Statistical Modeling and Forecasting of Seismic Events

Abstract

Understanding earthquakes remains a challenge in our world which is strongly affected by natural forces. This study analyzes earthquake data across various seismic regions, focusing on the frequency and intensity of seismic activity. Monthly maximum magnitudes are modeled using extreme value distribution theory. We use nonparametric methods, such as locally fitted regressions and splines, alongside parametric ARIMA models to assess temporal patterns. Machine learning techniques are incorporated for anomaly detection, and earthquake occurrences are modeled using a Poisson process based on interarrival times. These methods provide insights into earthquake dynamics and may improve risk assessments.