A Real time Integrated Framework for Seismic Activity Monitoring and Probabilistic Forecasting using Machine Learning and Geospatial Analytics: The EASO System

This paper presents a robust, automated computational framework for the real time monitoring and statistical analysis of seismic activity, focusing on the seismically active region of Epirus, Greece. To address the inherent latencies of traditional monitoring, this study introduces the Epirus Advanced Seismic Observatory (EASO). Utilizing the Python programming language and integrating live data from the Euro Mediterranean Seismological Centre (EMSC) API, EASO implements an end to end pipeline for data acquisition, processing, and visualization. The methodology employs the Density Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm to autonomously identify seismogenic structures and spatial clusters without prior geological constraints. Furthermore, the framework integrates fundamental seismological laws, including the Gutenberg Richter magnitude frequency relationship and the Modified Omori Law (ETAS model), to estimate b-values and decay rates of aftershock sequences. A multi panel visualization dashboard was developed, providing a "full spectrum" probability forecast for various magnitude thresholds (4.0–7.0 Richter) within a seven day window. The results demonstrate that the integration of unsupervised machine learning with traditional stochastic models significantly enhances the interpretability of regional seismic swarms. This study concludes that EASO provides a critical tool for real time visual forensics, risk assessment, and public information, offering a scalable solution that can be adapted to other high risk tectonic zones globally.