Radar Stereo; Interferometry - Remote Sensing Application - Completely Remote Sensing, GPS, and GPS Tutorial
Radar Stereo; Interferometry

When flown on aircraft, a radar can point in different look directions and angles during successive passes to simulate the necessary parallax conditions for relief displacements across track. Side-looking Airborne Radar (SLAR) observations, looking out from the same side in successive passes, produce effective stereo imagery. An example showing the Appalachian Fold Belt in central Pennsylvania, using a SAR X-band system, demonstrates the quality of radar stereo.

SAR radar stereo pair image of the Appalachian Fold Belt in central Pennsylvania; when printed out, these need to be cut apart.

Stereo radar images from space can produce acceptable stereo models using multiple look angle (off-nadir) viewing. Vertical and horizontal resolutions are hard to achieve by optical systems from comparable orbits. We can significantly reduce the vertical exaggeration because of low B/H (base to height ratio) values. The stereo pair below, showing volcanic terrain in Chile, acquired with the SIR-B L-band SAR from two incidence angles (54 and 45), has moderate exaggeration of a high relief scene.

Stereo pair representing two incidence angles under which the L-band radar on SIR-B operated while passing over Chile.

Multiple incidence angle imagery serves well in preparing perspective views, as demonstrated in this SIR-C oblique rendition of a group of volcanic calderas on two of the Galapagos Islands.

Colorized SIR-C oblique perspective view of a group of volcanic calderas on two of the Galapagos Islands.

Synthetic Aperture Radar images, generated simultaneously from two antennas on the same platform, have such a small base that they do not, in themselves, function as a stereo pair. But being composed of coherent radiation, signals from each, matched to equivalent pixels in the other, show varying degrees of phase differences, indicative of partial interference. Variations in the signal travel times between two adjacent points induce the interference, and thus, show slightly different heights. When we remove Earth curvature effects and make other adjustments, residual differences become a measure of terrain elevations. SAR Interferometry is a developing technique that you can learn more about by perusing another University of Texas review dealing with this subject. A mathematically-based exposition of interferometric principles is embedded in this JPL site.

The next two paragraphs were extracted verbatim from a CCRS Radarsat web site:

"Interferometric SAR (InSAR) techniques make use of the coherence of the radar signal and the fact that the signal phase is is equal to twice the path length between the sensor and the earth's surface. The phase difference between measurements generated from two SAR images with the sensor separated by a baseline, allows measurement of the slant range difference to fractions of a radar wavelength. The slant range difference can be geometrically related to the terrain height."

"There are two basic SAR interferometry methods. In the first, two antennas are placed on the same platform and simultaneously acquire images of the scene from two different angles. The relative phase difference may then be used to construct a digital elevation model. The CCRS airborne SAR was modified to operate in this mode. In the second, a pair of images from the same sensor are taken at different times. This is now a well-proven concept and has been demonstrated with several spaceborne SAR sensors, now including RADARSAT. For this repeat-pass interferometry, the scenes are acquired at different times, so there is a time difference as well as viewing geometry to consider. The passes must have rather similar geometry in order to allow extraction of the relative phase difference. This usually requires that the satellite be on an exact repeat orbit. For RADARSAT, this means that a candidate interferometric image pair is available for a particular site every 24 days. The difference in geometry allows the extraction of topographic information, in the same way as with the single pass airborne system."

These four images, also taken from the CCRS website, show steps involved in producing an interferogram image of part of Bathurst Island in northern Canada. Read the captions for a brief identification of each step:

Initial interferogram, incorporating magnitude information. A coherence image.
A range and azimuth correction image The final interferogram, colored using the IHS system, in which I = magnitude, H = coherence, S = phase correction.

You can get some idea of these principles from this illustration showing how SIR-C interferometry produced topographic information for this Long Valley, CA test area.

A set of four images of the Long Valley area in the eastern Sierra Nevadas of California; read text for description of information in each panel.

The upper left panel is a Horizontal-Horizontal polarized L-band image, in which many bright-toned patches correspond to hilly rock and dark patches to smooth valley floors. Lake Crowley (dark) is at the lower left. The ridge is part of the Long Valley volcanic caldera (inactive). The upper right panel shows an interferogram derived by combining two L-band images taken in April and October, 1992, respectively. The colors relate to differences in signal phase, caused by elevation differences. In the lower left is a topographic map derived from the interferogram. Its total relief is 1,300 m (4264 ft). The lower right panel presents a 3-D perspective of this map, made by "draping" a C-band image over the L-band topographic data.

Here is another interferogram that shows the color band pattern around the Landers (California) fault line (superimposed). Each color cycle represents a rise in elevation of 2.8 cm.

Landers earthquake effects, shown as color bands indicating the upward movement of the surficial rocks on either side of the fault.

Radar interferometry using SAR data has been monitoring Yellowstone National Park. The hot region (responsible for the heated geysers) is associated with an underlying caldera - active in the last million years. The region is rising, as shown in the interferograms calculated for the six periods shown:

Interferograms for Yellowstone National Park

Radar Interferometry monitors other kinds of natural phenomena involving elevation changes and other aspects of motion. Here is an interferogram for a glacier in the Himalayas:

Changes in elevation accompany ice flow in a glacier.

As with so many other kinds of space images, the combination of a plan view image with elevation data - both acquired by SAR interferometry - permits construction of perspective views. This one shows the Karakax Valley in Nepal:

InSAR image of the Karakax Valley.

Source: http://rst.gsfc.nasa.gov