SRTM: NASA Shuttle Radar Topography Mission

The wonderful SRTM project has the potential to provide global elevation data at up to 30m (1 arc-second) resolution, worldwide!  For a description of the project, see the NASA/JPL SRTM site.

Acquiring the data

• 90m (3 arc-second) elevation for North and South America can be browsed and downloaded from the National Map Seamless Data Distribution System - just click "International Viewer"
  • instructions in case you need them: Quick Steps for Obtaining Seamless Data
  • or directly from their FTP site, files are named according to latitude and longitude, although this server is chronically overloaded and often unresponsive
  • if you use the FTP site, the rest of the world is also online: Africa, Eurasia, Australia, Islands
• The GLCF ESDI is also a convenient way to browse and download SRTM data, in GeoTIFF format - for more information, see the GLCF SRTM page.  They also offer direct FTP access.
• A very useful server is CGIAR-CSI: SRTM 90m Digital Elevation Data.  It also happens to offer their "version 4" SRTM which has the gaps filled.
  • Direct FTP link to the SRTM v4 data: ftp://srtm.csi.cgiar.org/srtm_v4/
• Higher-resolution 30m (1 arc-second) SRTM exists, but since it is only released for the USA, it is of limited use, generally inferior to existing USGS 30m DEMs in many ways.

Global 1km data: SRTM30

• SRTM data was used to update the older USGS GTOPO30 global DEM, by averaging to 30 arc-sec resolution and replacing GTOPO30 heixels between the latitudes of 60° North and 56° South.
• The resulting data can be directly downloaded and is now (as of 2004) the best available global 1km elevation dataset!
• See the before-and-after improvement in the images to the right - SRTM30 replaces the wild guesswork of previous data with actual measured values

History

• "the Memorandum of Understanding between NASA and NIMA for SRTM specifies that data processed at 3 arc-seconds (~100m) for anywhere on the globe will be unrestricted, as will 1 arc-second (~30m) data for the United States and its territories.  Distribution of one arc-second data for outside the United States will be approved by NIMA on a case-by-case basis for NASA Investigators, for their use only."
• SRTM data came out first for the USA, but that data is not much used, since regular surveyed DEMs of higher resolution and quality already exist for the USA
• International data was released beginning in June 2003
• In late 2005, SRTM version 2 was released.  SRTM V2 is "the result of a substantial editing effort by the NGA and exhibits well-defined water bodies and coastlines and the absence of spikes and wells (single pixel errors), although some areas of missing data ('voids') are still present."
• Subsequent improvements have been made by groups such as CGIAR (see 'Improved SRTM' below.)

Drawbacks

• there were many data quality issues with SRTM V1 data
• overall the issues are too complicated to describe here, but to summarize:
  • there is a vertical error of +/- several meters, including water areas, so that for example oceans are not flat, nor are they "sea-level"
  • there are a very large number of holes or gaps in the data, especially in mountainous areas
• SRTM V2 fixes the water area noise, but still has the gaps

Improved SRTM

• CGIAR-CSI: SRTM site offers cleaned up, interpolated SRTM data!
  • Their "version 2" was based on SRTM V1.  It has water areas clipped with shoreline vectors, and voids filled with interpolation.
  • Their "version 3" (as of September 2006) is based on SRTM V2:  "Dr. Andrew Jarvis and Edward Guevara of the CIAT Land Use project, Dr. Hannes Isaak Reuter (JRC-IES-LMNH) and Dr. Andy Nelson (JRC-IES-GEM) have further processed the original DEMs to fill in the no-data voids. This involved the production of vector contours, and the re-interpolation of these derived contours back into a raster DEM. These interpolated DEM values were then used to fill in the original no-data holes within the SRTM data. Three additional points have been added over processing in SRTM V2 available from CIAT: (i) the support for auxiliary information, (ii) the use of a void region specific processing over a tile based processing, and use of SWDB V2 water body database."
  • Their "version 4" (as of September 2008) says: "This latest version represents a significant improvement from previous versions, using new interpolation algorithms and better auxiliary dems. We are confident this is now the highest quality SRTM dataset available."
• Jonathan de Ferranti's Viewfinder Panoramas site hosts a page about elevation datasets for mountain peaks around the world.  This page includes SRTM-derived DEMs "in which the no-data void areas have been filled from the best available alternative sources. They are much more accurate than those created by interpolation."
  • The included mountain ranges are: the Himalayas (including Karakoram, Hindu Kush, Pamir and Tien Shan), Andes, Tatra, Caucasus, New Zealand, Japan, Corsica
  • His site also contains detailed interesting information about the factual and political nature of mountain peak altitude values, which are often very inaccurate with many sources giving conflicting information.

Tools

• If for some reason you don't want to use the CGIAR SRTM version 3, there are several software tools for patching the gaps in SRTM (V1 and V2) files, all of which read .hgt file format in the standard 1-arcsecond or 3-arcsecond tiles that come directly from the SRTM FTP site
• SRTMFill Utility (free, for Windows)
  • fast and easy program for patching NULL-data holes in SRTM DEMs by progressively in-filling from surrounding data, "quickly making usable data from unusable DEMs"
• DG SRTM Void Killer (for Windows, $43)
  • works by interpolating the gaps in the SRTM, then combining the result with GTOPO30 data
  • operates on a single tile at a time
  • can write back to the original SRTM .hgt file, or to its own DRC format
• BlackArt (free) is generally a tool for interpolating contour lines, but it also can also patch gaps in SRTM-1 and SRTM-3 data
• VTBuilder (free, cross-platform) can also load any number of SRTM .hgt files, merge and resample them, and fill the gaps with interpolation or lower-resolution data