Summary

1. Background
The production of digital elevation models (DEMs) is an expanding aspect of photogrammetry and a new and rapidly developing technology, radar interferometry. The growing availability of DEM data is opening new possibilities to earth scientists whether they be geologists, geomorphologists or pedologists. The land surface has long been recognized as the basis to which all geoscientific observations are tied. Only recently, however, have the technologies been developed to produce topographic data of large areas with sufficient horizontal and vertical resolution to meet the requirements of detailed spatial analysis. Earth scientists in the past have often been constrained by using a mosaic of existing DEMs that were derived by different resolutions and datum. The results were mostly inhomogeneous, inconsistent and, consequently, not comparable. Improved aerial photogrammetry (stereoscopy) through state-of-the-art Soft Photogrammetric Digital Workstations, and interferometric synthetic aperture radar, has meant that topography has gained a whole new meaning in spatial studies. Large-area DEMs with consistent quality and high precision can now be produced. NASA's recent Shuttle Radar Topography Mapping Mission (SRTM) ensured that high-resolution topographic data for the majority of the earth's surface will be available to all users within two years (http://www.jpl.nasa.gov/srtm/). In the Pacific Rim, TOPographic Synthetic Aperture Radar (TOPSAR) datasets were acquired over selected sites during the PacRim1 (1996) and PacRim2 (2000) AIRSAR missions for the purposes of advancing the understanding of radar technology, in particular that of radar polarimetry and radar interferometry. Details describing sites covered during these missions can be found on the JPL AIRSAR web page (http://airsar.jpl.nasa.gov/).

With the advent of geographical information systems (GIS) DEMs can be used together with other spatial datasets such as geological information, airborne magnetics, gamma-ray spectroscopy and hyperspectral datasets. The DEM provides a basic spatial reference system and images and vector data can automatically be draped over the DEM for more advanced analyses. Software packages such as ESRI's Spatial Analyst and 3D Analyst (http://www.esri.com/software/arcview/extensions/spatext.html), and IDL's RiverTools (http://www.rsinc.com/rivertools/index.cfm) includes tools for generating geomorphometric models such as slope, and modelling of drainage basins.

Freeware packages on the Internet such as:
LandSerf (http://www.geog.le.ac.uk/jwo/research/LandSerf/index.html),
MicroDem (http://www.nadn.navy.mil/Users/oceano/pguth/website/microdem.htm),
GRASS http://www.baylor.edu/~grass/index2.html and
DiGem (http://member.aol.com/oconrad/dgm/dgm_main.htm)
can also be downloaded for visualizing and performing analysis of generic elevation datasets.

2. Current Capabilities

2.1 Stereo Photogrammetry

Aerial photography
Precision DEMs are now being acquired by commercial groups using soft photogrammetric digital workstations, such as the HELAVA system. Stereo aerial photographs used for generating the DEM are either purchased from an archive or reflown if the existing photographs are unsuitable. In addition to the DEM, digital imagery of the aerial photos can be projected or draped over the terrain data to form an orthoimage. For example, using 1:86 000 scale photography scanned at 15mm, the resultant product has a ground resolution of 1.2 m and height accuracy of 3-5 m, depending on accurate ground calibration and quality of the stereo model.

SPOT-PAN
DEMs can be produced from SPOT-PAN image stereopairs through stereoplotting and automatic correlation. The cross-track pointing capability of SPOT allows the second image of the stereo pair to be acquired as little as 2 days after the first with base-to-height ratios of 0 to 1. McMullen Nolan and Partners have shown that when using high contrast imagery with minimal time difference between the images forming the stereo pair, an optimal base:height ratio of 0.7 and suitable ground survey information, SPOT DEMs of low relief terrain typically achieve 7-12 m RMSE horizontal and vertical errors.

RADARSAT
Stereo pairs for DEM generation are created using one of RADARSAT's Fine, Standard, Wide or ScanSAR Narrow beam modes. The finest resolution of 8x8 m horizontal and 15-20 m vertical for 50 m postings is provided in the Fine beam mode but it covers the smallest area (50 km width). These products are suitable for a variety of applications at map scales smaller than 1:100,000. Further details regarding stereo accuracy and costs for each mode can be viewed at http://www.vexcel.com/radar/stereo.dem.html

ASTER
The Advanced Spaceborne Thermal Emission and Reflection Radiometer mounted on the TERRA spacecraft launched in December 1999 obtains high-resolution (15 to 90 square metres per pixel) images of the Earth in 14 different wavelengths of the electromagnetic spectrum, ranging from visible to thermal infrared light (http://asterweb.jpl.nasa.gov/).

Unlike the other instruments aboard TERRA, ASTER will not collect data continuously; rather, it will collect an average of 8 minutes of data per orbit. All three ASTER telescopes (VNIR, SWIR and TIR) are pointable in the cross-track direction. Given its high resolution and its ability to change viewing angles, ASTER will produce stereoscopic images and detailed terrain height information. Stereo data have a spatial resolution of 15 m and the base:height ratio, based on the nadir and back view (at 27.6 degrees off nadir, along track), is 0.6. Standard product DEMs will come in two basic modes, relative and absolute. Both are produced using automated, correlation-based, commercial softcopy photogrammetric software. Both modes will have a X-Y resolution of 30 m and a vertical resolution (smallest Z increment measured) of 1 m. Absolute DEMs require user-supplied ground control points (GCPs). Based on pre-launch studies, with 4 or more GCPs/scene with XYZ accuracies of 5-15 m, RMSE xyz accuracies of 7-30 m for the absolute DEMs are predicted.

