Digital Elevation Models Compared: SRTM, ALOS, Copernicus DEM, and When to Use Each

"Which DEM should I use?" is one of the most common questions in geospatial analysis, and the answer has changed significantly over the past five years. For nearly two decades, SRTM was the default global elevation dataset. Today, the Copernicus DEM has largely taken that position — but SRTM isn't obsolete, and the choice depends on what you're trying to do.

The Free Global DEMs

SRTM (Shuttle Radar Topography Mission)

• Source: Space Shuttle Endeavour, single 11-day mission in February 2000
• Technology: C-band InSAR (single-pass interferometry with two antennas)
• Resolution: 1 arc-second (~30m) globally; 3 arc-second (~90m) historically
• Coverage: 60°N to 56°S (no polar coverage)
• Vertical accuracy: ±9m (90% linear error) globally; ±6m in flat terrain
• Epoch: February 2000

Strengths: Well-characterized, extensively validated, consistent global coverage within its latitude range. The "known quantity" in the DEM world.

Weaknesses: 25-year-old data. Significant void areas (steep terrain, water bodies). Stripe artifacts in some regions. DSM, not DTM (includes vegetation and building heights).

NASADEM

• Source: Reprocessed SRTM data with improved algorithms
• Resolution: 1 arc-second (~30m)
• Coverage: Same as SRTM
• Improvements: Void filling using ASTER and ICESat data; reduced noise; improved water body handling
• Epoch: Still February 2000 (same raw data)

NASADEM is essentially SRTM done better. If you would have used SRTM, use NASADEM instead — it's strictly superior for the same epoch.

ALOS World 3D (AW3D30)

• Source: ALOS PRISM (optical stereo satellite)
• Technology: Optical stereophotogrammetry
• Resolution: 1 arc-second (~30m); commercial 5m version available
• Coverage: Near-global (polar gaps)
• Vertical accuracy: ±5m (RMSE) globally
• Epoch: Approximately 2006-2011

Strengths: Better accuracy than SRTM, especially in steep terrain. Optical-derived, so no radar-specific artifacts (layover, foreshortening).

Weaknesses: Cloud-affected areas may have lower quality. Some regions filled with SRTM data.

Copernicus DEM (GLO-30)

• Source: TanDEM-X mission (DLR/Airbus)
• Technology: X-band single-pass InSAR
• Resolution: 1 arc-second (~30m); commercial 10m and 30m versions
• Coverage: Near-global (including polar regions)
• Vertical accuracy: ±4m (RMSE) globally; ±2m in flat/moderate terrain
• Epoch: Approximately 2012-2015

Strengths: Best accuracy of any free global DEM. Most recent epoch. Near-global coverage including high latitudes. Fewer voids than SRTM. Edited to reduce artifacts.

Weaknesses: Relatively new, so less extensively validated in all environments than SRTM. Still a DSM (includes canopy/buildings).

DSM vs. DTM: The Canopy Problem

All four global DEMs are Digital Surface Models (DSMs) — they measure the elevation of the highest surface, which includes tree canopies, buildings, and other above-ground features.

In forests: The DSM measures the canopy top, not the ground below. A 30m tall forest on flat terrain will appear as a 30m hill in the DSM. For hydrology (flow routing, watershed delineation), this creates significant errors.

In cities: Buildings appear as elevated features, affecting slope and aspect calculations.

In open terrain: DSM ≈ DTM (the surface IS the ground). All four DEMs work well for bare or sparsely vegetated terrain.

Getting a DTM from a DSM

Options for removing canopy and buildings:

• ICESat-2 / GEDI LiDAR: Provide ground elevation samples that can be used to correct DSMs
• Canopy height subtraction: DSM − canopy height map = approximate DTM
• Morphological filtering: Algorithms that attempt to separate ground from above-ground features
• National DTMs: Many countries have LiDAR-derived DTMs at higher resolution and accuracy than global DSMs

Comparison by Application

Hydrology (Watershed, Flow Direction)

Best choice: Copernicus DEM (most accurate) + hydrological conditioning (pit filling, stream burning)

Caveat: All global DSMs produce significant errors in forested areas. If your watershed is forested, consider using a national LiDAR-derived DTM if available.

Orthorectification (Terrain Correction)

Best choice: Copernicus DEM (most accurate, most recent)

For SAR terrain correction (radiometric terrain correction of Sentinel-1), the DEM accuracy directly affects the quality of the result. The Copernicus DEM's superior accuracy produces the best orthorectification.

Change Detection (Before/After)

Best choice: SRTM for the "before" epoch (2000); Copernicus DEM for the "after" epoch (~2014)

Differencing DEMs from different epochs reveals elevation changes — useful for glacier volume change, mining volume estimation, and landslide detection. Note that systematic differences between DEM sources can introduce artifacts.

Slope and Aspect Analysis

Best choice: Copernicus DEM (best accuracy = most reliable slope/aspect)

Slope errors are amplified by DEM errors — a 9m vertical error (SRTM) over a 30m pixel produces up to ~17° slope error. The Copernicus DEM's 4m accuracy halves this error.

Visualization (3D Terrain)

Any DEM works for visualization. SRTM is common because it's widely supported and familiar. Copernicus DEM provides slightly better visual detail.

Historical Analysis

SRTM/NASADEM: The only option for ~2000 elevation baseline. Useful for glaciology, coastal change, and other applications requiring historical reference.

Practical Tips

Check the void mask: All DEMs have data gaps (voids). Always check whether your area of interest has complete coverage. Void-filled versions exist but filled areas have lower accuracy.

Understand the CRS: Global DEMs are typically in geographic coordinates (WGS84) with elevations referenced to EGM96 or EGM2008 geoid. If your project uses ellipsoidal heights, conversion is needed.

Resolution isn't everything: A 30m DEM with ±4m vertical accuracy (Copernicus) is more useful for most applications than a 30m DEM with ±9m accuracy (SRTM), even though the spatial resolution is identical.

National DEMs may be better: Many countries have LiDAR-derived DEMs at 1-5m resolution with sub-meter vertical accuracy. These are far superior to any global DEM for local applications. Check your country's national mapping agency before defaulting to a global product.

The Copernicus DEM should be the default choice for most new projects. It's the most accurate, most recent, has the best coverage, and is freely available. SRTM retains value for historical reference and for established workflows. The key is understanding what each DEM actually measures (surface, not ground), its accuracy limitations, and how those limitations affect your specific application.