Compare elevation data products
Compare elevation products by their surface representation, reference systems and documented accuracy context before comparing grid spacing. The twenty rows below keep those characteristics separate so a smaller cell size does not stand in for evidence that a product fits your study.
Match the surface to the question
Decide whether your work needs bare ground or a surface that can include vegetation and structures. Write that requirement down before selecting a product. USGS distinguishes the surface types and source products within 3DEP; a product-family name alone is not a complete terrain specification.1
Then consider coverage and collection period. A product may offer broad geographical coverage while combining source observations from different times. The record’s acquisition field is therefore separate from the date on which its documentation was checked. Unknown acquisition details stay unknown rather than borrowing the publication date of a web page.
Use the searchable elevation table to narrow this list by product or descriptive text. It does not select a scientifically suitable product for you. For a step-by-step selection route, start with the elevation hub and keep your project requirements beside the table.
| ID | Product | Surface | Spacing | Vertical reference | Accuracy context | Terms |
|---|---|---|---|---|---|---|
| E01 | USGS 3DEP 1-metre DEM | Bare earth | 1 metre | Usually NAVD88; regional exceptions require tile metadata | QL2-or-better lidar source; no per-tile accuracy established | US public domain (USGS-produced) |
| E02 | Copernicus GLO-30 DGED | DSM including buildings and vegetation | 1 arc-second; nominal 30 metres | EGM2008, EPSG:3855 | Product specification: absolute vertical <4 m LE90; not local validation | Copernicus DEM ESA–User Licence (instance-specific) |
| E03 | ALOS AW3D30 | DSM | Nominal 1 arc-second; longitude angular spacing increases at high latitudes | EGM96 orthometric | Quality fields supplied; no uniform measured local accuracy established | JAXA Terms of Use of Research Data |
| E04 | ArcticDEM 2-metre strips | DSM | 2 metres | WGS84 ellipsoid | Strips unregistered to ground control/altimetry | NextView / EOCL public DEM release exception |
| E05 | GEBCO 2026 grid | Land elevation and bathymetry | 15 arc-seconds | Assumed mean sea level; some source datums differ | Grid spacing is not source-data resolution or accuracy | Public domain (GEBCO terms of use) |
| E06 | USGS 3DEP 1/3-arc-second DEM | Bare-earth seamless terrain | 1/3 arc-second; approximately 10 m north–south | Usually NAVD88; local/NGVD29 exceptions outside CONUS | Varies by source project; no uniform local value established | US public domain (USGS-produced) |
| E07 | USGS 3DEP 1-arc-second DEM | Bare-earth seamless terrain | 1 arc-second; approximately 30 m north–south | Usually NAVD88; local/NGVD29 exceptions outside CONUS | Varies by source project; no uniform local value established | US public domain (USGS-produced) |
| E08 | USGS Alaska 5-metre IfSAR DEM | Bare-earth DTM | 5 metres | Tile metadata required; do not assume CONUS vertical datum | Not established for selected tile | US public domain (USGS-produced) |
| E09 | USGS Alaska IfSAR DSM | DSM; vegetation and structures included | 5 metres | Not established in inspected sources | Not established for selected tile | US public domain (USGS-produced) |
| E10 | Copernicus GLO-90 | DSM; buildings and vegetation retained | 3 arc-seconds; nominal 90 m | EGM2008 (EPSG:3855) | Product specification <4 m absolute vertical LE90; not local validation | Copernicus DEM ESA–User Licence (instance-specific) |
| E11 | Copernicus EEA-10 | DSM; buildings and vegetation retained | DGED 0.4 arc-second; INSPIRE nominal 10 m | EGM2008 (EPSG:3855) | Product specification <4 m absolute vertical LE90; not local validation | Copernicus DEM ESA–User Licence (instance-specific) |
| E12 | ArcticDEM mosaic | DSM mosaic | 2 m primary; coarser mosaics also offered | WGS84 ellipsoid | Mosaic registration and source coverage vary; no local benchmark here | NextView / EOCL public DEM release exception |
| E13 | REMA strips | DSM strips | 2 metres | WGS84 ellipsoid | Strips not registered to altimetry; mosaic claim does not apply | NextView / EOCL public DEM release exception |
| E14 | REMA mosaic | DSM mosaic | 2 m; reduced resolutions 10/32 m and whole-domain grids | WGS84 ellipsoid | Provider reports <1 m absolute uncertainty over most registered area; not local validation | NextView / EOCL public DEM release exception |
