震源机制解测定在常规分析中的应用

【作者】 任枭；

【导师】 刘瑞丰；

【作者基本信息】 中国地震局地球物理研究所 ， 固体地球物理学， 2004， 硕士

【摘要】 “九五”完成后，我国地震观测系统实现了由模拟系统向数字系统的更新换代，中国地震观测全面进入数字时代。宽频带地震仪在各级地震台网中得到了广泛应用，产出了大量丰富的数字资料。就国家数字地震台网而言，每天会产生3G数据量的资料，每月会记录分析600多个地震事件。随着数字化地震台网在全国的陆续建成，利用数字化台网记录的大量地震波形资料确定地震的震源机制解，将结果以常规报告的形式给出，从而更好的为科研工作服务，已成为当前的一项基础需求。本文选题就是以国家数字地震台网宽频带数字地震资料的常规分析处理为出发点。 为了对新建成的国家数字地震台站的台基状况有个基本的了解，便于在选取资料时有所参考，对国家数字地震台网中台站的台基做了噪声功率谱的研究。(其中，CDSN台的台基噪声情况前人已作过研究和分析，所以只对使用国产仪器的37个地震台站的台基噪声功率谱做了计算和分析。)在研究结果中，根据噪声功率谱的全频段平均值，把37个国家数字地震台站的台基质量分为三类，在文中给出了分类结果和相应的地动噪声曲线图。在此基础上，使用地震分析处理软件SAC完成了对所使用的宽频带数字地震资料的预处理工作。 利用剪切位错点源作为震源模型，采用横向均匀的水平层状介质模型作为地壳近似模型，使用F-K波数积分法计算格林函数并拟合理论地震图。计算过程中，对于高频部分使用Filon积分，低频部分使用Bessel函数的多项式近似方法(补充使用了梯形积分法)。也就是说，使用该方法合成理论地震图，既可以选用高频的Pnl波参加拟合，又可以选用长周期的Lg波和Rayleigh波参加拟合。另外，如果加入P波、SV波或SH波的初动符号，则可以作为约束条件减少计算所需时间，提高计算结果的可靠性。因此，该方法适合用在数字地震台网所记录资料的日常处理工作中。通过合成地震图与实际记录图相比较，在反复调试使两者吻合得很好的情况下，得到了本文中所用的我国的地壳速度结构模型；同时使用grid搜索法得出最佳断层面解(θ，δ，λ)，进一步确定了地震矩张量的分量，并由此得出震源机制解。 在已有理论方法的基础上，利用国家数字地震台网资料选取了2002-2003年间发生在国内M_S≥5.5的5个地震事件的宽频带数字资料进行震源机制解测定，并将测定结果与HRV的矩心矩张量解、USGS的地震矩张量解、分析预报中心的速报震源机制解分别进行了对比，通过实际计算表明测定结果是可靠的。从而初步验证了该方法的可用性和准确性。更多还原

【Abstract】 The finished seismic observation system of the 9th five-year plan, has upgraded from analog to fully digital system. Broad band seismic instruments are used by all levels of seismic networks, and plenty of data material has been saved by far. The National Digital Seismic Network (NDSN) usually saves 3G data every day and analyzes over 600 seismic events every month. With digital networks developing in all-country, it is necessary for seismic networks to use plenty of data to determine focal mechanism in routine analysis and put it in routine report, so that it can give better service for scientific research. Composing this paper is based upon a routine work in the data analysis and processing works of NDSN.In order to know about the situation of new stations (except CDSN stations, because they have been analyzed by other researchers) in NDSN and select conveniently data, we compute and analyze the ground motion noise power spectrum. As a result, we classify these stations into three grades by comparing the average of noise power spectrum in whole-band range and present the resultant figures in this paper. And then, We preprocess the data with seismic analysis tools SAC.In this paper, we use simple point source model-shear dislocation model andapproximately regard the earth crust as horizontally layered medium that is transverse isotropy. We use F-K (frequency-wavenumber) integral algorithm to compute Greens’ function and synthesize the theoretical seismograms. In this method, the frequency-wayenumber response is evaluated by using the Filon quadrate at.high frequency and the polynomial approximation of Bessel functions (implementing trapezoidal integration) at low frequency. That is to say, using this algorithm, both short-period Pnl wave and long-period Lg wave Rayleigh wave can be successfully modeled. Moreover, the computing time can be reduced if there is P, SV, or SH polarity data available. Thus, this technique is ideally suited for studying focal mechanism with broadband data of NDSN in routine works. When the different between the practical seismograms and the theoretical seismograms is the best fit, we produce suitable crustal velocity model of China and use it in this paper. In addition, through grid search procedure we get optimal fault parameters and determine the seismic moment tensor, so that we estimate the sourcemechanism.Based on the existing theory and methods, we use broadband data of five seismic events thatare occurred from 2000 to 2003 and scaled Ms 5.5, to determine the focal mechanism, thencompare these results respectively with CMT of HRV, seismic moment tensor solution of USGS, source parameters of Center for Analysis & Prediction. The result shows we have certificated usability and veracity of these theories and methods.更多还原