IEEE Geoscience and Remote Sensing Magazine - March 2016 - 20

CONCLUSIONS AND PERSPECTIvES
The arrival of remote sensing has caused a true revolution in displacement measurement by significantly
improving the spatial coverage and the measurement
accuracy. Spectacular results have been obtained in numerous fields, such as the study of urban subsidence;
coseismic, interseismic, and postseismic motion; glacier
flow; and volcanic deformation. With the continuous
launching of Earth-observation satellites and the increasing availability of the amount of remote sensing
data, data fusion becomes necessary and plays a more
important role in displacement measurements. However, using all of the available measurements cannot always provide satisfactory results, but it always presents
difficulties such as unknown weighting coefficients and
high computational cost. Intelligent fusion strategies
and methods, involving how to benefit from the large
volume of data in an efficient way to reduce the displacement uncertainty and to improve our knowledge on the
physical process of the phenomenon under observation,
constitutes a living topic in many works. Meanwhile,
more and more attention is paid to the displacement
uncertainty characterization and quantification. The
consideration lies not only on independent random
uncertainty but also on correlated or systematic uncertainty. The uncertainty management approach has also
been extended from a probabilistic approach to a possibilistic approach. On the other hand, the techniques in
displacement measurement by remote sensing are still
being improved to integrate, as much as possible, the
benefit of the high spatial resolution and the increasing frequency of data acquisition for terrestrial displacement measurements. Moreover, efforts have been made
to combine different techniques, e.g., the combination
of PS and SBAS methods, of correlation and DInSAR,
seeking to make the best use of the information contained in the data by exploiting the complementarity of
different techniques.
Besides the achievement in displacement measurements fusion, challenges are also present. Even though
rapid development has been obtained in the recent
years, the fusion of heterogeneous measurements, from
SAR, optical images, GPS, and other sources of information, still remains a delicate problem. No fusion method
or strategy is completely operational to deal with diverse
characteristics and uncertainty levels of the heterogeneous measurements in an inversion system. No efficient
solution has been proposed to the determination of the
contribution of each individual measurement as well as
their covariance. From the computational point of view,
even with the availability of supercomputing facilities,
we can still be quickly limited by the memory and storage
capacity, as well as the computation time, given the high
spatial resolution and the strong repetitiveness of acquisitions. For accuracy and uncertainty consideration,
on one hand, the quantification and the improvement
20

of the accuracy are always very challenging, given that
in most cases the ground truth is not available. On
the other hand, it is always difficult to characterize
the uncertainty in displacement measurements and to
choose an appropriate uncertainty management approach. In satellite imagery, uncertainty comes from
different perturbations generated along the wave propagation path, at the backscattering surface, as well as
from noise generated in the electronic processing. In
addition, imperfect corrections (i.e., atmospheric and/
or geometric corrections) performed in the displacement extraction chain also introduce systematic uncertainties. These diverse sources result in uncertainties of
complex characteristics. Moreover, in the case of model
inversion, it is very difficult to propagate the uncertainty. The retrieved deformation models are often provided without uncertainty information. The evaluation of
these models obtained with more or less different measurements is thus a challenging task. For example, in the
case of the Kashmir earthquake in 2005, [12], [31], [33],
and [116] obtained different fault rupture models by using different surface displacement measurements. Without ground truth, it is impossible to assess these models
in an objective way.
Given the current status and the future development
of displacement measurement fusion, sophisticated
statistic tools, such as the Kalman filter, the Bayesian
theory, and so on, can be expected to further improve
the results. Meanwhile, it will be important to modify the processing algorithms and to adapt our way of
working. Inspired from the ocean reanalysis, different
measurements with different spatial coverage, different
spatial resolution, different time spans, bringing different information, including the correlation of SAR,
optical images, DInSAR, GPS, and other in situ measurements, can be homogenized through a realistic
physical model to produce spatially and temporally
regular displacement maps (namely, displacement reanalysis) to record the properties of the displacement
over time. Later, instead of keeping and processing different types of measurements of large volume, these
displacement reanalyses present numerous advantages. Currently, the displacement measurement by remote sensing is still mainly applied to past events that
have taken place before the data processing. With the
launching by the European Space Agency of the Sentinel series, remote sensing data can be acquired nearly
everywhere on Earth at least every six days. By adding the data issued from other satellites, TerraSAR-X
and TanDEM-X, Landsat 8, the four satellites COSMOSkyMed, and ALOS-2, real-time monitoring by time
series will become possible. The combination in real
time of displacement measurements from remote sensing imagery and physical models is possible. It will
thus be possible to predict the evolution of an event
such as a magma reload of a reservoir located beneath
ieee Geoscience and remote sensing magazine

march 2016

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - March 2016