| E15 | NASADEM HGT | Radar-derived terrain/surface with merged fills; not universally bare earth | 1 arc-second | EGM96 geoid for merged HGT; SRTM-only float product differs | Reprocessed product; no single local accuracy guarantee | CC0 (unmarked NASA-led mission data; check exceptions) |
| E16 | SRTM GL1 v3 | Radar-derived elevation; vegetation/buildings may influence heights | 1 arc-second | EGM96 | USGS describes mission accuracy; no local validation here | CC0 (unmarked NASA-led mission data; check exceptions) |
| E17 | CGIAR SRTM 90 m | Void-filled SRTM-derived elevation | 3 arc-seconds; nominal 90 m at equator | Not established in inspected sources | Page reports source vertical error <16 m; not independent CGIAR/local validation | CIAT SRTM distribution terms |
| E18 | MERIT DEM | Error-reduced terrain estimate | 3 arc-seconds; nominal 90 m at equator | EGM96 geoid | Provider reports 58% of land better than 2 m after correction; not uniform accuracy | CC BY-NC 4.0 or ODbL 1.0 (choice; additional condition) |
| E19 | FABDEM | Terrain estimate with forest/building biases removed | Uniform 1 arc-second; approximately 30 m at equator | EGM2008 EPSG:3855 | No single numerical local accuracy claim established here | CC BY-NC-SA 4.0 (commercial permission separate) |
| E20 | ASTER GDEM v3 | Optical-stereo surface elevation | 1 arc-second; nominal 30 m at equator | EGM96 | Known inaccuracies/artifacts; no uniform accuracy guaranteed | NASA/METI data-use conditions; no separate licence established |
Read spacing and accuracy as different fields
Grid spacing describes the arrangement of samples. A stated accuracy measure describes an error assessment under particular conditions. Keep the metric, terrain context and edition together rather than turning every numerical entry into a single comparable score.
This table retains the publisher’s accuracy basis where it was established. “No per-tile accuracy established” means that the inspected family-level evidence does not establish that property for the file you will use. It is not an assertion that a product has no quality documentation. Open the linked source and the selected tile’s metadata for the more specific evidence.
Do not convert arc-seconds to an unexplained fixed metre figure and use it as a universal resolution comparison.2 Preserve the published spacing notation in the acquisition record. If your analysis requires a projected grid, document the later projection and resampling choices as processing decisions rather than changing the source description.
Check vertical reference and units
A shared-looking horizontal grid does not establish compatible heights. Keep the vertical reference and vertical units alongside the horizontal reference. NOAA explains how mixing vertical datums can create discontinuities in a combined surface.3 A visible seam should therefore lead back to these fields as well as to collection dates and processing choices.
Where a source family has regional exceptions, the table says to inspect the actual tile. Avoid replacing that condition with the most common datum for the family. The selected product may require a different reference or an explicit transformation whose coverage and uncertainty must be considered separately.
The DEM inspection guide prepares a metadata command, while the mosaic guide describes a constrained combination workflow. Neither route supplies missing reference information or performs an accuracy validation by reading a file header.
Retain the terms with the product edition
The licence column is a short source label. Read the complete source-linked reuse conditions before downloading or sharing a product. Some entries identify a particular instance or release; others include non-commercial restrictions, geographic exceptions or separate permissions. A freely accessible download is not itself a blanket redistribution statement.
Download the complete JSON or CSV with the method and source ledger. These files preserve twenty product attributes plus source and date information. The original compilation is CC BY 4.0; the underlying elevation products keep their own terms.
Finish by writing a short selection note: product and edition, why its surface fits, coverage evidence, reference systems, unresolved fields and the official item link. That note makes a later change of product explainable without inventing a winner from incomplete or differently defined accuracy figures.
Not covered on this page
- Measured vertical accuracy, terrain suitability or a universal product ranking.
- A transformed or resampled elevation dataset, or validation of an individual tile.
- Legal advice, nautical navigation or engineering design from this comparison alone.
Sources
Code on this page is original and MIT licensed.
How we test: methodology. Found a mistake? Corrections.