|
Local Magnitude Determination Using the P-arrival
R. M. Allen University of Wisconsin-Madison H. Kanamori California Institute of Technology AGU 2002 fall meeting Local earthquake magnitude, $M_L$, is usually determined using peak ground motion observations made relatively late in the seismic wavetrain. Here we present a technique for estimation of earthquake magnitude using the P-arrival only, $M_P$. Such rapid magnitude determination makes it possible to estimate ground motion parameters prior to significant ground shaking (even at the epicenter for deeper earthquakes), making it ideal for earthquake early warning in southern California. Using 53 historical earthquakes in southern California, ranging in magnitude from 3.0 to 7.3 and including Landers, Northridge and Hector Mine, we test the accuracy of the method by comparing $M_P$ to $M_L$ as determined by TriNet. For earthquakes with magnitudes $<$ 5.0 the best $M_P$ estimates require only 2 sec of broadband data. For larger events, low-passing the data at 3 Hz provides the best estimates of $M_P$, which are possible with 4 sec of data. Using data from all available broadband seismic stations within 100 km of all 53 earthquakes the rms error in $M_P$ is 0.5 magnitude units. To test the speed with which a magnitude estimate could be provided in southern California we select a subset of earthquakes occurring beneath the more dense portions of TriNet. The subset consists of 28 events with magnitudes ranging from 3.0 to 5.1. Of these 28 test events, magnitude estimates are available at the same time as the S-arrival at the epicenter for 12 of them with an rms error of 0.76 magnitude units. Three seconds later estimates are available for 26 events with an rms error of 0.48. The algorithms developed, named ElarmS, provide a basis for earthquake early warning in southern California and are currently being tested with the TriNet realtime system. |