COMPARISON OF SOURCE TIME FUNCTIONS RETRIEVED FROM VARIOUS REGIONAL WAVES AND CODA USING MODERATE EARTHQUAKES IN EAST AND CENTRAL ASIA

Abstract:

<span">Source time functions may vary with various P and S wave types. Regional Lg waves have been used to estimate radiated energy and apparent stress. In nuclear explosion seismology, the practice of discriminating Earthquake and explosions using regional wave spectra and spectral ratios relies on a fundamental assumption that Earthquakes and explosions excite various regional waves in a systematically-different manner. We have been carrying out a systematic study to retrieve source time functions (STFs) from seismic sources using the empirical Green’s function (EGF) approach. In phase 1 of the study, we focus on retrieving STFs from moderate earthquakes in east and central Asia using regional Lg, Pg, Lg coda, and Sn and Pn waves to see whether or not the shape of the STFs varies with the wave type. We explored various methods to reduce the noise in the deconvolved STFs which tend to be higher for the relatively weaker Pg, Sn, Pn and coda waves. For example, an array-stacking method is used to enhance source pulses and reduce the level of side-lobes. Preliminary results suggest that Lg and Lg coda STFs may be similar, as generally believed. Pg STF is less similar to Lg STF. For example, when a stacked Lg STF is clearly asymmetric with a sharp-rise time, the stack Pg STF seems to be fairly symmetric. In general, our confidence on this kind of dissimilarity is still limited by the non-diminishing deconvolution noise in the retrieved Pg STFs, and by the small number of events studied. We are trying to further reduce the deconvolution noise for each regional wave, and find and analyze more moderate events. We will present newly-retrieved STFs from each of the various regional waves and coda, and compare the pulse widths and shapes of the STFs from different waves. We will also present estimates of source-radiated energy and apparent stresses using the estimated Lg STFs without using various simplified source models and Q corrections.