Rapid Source Estimation from Global Calibrated Paths

Abstract

On 9 October 2006, an explosion occurred in North Korea as predicted by that country's press office (Figure 1). The highfrequency P waves relative to S waves allowed easy identification (Richards and Kim 2007a, 2007b). However, estimating the yield of the explosion proved more difficult. Richards and Kim (2007a, 2007b) employed the current procedure for estimating the size of events from inferences from magnitude-yield curves, usually mb or a regional proxy, mb ( Lg ). Linear plots of mb vs. yield are based on measured mb for known yields, i.e. , Amchitka and Nevada Test Site (NTS), etc. In this particular case, they used Murphy's (1977) formula, which states mb = 4.75 + log Y , where the initial amplitudes of short-period (SP) P waves at a large number of azimuthally distributed measurements are made. The same mb measure is used for earthquakes where it is assumed that by this averaging, we can eliminate radiation pattern effects, since the amplitudes can vary by a factor of 10 (Butler and Ruff 1980). However, the radiation patterns of down-going energy are distinctly different between strike-slip and dip-slip mechanisms. The former are weak at ray parameters emitting P waves teleseismically. Thus one would expect mb is to be biased low for strike-slip events if only distance P waves are available. This observation has been noted in Zhu et al. (2006) where regionally determined Mw is compared with published mb . Thus, we expect smaller mb from strike-slip events of about 0.4 magnitudes based on the above study. This feature can be observed directly (Figure 2) by examining the various mechanisms given in Figure 1. Note that the strikeslip event ( Mw = 4.5) is relatively weak at the ILAR array, at …