Digital representations of the surface of the earth are a key data set for many GIS projects, but locating, identifying, downloading and manipulating digital elevation data is not for the faint of heart. There are many different skills required and hundreds of tools, systems and instruments from which to choose. In this article, author Diana Sinton highlights available resources and need-to-know information.
Introduction to the digital elevation model
The most common form of digital representation of the surface of the earth is presented as values of elevation above sea level, often derived from sampled point measurements and represented in raster formats as a digital terrain model or digital elevation model (DEM), or as a vector triangulated irregular network (TIN). Apart from generating a topographical surface itself, these data are also the basis for deriving slope gradient, slope aspect and hillshade relief. Digital elevation data are central to transportation planning, land use planning, and geological and hydrological analyses, among countless others. For this article, we’ll focus on DEMs as a generic format of elevation data in digital form.
For many years, the most common source and scale for a DEM were the 10-meter and 30-meter resolution data organized and distributed by the US Geological Survey to align with their 7 ½ minute topographic quad sheets. These original DEMs were derived from traditional photogrammetric methods or reverse-engineered from contour lines. Errors and inaccuracies abound. Nine times out of ten, one’s area of interest was situated at the intersection of four quad sheets, so there was great rejoicing when it became possible to download “seamless” elevation data, foregoing the need to edge-match or mosaic multiple data sets together.
Measuring the horizontal resolution of elevation data often refers to spherical units of arc seconds, or 1/3600 of a degree. One arc second represents approximately a 30-meter grid cell. Accordingly, a one-third arc second of measurement is approximately ten meters in distance, and a one-ninth arc second is three meters. However, these measurements hold true at the equator, when both latitudes and longitudes are evenly spaced. Once distances are measured towards the poles, longitude measurements begin to converge and regular grid spacing becomes distorted. By the time one is measuring in arc seconds at 49 degrees latitude, an arc second of longitude has shrunk to 20.25 meters and grid cells have become elongated in shape.
Becoming familiar with the arc second system of horizontal measurements is a worthwhile investment of time when navigating elevation data sites, but it may be even more important to understand the absolute and relative vertical errors within DEM data. The original production goal of the 7 ½ minute USGS quads included a vertical accuracy standard of 7 meters, and up to 15 m variability was permitted (USGS Data Users Guide, pdf).
DEM meets Big Data in the US
Fast forward to 2015 and digital elevation information has intersected with the Big Data movement. In the United States, the National Elevation Dataset (NED) has replaced the former system of quad-based DEMs. Significant efforts have been made to ensure that the horizontal and vertical datums, elevation units and projections or coordinate systems have been made consistent or, where needed, optimized for that locale. Root mean square errors for vertical accuracy have fallen to less than 2 meters within much of the NED collection. Light Detecting and Ranging, aka LIDAR, data, and interferometric synthetic aperture radar, aka IfSAR, have become the standard approaches for high resolution data collection, and this has allowed for improvements and upgrades throughout the United States. Unlike the bare-earth presumption of DEM data, these new sources also provide detailed data for what is on the surface of the earth, for example the heights of vegetation and structures. The use of new technologies has been particularly important in states such as Alaska, where conditions had never previously permitted consistent and high quality data to be collected.
Of course there are times when it is both desirable and necessary to access older data, particularly when needing to make comparisons between before-and-after geomorphic changes following earthquakes and volcanic eruptions. For such purposes, the USGS also maintains a collection of historic DEMs.
Global data resources
When elevation data outside of the U.S. is needed, two important sources include data derived originally from NASA’s Shuttle Radar Topography Mission, as well as the more Advanced Spaceborne Thermal Emission and Reflection Radiometer global digital elevation model, now at Version 2. Since its original collection in the year 2000, the SRTM data has been corrected and revised, and its 90-meter resolution coverage is some of the most comprehensive world-wide. ASTER's Global DEM data has also undergone revisions and corrections, and its one arc second, 30-meter, resolution extends to even broader global coverage.
New satellite technologies and demand for higher resolution and more consistent data are driving the growth in digital elevation data advancement today. In 2010, DLR, Germany’s national aeronautics and space research center, launched the TanDEM-x satellite to partner with the already-orbiting European TerraSAR-X and is now producing data designed to be high resolution, with great vertical accuracy, and as consistent and reliable as possible in their coverage.
In the U.S., the current 3D Elevation Program has brought together multiple funding entities to produce and distribute nation-wide LIDAR data coverage, with IfSAR-based data in Alaska. Acquiring and processing these data will take years, but there is wide agreement that it is a wise investment with extensive benefits for the public and private sectors alike. The specter of sea level change has also compelled NOAA to prioritize LIDAR-based topographic data for coastal regions.
Locating, identifying, downloading and manipulating digital elevation data is not for the faint of heart. New interfaces for data discovery such as Reverb|ECHO come complete with 317 platforms, 658 instruments and 717 sensors from which to choose. Even the simpler National Map and Earth Explorer assume that users are familiar with the optimal spacing of LIDAR point clouds, arc second measurements, and the deciphering of acronyms. OpenTopography is specifically designed to lower the access barriers to high resolution data, but to date the availability is limited.
My advice? Give yourself plenty of time to sort out what’s available for your area of interest and what you really need for your project or application. Being able to find exactly the data you seek, download it, figure out and manipulate its compression format, modify its projection or coordinate system and successfully add it to your project is likely to require persistence, patience and the knowledge of a rocket scientist. Or two.