Ministry of Education, Culture, Sports, Science and Technology Web Site. Top of the page.

Overview of research topics (research topics selected for 2021)

Research topic (1): Synergy effect Through Human and Artificial Intelligence Towards New Era in Seismology (Principal Investigator: Hiromichi Nagao, Earthquake Research Institute, The University of Tokyo)

Research overview

This project facilitates the interaction and cooperation between artificial intelligence and natural intelligence. The goal is to improve earthquake and tremor detection techniques and to advance earthquake modeling techniques by leveraging the synergy between deep learning and human experience in order to accelerate new earthquake research and prepare for earthquake disasters. We will make use of information science and technology such as artificial intelligence to analyze both digital and image seismic waveform data, and to develop new earthquake and tremor detection techniques and subsurface modeling technologies.

Message from the Principal Investigator

Mathematical techniques are being introduced into earthquake research more commonly than before, with recent and rapid developments in information science and technology such as artificial intelligence completely changing how earthquake-related big data analysis is performed. For example, deep learning methods for detecting seismic waves in seismic waveform data often demonstrate performance exceeding that of an experienced seismologist. In this project, we will focus our efforts on developing new research that combines earthquake knowledge from the various perspectives of experts in seismology and information science, on efforts to raise awareness among citizens, and on developing and training the next generation of researchers.

Hiromichi Nagao, Earthquake Research Institute, The University of Tokyo

Related links

Research topic (2): Exploration of subsurface faults by big data analysis of seismic waveforms using signal processing and machine learning (Principal Investigator: Takahiko Uchide, National Institute of Advanced Industrial Science and Technology (AIST))

Research overview

This project aims to detect subsurface faults based on seismic waveform data and to objectively identify the fault geometry in order to help build fault models to perform strong ground motion assessments and enhance our understanding of tectonics. We will conduct research on the automation of seismic data processing, the automatic identification of subsurface faults based on hypocentral distributions, and the estimation of the geometries of subsurface reflectors from later phases. We will also establish a research infrastructure for applying information science in earthquake research through such means as releasing programs we develop in this project.

Message from the Principal Investigator

This research topic aims to support the creation of earthquake occurrence scenarios that are more realistic by searching for subsurface faults based on information such as seismic activity and seismic wave reflections. This research topic is related to active fault investigation and artificial intelligence research, both of which have been performed in our institute. We will contribute to earthquake and artificial intelligence research based on our broad range of expertise with young postdoctoral researchers.

Takahiko Uchide, Geological Survey of Japan, AIST

Related links (linked websites are in Japanese)

Affiliations of participating researchers

Participating researcher websites, etc.

Research facility used for this research topic

Publications with articles related to this research topic

Research topic (3): Innovation on geodetic data analysis toward fault slip monitoring based on data assimilation (Principal Investigator: Masayuki Kano, Tohoku University)

Research overview

This project aims to construct a fault slip monitoring system to determine the state of fault slips in plate subduction zones and predict short-term slip changes, and to evaluate the impact on future major earthquakes. In order to accomplish this goal, we will develop an innovative geodetic data analysis technique based on statistics and machine learning, and enhance existing fault slip monitoring techniques through improving our ability to detect crustal deformation and deepening our understanding of the characteristics of observation noise. We aim to develop a fault slip monitoring system by combining these results with data assimilation techniques that take the physics of fault friction into account.

Message from the Principal Investigator

My name is Masayuki Kano (Tohoku University), and I am the lead researcher on this project. In this topic, we will develop an innovative geodetic data analysis technique using information science, in order to construct a system to monitor fault slips in subduction zones. Of the five topics selected this year, our project is the only one to focus mainly on geodetic data. We will promote this topic so that we can bring new developments in our understanding of phenomena related to earthquake and volcanoes, through developing a geodetic data analysis technique that can be used in various types of research. There are currently very few young researchers and graduate students working with geodetic data, and training researchers who can support future geodetic data analysis from a mathematical perspective is a pressing issue. Out project is also focusing on developing the next generation of researchers. We are regularly holding study sessions on fundamental topics related to geodetic data analysis. We welcome joint collaboration.

