Earthquake and Tsunami Nowcasting and Forecasting Using Shannon Information Theory

Date: 4/24/2019

Time: 06:00 PM

Room: Grand Ballroom

What is the information content of a sequence of earthquakes? In 1948 Claude Shannon described a method for understanding the quantity of information provided by a sequence of events derived from an alphabet of symbols. Shannon’s statistical communication theory is now widely recognized as one of the fundamental discoveries of the 20^th century, and is the foundation for modern computer technology.

We have begun to apply Shannon’s methods in an attempt to extract the information content of earthquake events. Clearly, both nowcasting and forecasting are techniques that try to understand the information content of earthquake sequences and the implication for future earthquakes. We note that nowcasting refers to the estimation of the current uncertain state of a dynamical system, whereas forecasting is a calculation of probabilities of future state(s).

In this talk we describe a new method for nowcasting and forecasting that incorporates some of the results of statistical communication theory. We describe preliminary results for global catalogs and regional catalogs such as the Japanese islands, and discuss implications of future and ongoing research, particularly as it might apply to both great earthquakes and tsunamis.

We also adopted and further developed the earthquake nowcasting method to Japan. For southern Kanto, we used earthquakes since 1926 with depths shallower than 150 km and set M_L=6.0 and M_S = 4.2. The obtained result as of December 20, 2018 is EPS of 8%. This is interpreted as an indication that if a large (M6+) earthquake occurred on this date, the recurrence interval would be relatively short. Errors that arise from statistics-based estimation such as this must be considered so that we introduced the calculation of uncertainties to the earthquake nowcasting, using a bootstrap approach. Taking into account the contribution of these errors to the EPS calculations, we conclude that the earthquake nowcasting generates stable estimates of EPS for southern Kanto.

Presenting Author: John B. Rundle

Authors