| 摘要: | 本研究選用中央氣象局自由場強震觀測網在1993至2014年間，收錄芮氏規模2-7 且震源深度小於30公里的地震加速度資料，來計算震源、路徑及場址效應之地動參數(κ0、Q、Mw、M0及△σ)，最後將逆推所得之參數，利用點震源隨機式模擬法(SMSIM)及隨機式有限斷層法(EXSIM)模擬波形，來討論參數之可行性。
首先將地震觀測水平向加速度資料的高頻訊號部分，利用頻譜擬合法(Anderson and Hough, 1984)計算臺灣的S波近地表衰減運算子(kappa)，每一筆地震紀錄之震源距及所求κ不同，由每站所獲得的kappa值與距離之關係(κ(R)=αR+κ0)，當震源距為0公里，則κ(0)=κ0，則κ0值可視為場址參數之一，共獲得689個TSMIP測站之κ0值，由κ0值分佈情形，跟地質有明顯對應，高κ0值出現在台北盆地、宜蘭平原、西南部平原及恆春地區，而低κ0值則出現在山區地層堅硬地區，且在西南部沖積平原處其κ0值變化小，顯示影響κ0值的深度變化較小，但盆地、丘陵及山邊等構造複雜處，其κ0值變化也大。
第二部份，使用非參數逆推方法(Oth et al., 2011)，將不同區域(臺灣及南部地區) 挑選之S波到時資料逆推獲得震源、路徑及場址效應之頻譜，再將三種效應進一步透過Boore (2003)之參數設定求得地震的地震矩M0、震矩規模Mw、拐角頻率 fC、應力降Δσ、地殼的非彈性衰減Q( f )與各測站的場址效應。
最後將本研究所獲得之強地動參數，當ML<6時，使用SMSIM模擬，ML≧6，且有震源資訊(斷層面解及滑移量等)時，使用EXSIM模擬，由於本研究所獲得之κ0值與SMSIM及EXSIM之設定不同，故修改模擬流程，先利用SMSIM模擬無κ0值及場址效應時波形，再將波形轉至頻率，在頻率域中加入由逆推所得場址效應，及由高頻開始衰減的κ0值；當SMSIM模擬中小地震時，使用區域參數模型時，則可獲得最佳模擬結果，如沒區域參數模型時，使用臺灣地區模型也能獲得不錯之模擬結果；當具有震源資訊(斷層面解及滑移量等)之大地震時，則利用EXSIM模擬，透過三步驟模擬，可獲得不錯之模擬結果，故本研究所得之參數模型可用於地動預估。
;Strong ground motion parameters (κ0、Q、Mw、M0 and △σ) were measured from seismograms of Taiwan Strong Motion Instrument Program (TSMIP) with local magnitudes (ML) between 3.0–7.1 that occurred between 1993 and 2014 in this study. Meanwhile, inversion technique was also used for testing input for stochastic simulation method.
First, the high frequency decay parameter, kappa (κ) was computed by fitting the Fourier amplitude spectra of each station from TSMIP network. The relation between κ values and the hypocentral distance (Rhyp) were calculated from SH-waves for each individual station. Incidentally, the κ value at Rhyp=0 (denoted as κ0) can be used as site parameter, the range of κ0 for TSMIP stations were from 0.0185 - 0.0939 s in this study and the distribution is highly corresponding to geology and velocity. For instance, low κ0 values that below 0.06 s were occurred in and around the Central Mountain and foothill region, which was basically located in the middle of Taiwan. In contrast, high κ0 values that upon 0.06 s were observed at the alluvial areas, i.e., the Taipei basin and the Ilan plain in the northern Taiwan, the Chianan plain in the southwestern Taiwan, and the longitudinal valley in the eastern Taiwan. The site-specific κ0 values from 426 stations were correlated with the averaged shear wave velocity of the top 30 m of strata (VS30), and the relationship could be described by κ0 =0.163 – 0.077·ln(VS30) ± 0.053 and a high linear correlation (R2 = 0.63) was found.
The second part, the generalized inversion technique (GIT) (Oth et al., 2011) was used for inversion purpose of seimogenic parameters from SH-wave in the frequency range 0.1 to 40 Hz (interval 0.1 Hz) for whole Taiwan region (Taiwan model) and the southern region (regional model). The attenuation characteristics, earthquake source parameters and site amplification functions could be decomposed step by step from GIT. In this study, the characteristics of the site amplification are referred to horizontal-to-vertical (H/V) Fourier spectral ratios of earthquakes for a referent rock site. The basic three effects were are set with the parameters of Boore (2003) to determine the rest seismic moment(M0)、corner frequency(fc)、stress drop(△σ) and Q(f).
Finally, the strong ground parameters obtained by this study are verified from stochastic simulation method. Accordingly to size of seismic source, point source technique (SMSIM) was used for ML<6 and finite-fault technique (EXSIM) with correlated source information (fault plane solution and slip, etc.) was used for ML≧6 events. However, the definition of κ0 obtained from this study is differed from that of previous studies (Boore, 2003; Boore, 2009, SMSIM and EXSIM). Several changes were made in this study, first, the κ0 and site amplification were exclude from waveforms that were generated from SMSIM. Second, different shape of high-frequency decay (related to κ0) spectrum generated from this study and also site amplification function was superposed in the SMSIM simulation spectrum in frequency domain. Finally, the regional parametric model was used to obtain the best simulation results for SMSIM technique. However, if there is no regional parameter model, Taiwan model can also get a good simulation results. When source information could be found for large earthquakes, EXSIM could provide better simulation results. Therefore, the parameters model calibrated in this study can be used to predict ground motion. |