Masayuki Kano, Tohoku University

Related links (linked websites are in Japanese)

Participating researcher websites, etc.

Publications with articles related to this research topic

Research topic website

Research topic (4): Research and development for semi-real-time spatio-temporal forecasting of seismic activity and associated ground motion considering incompleteness of seismic data (Principal Investigator: Hisahiko Kubo, National Research Institute for Earth Science and Disaster Resilience)

Research overview

An enormous amount of seismic data has already been observed and accumulated. However, since not everything that happens in nature can be observed with limited resources, the seismic data is inherently incomplete, and there are limitations to the forecasting based on these data. To realize semi-real-time spatio-temporal forecasting of seismic activity and associated ground motion after a large earthquake, we will conduct research and develop elemental technologies to overcome the incompleteness of seismic data and establish prediction approaches that connect these elements, by combining information science insights with earthquake observation data and domain knowledge in seismology and earthquake engineering.

Message from the Principal Investigator

Japan is one of the most earthquake-prone countries in the world and has suffered enormous damage from earthquakes many times. To mitigate earthquake damage as much as possible, we need to establish and update forecasting technologies. We will focus on the issues of forecasting the transition of seismic activity after a large earthquake and the associated ground motion. This research topic is mainly composed of young researchers in seismology and earthquake engineering. We will boldly take up the challenge by absorbing the latest knowledge in statistics and machine learning.

Hisahiko Kubo, National Research Institute for Earth Science and Disaster Resilience

Related links (linked websites are in Japanese)

Participating researcher websites, etc.

Publications with articles related to this research topic

Research topic (5): New developments in space-time earthquake forecasting and monitoring: from long-term to real-time (Principal Investigator: Jiancang Zhuang, The Institute of Statistical Mathematics)

Research overview

Seismic activities, such as the Kumamoto earthquake sequence, major earthquakes near the Tohoku offshore, the Nankai Trough earthquake, and the 1938 Fukushima offshore earthquake, are complex and diverse. We aim to develop, deploy, and implement useful models that can be used to conduct short-term probability forecasting and real-time monitoring of crustal deformation and seismic motion, by taking into consideration the potential for recurring major earthquakes and intense earthquake swarms. To achieve this object, we will leverage state-of-the-art of high-performance computing methods in statistical science, including statistical seismology and multivariate time series analysis, to provide long-term, medium-term, and short-term earthquake forecasts and real-time monitoring with associated confidence levels. Initially, by refining the space-time ETAS model to overcome challenges such as the heterogeneity of seismic source data and incorporating into the model factors like crustal changes and abnormal seismic activity, we will then implement an online system capable of providing probability forecasts on different time scales, including long-term, medium-term, and short-term, as well as composite probability forecasts. The outputs of our online probability forecasts will be utilized for optimizing existing and new observation network configurations and developing preliminary scenarios for earthquake early warnings, as well as serve as preliminary information to enhance system reliability. Finally, the heterogeneity of space-time monitoring in terms of scope and accuracy will be evaluated by comparing observed seismicity with the forecasts.

Message from the Principal Investigator

The difficulty of deterministic "earthquake prediction" is an inherent challenge in the field of complex systems science. We have up to now accumulated a wide range of relevant data on a large scale, with which there is a growing need to pursue probabilistic forecasts. This project aims to conduct such probabilistic forecasts based on quantitative models of seismicity and crustal activity, so that we can further promote researches on the utility of seismological knowledge. Particularly, by composing multiple sources of earthquake probability forecasts in various timescales, including long-term, medium-term, and short-term, we can implement real-time forecasts with higher probability gains. This enables us to forecast seismicity based on anticipated seismological scenarios and deploy emergency observation networks swiftly. From the perspectives of multi-timescale, including long-term, medium-term, short-term, and immediate, this project will not only leverage methods from information science and statistical science, but also make use of the advantages of other research areas, to contribute to the advancement of seismology research and disaster prevention.

Jiancang Zhuang, The Institute of Statistical Mathematics

Related links (linked websites are in Japanese)

Research topic website

Participating researcher websites, etc.