IEEE Geoscience and Remote Sensing Magazine - March 2020 - 95

processes, such as monitoring the hazards and risks specifically related to landslides and buildings [67]. With
increasing scientific and economic interest in remote
sensing, SAR technology is experiencing a period of extraordinary growth. In the last decade, emerging countries, such as India, Korea, China, and Argentina, have
invested in these technological advances by developing
innovative SAR satellites. In fact, new satellite missions
were recently launched: RISAT [68], KOMPSAT-5 [69],
GAOFEN 3 [70], PAZ-1 [71], and SAOCOM [72]. In addition, future missions include the Italian COSMO-SkyMed
II generation [73], German Tandem-L [74], Canadian RADARSAT Constellation Mission [75], and a joint United
States-India mission [76].
The increasing interest in SAR satellite resources is reflected in commercial opportunities: many industries and
innovative start-ups are investing in the development of
small satellites and CubeSats, such as Urthecast, Capella
Space, and ICEYE. Urthecast is designing the Opti-SAR mission: the world's first fully integrated, multispectral optical
and SAR constellation of Earth observation satellites formed
by 16 satellites: eight tandem pairs in two orbital planes [77].
Capella Space is planning to develop a high-revisit constellation (approximately 36 satellites), which will provide hourly
imagery of any location on Earth [78]. The first satellite, Capella-1, operates an X-band SAR system and has a resolution
of 0.5 m. In January 2018, ICEYE launched ICEYE-X1 [79],
the world's first SAR satellite of fewer than 100 kg, to deliver
on-orbit images from space, marking a new era for the SAR
imaging industry.
BASIC RATIONALE OF SYNTHETIC APERTURE RADAR
INTERFEROMETRY
SAR interferometry exploits the coherent nature of active
radar sensors. The key principle consists of evaluating the
path (i.e., target-to-sensor distance) variation from two
(complex) radar images acquired at slightly different positions and/or acquisition epochs [80]-[82]. Access to the
phase information is crucial to be able to measure the
path difference on the order of a fraction of the transmitted wavelength. This implies centimeter- to millimeterlevel accuracy on the differential path for current satellite
and airborne SAR sensors typically working at the L, C,
and X bands.
Cross-track SAR interferometry, or simply InSAR, exploits the spatial diversity induced by a slight orbital offset,
referred to as spatial (or perpendicular) baseline, to impinge
on the phase difference of the contribution associated with
the scatterer height: the spatial baseline scales up (i.e., amplifies) this topographic contribution. This principle has
been used by the SRTM and TanDEM-X missions to simultaneously acquire SAR data so as to derive the most accurate
DEM on a global scale [2], [3].
DInSAR considers the path difference of a pair of images, acquired with repeated passes of a (generally singleantenna) SAR sensor, over the same scene at different times
MARCH 2020

IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE

[6], [83]. The presence of a slight surface deformation or,
better, a line-of-sight (LOS) component of the deformation
superimposes an additional phase contribution to the possible topographic "signature." Therefore, the DInSAR interferometric phase signal is given by
	

D{ = D{ T + D{ D + D{ n, (1)

where D{ T represents the interferometric phase contribution related to the topography of the scene through the
(perpendicular) spatial baseline; D{ D is the phase variation induced by possible target displacement, and D{ n is
the thermal noise contribution.
For most displacement-monitoring uses, DInSAR would
require two images collected with repeated passes over the
same orbit (zero spatial baseline). Such a condition is rarely
achieved in reality; therefore,
DInSAR relies on the use of an
external reference DEM. When
DInSAR CONSIDERS THE
the orbits are known, the DEM
PATH DIFFERENCE OF A PAIR
can be exploited to estimate
OF IMAGES, ACQUIRED
and cancel out the topographWITH REPEATED PASSES OF
ic phase contribution; this step
is usually referred to as zeroA (GENERALLY SINGLEbaseline steering (zbs). NevertheANTENNA) SAR SENSOR,
less, due to the typically low
OVER THE SAME SCENE AT
resolution and accuracy of the
DIFFERENT TIMES.
external DEM, a residual topographic (RT) phase term must
be taken into account following the zbs operation, especially for VHR applications such as
building monitoring. To highlight the effects of zbs, the following substitution is implemented: D{ T ! D{ z, where the
pedix z represents the residual height.
Even in the optimistic case of accurate DEM knowledge,
or, conversely, when high spatial resolution is not a requirement, a limiting error source in multipass analysis is related
to the propagation of electromagnetic waves through the
atmosphere. The latter is heterogeneous, that is, spatially
variable, mainly because of the horizontal and temporal variation of the refractive index due to the presence of
water vapor. In the expression of D{, such a phenomenon
leads to the presence of an additional spurious phase-error
source, D{ A, denoted as atmospheric phase screen (APS) or
atmospheric phase delay (APD):
	

D{ = D{ z + D{ D + D{ A + D{ n . (2)

APD is typically highly spatially correlated [84], that
is, acting on a low spatial frequency scale; in most cases, it can be considered uncorrelated over the images
acquired at different epochs. Consequently, it cannot
be easily filtered out from single interferograms because it is generally indistinguishable from the (wanted) spatial deformation signal, which also shows spatial correlations.
95
IEEE Geoscience and Remote Sensing Magazine - March 2020

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - March 2020
