IEEE Geoscience and Remote Sensing Magazine - March 2020 - 23

that readers may have a comprehensive understanding of
various methods and learn about the design ideas behind
them. Several main representative methods are described
in detail. Third, we summarize the differences in data
requirements, method adoption, and acquisition results
for time-series InSAR techniques in different application
scenarios and discuss some limitations of current timeseries InSAR techniques.
CHARACTERISTICS OF TIME-SERIES
INTERFEROMETRIC SYNTHETIC APERTURE RADAR
Time-series InSAR is actually a general term for a class of
techniques, also known as time series, multibaseline, multitrack, multitemporal, repeat pass, and repeat track in other
literature. Compared with traditional differential InSAR
(DInSAR) [1], [2], the time-series InSAR technique has two
distinguishing features. First, it requires a large number of
SAR images (generally greater than 20) in the same region
at different times. DInSAR can be divided into two-, threeand four-pass interferometry. As the name implies, only
two, three, and four SAR images are needed [3]. Second, it
extracts signals only from the points with stable scattering
properties, i.e., sparse points [4], while DInSAR operates
on the entire image, i.e., a 2D grid. Obviously, it is easier to
ensure high coherence in specific points than it is to maintain high coherence on the entire surface. In DInSAR, the
spatial baseline is limited, and generally an interferogram
with a short baseline is required [5]. The temporal baseline
is also an important limitation, particularly in areas covered by vegetation, because the long time interval may lead
to severe decorrelation phenomena [6].
With the time-series InSAR technique, many SAR images in the monitored area can be used, and the limitation
of the spatial and temporal baseline is much smaller than
that of DInSAR. Based on the statistical characteristics of
atmospheric propagation signals, the time-series InSAR
technique can, to a large extent, mitigate the effects of atmospheric propagation delay [7], [8]. DInSAR can estimate
the atmosphere phase only under limited circumstances,
considering the linear relation between the phase delay and
the topography, which is of less use in cloudy atmospheres
[9]-[12]. The time-series InSAR technique overcomes the
degradation problem that DInSAR faces, especially decorrelation and atmospheric delays.
In addition, the time-series InSAR technique has the
ability to monitor a larger range than do traditional single-point deformation measurement techniques, such
as leveling [13] and GPS [14]. Presently, the time-series
InSAR technique is ascendant and has the potential to
become one of the most important means for large-scale
deformation monitoring [15], [16].
DEVELOPMENT
Since the introduction of persistent scatterer (PS) interferometry (PSI) [4], many organizations have proposed their
own time-series InSAR techniques. The development of
MARCH 2020

IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE

the time-series InSAR technique has benefited from three
developments in particular.
The first is the progress of SAR satellite imaging technology and computer data processing capabilities. Since
NASA launched Seasat-A, the first satellite carrying L-band
SAR in 1978, many SAR satellites have been launched
worldwide, providing much data for theoretical research
on and application of interferometry. For example, the
European Space Agency launched the C-band SAR satellites ERS-1, ERS-2, and ENVISAT in 1991, 1995, and 2002,
respectively; Canada launched the C-band SAR satellites
RadarSAT-1 and RadarSAT-2 in 1995 and 2007, respectively (the RadarSAT Constellation Mission, consisting of
three small, identical C-band SAR satellites, was launched
in 2019 and provides a four-day revisit); Japan launched
the L-band SAR satellites ALOS-1 and ALOS-2 in 2006
and 2014, respectively; Germany launched the X-band
SAR satellites TerraSAR-X and TanDEM-X in 2007 and
2010, respectively (they can form a bistatic system); and
Italy launched COSMO-SkyMed, an X-band SAR satellite
system with four constellations between 2007 and 2010
(their joint revisiting time can be up to four days). Figure 1
shows the SAR satellites mainly used as the source of timeseries InSAR data (see [17] for more details). Spaceborne
SAR imaging is developing toward multimode, high resolution, and short revisiting time. The time-series InSAR
technique usually deals with a large amount of data and
has great processing capacity and storage requirements
for processing platforms. Fortunately, the development of
current computer technology has made its performance
affordable for data processing.
Second, the time-series InSAR technique has the advantages of high accuracy (millimeter accuracy), large
scale (hundreds of square kilometers), and low cost. Currently, the highest accuracy of time-series InSAR techniques has reached the millimeter or even submillimeter level [18], [19]; this results from the sensitivity of the
phase signal to the surface deformation [20]. This accuracy allows for the use of most deformation monitoring
applications. The range that the time-series InSAR technique can monitor depends on the size of the SAR image,
from hundreds to thousands of square kilometers (e.g.,
30 km × 50 km using the StripMap mode of TerraSAR-X and
250 km × 250 km using the interferometric wide swath of
Sentinel-1). It is beneficial to observe a wide range of geological phenomena: because the marginal cost of increasing the monitoring scope is very low, the cost of manual
deployment is reduced significantly. The range that can be
monitored depends on the coverage of the time-series data.
Compared with other deformation monitoring methods,
the time-series InSAR technique has unique advantages in
retrieving surface deformation.
Finally, the demand for large-scale deformation monitoring
is increasing. Many geological phenomena (manmade and
natural) such as subsidence, volcanoes, and landslides are
associated with the deformation of corresponding objects
23

IEEE Geoscience and Remote Sensing Magazine - March 2020

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