Optimization of Network-Based Earthquake Early Warning Systems on the Korean Peninsula

Most countries and local governments provide earthquake services in the public domain, and they must have high accuracy. If a missed alarm of the Earthquake Early Warning (EEW) system causes many casualties, or if the industrial system is temporarily suspended owing to a false alarm, causing economic losses, it inevitably becomes the responsibility of the government. Therefore, most countries approach the technological improvement of EEW systems carefully by performing simulations and conducting long-term tests to ensure their reliability. In this study, we extract characteristics of the initial P-wave amplitude from an earthquake on the Korean Peninsula and perform trend analysis. We found a common optimal threshold on the Korea Meteorological Administration's seismic observatory network from trend analysis. We then evaluated the performance of the optimized algorithm based on the simulation. The performance evaluated the actual events recorded corresponding to the number of matched, missed, and false events. As the result of the evaluation, the optimized module combination had a significantly lower occurrence of false events than the previous version. Therefore, we expected that the proposed optimization should contribute to improving alarm stability in real-time EEW.