As the quality and availability of remotely sensed elevation models increase, minimizing the trade-off between homogeneous coverage and accuracy level matters.
By Vanessa Casals and Gertrude Riegler, Airbus Defence and Space (www.geo-airbusds.com), Toulouse, France.
Whether planning an infrastructure network, conducting a military intervention mission in unknown terrain or analyzing terrain features for oil and gas exploration, precise, reliable elevation data support a wide range of applications and form the foundation of any accurate geospatial product.
A digital elevation model (DEM) is a generic term that describes two distinct topographic models: digital surface models (DSMs) and digital terrain models (DTMs). DSMs include the heights of natural or man-made features, such as vegetation, buildings and roads, whereas DTMs represent bare-earth elevation, with all vegetation and man-made objects removed.
Today˜s new high-resolution Earth observation (EO) satellites”an ever-growing number, with data commercially available from providers around the globe”directly affect elevation models in two ways. The availability of high-quality input data allows providers to quickly create increasingly precise elevation models. At the same time, the demand for homogeneous large-area (or even global) elevation models increases, as they're used as a base layer for orthorectification or data-fusion applications.
Data collection for creating more precise DEMs is becoming increasingly easy. Several satellites even feature specific capabilities for stereo and tri-stereo collection, with some operated in a formation for simultaneous data acquisition.
According to researchers at the University of Porto in Portugal, satellite-based DSMs provide an evident advantage over aerial images in terms of the cost and time and other complications of the image collection. The same assessment, available online at http://bit.ly/1qKgTJ7, finds the level of detail of DSMs automatically extracted from PlÃ©iades tri-stereo satellite imagery is very close to that of aerial images.
Such research confirms that elevation models extracted from new-generation satellite sensors offer quality that's comparable to models created from aerial data. Moreover, considering a satellite like PlÃ©iades can collect stereo segments 280 kilometers long (or 120-kilometer tri-stereo segments) and even perform stereo/tri-stereo collection over contiguous segments to widen the field of view, it's obvious how the availability of such input data will change DEM generation in the coming years. When it comes to performing a quick 3-D coverage of a major metropolitan area, regardless of location, all necessary input data can be collected in a single day.
Determining Heterogeneous User Requirements
From large-area mapping of remote or inaccessible terrain to the detailed preparation of construction sites, elevation models are used for a wide range of applications. Today's providers are trying to be as diverse as their customers' requirements. For example, Airbus Defence and Space, which operates a comprehensive constellation of Earth observation satellites, offers its GEO Elevation product suite”with a range from 30 meters down to 1 meter raster, with three intermediate levels at 12 meters, 8 meters and 4 meters”as DSMs and DTMs with a variety of editing options.
Depending on the size of an area of interest and a project's purpose, users can select an applicable elevation model. However, a trade-off needs to be considered between resolution and coverage (or project duration if a new DEM is generated), especially when a large area of interest is involved.
Because today's sensors combine coverage and accuracy, individual DEM creation can be an option for medium- or even large-area coverage. For example, the agile high-resolution twin satellites SPOT 6 and SPOT 7 can offer stereo and tri-stereo imagery acquisitions of entire countries in a few days.
Exploring New Global Solutions
The availability of high-quality, homogeneous large-area elevation data is important when it comes to precisely orthorectifying the growing amount of imagery acquired by the new generation of EO satellites. The accuracy of the base DEM is vital to successful performance, particularly when the integration of multisource imagery and data is based on it. For many applications, fused data provide improved reliability, increased confidence and reduced ambiguity.
Most users will resort to the best-available elevation model for their purpose, particularly when it comes to large-area or cross-border coverages or disconnected areas of interest”or, in the worst case, a combination of all of these. Homogeneity becomes a key factor, and whenever several models are available, the aforementioned trade-off between coverage and precision is a major decision factor.
Since the turn of the century, the Shuttle Radar Topography Mission (SRTM) has been the large-area DEM standard for most applications. In February 2000, SRTM collected interferometric synthetic aperture radar data over land between 60° north and 56° south latitudes on a mission co-sponsored by NASA and the National Geospatial-Intelligence Agency. The SRTM DEM offers 90-meter resolution at the equator and is provided in mosaicked 5° x 5° tiles in GeoTiff format for easy download and use.
The well-known ASTER Global Digital Elevation Map (GDEM), developed by U.S. and Japanese partners, covers land surfaces between 83° north and 83° south and is composed of 22,600 1° x 1° tiles. The ASTER GDEM is in GeoTiff format and features 30-meter grid spacing. Estimated accuracies are 20 meters at 95 percent confidence for vertical data and 30 meters at 95 percent confidence for horizontal data.
Elevation30 is another leading geographic reference database designed for large-area coverage. Initially developed as Reference3D by the French survey and mapping agency IGN and Spot Image, now part of Airbus Defence and Space, Elevation30 offers 80 million square kilometers worldwide available off the shelf. A DEM in DTED level 2 format, delivered with the corresponding orthoimage (ground control source) and a set of quality masks, has a vertical accuracy down to 8 meters.
Most recently, the German Aerospace Center (DLR) and industry partner Airbus Defence and Space announced the availability of their globally homogeneous WorldDEM elevation model. Providing elevation information with 2-meter vertical accuracy at 12-meter grid spacing, the product will become available for Earth's entire 150-million-square-kilometer landmass by mid-2015. Vast areas worldwide are available today (see www.geo-airbusds.com/worlddem-db).
Widespread Use Increasing
High accuracy and homogeneous coverage is set to boost effectiveness for military applications such as fighter aircraft, helicopters and unmanned aircraft systems (UASs). Such coverage also will deliver more precise applications in various sectors, from commercial aviation to oil, gas and mining.
For example, oil and gas industry experts appreciate accurate, global coverage for a range of applications that can make a significant contribution to upstream oil and gas projects throughout a project's workflow. For follow-up studies, where higher resolutions and accuracies are required, bespoke models at 1-meter grid spacing can be generated using high-resolution stereo satellites, particularly with the use of ground control points.
Globally available datasets also support international cooperation and cross-border mission planning no matter where an area of activity is located. During emergencies, such as natural disasters, the availability of a standardized, highly accurate DEM becomes a major advantage to provide reliable information to rescue teams on the ground.
Flight safety is another area where such datasets play a major role by providing more precise altitude data for flight path and landing area planning, especially over remote regions. Such databases are becoming a standard on-board element for aircraft and UASs. Cartographic authorities around the world also rely on precise elevation data to improve and update their standard cartographic maps or perform geological structure studies based on large-area elevation models.