Introducing the Data-Oriented Approach for Earthquake Disaster Prevention Policy in Japan: Prospects and Challenges

• In Event: Precision and Uncertainty in a World of Data - II

Abstract

This paper focuses on how “the data-oriented approach”, characterized by pattern recognition and classification using big data or machine learning, has affected seismological research and earthquake disaster prevention policy in Japan. Recently, the data-oriented approach has come to play key roles of innovation in various areas – including earth science – and the Japanese government is planning to incorporate it into the next Earthquake and Volcano Hazards Observation and Research Program. In this program, Japan is going to focus on data assimilation to make better long-term earthquake prediction possible. In addition, this program refers to the potential expansion of available data due to the increase of monitoring spots. On the other hand, the program also mentions the fundamental limitation of the data-oriented approach for earthquake prediction due to lack of data older than 100 years. Also, like other extremely rare but serious disasters, there is a mismatch between the pattern of event occurence and the pattern of observable data. Thus, scientists have to complement the data by historical or geological evidence. With such recent developments in mind, we examined the prospects and challenges of introducing the data-oriented approach into seismological research and disaster prevention policy-making.
According to interviews and literature survey, the data-oriented approach has not played important roles yet, partly due to the academic tradition in Japanese seismology to value the analysis of seismological mechanism. Policy makers in Japan have also been influenced by such tradition, but the situation is changing, as seen in the aforementioned new program.

Authors

• Noel Kikuchi, National Graduate Institute for Policy Studies

• Keiko Matsuo, Japan Science and Technology Agency

• Yasushi Sato, Niigata University