Chong XU
National Institute of Natural Hazards,
Ministry of Emergency Management

Chong Xu is the Director of the Committee on Earthquake Hazard Chain, Seismological Society of China, the Director of the Key Laboratory of Compound and Chained Natural Hazards Dynamics, Ministry of Emergency Management of China, the Director of the Geological Hazards Research Center, National Institute of Natural Hazards, and Research Professor in National Institute of Natural Hazards, Ministry of Emergency Management of China. He got his Ph.D. at Institute of Geology and Geophysics, Chinese Academy of Sciences in 2010. His research interests include earthquake-triggered landslides, rainfall-induced landslides, and geological hazards using GIS, remote sensing, and statistical analysis techniques. By virtue of the study of earthquake-triggered landslides over 10 years, he has obtained a series of scientific research achievements of this subject, including database construction and criteria of earthquake-triggered landslides, quantitative description of landslide distribution, addressing issues of earthquake geology, landslide volume modeling, and hazard assessment. He chaired four NSFC projects (Youth, General, and International Joint Research) and two sub-projects of National Key R&D Program of China. He has published over 300 papers with over 10000 citations. He was selected into Elsevier's "China's highly cited scholars" since 2020. His research achievements, mainly on landslides triggered by recent large earthquakes, have been extended to about 50 research institutes and universities worldwide and are adopted directly by more than 100 research teams.

SE Distinguished Lecture | 31 July (Mon) 08:15 AM – 10:00 AM | Level 3 MR332

Probability Hazard Assessment of Earthquake-Triggered Landslides in China

Abstract: Earthquake-triggered landslide is an important type of earthquake secondary disasters occurring in mountainous areas, which has the characteristics of large scale, wide distribution and heavy losses. In some earthquake cases, the loss caused by earthquake-triggered landslide even exceeds the direct earthquake losses. Therefore, in recent decades, earthquake-triggered landslides have received extensive attention, and related research mainly includes information acquisition, database construction, distribution law analysis, vulnerability, hazard and risk assessments, and long-term evolution. Among them, earthquake-triggered landslide hazard assessment plays a scientific and technological supporting role in pre-disaster planning, emergency rescue during disasters, and post-disaster recovery and reconstruction. Therefore, there are many researches engage in this aspect in recent years. Previous studies mainly focused on landslide hazard assessment in the area affected by an individual earthquake, and the research methods mainly included expert experience, Newmark, statistical analysis and other methods. It is difficult to avoid human experience errors in expert knowledge methods. Because Newmark method generalizes many parameters, it can basically be considered as a method combining physical ideas, but more inclined to the expert knowledge method. The statistical analysis method needs abundant earthquake-triggered landslide data sets. With the rapid development of remote sensing technology and the production of detailed and complete earthquake-triggered landslide data sets, this method has been widely used in recent years. There is little research on earthquake-triggered landslide hazard assessment in large regions, especially in the whole country. Moreover, in most previous studies, the earthquake-triggered landslide hazard value is relative, and the hazard level does not have the significance of earthquake-triggered landslide occurrence probability. Therefore, there is an urgent need for a model that can express the probability hazard of earthquake-triggered landslides to carry out a large-scale, even nationwide, probabilistic hazard assessment of earthquake-triggered landslides.

In this talk, I will introduce a solution to this problem. My research team used Bayesian probability method and machine learning model to carry out the probability hazard assessment of earthquake-triggered landslides in China, and made a new generation of probability hazard distribution map of earthquake-triggered landslides in China. Based on more than 20 earthquake events, detailed and complete landslide database construction was carried out. The results show that these earthquakes triggered about 500,000 landslides. Select multiple earthquake-triggered landslide influencing factors such as peak ground acceleration (PGA), geology, geomorphology, and precipitation, and combine Bayesian probability method with a machine learning model to build a multi-factor impact model of earthquake-triggered landslide probability hazard model, and then obtain the weight of each factor. This means that the training of the model has been completed. In the subsequent application stage of the model, the absolute probability of earthquake-triggered landslide occurring in the whole region of China under the conditions of different PGA can be calculated conveniently and quickly. Combined with the distribution map of PGA in China, the probability hazard distribution of earthquake landslides under the background of PGA in China can be calculated. In addition, the model can also be used to quickly calculate the probability distribution of co-seismic landslides of various magnitudes of earthquakes, so as to provide support for earthquake emergency response.