On estimation of earthquake magnitude in Earthquake EarlyWarning systems

Determination of earthquake magnitude from the initial P-wave portion of ground motion is the key problems for Earthquake Early Warning (EEW). We analyzed performance of scaling relations between one of the frequently used early warning parameters, so-called characteristic period τ_C and moment magnitude M_W in respect of (a) characteristics of datasets accumulated in various regions (earthquake depth, characteristics of network) and (b) variation of initial conditions applying for determination of the parameter τ_C (length of Pwave window, number of used stations). The used data contain strong-motion records from 110 earthquakes (moment magnitude range 4.4–7.6) occurred in Japan and Taiwan. We show that, although the standard error of regression τ_C = f (M_W) in general becomes smaller with the increase of the length of P-wave window (PL) and the number (N) of averaged observations (stations), the uncertainty in estimation of MW given observed tC does not decrease further for PL > 3–4 seconds and N > 3–4 stations. The information about earthquake depth also plays an important role in reducing the uncertainty of magnitude estimations. The table of confidence limits for estimations of the moment magnitude M_W based on observations of the characteristic period τ_C is provided.