.2	1 arc seconds	SRTM, Canada	30m
13	19.1	0.6	2/3 arc seconds
14	9.6	0.3	1/3 arc seconds	3DEP, Austria, Australia, New Zealand, Norway	10m
15	4.8	0.2	1/5 arc seconds	Mexico, Arctic
16	2.4	0.1	1/9 arc seconds	3DEP, United Kingdom	3m

Data updates

Terrain tiles version 1 was built during 2016Q2 and released in 2016Q3 based on available sources at that time. Version 1.1 was built during 2017Q3 and released 2017Q4. Regular updates are not planned.

Future updates will be on an as-needed basis for smaller regions to incorporate additional 3DEP coverage in the United States and additional country specific data sources globally.

We are always looking for better datasets. If you find a data issue or can suggest an open terrain datasets, please let us know by filing an issue in tilezen/joerd.

Known issues

Many classical DEM and LIDAR related issues occur in terrain tiles. It is not uncommon to see large variations in elevation in areas with large buildings and other such structures. Resampling and merging artifacts are also observed along coastlines (where different datasets are seamed together).