IEEE Geoscience and Remote Sensing Magazine - March 2020 - 74

Generally speaking, the accuracy of offset tracking-derived 2D displacements along the LOS and azimuth directions depends on the spatial resolution of the used SAR images
(e.g., 1/10 to 1/30 pixel) [24], [29]. Consequently, high-resolution SAR images are preferred for OT-based 2D mining
displacement retrieval. In addition, the OT technique retrieves 2D displacements based on the amplitude features of
SAR images. Therefore, images acquired by short-wavelength
(e.g., X-band) SAR sensors are possibly preferred, as short
wavelengths are generally sensitive to structural attributes
(e.g., surface roughness) and may result in more texture features in the images than long wavelengths, especially for areas covered by short vegetation or bare soils [30]. It should
be pointed out that the accuracy of OT-derived displacement
measurements from SAR images is usually much lower with
respect to phase-based measurements in high-coherent areas [31] and likely performs less efficiently with respect to
the classical use with high-quality optical images (e.g., high
spatial resolution, cloud-free, and high brightness contrast).
SAR/InSAR METHODS FOR
TS DEFORMATION MONITORING
TS-InSAR
Only two coregistered SAR images (i.e., a single SAR image
pair) are used in the aforementioned methods (i.e., DInSAR,
MAI, and OT). This means that only the surface deformation that occurs between these two SAR acquisition dates
can potentially be obtained. To reveal the temporal evolution of surface deformation, TS InSAR techniques based on
single-track SAR images (greater than two) have been developed in the past decades. The TS-InSAR methods can also be
classified into two types: 1) single-reference TS-InSAR, such
as persistent scatterer (PS) InSAR [32]-[35]; and 2) multireference TS-InSAR, such as small baseline subset (SBAS)
InSAR [36], [37], although a method that incorporates both
was recently proposed [38]. We refer the reader to [39] for a
systematic review of TS-InSAR techniques.
In this section, we give an overview of the basic theory
and data processing of the SBAS-InSAR method, considering that SBAS-InSAR generally shows a better performance in TS mining deformation monitoring than does
PS-InSAR [40]. If we assume that M InSAR pairs are generated from K + 1 single-track SAR images over the same
area by setting small spatial and temporal baselines,
then M unwrapped interferograms can be obtained, i.e.,
Tz = 6Tz 1, Tz 2, f, z M@, using DInSAR processing. At this
stage, the unwrapped phase is mainly composed of residual topographic, atmospheric, and deformation phases [36].
To solve TS deformation, the mean deformation velocities
between time-adjacent SAR acquisitions V are assumed,
and an observation system involving the mean deformation velocities, residual topography Tz, and unwrapped
phase Tz is constructed as
	
74

Tz = B $ X + f, (5)

where X = 6V Tz@T, B is the coefficient matrix depending
on the formation of the available InSAR pairs [36], and f
denotes the error term of the system. Finally, a least-squares
solver (for a full-rank system) or singular-value decomposition (for a rank-deficient system) is used to solve the mean
deformation velocities between the time-adjacent SAR acquisitions and further estimates the TS LOS deformation
by accumulation.
It should be noted that, owing to denser measurement
points than those of PS interferometry, the recently developed "squeezed" SAR (SqueeSAR) method [41], which can
jointly process PS and distributed PSs (DPSs) at full resolution, is also a promising tool for retrieving ground surface
TS displacements, especially in nonurban areas (e.g., mining areas with short vegetation). Readers are referred to [41]
for more details about SqueeSAR.
TS-OT AND TS-MAI
Following the same core idea of SBAS-InSAR, TS analysis
methods based on OT and MAI measurements have also
been developed. For instance, Casu et al. [42] proposed a
method called PO-SBAS to retrieve 2D large TS displacements from multiple OT-derived displacement measurements and tested this method over the Sierra Negra caldera
in the Galápagos Islands in 2011. Following this, the POSBAS method has been successfully applied to the large 2D
deformation retrieval process associated with landslides
[43], crustal movements [44], and glacier motion [45]. In
recent years, MAI-based TS analysis methods [46]-[48] have
also been proposed to retrieve azimuth TS deformations.
SAR/InSAR-BASED 1D/2D MINING DISPLACEMENT
MEASUREMENTS
1D LOS MINING DISPLACEMENT MEASUREMENTS
Since Carnec et al. [7] first demonstrated the capability
of DInSAR in constructing a 1D LOS deformation map in
1996, many studies have been conducted to further validate
and explore the potential of the DInSAR technique in 1D
mining displacement measuring. For instance, Wright and
Stow [49] obtained LOS deformation maps over the Selby
coalfield in the United Kingdom from interferometric data
(i.e., topographic phase components were not differentiated)
derived from 35-day European remote sensing (ERS) satellite SAR images in 1999. They confirmed that the concentric
interferometric fringe features were associated with underground mining activities from the precise geolocation alignment between the fringe features and underground working
panels. After that, more tests on 1D mining displacement
measurements using DInSAR were carried out in different
countries (e.g., [50]-[54]). Meanwhile, the accuracy of DInSAR in mining displacement measurements has been confirmed by comparing the results with traditional geodetic
measurements (e.g., [55] and [56]). The results have shown
good agreement between InSAR and the geodetic measurements, with an accuracy level of roughly 1 cm per year [56].
IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE

MARCH 2020
IEEE Geoscience and Remote Sensing Magazine - March 2020

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