A simple approach to estimate earthquake magnitude from the arrival time of the peak acceleration amplitude

Wednesday, 17 December 2014: 11:20 AM
Shunta Noda1,2 and Shunroku Yamamoto1, (1)RTRI Railway Technical Research Institute, Center for Railway Earthquake Engineering Research, Kokubunji, Japan, (2)USGS, Earthquake Science Center, Baltimore, MD, United States
In order for Earthquake Early Warning (EEW) to be effective, the rapid determination of magnitude (M) is important. At present, there are no methods which can accurately determine M even for extremely large events (ELE) for EEW, although a number of the methods have been suggested. In order to solve the problem, we use a simple approach derived from the fact that the time difference (Top) from the onset of the body wave to the arrival time of the peak acceleration amplitude of the body wave scales with M.

To test this approach, we use 15,172 accelerograms of regional earthquakes (most of them are M4-7 events) from the K-NET, as the first step. Top is defined by analyzing the S-wave in this step. The S-onsets are calculated by adding the theoretical S-P times to the P-onsets which are manually picked. As the result, it is confirmed that logTop has high correlation with Mw, especially for the higher frequency band (> 2Hz). The RMS of residuals between Mw and M estimated in this step is less than 0.5. In case of the 2011 Tohoku earthquake, M is estimated to be 9.01 at 150 seconds after the initiation of the event.

To increase the number of the ELE data, we add the teleseismic high frequency P-wave records to the analysis, as the second step. According to the result of various back-projection analyses, we consider the teleseismic P-waves to contain information on the entire rupture process. The BHZ channel data of the Global Seismographic Network for 24 events are used in this step. 2-4Hz data from the stations in the epicentral distance range of 30-85 degrees are used following the method of Hara [2007]. All P-onsets are manually picked. Top obtained from the teleseimic data show good correlation with Mw, complementing the one obtained from the regional data.

We conclude that the proposed approach is quite useful for estimating reliable M for EEW, even for the ELE.