论文部分内容阅读
给出了三分量地震台地震数据自动在线分析系统的基本方法和算法以及初步实验结果。数据处理分为4步:(1)用基于能量的γ检测器进行初步检测和信号到时估算;(2)精取信号参数(到时、振幅、周期和尾波持续时间);(3)用偏振分析估算地震射线的方位角和入射角;(4)对于一个固定的震源深度,由远震P波近似估算事件的参数(震中坐标、发震时刻和震级)。在哈萨克斯坦东部的一个实验地震台上,用命名为“SEISMOSTANSIA”的软件包进行在线环境下的测试。测试结果表明,在24小时周期内能处理多至100个输入信号的数据流。自动检测的结果与可视模拟地震图得到的台站报告资料比较表明:(1)按现在的闹值设置,自动系统实际上每天检测到了分析者看到的所有远震信号,同时有10～15个信号没有被可视系统确认。(2)在70％的情况下,自动与人工处理得到的到时残差,在SNR(信噪比)>10时,为0.5s或更小,当SNR<10时,为1s或更小。(3)在70％的情况下,自动与人工处理得到的信号周期残差为0.22s或更小。资料分析表明,确认在低SNR情况下的假报警的一个有效方法是P波和S波的偏振线性水平。通过对内华达试验场的一系列爆炸方位角的分析,其结果表明,系统偏差构成了与真方位角的残差的主要部分。一般情况下,方位角残差是由于震中和台站介质特性引起的,总之取决于方位角和震中距。方位角残差分析指出,其偏差周期性地依赖于方位角。这个周期值大约为120°。从三分量地震台给数据中心报告中得到的其他资料(方位角和入射角)明显地改善了那些与全球地震台网数据处理的相关震相的可靠性。
The basic method and algorithm of the three-component seismic data automatic on-line analysis system and the preliminary experimental results are given. The data processing is divided into four steps: (1) preliminary detection and signal arrival estimation with an energy-based gamma detector; (2) fine gain signal parameters (arrival, amplitude, period and coda duration); (3) Polarization analysis is used to estimate the azimuth and incident angle of the seismic ray. (4) For a fixed focal depth, the parameters of the event (epicenter, seismogenic moment and magnitude) are approximated by teleseismic P-waves. In an experimental seismic station in eastern Kazakhstan, the software package named “SEISMOSTANSIA” was used for online testing. The test results show that it can process data streams of up to 100 input signals over a 24-hour period. The comparison between the results of automatic detection and those reported by the station of visual analogue seismograms shows that: (1) According to the current alarm setting, the automatic system actually detects all the teleseismic signals seen by the analyst every day, 15 signals have not been confirmed by the visual system. (2) The time-to-residual resulting from automatic and manual processing at 70% is 0.5 s or less at SNR (Signal to Noise)> 10 and 1 s or less at SNR <10 . (3) In 70% of the cases, the signal cycle residuals obtained by automatic and manual processing are 0.22s or less. Data analysis shows that an effective way of identifying false alarms at low SNR is the polarization linearity of P and S waves. The results of a series of explosion azimuths at the Nevada test site show that systematic deviations form the major part of the residual with true azimuth. In general, the azimuth residual error is due to the characteristics of the epicenter and the station medium, in short, depending on the azimuth and the epicentral distance. Azimuth residual analysis pointed out that the deviations periodically depend on the azimuth. This period is about 120 °. Additional information (azimuth and angle of incidence) reported from the three-component seismometer to the data center significantly improves the reliability of those phases associated with global seismic network data processing.