MACHINE LEARNING FOR FAST AND RELIABLE SOURCE LOCATION ESTIMATION IN EARTHQUAKE EARLY WARNING

Abstract

Earthquakes are among the most destructive natural disasters, causing significant loss of human lives, infrastructure, and economic resources. Fast and accurate estimation of earthquake source location is therefore essential for effective Earthquake Early Warning (EEW) systems. Conventional source-location methods generally depend on seismic-wave travel-time calculations followed by mathematical inversion, which can require considerable computational time and may be difficult to apply under real-time conditions. This research proposes a machine-learning-based approach for fast and reliable earthquake sourcelocation estimation. Historical seismic observations containing P-wave and S-wave arrival times, signal amplitude, latitude, longitude, depth, and magnitude are processed through data cleaning, missing-value handling, normalization, and feature extraction. Five supervised machine-learning algorithms, namely Random Forest (RF), Support Vector Machine (SVM), K-Nearest Neighbours (kNN), Logistic Regression (LR), and Naive Bayes (NB), are trained and evaluated. The models are assessed using accuracy, precision, recall, F1-score, and confusion matrix analysis. The proposed framework aims to identify the most effective model for rapidly estimating earthquake source locations while maintaining reliable prediction performance under noisy and high-volume seismic observations. The approach demonstrates the potential of machine learning to reduce computational requirements and provide faster earthquake information for early-warning and emergency-management applications. Keywords: Earthquake Early Warning, Machine Learning, Source Location Estimation, Random Forest, Support Vector Machine, K-Nearest Neighbours.