IEEE Geoscience and Remote Sensing Magazine - March 2020 - 18

interpretation and visualization of the results for customers and general users who are not SAR experts. In general,
we expect further growth of the SAR service industry but
also ongoing consolidation and concentration in different markets.
CONCLUSIONS
The development of surface motion from SAR can roughly
be divided into three main periods: data sparsity, availability, and ubiquity. We define the period of data sparsity as the time before the launch of Cosmo SkyMed and
TerraSAR-X in 2007. As data are the fuel on which our
science runs, the availability of data is very important.
However, the period of data sparsity was not a period of
science sparsity. The most important and ground-breaking
papers were published during this period. The foundation
of surface-motion estimation from SAR was laid, and the
possibilities have already been demonstrated at various
test sites.
However, data sparsity and difficulties in acquiring data
from self-selected sites certainly hindered the commercial
development and use of the
new techniques by researchers from other domains. This
FOR THE SERVICE
changed around the years 2007
INDUSTRY, THE INCREASE
and 2008, when more data
IN AVAILABLE DATA
became scientifically, and also
PROVIDERS AND THE
commercially, available. The
EXPECTED DECREASE IN
availability of high-resolution
DATA COSTS ARE A
SAR data further increased the
PROMISING TREND.
interest of a wider user base.
In contrast to early research,
where test sites were selected
based on data availability, it was now possible to acquire data
from research sites of interest.
In this period, we see an increase in publications and
commercial activities. D-InSAR is now applicable with highresolution data, allowing new applications; scientists and
end users recognize this opportunity and use the data in
more fields. Thanks to the available high-resolution data,
surface-motion estimation from amplitude-based approaches is now applicable in many more use cases.
Finally, we now see the beginning of the phase of data
ubiquity. This is currently driven by the global availability
of Sentinel-1 data. It will gain further momentum based on
the large number of upcoming and planned systems not
only from traditional data providers but also from a series
of newcomers and start-ups.
Although motion estimation from SAR developed
rapidly in the last decades, we still face old problems and
new challenges. One issue is the huge variety of different
approaches to surface-motion estimation from SAR and
the resulting confusion and uncertainty for users. In this
context, the lack of clear quality estimations for the results
increases the problem of variety, as it is difficult to assess
the results. Depending on the method-again, reflecting
18

the fact that the variety of techniques is an opportunity as
well as a challenge-the variety of parameters during processing can significantly affect the results. In combination
with the third major problem we are facing, the difficulty
in clearly presenting and communicating the measurement results and their meaning, these problems result in a
constant need to reinvent the wheel and revalidate the results. As experts in the field, we recognize the tremendous
capabilities of SAR for surface-motion estimation as well as
its shortcomings, but we still do not succeed at clearly and
convincingly communicating those.
A more recent technical issue involves dealing with the
larger volume of data in the phase of data ubiquity. With
the increasing amount of data, processing capabilities are
not increasing to the same degree. In this regard, big data
processing, covering surface motion close to real time for
large regions and countries, is certainly a major research
and commercial trend. This includes not only managing
large data sets and dealing with the immense computing requirements for processing but also developing new
methods for processing, quality estimation, and interpretation of the results. Advances in machine learning can
help correctly separate dangerous motion outliers that
could be precursors of fast landslide events or sinkholes
from outliers caused by noise or the measurement of nonstable scatterers.
Furthermore, we expect more data-driven research. In
the phase of data sparsity, the research area followed the
data, but, in the phase of data availability, data over previously known test sites, such as known landslide-prone areas, can be ordered. We expect that, in the future, the data
will be assessed first, revealing already moving sites that
might then be analyzed in more detail. This approach may
give us new insights into our dynamic Earth, but it will also
change the way we research and work. It is time to prepare
and get ready!
ACKNOWLEDGMENTS
This work was supported by the Natural Science Foundation of China under grants 41571435 and 61331016.
AUTHOR INFORMATION
Mingsheng Liao (liao@whu.edu.cn) received his B.S. degree in electronic engineering from Wuhan Technical
University of Surveying and Mapping (WTUSM), China,
in 1982, his M.A. degree in electronic and information engineering from the Huazhong University of Science and
Technology, Wuhan, in 1985, and his Ph.D. degree in photogrammetry and remote sensing from WTUSM in 2000.
He has been a professor with the State Key Laboratory of
Information Engineering in Surveying, Mapping, and Remote Sensing, Wuhan University, since 1997. He is also the
coprincipal investigator and principal investigator of the
ESA-MOST cooperative Dragon 1 (2004-2008), 2 (2008-
2012), 3 (2012-2016), and 4 (2016-2020) projects. He has
published more than 100 peer-reviewed journal papers
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