IEEE Geoscience and Remote Sensing Magazine - March 2013 - 19

classifier able to refine up to 16 classes [15]. The result
is shown on the bottom of Fig. 9. More recently, polarimetric decomposition techniques have been successfully
used to separate and remove the disturbing vegetation
contribution and allow estimation of the soil moisture
content on the isolated surface components. Fig. 10
shows the soil moisture maps obtained from polarimetric L-band data acquired at three different dates in the
frame of the AGRISAR experiment in 2006 [51]. At the
time of the first acquisition in April, the crop layer was
still short and light. Its height and density increased during the next acquisitions performed in June and July. The
underlying soil moisture estimation accuracy was on the
order of 4 vol% and 11 vol% RMSE indicating the potential of the decomposition approach.
IV. Interferometry
SAR interferometry is a powerful and well-established
remote sensing technique that enables the highly accurate
measurement of important geophysical parameters such
as surface topography, ground deformation and subsidence as well as glacier movements [13], [52]-[54]. The key
idea of SAR interferometry is to compare for a given scene
the phase of two or more complex radar images that have
been acquired from slightly different positions or at different times. Since the phase of each SAR image pixel contains
range information that is accurate to a small fraction of the
radar wavelength, it is possible to detect and measure tiny
path length differences with centimetric or even millimetric accuracy. This outstanding accuracy is independent of
the distance between the sensor and the scene which makes
SAR interferometry highly relevant for both air- and spaceborne remote sensing. Over the last decades numerous terrestrical applications have been demonstrated using either
airplanes [18], [55]-[59] or satellites [16], [60]-[70]. A drawback and fundamental challenge of SAR interferometry is,
however, that the measured range difference is ambiguous
with the wavelength. This ambiguity is typically resolved
by using some external information together with appropriate regularity assumptions about the imaged scene, a
process known as phase unwrapping [71].
The radar images for SAR interferometry are typically
acquired either from mutually displaced flight tracks or
from one and the same flight track but at different times.
The former is known as across-track interferometry and
enables, besides several other applications, a precise measurement of the surface topography. The latter is known
as along-track or differential interferometry. By varying
the temporal baseline between the interferometric acquisitions, velocities ranging from several meters per second
down to a few millimeters per year can accurately be measured. Important applications covering the whole range of
potential time scales are the detection of moving objects
like cars or ships [70], [72], the observation of ocean surface
currents [57], [73], the measurement of sea ice drift and glacier flow [62], [63], [67], the study of seismic deformations
march 2013

ieee Geoscience and remote sensing magazine

(a)

(b)

Surface

Specular Volume Double Bounce

FIGURE 9. (a) Freeman and Durden decomposition applied on

quad-polarimetric L-band data acquired over the Oberpfaffenhofen
test site. Red, green and blue correspond to fd , fv and fs . (b) 16-class
Wishart classification initialized by using the three classes derived
by the dominant Freeman and Durden fd , fv and fs amplitudes.
and volcanic activities [16], [66], as well as the monitoring
of land subsidence [69]. Further potential arises from a
comparison of the coherence between several data acquisitions, which can be used for land classification and change
detection [64], [74], [75].
A. Across-Track Interferometry
Conventional SAR imaging implies a projection from the
3-D object space to a planar 2-D radar image where the
image coordinates are range and azimuth. This means
that a single image pixel contains the focused radar echoes
19

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - March 2013