scholarly journals Source time function clustering reveals patterns in earthquake dynamics

2020 ◽  
Author(s):  
Jiuxun Yin ◽  
Zefeng Li ◽  
Marine Denolle
2013 ◽  
Vol 5 (2) ◽  
pp. 1125-1162 ◽  
Author(s):  
S. C. Stähler ◽  
K. Sigloch

Abstract. Seismic source inversion is a non-linear problem in seismology where not just the earthquake parameters themselves, but also estimates of their uncertainties are of great practical importance. Probabilistic source inversion (Bayesian inference) is very adapted to this challenge, provided that the parameter space can be chosen small enough to make Bayesian sampling computationally feasible. We propose a framework for PRobabilistic Inference of Source Mechanisms (PRISM) that parameterises and samples earthquake depth, moment tensor, and source time function efficiently by using information from previous non-Bayesian inversions. The source time function is expressed as a weighted sum of a small number of empirical orthogonal functions, which were derived from a catalogue of >1000 STFs by a principal component analysis. We use a likelihood model based on the cross-correlation misfit between observed and predicted waveforms. The resulting ensemble of solutions provides full uncertainty and covariance information for the source parameters, and permits to propagate these source uncertainties into travel time estimates used for seismic tomography. The computational effort is such that routine, global estimation of earthquake mechanisms and source time functions from teleseismic broadband waveforms is feasible.


1997 ◽  
Vol 87 (4) ◽  
pp. 999-1010
Author(s):  
F. Courboulex ◽  
M. A. Santoyo ◽  
J. F. Pacheco ◽  
S. K. Singh

Abstract We analyze source characteristics of the 14 September 1995, Copala, Mexico, earthquake (M = 7.3) using teleseismic, regional, and local seismograms. In the analysis of the teleseismic and the regional seismic waves, we apply the empirical Green's function (EGF) technique. The recording of an appropriate aftershock is taken as the EGF and is used to deconvolve the mainshock seismogram, thus obtaining an apparent far-field source-time function at each station. The deconvolution has been done using surface waves. For teleseismic data, we apply a spectral deconvolution method that enables us to obtain 37 apparent source-time functions (ASTFs) at 29 stations. In the analysis of the regional broadband seismograms, we use two different aftershocks as EGF, and the deconvolution is performed in the time domain with a nonlinear method, imposing a positivity constraint, and the best azimuth for the directivity vector is obtained through a grid-search approach. We also analyze two near-source accelerograms. The traces are inverted for the slip distribution over the fault plane by applying a linear inversion technique. With the aid of a time-window analysis, we obtain an independent estimation of the source-time function and a more detailed description of the source process. The analysis of the three datasets permits us to deduce the main characteristics of the source process. The rupture initiated at a depth of 16 km and propagated in two directions: updip along the plate interface toward 165° N and toward 70° N. The source duration was between 12 and 14 sec, with the maximum of energy release occurring 8 sec after the initiation of the rupture. The estimated rupture dimension of 35 × 45 km is about one-fourth of the aftershock area. The average dislocation over the fault was 1.4 m (with a maximum dislocation of 4.1 m located 10 km south of the hypocenter), which gives roughly 1 MPa as the average static stress drop.


2020 ◽  
Author(s):  
Wenzheng Gong ◽  
Xiaofei Chen

<p>Spectra analysis is helpful to understand earthquake rupture processes and estimate source parameters like stress drop. Obtaining real source spectra and source time function isn’t easy, because the station recordings contain path effect and we usually can’t get precise path information. Empirical Green’s function (EGF) method is a popular way to cancel out the path effect, main two of which are the stacking spectra method (Prieto et al, 2006) and the spectral ratio method (Viegas et al, 2010; Imanishi et al, 2006). In our study, we apply the latter with multitaper spectral analysis method (Prieto et al, 2009) to calculate relative source spectra and relative source time function. Target event and EGFs must have similar focal mechanism and be collocated, so we combine correlation coefficient of wave at all stations and focal mechanism similarity to select proper EGFs.</p><p>The Bucaramanga nest has very high seismicity, so it’s suitable to calculate source spectra by using EGF method. We calculate the source spectra and source time function of about 1540 earthquakes (3-5.7ml, 135-160km depth) at Bucaramanga nest in Colombia. Simultaneously we also estimate corner frequency by fitting spectral source model (Brune, 1970; Boatwright, 1980) and stress drop using simple model (Eshelby, 1957) of earthquakes with multiple station recordings or EGFs. We obtain about 30000 events data with 12 stations from National Seismological Network of Colombia (RSNC).</p><p>The result show that the source spectra of most earthquakes fitted well by omega-square model are smooth, and the source spectra of some have obvious ‘holes’ near corner frequency, and the source time function of a few earthquakes appear two separate peeks. The first kind of earthquakes are style of self-arresting ruptures (Xu et al. 2015), which can be autonomously arrested by itself without any outside interference. Abercrombie (2014) and Wen et al. (2018) both researched the second kind of earthquakes and Wen think that this kind of earthquakes are style of the runaway ruptures including subshear and supershear ruptures. The last kind of earthquakes maybe be caused by simultaneous slip on two close rupture zone. Stress drop appear to slightly increase with depth and are very high (assuming rupture velocity/s wave velocity is 0.9). We also investigate the high-frequency falloff n, usually 2, of Brune model and Boatwright model by fitting all spectra, and find that the best value of n for Boatwright model is 2 and for Brune model is 3.5.</p>


Sign in / Sign up

Export Citation Format

Share Document