On optimization of parameters used in Earthquake Early Warning systems for rapid evaluation of earthquake magnitude and level of shaking

Speaker: Vladimir Sokolov, Friedemann Wenzel
Determination of earthquake magnitude and the strength of shaking at a site from the initial P-wave are the key problems for Earthquake Early Warning (EEW). Several techniques have been proposed recently for this purpose. The most of the techniques uses (a) characteristic parameters of ground motion determined within specific time window started after the onset of P-wave and (b) regression relationships between these parameters and magnitude or strength of shaking. The relationships were initially developed based on empirical data from the regions characterized by large datasets (Japan, California and Taiwan) and recently were checked for a few other seismically active regions (e.g. Italy and Pacific Northwest). Bearing in mind recent progress in real-time procedures of estimation of earthquake size and strength of shaking, at present there is an opinion that reliability of procedures can be improved by combining estimations obtained from different used ground-motion quantities obtained as function of time.

The good EEW technique should produce fast and reliable estimations of earthquake magnitude and strength of shaking at the given location simultaneously. At the same time, the technique should produce stable results regardless the regional differences. In this study, after analysis of various input (ground-motion quantities) parameters used in Earthquake Early Warning schemes, we selected so-called “”characteristic period”” (Kanamori 2005) and JMA instrumental intensity. We evaluated performance of input-output scaling relations in respect of (a) characteristics of datasets accumulated in various seismic regions (Japan, Taiwan, California); (b) variation of characteristics of input parameters (e.g. length of P-wave windows, number of stations used); (c) possible combination of the input parameters.