Terrain Data – which to use?

      Tuesday 20 March 2018 |  by


I often get asked the best terrain data/DEM to use and how to access it. Here’s a run-down with an explanation


Medium resolution (30m):


Easiest/global coverage:

NASA’s Shuttle Radar Topography Mission (SRTM) ‘Plus’/Version 3.0

Download here  (Quick NASA Earthdata registration needed)


Harder, more accurate/not quite global coverage (Expanded on below):

ALOS World 3D – 30m (AW3D30)

Browse here / Download here (Quick email registration needed. Google Chrome issues but works with Firefox)



Fine resolution (1m+) Digital Surface Modelsee below for DEM/DSM


UK (near-complete):

LiDAR DSM Composite

Browse here / Download here


USA (1m & 3.3m sparse, 10m near-complete):


Browse / Download here


DEM, DSM, DTM difference?

Without getting into pedantics:

  • Digital Terrain Models (DTM) are the height of the earth’s surface in relation to other points
  • Digital Elevation Models (DEM) are the height of the earth’s surface specifically in relation to a standard global/regional elevation (normally sea level). They are a subset of DTMs
  • Digital Surface Models (DSMs) are the height of the surface including objects (buildings, vegetation, etc). See below for clarification:

Medium resolution datasets explanation:


Section summary: AW3D30 is currently the most accurate DEM, though a new NASA DEM product is expected imminently (time of writing: Mar 2018).


There are currently 3 main open-source DEMs in use globally: AW3D30, SRTM and ASTER Global DEM Version 2 (GDEM2). These all have different versions, and it is important to briefly summarise their histories to explain their differences and future implications. All 3 datasets were initially charged for, but over time, their transition to being freely available has influenced their roles.

SRTM was collected over 11 days in 2000, though lacked global coverage (~80% coverage) and had “voids” (black spots where no direct line-of-sight existed). Version 2 (AKA “finished” product ~2006) “filled” these voids by interpolation and using other elevation data.

Though ASTER was producing scene-scale DEMs from 2003, the global GDEM V1 was only released in 2009 with a 99% land coverage (though ASTER data then had a charge associated). This had the same resolution as SRTM data (30m over US, 90m for security reasons over rest-of-world). GDEM V2 which incorporated more ASTER data and dealt with a number of issues was later released 2011 (at 30m resolution?). In 2012, SRTM data was made freely available.

ASTER data however was only later freely distributed 2016. As such, SRTM was updated to V3/SRTM ‘Plus’ in 2013, fundamentally improving via enhanced void-filling (using GDEMs amongst other datasets). However, in 2014, the US announced they would roll-out 30m (1 arc second) DEM data, with the increased 30m resolution being adopted in SRTM V3 from 2014-2015.

The ALOS World 3D (AW3D) product development began in 2014, with their 30m AW3D30 product released freely in 2016 (a similar time to ASTER data).

Being based upon their native 5 x 5m satellite product (the fee-based AW3D5) vs 30 x 30m of the alternatives, AW3D30 unsurprisingly appears to have the highest elevation accuracy. Santillana & Makinano-Santilla‘s direct comparison yielded the following vertical Root Mean Square Errors:

  • AW3D30  5.68m
  • SRTM V3  8.28 m
  • GDEM2  11.98m

As mentioned above, AW3D30 does not provide a global coverage. In addition, NASA are currently very near completion of a new global (free) DEM making use of SRTM, GDEM2 and multiple other DEM products.

Initially scheduled for 2017 NASADEM (see here for a dumbed down explanation) is currently in beta – provisional data can be downloaded by authorised users here.

Upon investigation, Kemen provides a good summary of the initial data, pointing to a number of issues which might account for the delay. At the time of writing (March 2018), no announcement has been made as to when the final product will be made available.


This article will be updated in the near future

Tags: , , , , , , ,