IEEE Geoscience and Remote Sensing Magazine - March 2020 - 147

integrating machine learning with EO technologies could
be one encouraging solution to automatic landslide detection. To address this first big question, there is an urgent
need to answer the following additional questions: 1) At
what percentage are the detected landslides true positives?
2) What is the percentage of the missing landslides (false
negatives)? 3) In which scenarios are landslides more likely
be successfully detected?
Big question 2: When will landslides occur? A range of stateof-the-art landslide initiation and runout models have enabled us not only to estimate the location and geometry
of potential landslides but also to assess their potential
impacts. Predicting when landslides will occur remains
a grand challenge. There have been a limited number of
successful case studies, including those for the 2017 Heifangtai landslide. In these cases, deformation anomalies
(acceleration or a change in pattern) were observed prior
to failure and have been recognized as precursors. However, accurate EWSs require the identification of a diagnostic signature that can be somewhat uniquely related to
impending failure. The degree to which this signature is
unique defines the confidence with which a warning can
be issued, which represents a much stricter definition of
precursor. Further research is required to constrain the relationship between accelerated displacement and landslide
failure and thus to establish these diagnostic signatures
with more confidence. We suggest that widespread and
long-term deformation monitoring, combined with landslide observations, will enable considerable progress in addressing this problem.
Big question 3: What is the best way to reduce landslide disaster risk? The experience of the cooperation between experts
and local communities in the Dangchuan 4# landslide has
improved our understanding of best practices for community-based disaster risk management. How to best coproduce
a site-specific warning system with both local experts and
members of at-risk communities to reduce landslide disaster
risk remains an open challenge for the entire community.
ACKNOWLEDGMENTS
This work was supported by the National Natural Science
Foundation of China under grants 41801391, 41874005,
and 41929001; the National Science Fund for Outstanding Young Scholars of China under grant 41622206; the
Fund for International Cooperation under grant NSFCRCUK_NERC; Resilience to Earthquake-Induced Landslide Risk in China under grant 41661134010; the open
fund of State Key Laboratory of Geodesy and Earth's Dynamics (SKLGED2018-5-3-E); Sichuan Science and Technology Plan Project under grant 2019YJ0404; State Key
Laboratory of Geohazard Prevention and Geoenvironment Protection Independent Research Project under grant
SKLGP2018Z019; the Spanish Ministry of Economy and
Competitiveness, the State Agency of Research, and the
European Funds for Regional Development under projects
TEC2017-85244-C2-1-P and TIN2014-55413-C2-2-P; and
MARCH 2020

IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE

the Spanish Ministry of Education, Culture, and Sport
FURTHER RESEARCH IS
under project PRX17/00439.
REQUIRED TO CONSTRAIN
This work was also partially
THE RELATIONSHIP
supported by the U.K. Natural
BETWEEN ACCELERATED
Environment Research CounDISPLACEMENT AND
cil through the Center for the
Observation and Modeling of
LANDSLIDE FAILURE AND
Earthquakes, Volcanoes, and
THUS TO ESTABLISH THESE
Tectonics under come30001
DIAGNOSTIC SIGNATURES
and the Looking Inside the
WITH MORE CONFIDENCE.
Continents From Space and
Community Earthquake Disaster Risk Reduction in China
projects under NE/K010794/1 and NE/N012151/1, respectively, and by the European Space Agency through the ESAMOST DRAGON-4 project (32244 [4]). Roland Bürgmann
acknowledges support by the NASA Earth Surface and Interior focus area.
Keren Dai, Zhenhong Li, Bürgmann, David G. Milledge,
and Roberto Tomás contributed to writing this article. Dai,
Li, Zheng Wang, Tengteng Qu, Chaoying Zhao, and Jingnan
Liu carried out the SAR data processing and related experiments and analysis. Qiang Xu, Xuanmei Fan, and Tomás
contributed to the geological analysis and result interpretation. Chaoyang He and Xu collected and analyzed the
crackmeter data for landslide warning in the Heifangtai
loess landslide. Li, Xu, Dai, and Jianbing Peng developed
the concept of the Earth observation-based landslide EWS
proposed in this article with contributions from Milledge,
Bürgmann, Qin Zhang, Li, and Liu. All relevant data are
available from the authors upon reasonable request.
AUTHOR INFORMATION
Keren Dai (daikeren17@cdut.edu.cn) received his B.Eng.
degree in remote sensing technology and application and
his Ph.D. degree in photogrammetry and remote sensing
from Southwest Jiaotong University, Chengdu, China, in
2012 and 2017, respectively. He was a visiting Ph.D. student at Newcastle University, Newcastle upon Tyne, United
Kingdom, from 2014 to 2015 and has been a postdoctoral
researcher at Chang'an University, Xi'an, China, since 2019.
He is currently a professor at the College of Earth Science,
Chengdu University of Technology, as well as a research
scientist with the State Key Laboratory of Geohazard Prevention and Geoenvironment Protection. He specializes
in landslide detection, mapping, and monitoring with interferometric synthetic aperture radar and multispectral
remote sensing. He was awarded the Promising Young
Scientist Award by the International Consortium on Geodisaster Reduction in 2017.
Zhenhong Li (zhenhong.Li@newcastle.ac.uk) received
his B.Eng. degree in geodesy from Wuhan Technical University of Surveying and Mapping (now Wuhan University), China, in 1997 and his Ph.D. degree in GPS, geodesy,
and navigation from University College London in 2005.
147

IEEE Geoscience and Remote Sensing Magazine - March 2020

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