Digital Elevation Models Explained: SRTM, ASTER and Copernicus
SRTM, ASTER GDEM and Copernicus DEM are the main global elevation datasets. Learn how they differ.
A Digital Elevation Model (DEM) is a raster dataset giving the ground elevation at regularly spaced intervals — the invisible scaffolding behind terrain maps, flood models, flight simulators and your phone's hiking app. Several global DEMs exist, collected by different sensors at different times, and choosing the right one matters. This guide compares the main products: how they were made, how accurate they are, and where to get them.
What a DEM actually is
A DEM is a grid of elevation values, each cell representing the height of the ground at that location — like a spreadsheet draped over the terrain. Cell size (resolution) might be 30 m, 10 m or 1 m; smaller cells capture more detail but make bigger files. DEMs come in two flavours: DSM (Digital Surface Model), which includes trees, buildings and anything else on the surface, and DTM (Digital Terrain Model or 'bare-earth'), which strips those away. Which you want depends on the job: a DSM is right for radio line-of-sight or drone flying; a DTM is right for flood modelling or archaeology. Most global products are DSMs.
SRTM: the 2000 Space Shuttle mission
The Shuttle Radar Topography Mission flew aboard Space Shuttle Endeavour for 11 days in February 2000, using radar interferometry from orbit to map about 80% of Earth's land surface (60° N to 56° S). The result — SRTM — was the first near-global elevation dataset and remains the most widely used. The 1 arc-second product has 30 m resolution with vertical accuracy around 16 m in most terrain; a 3 arc-second (90 m) version covers the same area in smaller files. Radar sees through clouds but not vegetation or buildings, so SRTM is a surface model that rides a few metres high over forests and cities. For most of the world, SRTM is still the default 'good enough' DEM.
ASTER GDEM: full global coverage
ASTER GDEM is derived from stereo optical imagery collected by the ASTER instrument aboard NASA's Terra satellite. Its headline advantage is coverage: the entire land surface, including the high latitudes SRTM missed. Resolution is 30 m, similar to SRTM, but vertical accuracy is somewhat lower — around 20 m — and optical stereo struggles in featureless terrain (deserts, ice sheets) where the matching algorithm finds nothing to lock onto, producing characteristic artefacts. Version 3 improved processing significantly. Choose ASTER when you need coverage above 60° N or below 56° S and accept its quirks, or use it to fill SRTM's gaps.
Copernicus DEM: the current accuracy leader
Released by the European Space Agency, the Copernicus DEM is derived from the commercial WorldDEM product (itself from the TanDEM-X radar mission) and published at 30 m and 90 m resolutions with global coverage. It is widely regarded as the most accurate freely available global DEM, with vertical accuracy around 4 m in many regions — several times better than SRTM. It also has cleaner coastlines and fewer voids. If you need the best freely available global elevation data today and your area is covered, Copernicus DEM is the default choice; its main limitation is simply that some workflows and older tools still default to SRTM out of habit.
ALOS and other sources
JAXA's ALOS World 3D (AW3D30) is another 30 m global product from optical stereo, competitive with SRTM in quality and useful as an independent check. National mapping agencies publish better data for their own countries — the USGS 3DEP program provides 10 m and even 1 m DEMs for much of the United States, and several European countries publish open 1–5 m LiDAR DEMs. For the highest detail, airborne LiDAR surveys produce 1 m or finer DEMs but only over limited areas; they are the standard for engineering, flood modelling and forestry. Always check what your country publishes before settling for a global product.
Resolution vs accuracy: not the same thing
Beginners often conflate resolution (cell size) with accuracy (how close the heights are to reality). A 30 m DEM can have 16 m vertical accuracy (SRTM) or 4 m (Copernicus) — same resolution, very different quality. Conversely, resampling a coarse DEM to fine cells does not create detail; it just makes bigger files with the same information. When comparing products, look at three numbers: resolution, vertical accuracy (often quoted as RMSE or at 90% confidence — different statistics, not directly comparable), and vintage. A 2020 DEM of a fast-changing landscape (glaciers, mines, cities) beats a 2000 DEM regardless of nominal accuracy.
Voids, artefacts and what to watch for
Every DEM has flaws. Radar shadows behind steep mountains create voids (missing data), especially in SRTM's mountain coverage. Optical stereo produces pits and spikes in clouds, snow and deserts. Coastlines are tricky: some products flatten water to zero, others include wave noise. Urban areas ride high on buildings. Before trusting a DEM for analysis, hillshade it — rendering the terrain with artificial lighting makes artefacts, steps at tile boundaries, and voids visually obvious in seconds. Filling small voids by interpolation is standard practice; large voids should make you switch products.
Where to download DEM data free
USGS EarthExplorer hosts SRTM and ASTER GDEM free after a simple registration. The Copernicus DEM is available through the Copernicus Data Space Ecosystem and via the AWS Open Data program (no account needed for public buckets). OpenTopography provides high-resolution LiDAR DEMs for research areas. National portals (USGS 3DEP, UK DEFRA Data Services, various European geoportals) serve country-specific data. Files are typically GeoTIFFs in tiles — download the tiles covering your area of interest, and use GDAL or QGIS (both free) to merge, reproject and hillshade them.
Choosing the right DEM
A practical decision tree: need the best free global data? Copernicus DEM 30 m. Working above 60° N? Copernicus or ASTER (SRTM has no data there). Need the US at high detail? USGS 3DEP 10 m or 1 m. Engineering or flood modelling? LiDAR DTM if available, otherwise the best national product. Quick global visualisation? SRTM 90 m keeps files small. And always ask DSM-or-DTM first: if your analysis needs bare earth (hydrology, archaeology, construction), a surface model with forests and buildings baked in will give wrong answers no matter how 'accurate' its spec sheet claims.
Hands on
Query the elevation finder for a spot you know well, run a slope calculation on it, then practice reading a topographic map of the same area.
Frequently asked questions
Which global DEM is best?
Copernicus DEM (30 m) is currently the most accurate freely available global product, at around 4 m vertical accuracy in many regions. SRTM remains the most widely supported, and ASTER fills the high latitudes SRTM misses.
Can I download DEM data for free?
Yes. USGS EarthExplorer hosts SRTM and ASTER; the Copernicus DEM is on the Copernicus Data Space Ecosystem and AWS Open Data; OpenTopography serves research LiDAR. All are free, most after a simple registration.
What is the difference between a DSM and a DTM?
A Digital Surface Model includes everything on the surface — trees, buildings, towers. A Digital Terrain Model strips those away to show bare earth. Use DSMs for line-of-sight and aviation, DTMs for hydrology, archaeology and construction.
Why does my DEM have holes in the mountains?
Radar-based DEMs like SRTM get no return signal in radar shadows behind steep ridges, leaving voids. Fill small ones by interpolation; for large void areas, switch to a product with better coverage such as Copernicus DEM.
Sources & data
Authoritative references used to research and verify this article: