GISdevelopment.net --> Application --> Miscellaneous

Applications of SAR interferometry: Limits options and perspectives

Hiroshi Kimura
Gifu University, Yanagido, Gifu-shi, 501-1193, Japan


Synthetic Aperture Radar (SAR) interferometry is a promising tool for mapping topography and observing displacement fields caused by Earthquake, volcanic activity, ice movement and so forth. Especially the latter is referred as a differential interferometry, and is used to develop a model of a relevant phenomenon. Recently, this technique is being extended to monitor surface change and vegetation structure. However, in case of a repeat pass interferometry using a single antenna SAR such as JERS, ERS and RADARSAT, there are some error sources to limit the accuracy. They are as follows:
  1. Orbit data or baseline
  2. Temporal decorrelation or elapsed time
  3. Geometric distortions (Layover and shadow)
  4. Atmospheric effects
Accurate orbit data is essential to SAR interferometry, because the baseline is critical parameter for a quantitative measurement but changes with every interferogram. The error of the baseline propagates to the error of topographic height and surface movement measurement. In case of JERS SAR interferometry with less accurate orbit data, ground control points (GCPs) and/or digital elevation model (DEM) are used to optimize the baseline. Temporal decorrelation of a land surface that causes fluctuation of interferometirc phase is dependent on elapsed time and radar wavelength. In general, the shorter the elapsed time is and the longer the wavelength is, the higher the temporal correlation is. With regard to the application to a vegetated area, it is demonstrated that L-band (JERS) is much superior to C-band (ERS and RADARSAT). In addition, in case of the differential interferometry, the elapsed time should be chosen properly to measure the surface movement, because the differential interferometry is sensitive to the movement in the slant range direction only. The shorter wavelength presents more fringes over the displacement area. However, too many fringes tend to make an interpretation difficult.

Geometric distortions such as layover and shadow are inherent in SAR images. Layover and shadow have an opposite relation to an incidence angle. If the incidence angle increases, chance of layover decreases, while that of shadow increases. In either case, phase unwrapping becomes difficult. To overcome this, combining interferograms from different viewing geometries like incidence angle and ascending/descending orbits is useful. It is reported that the optimal incidence angle over alpine terrain is 45 degrees from analysis of shadow and layover proportion derived from DEM. It is known that artifacts in interferograms are often linked to heterogeneity in the troposphere and ionosphere. In case of differential interferometry, they tend to confuse with fringes attributed to real surface displacements. At present, differential delay due to water vapor in the atmosphere is thought to be dominant, because some examples of improvement after removal of the water vapor effect are shown. When there are no data on the detailed distribution of water vapor, averaging multiple interferograms reduce the effect of water vapor. This works to detect transient movement, but does not well to detect progressive and unstable movement, because averaging weakens sensitivity to movement in a specific period. The multi-pass method has the effect of averaging.

In 2003, the Advanced Land Observation Satellite (ALOS) is planed for launch. The Phased Array type L-band SAR (PALSAR) and two high-resolution optical sensors will be on board the ALOS. The PALSAR has improved capabilities of variable incidence angle and full polarizations. National Space Development Agency of Japan (NASDA) plants to release the second Research Announcement after the launch. In this paper, examples of SAR interferometry presenting advantages and limits, and perspectives on use of the PALSAR data are shown.
© GISdevelopment.net. All rights reserved.