ear real time estimation of magnitudes and moments for local seismic events

Abstract

The general Popularity of magnitude as a convenient and robust measure of earthquake size makes it tempting to examine whether this parameter can be reliably estimated in near real time. In this study we dernonstrate that this is indeed the case conditioned on the design of the signal detector being of STA/LTA type where STA is a short term signal power or rms estimate. Using real data we dernonstrate the Random Vibration Theory relation that Amax (21nN) 1/2 Arms , is valid for non stationary seismic signals. Using Rayleigh's theorem we also estabmlaixshed a relation brmetween Arms and the flat portion of the source spectra. These Amax and Arms estimation procedures are used for determining conventional magnitudes and moment magnitudes for 29 events as recorded by the Norwegian Seismograph Network (NSN). We used here a procedure outlined by Sereno et al. (1988) and also their geometrical spreading and attenuation parameters derived from analysis of NORSAR recordings. Our magnitude and moment magnitude estimates for 5 different frequency bands are in good agreement with the ML estimates derived from the conventional magnitude formulas in combination with empirical correction tables. Surprisingly, the Amax and Arms magnitudes produced consistent negative biased by ca. 0.4 units estimates even in the extreme 4 8 Hz band. In view of the good agreement between various types of magnitude estimates, we constructed conventional magnitude correction tables spreading and attenuation parameters from Sereno et al (1988) for a variety of signal frequency bands. Near real time Amax ad/or Arms or correspondingly event magnitudes would be of significance in automatic phase association analysis, bulletin production for local and regional seismic networks and the earthquakes monitoring performances of such networks.