Seismic Pattern Changes Before the 2011 Tohoku Earthquake Revealed by exploratory data analysis

The ability to predict earthquakes is invaluable, especially in high-risk seismic zones, yet precise predictions remain elusive. One potential reason is the limited integration of statistical approaches in earthquake research. In this study, I employed exploratory data analysis (EDA), a data-driven parametric statistical method, to investigate seismic records from Japan, using data provided by the Japan Meteorological Agency. The intervals between earthquakes closely followed an exponential distribution, defined by a single parameter, <i>λ</i>, representing event frequency. In contrast to the conventional Gutenberg-Richter law, earthquake magnitudes conformed to a normal distribution, characterised by two parameters: <i>µ</i> (mean) and <i>σ</i> (scale). After establishing these distributions and their parameters, significant shifts became evident. Before the 2011 Tohoku Pacific Ocean earthquake, notable changes emerged: earthquake intervals initially shortened, possibly reflecting energy destabilisation around asperities at plate boundaries, followed by an increase in the magnitude scale. These patterns suggest tectonic plates are heterogeneous, with varying boundary rigidity. Additionally, moving averages of magnitude exhibited substantial fluctuations, reaching unusually high levels. Identifying such anomalies accurately required understanding baseline distributions under normal conditions. Broadening the use of EDA across diverse seismic datasets could improve prediction accuracy. This study underscores the importance of statistical methodologies in seismic research and provides critical insights for enhancing seismic risk assessment.<br>