Evaluating P-Wave detection algorithms for earthquake early warning: insights from GeoNet data in Canterbury, Aotearoa New Zealand

• Published in: Earth science informatics 2025, Vol.18 (1), p.46
• Main Authors: Chandrakumar, Chanthujan, Tan, Marion Lara, Holden, Caroline, Stephens, Max T., Prasanna, Raj
• Format: Article
• Language: English
• Subjects: Algorithms; Datasets; Early warning systems; Earth and Environmental Science; Earth Sciences; Earth System Sciences; Earthquakes; Ground motion; Information Systems Applications (incl.Internet); Motion perception; Ontology; P waves; Performance evaluation; Seismic activity; Simulation and Modeling; Space Exploration and Astronautics; Space Sciences (including Extraterrestrial Physics
• ISSN: 1865-0473; 1865-0481
• DOI: 10.1007/s12145-024-01519-9

What is the most effective P-wave detection algorithm for an Earthquake Early Warning (EEW) system that minimises false, late and missed detections? This study evaluates the performance of four distinct P-wave detection algorithms in terms of their detection accuracy. Utilising a comprehensive MEMS-based ground motion dataset from the GeoNet network, this study analyses the algorithms’ performances by introducing four distinct pick deviation categories. Among the evaluated algorithms, the wavelet-based P-wave picker is identified as the most suitable and accurate for EEW systems, achieving a 98.3% success rate with a mean deviation of 0.12 s and a standard deviation of 0.63 s compared to the manual pick. This algorithm proves effective for both community-engaged and traditional EEW systems. The methodology used for performance comparison in this research is applicable to other regions and datasets, aiding in selecting more accurate and reliable P-wave detection algorithms. The study suggests extending this performance analysis to encompass a broader spectrum of traditional and modern algorithms in future research.