IASPEI - Picture Gallery

IUGG XXIV General Assembly July 2-13, 2007 Perugia, Italy
(S) - IASPEI - International Association of Seismology and Physics of the Earth's
Interior
JSS003 Oral Presentation 1856
An evolutionary approach for real-time magnitude estimation for
earthquake early warning
Dr. Maria Lancieri
RISSC-LAB INGV, Osservatorio Vesuviano
Aldo Zollo
Regional earthquake early warning systems relies on the possibility of achieving fast and reliable
estimates of location and magnitude/moment of an occurred, potentially destructive event, in order to
predict peak ground motion quantities at a distant target infrastructure. Recent studies showed the
possibility of predicting the final event magnitude using measurements of the predominant frequency
and/or the low-frequency peak displacement amplitude in the very early portion of P-wave signals. In
particular, the recent analysis of near-source strong motion records from the European Data Base
revealed a clear correlation between distance corrected peak displacement amplitudes and magnitude
for events in the magnitude range 4-7.4 (Zollo, Lancieri and Nielsen, 2006). In the present work we
analyze a Japanese strong motion data-set (extracted from K-Net and KiK-net) with the aim to
investigate the relationship between the early P- and S- wave peak displacement and magnitude and its
use for an evolutionary estimate of magnitude for earthquake early warning application. About 2700
records from 256 Japanese earthquakes have been analyzed with magnitude ranging from 4 to 7 with
maximum depth of 50Km and hypocentral distance smaller than 60Km. The records come from the
Kyoshin strong motion network, which collects data from 1000 strong-motion station deployed over the
whole Japanese archipelago, with an average inter-station distance of 25 Km. The accelerometric
records have been integrated twice, and band pass filtered between 0.075 and 3Hz. In order to correct
the early peak amplitudes for the distance attenuation effect, we evaluated an attenuation relationship
between the logarithm of the distance, the magnitude and the peak displacement read on short time
windows after the P-wave and the S-wave arrivals. While using a duration of 2-sec after the first P-
arrival the peak displacement appears to saturate with magnitude around M=6, this effect is removed
using windows larger than 3-sec for which a linear relationship is found between the logarithm of
displacement peak amplitude and magnitude. On the other hand the saturation effect is not visible on
S-wave peak even for very short time windows (1 sec). Assuming the existence of a strong motion
network, densely deployed in the epicentral area of an impending earthquake, we illustrate a Bayesian
approach to estimate the magnitude and its uncertainty from the P- and S-peak information available at
each triggered station as a function of time from the event occurrence. We show the application of the
proposed procedure to the Japan earthquake data-set by simulating the real-time estimation of several
moderate to large magnitude events along with time evolution of its confidence level.
Keywords: magnitude estimation, evolutionary, bayesian approach