| 摘要: | 設置於加拿大之Yellowknife Array(YKA)是一展距約20km?20km之小型短週期地震儀陣列，記錄了許多由台灣附近中大規模地震發出之深度波相(depth phase)訊息，深度波相對於震源深度可提供極明確的確認資訊。台灣島由南至北大約400公里，發生於台灣附近的地震事件在區域地震觀測記錄上難以發現這類波相，利用此類深度波相（如pP、pwP、sP）可以提供台灣地震觀測網外地震事件深度之控制而避免深度和震央距離兩者間難以取決的難題。 利用遠域地震儀記錄之初達P波與深度波相之到時時間差來估算地震發生深度是一簡單又直接的方法。由於通常遠地記錄中的深度波相振幅較小，本研究中利用波形叠加來加強深度波相的訊號，再以順推波形模擬計算不同震源深度之各波相波形變化加以輔助深度波相的判別，最後利用波形交互比對計算找出觀測記錄與不同深度模擬波形之最佳擬合波線以推估發震深度之最可能範圍。 本研究首先以兩個發生於台灣島內的地震事件為例，利用本研究方法分析所得震源深度與中央氣象局定位結果和台灣寛頻地震網(Broadband Array in Taiwan for Seismology，BATS)分析結果都相當符合，再配合近震強震資料之P波及S波到時差資料以順推網格搜尋方式進行重新定位分析，結果顯示本方法確實可有效推算震源深度之合理範圍。此外，應用本研究方法分析位於台灣東北部較大規模地震事件，對於測站包覆性不佳而無法以近震資料進行定位分析，或是遠離地震觀測網的事件，依本研究方法分析皆可估算出適當的深度範圍。而應用於2006年屏東地震序列6個較大規模事件之分析也都得到穩定的分析結果。除了第2主震外，依本方法分析所得之深度皆較氣象局原定位結果略淺。由於屏東地震序列之第2主震破裂過程較為複雜，配合近震之強地動資料亦推估此一事件震源應為多次破裂，而本方法決定之深度應為最大破裂產生之深度範圍。配合其他事件之分析結果推論此事件之震源破裂方向可能由淺部持續往深部進行。 深度波相與直達P波之到時差不受絶對時間及發震時間誤差影響，對震源深度可提供極佳的控制；波形模擬計算不同深度之波形記錄則可協助深度波相的判別，本研究綜合了波形擬合方式的優點及深度波相對震源深度絶佳的控制來進行震源深度範圍的估算。本方法雖然主要應用的地震規模範圍著重於ML5.0~7.5間之中大規模之地震事件，對於破裂過程相對複雜(如多重破裂事件)的地震，有時因理論模擬波形與觀測波形不易比對而較難有理想的結果。但整體而言，對於遠離地震觀測網外的地震事件皆能分析得出最佳的震源深度範圍，有助於中大型地震定位時震源深度的範圍判定，對於位於外海有可能引發海嘯的地震相信也能提供良好的深度資訊以助防災預警工作的進行。 The Yellowknife array (YKA), a small-aperture array, recorded many depth phase seismograms from earthquakes which occurred in Taiwan region. Such kinds of phases are unavailable to observe on Taiwan’s local seismic records. The depth phase(pP、pwP、sP、sS and so on) reflected from the surface then traveled to the recording station can provide unequivocal confirmation of focal depth. The pP-P interval depends strongly on focal depth and is independent of clock errors. It can give a good constrain for earthquakes locating that occurred out of seismic network and eliminate a trade-off relationship between the source depth and epicenter determination. It is a simple and direct way to estimate focal depth using the pP-P interval on teleseismic records. However, the pP phase propagates a little bit longer path than P and dissipates more energy, so that the amplitude of pP phase is normally smaller than direct P. In this study, we proposed to use slant stack procedures to reduce the effects of random noise in the data and to enhance the depth phase signal. The surface reflections arrive behind the direct arrival and may be interfered by scattered P wave energy from both the source and receiver sides. The shape of the interference packet changes with depth and this information supplies a means of estimating the source depth. Comparing the observations with suitable synthetic seismograms will help us to identify the accurate source depth. For this reason, we adopt a forward modeling way to calculate synthetic waveforms in different focal depths to help us to recognize depth phase arrival. Finally, the source depth search can be quantitatively represented by a cross-correlation analysis to find the accuracy focal depth range. First, we used two events which occurred in the seismic network to check our method is authentic. The result of this analysis showed our method indeed helped us easily control the range of depth and obtain a robust result. Moreover, using forward grid search to look for the best hypocenter location by S-P time difference of TSMIP records provided information about shallow 1-D S wave velocity stracture under the epicenter region at the same time. Second, we applied our method and procedure to analyze other earthquakes which ML>4.5 and occurred offshore northeastern and southwestern Taiwan. All events are outside of Taiwan seismic network. Some earthquakes were far away from the seismic network and some did not have a good station distribution. The analyzed results of these events showed we can easily control the range of depth and estimate optimal results. The events in offshore southwestern Taiwan were selected from Pingtung Earthquake sequence. In this study, we chose 6 events for analyzing and the best depth solutions were accurately obtained. Analyzed results show that most depths of the offshore Pingtung earthquake sequence are slightly shallower than that reported by CWBSN except for the 2nd event that was the biggest event of the Pingtung earthquake sequence. Combining analyses from near source and teleseismic observations, we concluded that the rupture properties of the 2nd event began at a shallow depth, continued to grow to a depth of 56km where the largest rupture occurred and released the most energy. Considering the arrival times of surface-reflected phases to locate an event depth, it can provide an excellent constrain in source depth. Moreover, accompanying with the forward waveform simulation in different source depth to help us to identify the accurate depth phase, we took the superiors of both methods to have a result with good stability and high depth resolution. Although this method is mainly applied for the events with a ML from 5.0 to 7.5, which features a complicate rupturing process and sometimes hardly to obtained an idea result from the comparing with theoretical simulation and observed waveforms. However, in overall speaking, it is a reliable method for the events outside of the array in source depth calculation. A simple but robust procedure to identify the depth of a seismic event has been developed and successfully demonstrated. |