Standard product DEMs from ASTER are produced at no charge, but production plans are limited to only one 60x60 km scene per day. The idea is that users with their own soft copy photogrammetric software will be able to produce their own DEMs when the stereo data become available. The data are of sufficient quality to allow for 15 m posting by users with their own software, and will be available at no charge from early July 2000. The overall mission plan is to acquire and archive a global data set between 82 degrees North and South of cloud free stereo over a period of 2 and 5 years. The nominal maximum number of stereo pairs that can be acquired/day is 771, but this maximum will probably not be realized; and no more than about 350 scenes/day will be processed.

2.2 Radar interferometry
Radar interferometry is an innovative technique that enables very high-resolution topographic maps of the earth's surface to be generated using spaceborne and airborne radar instruments. The technique has the advantages of automatic processing of the data, and operation in cloud, smoke and rainfall conditions, night and day. A transmit antenna mounted on a spacecraft or plane illuminates the terrain with a radar beam that is scattered by the surface. This radar echo has two components: amplitude (brightness) and phase (distance to the target). Two receive antennas with a fixed baseline record the radar echo from slightly different positions resulting in two different radar images. The two signals received at both ends of the baseline (referred to as the interferometric baseline) show a phase shift due to differing lengths of the signal paths. The phase difference, determined by effectively subtracting the measured phase at each end of the baseline, is sensitive to both viewing geometry and the height of the terrain. If the viewing geometry is known to sufficient accuracy, then the topography can be inferred from the phase measurement to a precision of several metres. The accompanying amplitude information is measured to construct a radar image showing details describing the surface's roughness and dielectric properties.

Interferometric data can be generated from either single-pass or repeat-pass systems. Single-pass systems such as the TOPSAR and SRTM instruments use the two-antenna system, as described above, to record both images simultaneously. Repeat-pass interferograms are generated from separate passes over the same target such as from the European ERS-1 and ERS-2 systems http://www.eurimage.com/.

TOPSAR (TOPographic Synthetic Aperture Radar)
The JPL TOPSAR system is a 5cm wavelength, C-band interferometer operating on NASA's DC-8 research aircraft as an adjunct to the polarimetric Airborne Synthetic Aperture Radar (AIRSAR) system. TOPSAR is implemented via two antennas flush-mounted nearly vertically on the left side of the aircraft with a 2.6 m baseline spacing. The boresights of the antennas are depressed 45° with respect to the horizontal. The lower antenna is used for transmission whereas both antennas are used for reception. The height accuracy of TOPSAR digital elevation models has been shown by Madsen et al (1995) to be 1 m RMSE in flat terrain and 3 m in mountain areas with a 2 m RMSE overall. Typical data acquisitions are for areas of 10 km across-track (range direction) and 60 km along track (azimuth direction). The output of high precision TOPSAR datasets is accomplished by comprehensive navigational systems to determine the precise position of the DC-8 aircraft, and full motion compensation algorithms in the TOPSAR processor to accommodate large translational and altitude perturbations during the data recording process (Madsen et al, 1995). Apart from the digital elevation data, L- and P-band polarimetric and Cvv data are also recorded. All datasets are registered and the radar data ortho-corrected using the DEM.

SRTM (Shuttle Radar Topography Mapping Mission)
The SRTM recorded both radar images simultaneously using two antennas - a transmitting and receiving antenna in the cargo bay of the Shuttle Endeavor, and a second receiving antenna at the tip of a 60 m deployable mast. Both radar and phase data were recorded in C-band (5.56 cm) and X band (3.1 cm) frequencies. During the 11-day mission, data were acquired along 225 km wide swaths imaging all of Earth's land surface between 60° north and 56° south latitude, with data points spaced every 1 arcsecond of latitude and longitude (approximately 30 m). X-band coverage occurred along narrow 50 km wide swaths and cover 40% of the area mapped by the C-band data. The absolute horizontal and vertical accuracy of the C-band data will be 30 m and 16 m, respectively. Relative height accuracy will be 10 m. However, data of these specifications will not be readily available for public use. Within 2 years, data spatially degraded to 90x90 m horizontal resolution but retaining the initial height accuracy, will be available at the low cost of regridding the data to a 1°x1° area. The policy for distributing the higher resolution data is currently not clear. However, it is expected that data over politically insensitive areas will be accessible. X-band DEM data of the narrower 50 km wide swath have a horizontal resolution of 30 m and relative and absolute height accuracies of 6 m and 16 m, respectively. These data will be unclassified and available for public use from the German Aerospace Centre. Both C and X-band datasets will be geometrically corrected and projected to the WGS84 datum. Extensive information describing the data products and their availability is available on the SRTM Home Page (http://www.jpl.nasa.gov/srtm/) and DLR Home page (http://www.jpl.nasa.gov/dlr-mirror/srtm/).

Selected Reading

• Madsen, S. N., Martin, J.M. and Zebker, H.A. 1993. Analysis and Evaluation of the NASA/JPL Across-Track Interferometric SAR System. IEEE Trans. Geosci. Remote Sensing, vol. 33, no 2, pp. 383-391.
• Thompson, T.W., Zebker, H.A., Carande, R.E., Rosen, P.A., Madsen, S.N., Hensley, S., Rodriguez, E., van Zyl, J.J., Martin, J.M. and Miller, T.W. 1994. NASA/JPL TOPSAR Interferometric SAR. NASA/JPL Imaging Radar Homepage, http://southport.jpl.nasa.gov/topsardesc.html
• Van Zyl, J.J., Zebker, H.A., Hensley, S., Haub, D. and Wiesbeck, W., 1995. The new dual frequency (C- and L-Band) TOPSAR airborne interferometric SAR.
  http://www.jpl.nasa.gov/techreport/1995/95-0530.rfr.html
• Burgman, R., Rosen, P.A. and Fielding, E.J., 2000. Synthetic aperture radar interferometry to measure Earth's surface topography and its deformation. Annual Review Earth Planet Science 2000, 28:169-209.